Methods for improved predictions of polygenic phenotypes across diverse populations

Abstract

Systems and methods for predicting a medical trait in an individual are provided. The techniques include obtaining genomic data including information indicative of genetic variants present in a genome of the individual and calculating, using an at least one processor and a trained computational model, a trait score based on the information indicative of the genetic variants, wherein the trained computational model was obtained by training a first computational model using data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries. The trait score may be used to alter a course of medical treatment of the individual and / or to implement a clinical trial.

Description

Attorney Docket No.: M0437.70168WO00 METHODS FOR IMPROVED PREDICTIONS OF POLYGENIC PHENOTYPES ACROSS DIVERSE POPULATIONS CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] The present application claims the benefit under 35 U.S.C. § 119(e) to U.S. Provisional Patent Appl. No. 63 / 663,654, filed June 24, 2024, and titled “METHODS FOR IMPROVED PREDICTIONS OF POLYGENIC PHENOTYPES ACROSS DIVERSE POPULATIONS;” U.S. Provisional Patent Appl. No. 63 / 639,569, filed April 26, 2024, and titled “METHODS FOR IMPROVED PREDICTIONS OF POLYGENIC PHENOTYPES ACROSS DIVERSE POPULATIONS;” and U.S. Provisional Patent Appl. No.63 / 590,758, filed October 16, 2023, and titled “METHODS FOR IMPROVED PREDICTIONS OF POLYGENIC PHENOTYPES ACROSS DIVERSE POPULATIONS,” each of which is incorporated herein by reference in its entirety. FEDERALLY SPONSORED RESEARCH

[0002] This invention was made with government support under AG058002, MH109978, NS110453, AG062335, AG074003, AG067151, MH119509, HG008155, and AG054012 awarded by the National Institutes of Health. The government has certain rights in the invention. BACKGROUND

[0003] A polygenic score (PGS)—also known as a polygenic index, polygenic risk score, genetic risk score, or genome-wide score—is a numeric value used to quantify the estimated effect of many genetic variants on an individual’s phenotype. The individual’s phenotype may include a particular disease and / or other medical traits. Thus, a PGS can be used to quantify an individual’s medical trait(s) based on their genetics. In humans, a PGS is typically calculated using an additive sum of the effects of genetic variants identified from genome-wide association studies (GWASes). SUMMARY

[0004] In some aspects, the techniques described herein relate to a method of predicting a medical trait in an individual. The method includes: obtaining genomic data including information indicative of genetic variants present in a genome of the individual; calculating, using an at least one processor and a trained computational model, at least one trait score based on the informationAttorney Docket No.: M0437.70168WO00 indicative of the genetic variants, wherein the trained computational model was obtained by training a first computational model using data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries; and generating, using the at least one processor, a graphical user interface including a visualization of the at least one trait score, the at least one trait score being indicative of a presence of the medical trait in the individual; and displaying, using a display device, the generated graphical user interface.

[0005] In some aspects, the techniques described herein relate to a system, including: at least one computer hardware processor; and at least one non-transitory computer readable medium storing processor-executable instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of predicting the medical trait in the individual.

[0006] In some aspects, the techniques described herein relate to at least one non-transitory computer readable medium storing processor-executable instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of predicting the medical trait in the individual.

[0007] In some aspects, the method further includes altering a course of medical treatment related to the medical trait based on the at least one trait score.

[0008] In some aspects, altering the course of medical treatment includes reducing a number of medical interventions, reducing a frequency of medical interventions, and / or selecting a different type of medical intervention based on a value of the at least one trait score.

[0009] In some aspects, altering the course of medical treatment is based on the value of the at least one trait score relative to a threshold value.

[0010] In some aspects, altering the course of medical treatment includes increasing a number of medical interventions, increasing a frequency of medical interventions, and / or selecting a different type of medical intervention based on the at least one trait score.

[0011] In some aspects, altering the course of medical treatment is based on a value of the at least one trait score relative to a threshold value.

[0012] In some aspects, the method further includes identifying at least one therapeutic agent based on a value of the at least one trait score.Attorney Docket No.: M0437.70168WO00

[0013] In some aspects, the method further includes administering the at least one therapeutic agent to the individual based on the value of the at least one trait score.

[0014] In some aspects, the method further includes altering a recruitment strategy for a clinical trial related to the medical trait based on the at least one trait score.

[0015] In some aspects, altering the recruitment strategy for the clinical trial includes removing the individual from enrollment in the clinical trial based on a value of the at least one trait score.

[0016] In some aspects, altering the recruitment strategy for the clinical trial is based on the value of the at least one trait score relative to a threshold value.

[0017] In some aspects, altering the recruitment strategy for the clinical trial includes including the individual in enrollment in the clinical trial based on a value of the at least one trait score.

[0018] In some aspects, altering the recruitment strategy for the clinical trial is based on the value of the at least one trait score relative to a threshold value.

[0019] In some aspects, training the first computational model includes using batch screening iterative lasso (BASIL) regression.

[0020] In some aspects, training the first computational model includes performing a regression including a loss, the loss including a squared loss and / or a binomial loss.

[0021] In some aspects, obtaining the trained computational model further includes using the trained first computational model and performing a regression including a penalty applied to genetic variants with heterogeneous associations between single-ancestry populations.

[0022] In some aspects, performing the regression further includes including effects of a global ancestry of the individual on genetic variants with heterogeneous associations.

[0023] In some aspects, including the effects of the global ancestry of the individual includes including, for each instance of genomic data in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships, an interaction described by a number of copies of each genetic variant with heterogeneous associations in each instance of genomic data and principal components of a genomic matrix generated using each instance of genomic data.

[0024] In some aspects, performing the regression includes performing Elastic Net regression.

[0025] In some aspects, training the first computational model includes performing a regression including at least one penalty factor determined based on one or more biological priors.Attorney Docket No.: M0437.70168WO00

[0026] In some aspects, the one or more biological priors include one or more of the following: variant pathogenicity, predicted variant consequences, known causal variants, tissue-specific regulatory genomic annotations, cell-type-specific regulatory genomic annotations, and / or aggregation of one or more of the preceding effects.

[0027] In some aspects, the method further includes determining the genetic variants with heterogeneous associations between single-ancestry populations by: determining genetic variants associated with the medical trait for two or more single-ancestry populations using inverse- variance weighted meta-analysis of genome-wide association studies (GWAS) for each of the two or more single-ancestry populations.

[0028] In some aspects, determining the genetic variants with heterogeneous associations further includes using Cochran's Q test to determine genetic variants associated with -values smaller than a threshold value.

[0029] In some aspects, calculating the at least one trait score further includes using information indicative of one or more conventional risk factors and / or genetic variants associated with the individual.

[0030] In some aspects, obtaining the trained computational model further includes performing a regression including effects on phenotypes in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships of the one or more conventional risk factors and / or the genetic variants associated with the individual.

[0031] In some aspects, obtaining the genomic data includes obtaining genotyping data previously obtained by genotyping a biological sample obtained from the individual.

[0032] In some aspects, obtaining the genomic data includes obtaining genotyping data by sequencing a biological sample obtained from the individual.

[0033] In some aspects, obtaining the genotyping data includes obtaining microarray data, whole-genome sequencing data, whole-exome sequencing data, and / or genotype imputation from partially observed data.

[0034] In some aspects, training the first computational model further includes using one or more of a linear genetic effect, a genetic dominance effect, and / or a sex-based genetic effect associated with the medical trait.

[0035] In some aspects, obtaining the trained computational model further includes: determining an optimal regularization parameter value by: training a first plurality ofAttorney Docket No.: M0437.70168WO00 computational models using a plurality of different regularization parameter values and a subset of the data describing phenotypes, genotypes, and / or phenotype-genotype relationships; evaluating a predictive performance of each of the first plurality of trained computational models; and selecting the optimal regularization parameter value based on the evaluated predictive performance of each of the first plurality of trained computational models.

[0036] In some aspects, obtaining the trained computational model further includes training the first computational model using the optimal regularization parameter value.

[0037] In some aspects, the techniques described herein relate to a method of performing a clinical trial. The method includes: obtaining, for a first individual, a first trait score associated with a medical trait by: calculating the first trait score using an at least one processor, a trained computational model, and first genomic data including information indicative of genetic variants present in a genome of the first individual, wherein the trained computational model was obtained by training a first computational model using data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries; enrolling the first individual in the clinical trial based on a value of the first trait score; and altering a course of medical treatment for the first individual in accordance with the clinical trial.

[0038] In some aspects, the techniques described herein relate to a system, including: at least one computer hardware processor; and at least one non-transitory computer readable medium storing processor-executable instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of performing the clinical trial.

[0039] In some aspects, the techniques described herein relate to at least one non-transitory computer readable medium storing processor-executable instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of performing the clinical trial.

[0040] In some aspects, enrolling the first individual in the clinical trial based on the value of the first trait score includes enrolling the first individual in the clinical trial based on the value of the first trait score relative to a threshold value.

[0041] In some aspects, the method further includes: obtaining, for a second individual, a second trait score associated with the medical trait by calculating the second trait score using anAttorney Docket No.: M0437.70168WO00 at least one processor, the trained computational model, and second genomic data including information indicative of genetic variants present in a genome of the second individual; and declining to enroll the second individual in the clinical trial based on a value of the second trait score.

[0042] In some aspects, declining to enroll the second individual in the clinical trial based on the value of the second trait score includes declining to enroll the second individual in the clinical trial based on the value of the second trait score relative to a threshold value.

[0043] In some aspects, training the first computational model includes using batch screening iterative lasso (BASIL) regression.

[0044] In some aspects, training the first computational model includes performing a regression including a loss, the loss including a squared loss and / or a binomial loss.

[0045] In some aspects, obtaining the trained computational model further includes using the trained first computational model and performing a regression including a penalty applied to genetic variants with heterogeneous associations between single-ancestry populations.

[0046] In some aspects, performing the regression further includes including effects of a global ancestry of the first individual on genetic variants with heterogeneous associations.

[0047] In some aspects, including the effects of the global ancestry of the first individual includes including, for each instance of genomic data in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships, an interaction described by a number of copies of each genetic variant with heterogeneous associations in each instance of genomic data and principal components of a genomic matrix generated using each instance of genomic data.

[0048] In some aspects, performing the regression includes performing Elastic Net regression.

[0049] In some aspects, training the first computational model includes performing a regression including at least one penalty factor determined based on one or more biological priors.

[0050] In some aspects, the one or more biological priors include one or more of the following: variant pathogenicity, predicted variant consequences, known causal variants, tissue-specific regulatory genomic annotations, cell-type-specific regulatory genomic annotations, and / or aggregation of one or more of the preceding effects.

[0051] In some aspects, the method further includes determining the genetic variants with heterogeneous associations between single-ancestry populations by: determining genetic variants associated with the medical trait for two or more single-ancestry populations using inverse-Attorney Docket No.: M0437.70168WO00 variance weighted meta-analysis of genome-wide association studies (GWAS) for each of the two or more single-ancestry populations.

[0052] In some aspects, determining the genetic variants with heterogeneous associations further includes using Cochran's Q test to determine genetic variants associated with -values smaller than a threshold value.

[0053] In some aspects, calculating the at least one trait score further includes using information indicative of one or more conventional risk factors and / or genetic variants associated with the first individual.

[0054] In some aspects, obtaining the trained computational model further includes performing a regression including effects on phenotypes in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships of the one or more conventional risk factors and / or the genetic variants associated with the first individual.

[0055] In some aspects, obtaining the genomic data includes obtaining genotyping data previously obtained by genotyping a biological sample obtained from the first individual.

[0056] In some aspects, obtaining the first genomic data includes obtaining genotyping data by sequencing a biological sample obtained from the first individual.

[0057] In some aspects, obtaining the genotyping data includes obtaining microarray data, whole-genome sequencing data, whole-exome sequencing data, and / or genotype imputation from partially observed data.

[0058] In some aspects, training the first computational model further includes using one or more of a linear genetic effect, a genetic dominance effect, and / or a sex-based genetic effect associated with the medical trait.

[0059] In some aspects, obtaining the trained computational model further includes: determining an optimal regularization parameter value by: training a first plurality of computational models using a plurality of different regularization parameter values and a subset of the data describing phenotypes, genotypes, and / or phenotype-genotype relationships; evaluating a predictive performance of each of the first plurality of trained computational models; and selecting the optimal regularization parameter value based on the evaluated predictive performance of each of the first plurality of trained computational models.

[0060] In some aspects, obtaining the trained computational model further includes training the first computational model using the optimal regularization parameter value.Attorney Docket No.: M0437.70168WO00 BRIEF DESCRIPTION OF DRAWINGS

[0061] Various aspects and embodiments will be described with reference to the following figures. It should be appreciated that the figures are not drawn to scale. In the drawings, each identical or nearly identical component that is illustrated in various figures is represented by a like numeral. For purposes of clarity, not every component may be labeled in every drawing.

[0062] FIG.1-1 is a schematic block diagram of a trait score facility, in accordance with some embodiments of the technology described herein.

[0063] FIG. 1-2 is a flowchart of a process 200 for determining an individual’s trait score, including an inclusive polygenic score (iPGS), in accordance with some embodiments of the technology described herein.

[0064] FIG. 1-3 is a flowchart of a process 300 for generating a computational model for determining an individual’s trait score, including an iPGS, in accordance with some embodiments of the technology described herein.

[0065] FIG. 1-4 is a flowchart of another process 400 for generating a computational model for determining an individual’s trait score, including an iPGS, in accordance with some embodiments of the technology described herein.

[0066] FIG.1-5 is a flowchart of a process 500 for designing a clinical trial using trait scores, in accordance with some embodiments of the technology described herein.

[0067] FIG.1-6 is a diagram of an illustrative computer system 600, in accordance with some embodiments.

[0068] FIG. 2-1 shows a simulation study with synthetic data. Across three heritability and polygenicity scenarios (top), PGS models were fit with 110,000 individuals of synthetic African and synthetic European ancestry of 11 different compositions. The predictive performance was quantified in the held-out test set. Error bars represent 95% confidence intervals.

[0069] FIGs. 2-2A to 2-2B show inclusive PGS (iPGS) training with diverse ancestry enhances the transferability of polygenic scores in the UK Biobank. FIG. 2-2A shows principal- component projections of the unrelated individuals in the UK Biobank and population-label assignment. FIG. 2-2B shows relative average improvements of PGS model performance against the baseline model trained only with White British individuals (material and methods). Error bars represent 95% confidence intervals of average improvements.Attorney Docket No.: M0437.70168WO00

[0070] FIGs. 2-3A to 2-3G show a systematic predictive performance evaluation of inclusive PGS (iPGS) models and PRS-CSx across 60 anthropometric and hematological traits in the UK Biobank. FIG. 2-3A shows the predictive performance (R2) in White British (WB), South Asian (SA), and African (Afr) groups in the UK Biobank for four select models: (i) WB-only, (ii) inclusive, (iii) inclusive-FixN, and (vii) PRS-CSx. FIG.2-3B shows the number of approximately LD-independent (R2< 0.2 in the African population in the UK Biobank) variants with heterogeneous genome-wide association studies (GWAS) associations (material and methods). FIGs. 2-3C to 2-3G show the predictive performance of up to eight PGS models in White British (WB) and African (Afr) populations in the UK Biobank for five select traits. The refit models are trained only for the neutrophil and leukocyte counts, where genetic variants with heterogeneous GWAS effects were observed. The predictive performance for other models and ancestry groups is shown in FIGs. 2-8A to 2-8B and FIG. 2-9. BMI: body mass index. Vol.: volume. Dist.: distribution. Impd.: impedance. Frac.: fraction. Conc.: concentration. %: percentage. R: right. L: left. Error bars represent 95% confidence intervals.

[0071] FIGs. 2-4A to 2-4E show enhanced predictive performance with iPGS+refit that additionally accounts for ancestry-dependent genetic effects. FIG. 2-4A shows GWAS meta- analysis heterogeneity test in the UK Biobank for the neutrophil count. Genetic variants with heterogeneity ^ value < 5 × 10^^are highlighted. FIG. 2-4B shows GWAS effect size comparison between White British (x axis) and African (y axis) populations in the UK Biobank. The color indicates whether the variants show heterogeneous GWAS associations. Error bars represent the standard error of the GWAS effect-size estimates. FIG.2-4C shows allele frequency comparison for 5,890 genetic variants associated with neutrophil count (material and methods). Heterogeneous GWAS associations are shown. The shape and size represent the direction and the magnitude of GWAS associations in the African population in the UK Biobank. FIGs.2-4D to 2- 4E show phenotype mean values of neutrophil count (FIG. 2-4D) and leukocyte count (FIG. 2- 4E) stratified by decile of PGS in the held-out test set of individuals of the African population in the UK Biobank are shown. Error bars represent the standard error-of-mean estimates. PTVs: protein-truncating variants. PAVs: protein-altering variants. PCVs: proximal coding variants. Intronic: intronic variants. UTR: genetic variants on untranslated regions. Others: other non- coding variants.Attorney Docket No.: M0437.70168WO00

[0072] FIG. 2-5 shows the size of the inclusive PGS model. The number of genetic variants with non-zero coefficients in the inclusive PGS model trained on up to 284,661 individuals is shown as a histogram across 60 traits.

[0073] FIGs. 2-6A to 2-6D show a comparison of effect size between GWAS and inclusive polygenic scores (iPGS) for standing height. The effect size estimates from GWAS (x-axis) and iPGS (y-axis) are compared for (FIG. 2-6A) white British, (FIG. 2-6B) non-British white, (FIG. 2-6C) South Asian, and (FIG.2-6D) African. Error bars represent the standard error of the GWAS effect size estimates.

[0074] FIG. 2-7 shows a comparison of predicted PGS in the held-out test set individuals between the WB-only and the Inclusive PGS models. Across 60 traits, the Pearson’s correlation (R2) was evaluated between the two predicted PGS values across all unrelated held-out test individuals (“All”) as well as a subset of held-out test set individuals (WB: white British, NBW: non-British white, SA: South Asian, Afr: African, and Others). The dashed lines show the median value across 60 traits.

[0075] FIG. 2-8A shows a systematic predictive performance evaluation of inclusive PGS (iPGS) and PRS-CSx models across 31 anthropometric traits in UK Biobank. The predictive performance (R2) in white British (WB), non-British white (NBW), South Asian (SA), African (Afr), and Others in UK Biobank are shown for four iPGS models and two PRS-CSx models: (i) WB-only, (ii) inclusive, (iii) inclusive-FixN, (iv) MultiPop, NoAdmixed, and (vii) PRS-CSx, and (viii)PRS-CSx (Ntrain=256k). Error bars represent 95% confidence intervals. A subset of the results is shown in FIGs. 2-3A to 2-3G.

[0076] FIG. 2-8B shows a systematic predictive performance evaluation of inclusive PGS (iPGS) and PRS-CSx models across 29 hematological traits in UK Biobank. The predictive performance (R2) in white British (WB), non-British white (NBW), South Asian (SA), African (Afr), and Others in UK Biobank are shown for four iPGS models and two PRS-CSx models: (i) WB-only, (ii) inclusive, (iii) inclusive-FixN, (iv) MultiPop, NoAdmixed, and (vii) PRS-CSx, and (viii) PRS-CSx (Ntrain=256k). Error bars represent 95% confidence intervals. A subset of the results is shown in FIGs. 2-3A to 2-3G.

[0077] FIG. 2-9 shows the predictive performance of up to eight PGS models for five select traits. The refit models are trained only for the neutrophil and leukocyte counts, where genetic variants with heterogeneous GWAS effects were observed. WB: white British, NBW: non-BritishAttorney Docket No.: M0437.70168WO00 white, SA: South Asian, and Afr: African. Error bars represent 95% confidence intervals. A subset of the results is shown in FIGs. 2-3A to 2-3G.

[0078] FIG. 2-10 shows average improvements of an inclusive PGS (iPGS) model against a WB-only model. For the iPGS and WB-only model, the predictive performance (R2) was quantified using the individuals in the held-out test set for each of the 60 hematological and anthropometric traits. For each of the five population groups in UK Biobank, the improvementswere regressed on the baseline mode (R2WB-only) using Deming regression and the regression coefficients were reported as the average improvements. The 95% confidence interval of the estimated coefficients is 60.8% (95% CI: [31.2%, 90.3%]) for African ancestry individuals, 11.6% (95% CI: [8.9%, 14.4%]) for South Asian, 7.3% (95% CI: [5.0%, 9.5%]) for non-British white, 4.8% (95% CI: [3.7%, 5.9%]) for white British, and 17.8% (95% CI: [11.8%, 23.8%]) for the remaining individuals (Others).

[0079] FIG. 2-11 shows comparisons of heritability and predictive performance of the inclusive PGS model. The estimated heritability (h²) (x-axis) and predictive performance of the inclusive PGS model (y-axis) in the white British population in UK Biobank are shown. Error bars represent standard error.

[0080] FIG. 2-12 shows cumulative frequency distribution of minor allele frequency of the genetic variants in UK Biobank. The allele frequency was computed using different subsets of unrelated individuals in UK Biobank and different subsets of genetic variants. The “UKB variants” represents all of the 1,316,181 genetic variants considered in the study, consisting of directly genotyped and imputed genetic variants and imputed HLA allotypes. The union of variants selected in the PGS models were considered across 60 anthropometric and hematological traits for each of the four PGS models: WB-only (317,782 variants), Inclusive (347,362 variants), Inclusive- FixN (312,631 variants), and MultiPop, NoAdmixed (315,122 variants). The cumulative frequency distribution was assessed on the log-linear intervals of the minor allele frequency bin (x-axis).

[0081] FIG. 2-13 shows a comparison of allele frequency of genetic variants in white British and African in UK Biobank. The 347,362 genetic variants selected for the Inclusive PGS model in at least one of the 60 anthropometric and hematological traits were focused on and their allele frequency was compared in white British (x-axis) and African (y-axis). The contour lines of the density plot and genetic variants outside the contour line with the lowest density are shown.Attorney Docket No.: M0437.70168WO00

[0082] FIGs. 2-14A to 2-14D show ancestry-biased genetic associations for the neutrophil count in UK Biobank. FIG. 2-14A shows a Manhattan plot for the meta-analyzed p values from UK Biobank GWAS association analysis across white British, non-British white, South Asian, and African ancestry groups are shown. FIGs. 2-14B to 2-14C show a Manhattan plot for population- specific GWAS analysis for white British (FIG.2-14B) and African (FIG.2-14C) ancestry groups in UK Biobank. FIG.2-14D shows a Manhattan plot for GWAS heterogeneity p value across white British, non-British white, South Asian, and African ancestry groups in UK Biobank. The shape of the points represents the predicted consequence of genetic variants from Ensembl’s Variant Effect Predictor (VEP). PTVs: protein-truncating variants, PAVs: protein-altering variants, PCVs: proximal coding variants, Intronic: intronic variants, UTR: genetic variants on untranslated regions, and Others: other non-coding variants.

[0083] FIG. 2-15 shows allele frequency and minor allele frequency of genetic associations for the neutrophil count in UK Biobank. Cascade plots for population-specific GWAS analysis for white British (top) and African (bottom) ancestry groups in UK Biobank. The shape of the points represents the predicted consequence of genetic variants from Ensembl’s Variant Effect Predictor (VEP). PTVs: protein-truncating variants, PAVs: protein-altering variants, PCVs: proximal coding variants, Intronic: intronic variants, UTR: genetic variants on untranslated regions, and Others: other non-coding variants.

[0084] FIGs. 2-16A to 2-16C show a comparison of PGS coefficients in two PGS models for neutrophil counts. FIGs. 2-16A to 2-16B the coefficients (BETA) for the inclusive PGS model (FIG. 2-16A) and WB-only model (FIG. 2-16B). The shape of the points represents the predicted consequence of genetic variants from Ensembl’s Variant Effect Predictor (VEP). FIG. 2-16C shows a comparison of PGS coefficients in the WB-only model (x-axis) and the inclusive PGS model (y-axis). PTVs: protein-truncating variants, PAVs: protein-altering variants, PCVs: proximal coding variants, Intronic: intronic variants, UTR: genetic variants on untranslated regions, and Others: other non-coding variants.

[0085] FIGs. 3-1A to 3-1E show GenESIS integrates validated GxS effects on top of linear effects in a unified framework. FIG. 3-1A shows the number of predictor variables considered in iPGS (x-axis) and GenESIS model (y-axis). Each point corresponds to a trait category and whether the GenESIS model is selected based on the validation set metric. TBill: total bilirubin. DBill: direct bilirubin. FIGs. 3-1B to 3-1C show a comparison of GWAS effect size estimates betweenAttorney Docket No.: M0437.70168WO00 females (x-axis) and males (y-axis) for hip circumference in white British individuals. The genetic variants selected in GenESIS only for linear effects are shown in (FIG. 3-1B), and the selected variants with GxS interaction effects are shown in (FIG.3-1C). FIG.3-1D shows the distributions of GxS interaction effect test statistics, TDiff, for hip circumference in white British individuals. Genetic variants selected in the GenESIS model are shown, stratified by whether the variant has GxS interaction effects. FIG. 3-1E shows the number of traits where GenESIS is selected for trait prediction on the basis of the validation set metric. For each population group (y-axis), the number of traits per trait category is shown.

[0086] FIGs. 3-2A to 3-2D show enhanced PGS transferability in non-European populations with GenESIS. FIG. 3-2A shows the magnitude and statistical significance of the gain in predictive performance of GenESIS over linear-only iPGS across three population groups. 32 traits are shown where GenESIS was selected for at least one of the three populations (white British, South African, and African). The full results are shown in FIG. 3-6. FIG. 3-2B shows a fraction of GxS interaction terms in the selected predictor variables in the GenESIS model. FIG. 3-2C shows the predictive performance (R2) of covariate-only model, linear-only iPGS model, and GenESIS model for hip circumferences and body mass index in individuals of White British, South Asian, and African ancestry in the held-out test set. FIG.3-2D shows for hip circumference, the predictive performance of the GenESIS model (partial R2for covariate-adjusted phenotype, x- axis) was compared for individuals of African ancestry against all publicly available PGS model evaluations in the PGS catalog (y-axis). Error bars represent 95% confidence intervals.

[0087] FIGs. 3-3A to 3-3D show sparse GenESIS models offer interpretation. FIGs. 3-3A to 3-3B show the linear (FIG. 3-3A) and GxS interaction (FIG. 3-3B) effects in the GenESIS model for hip circumference. Genetic variants with large effects were annotated. The GxS effect size directions are represented for male individuals. FIG. 3-3C shows pleiotropic association of rs1260326, a protein-altering variant in GCKR gene for select traits. The GWAS-based effect size estimates are shown on the left and the statistical significance of the association on the right. Phenome-wide associations with nominal ^ < 1 × 10^^^are shown in Table 18. FIG.3-3D shows enriched ontology terms for genetic variants with GxS effects in GenESIS, nominated by the GREAT enrichment analysis.

[0088] FIGs. 3-4A to 3-4B show the number and magnitude of linear and gene-by-sex interaction effects in GenESIS. FIG. 3-4A shows a comparison of the number of predictorAttorney Docket No.: M0437.70168WO00 variables with linear (x-axis) and GxS interaction effects (y-axis) across 99 traits. FIG.3-4B shows a comparison of the linear and GxS effect sizes across 99 traits. The median of the absolute value of the effect size per standard deviation of trait values is shown along with the interquartile range (IQR). y=x is shown as a solid line and the linear regression fit, y = 0.752x, as a dashed line.

[0089] FIGs. 3-5A to 3-5B show validating GxS effects in GenESIS by sex-stratified GWAS for bilirubin traits. FIGs. 3-5A to 3-5B show the distributions of gene-by-sex (GxS) interaction test statistics (Example 2 Methods) for the genetic variants included in the GenESIS models stratified by whether the variant has linear or GxS interaction effects for total bilirubin (FIG. 3- 5A) and direct bilirubin (FIG. 3-5B).

[0090] FIGs. 3-6A to 3-6B show a systematic evaluation of the predictive performance of GenESIS and linear-only iPGS. FIG.3-6A shows the magnitude and statistical significance of the gain in predictive performance of GenESIS over linear-only iPGS across five population groups across 99 quantitative traits in UK Biobank. FIG. 3-6B shows a fraction of GxS interaction terms in the selected predictor variables in the GenESIS model.

[0091] FIG. 3-7 shows enhanced PGS transferability in non-European populations with GenESIS. Predictive performance (R2) of the covariate-only model, linear-only iPGS model, and GenESIS model is shown for hip circumferences and body mass index for the following five population groups in the held-out test set: White British, non-British White, South Asian, African, and other individuals.

[0092] FIG. 3-8 shows modeling male- and female-specific genetic effects in GenESIS. The predictive performance (R2) of GenESIS PGS models is shown for seven traits (rows) across five population groups (cols) in the UK Biobank resource. The “GxS_m'' model represents the primary analysis, focusing on sex-shared and male-specific effects. On the other hand, the “GxS_m&f” model represents the additional analysis with 3.9 million predictor variables, considering sex- shared, male-specific, and female-specific effects (Example 2 Methods). Error bars represent 95% confidence intervals.

[0093] FIG. 4-1A shows improvements in predictive performance (R2) for white British individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference between the predictive performance of the iPGS model vs. the baseline model is shown on the y-axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show theAttorney Docket No.: M0437.70168WO00 same predictive performance. The dashed line represents the average improvement of 6.28% across 177 quantitative traits.

[0094] FIG.4-1B shows improvements in predictive performance (AUROC) for white British individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference between the predictive performance of the iPGS model vs. the baseline model is shown on the y-axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show the same predictive performance. The dashed line represents the average improvement of 3.94% across 49 binary traits.

[0095] FIG. 4-2A shows improvements in predictive performance (R2) for non-British white individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference between the predictive performance of the iPGS model vs. the baseline model is shown on the y-axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show the same predictive performance. The dashed line represents the average improvement of 7.41% across 177 quantitative traits.

[0096] FIG. 4-2B shows improvements in predictive performance (AUROC) for non-British white individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference between the predictive performance of the iPGS model vs. the baseline model is shown on the y- axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show the same predictive performance. The dashed line represents the average improvement of 9.29% across 21 binary traits.

[0097] FIG. 4-3A shows improvements in predictive performance (R2) for South Asian individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference between the predictive performance of the iPGS model vs. the baseline model is shown on the y-axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show the same predictive performance. The dashed line represents the average improvement of 13.11% across 177 quantitative traits.Attorney Docket No.: M0437.70168WO00

[0098] FIG.4-3B shows improvements in predictive performance (AUROC) for South Asian individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference between the predictive performance of the iPGS model vs. the baseline model is shown on the y-axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show the same predictive performance. The dashed line represents the average improvement of 1.03% across 10 binary traits.

[0099] FIG.4-4A shows improvements in predictive performance (R2) for African individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference between the predictive performance of the iPGS model vs. the baseline model is shown on the y-axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show the same predictive performance. The dashed line represents the average improvement of 26.74% across 177 quantitative traits.

[0100] FIG. 4-4B shows improvements in predictive performance (AUROC) for African individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference between the predictive performance of the iPGS model vs. the baseline model is shown on the y-axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show the same predictive performance. The dashed line represents the average improvement of 3.75% across 5 binary traits.

[0101] FIG. 4-5A shows improvements in predictive performance (R2) for other individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference between the predictive performance of the iPGS model vs. the baseline model is shown on the y-axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show the same predictive performance. The dashed line represents the average improvement of 14.29% across 177 quantitative traits.

[0102] FIG. 4-5B shows improvements in predictive performance (AUROC) for other individuals in the held-out test set in UK Biobank. The predictive performance of the baseline model, trained only on white British individuals, is shown on the x-axis. The difference betweenAttorney Docket No.: M0437.70168WO00 the predictive performance of the iPGS model vs. the baseline model is shown on the y-axis. The dotted horizontal line represents when the two models (iPGS and the baseline model) show the same predictive performance. The dashed line represents the average improvement of 5.58% across 38 binary traits.

[0103] FIG. 4-6A shows improvements in predictive performance (R2) for white British individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y-axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of 0.45% across 364 quantitative traits.

[0104] FIG.4-6B shows improvements in predictive performance (AUROC) for white British individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y-axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of 3.83% across 49 binary traits.

[0105] FIG. 4-7A shows improvements in predictive performance (R2) for non-British white individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y-axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of -0.64% (a slightly decreased performance) across 364 quantitative traits.

[0106] FIG. 4-7B shows improvements in predictive performance (AUROC) for non-British white individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y- axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of 2.31% across 21 binary traits.Attorney Docket No.: M0437.70168WO00

[0107] FIG. 4-8A shows improvements in predictive performance (R2) for South Asian individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y-axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of 0.47% across 364 quantitative traits.

[0108] FIG.4-8B shows improvements in predictive performance (AUROC) for South Asian individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y-axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of 5.39% across 10 binary traits.

[0109] FIG.4-9A shows improvements in predictive performance (R2) for African individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y-axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of 8.72% across 364 quantitative traits.

[0110] FIG. 4-9B shows improvements in predictive performance (AUROC) for African individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y-axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of 25.12% across 5 binary traits.

[0111] FIG. 4-10A shows improvements in predictive performance (R2) for other individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y-axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of 0.58% across 364 quantitative traits.Attorney Docket No.: M0437.70168WO00

[0112] FIG. 4-10B shows improvements in predictive performance (AUROC) for other individuals in the held-out test set in UK Biobank. The predictive performance of the original iPGS model is shown on the x-axis. The difference in the predictive performance between the two iPGS models trained with and without the “training & validation refit” is shown on the y-axis. The dotted horizontal line represents when the two models show the same predictive performance. The dashed line represents the average improvement of 5.21% across 38 binary traits. DETAILED DESCRIPTION

[0113] A polygenic score (PGS) is a quantitative method used to predict an individual’s predisposition to a particular phenotype trait based only on their genome. The traits that may be described using a PGS are often of clinical relevance, and as a result, PGSes have a large number of potential applications in medical practice. For example, PGSes can be used to predict an individual's risk of developing certain diseases (e.g., coronary artery disease, type 2 diabetes, and / or breast cancer, as non-limiting examples). Additionally, PGSes may be used to improve preventative health and to provide personalized medical care. For example, individuals with a high PGS associated with a medical condition may benefit from more aggressive preventative measures or monitoring. In contrast, it may be beneficial to reduce unnecessary tests for “low-risk” individuals (e.g., with a low PGS) as reducing invasive testing and associated costs. PGSes may also inform personalized prevention and treatment strategies (e.g., for cancer, cardiovascular conditions, or other diseases, as non-limiting examples).

[0114] As another example, PGSes may inform pharmaceutical research and clinical trials. Researchers can use polygenic scores to understand how genetics influence drug response, aiding in the development of new medications. PGSes can also be used to identify individuals with higher genetic risk for a certain condition or medical trait, making them good candidates for clinical trials. Additionally, PGSes can be used to identify individuals with certain genetic predispositions ot disease conditions to assist in the development of cellular models (e.g., iPS cells) for pharmaceutical research. Furthermore, PGSes may be used in prenatal and infant health screening to identify health risks early in life. Such genetic screening using PGSes may provide valuable information for genetic counselors, who can then provide more precise advice to parents. Improved methods for calculating PGSes may therefore yield more clinically relevant PGSes such that clinicians may use PGSes to guide an individual’s medical treatment.Attorney Docket No.: M0437.70168WO00

[0115] Computational models for calculating PGSes are typically built using the results of genome-wide association study (GWAS) analysis. GWAS summary statistics, which provide information about an association between each genetic variant and a phenotype trait of interest, may be provided as an input. These statistics include the effect size (often a ^ coefficient) and ^- value for each genetic variant. As a first step, genetic variants of interest for the PGS model may be selected from the GWAS summary statistics. One approach is based on linkage disequilibrium (LD) clumping and p-value thresholding (C+T, also known as pruning and thresholding (P+T)), where correlated genetic variants are removed using LD clumping or LD pruning to ensure independence among the selected genetic variants. Thereafter, additional genetic variants are removed that do not meet a selected statistical significance threshold (e.g., genetic variants with ^ > 0.05). As a second step, each selected genetic variant is assigned a weight based on its effect size from the GWAS. This weight represents the contribution of the genetic variant to the phenotype trait. In the case of C+T, the effect size estimates from GWAS are used as the weights. Third, a PGS is determined for each individual in the sample data set. For each individual, the number of risk alleles that are present in their genome for each genetic variant is multiplied by the weight of that genetic variant. The PGS for that individual is then determined by summing across all genetic variants. In some approaches, the first and second steps are performed simultaneously. This procedure results in the selection of relevant genetic variants as well as an estimation of their PGS effect size.

[0116] An increased availability of genetic data in recent years has allowed researchers to analyze larger numbers of individual genomes and, correspondingly, to provide more reliable PGS models derived from increased statistical power. In parallel, there have been methodological improvements in training PGS models with higher accuracy. However, most PGS models suffer from limited transferability across populations. Typically, PGS models are trained from GWAS data collected from specific populations (e.g., Europeans, Africans, etc.), resulting in final PGS models that are population specific. It has been recognized that PGS models trained from a single population show limited predictive performance in individuals from different populations than the original training population. Because non-European individuals are typically underrepresented in genetic studies, they are later underserved by PGS models trained on European populations. Thus, this underrepresentation results in inequitable healthcare improvements derived from advancements in genetic research.Attorney Docket No.: M0437.70168WO00

[0117] Admixed individuals, whose genomes consist of haplotypes from more than one ancestry group and account for one in seven newborns in the United States, are often excluded in PGS model training due to technical limitations. While methods exist to apply GWAS analysis to admixed individuals, dependencies on the LD reference panels and computational complexities in representing LD for admixed individuals present challenges in variant effect size estimation in PGS modeling. The inventors have recognized and appreciated that including genetic data from admixed individuals in the training of PGS models represents an improvement in that using such genetic data yields PGS models with improved applicability across populations. In particular, the use of genetic data from admixed individuals offers improved PGS modelling for non-European populations and, as a result, improved equity in healthcare.

[0118] The inventors have further recognized and appreciated that population specific PGS models typically provide reduced PGS model accuracy for populations that are currently underrepresented in genetic studies (e.g., non-European populations). Accordingly, the inventors have developed an inclusive PGS (“iPGS”) model, described herein, in which genetic variants that exhibit genetic effects that are shared across different ancestries are characterized by a score determined by a single PGS model without relying on GWAS analysis of single-ancestry populations. Genetic variants showing ancestry-dependent genetic effects, if relevant to the phenotype trait being assessed, are then subsequently analyzed and included in the individual’s iPGS, allowing for joint modeling of ancestry-shared and ancestry-dependent genetic effects in a single PGS model.

[0119] In some embodiments, the techniques described herein include systems and methods for predicting a medical trait of an individual using a trait score and / or iPGS. The medical trait may be one or more of (i) continuous variable traits (e.g., blood pressure, weight, etc.), (ii) binary traits (e.g., whether the individual has a disease or a condition and / or a likelihood of the individual developing a disease or a condition), and / or (iii) a time-to-event (e.g., a prediction of how long the individual has before the onset of a disease, condition, or other trait). The medical trait may further include behavioral traits, physiological traits, and / or cognitive traits. The techniques include obtaining genomic data including information indicative of genetic variants present in the individual’s genome. The genomic data may be obtained by, prior to or contemporaneous to calculating the individual’s trait score and / or iPGS, genotyping the individual (e.g., by sequencing a biological sample obtained from the individual to obtain the genomic data). In someAttorney Docket No.: M0437.70168WO00 embodiments, genotyping may be performed using one or more of a DNA microarray, whole- genome sequencing, and / or whole-exome sequencing. Genotyping may also include performing genotype imputation using partially observed data.

[0120] In some embodiments, after obtaining the genetic data, the techniques include calculating a trait score based on the information indicative of the genetic variants. The trait score is calculated using a computational model obtained by determining effect sizes for one or more genetic variants associated with the medical trait. This determination may include training the computational model using supervised learning on data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries. Here, the term “admixed” refers to individuals with recent (<100 generations) admixture events between isolated populations. The supervised learning may be performed using, for example, penalized regression techniques including, but not limited to ^^-penalized regression (LASSO), ^^-penalized regression (Ridge), and / or ^^- and ^^-penalized regression (Elastic Net). Alternatively or additionally, the supervised learning may be performed using, for example, statistical boosting techniques. In some embodiments, batch-based or iterative techniques, including but not limited to batch screening iterative lasso (BASIL) regression, can be used to perform supervised learning using large-scale individual-level data.

[0121] In some embodiments, the techniques further include determining whether the medical trait is or is not described by an ancestry-dependent genetic effect. That is, the techniques include determining whether one or more genetic variants that are relevant to the medical trait have heterogeneous associations between single-ancestry populations (e.g., whether the genetic variants are more likely to be correlated with the medical trait in one population than another). If one or more genetic variants that are relevant to the medical trait have heterogeneous associations between single-ancestry populations, then a regression may be performed to obtain the computational model. The regression may include an Elastic Net regression including a penalty applied to the determined genetic variants with heterogeneous associations.

[0122] In some embodiments, the techniques next include generating a graphical user interface including a visualization of the calculated trait score. For example, the graphical user interface may include one or more charts and / or tables describing the calculated trait score and its impact on the medical trait of interest. Thereafter, the generated graphical user interface isAttorney Docket No.: M0437.70168WO00 displayed using a display device (e.g., a computer monitor, television screen, smartphone display, mobile device display, and / or a tablet display).

[0123] Following below are more detailed descriptions of various concepts related to, and embodiments of, methods and apparatus for the generation of inclusive PGS models. It should be appreciated that various aspects described herein may be implemented in any of numerous ways. Examples of specific implementations are provided herein for illustrative purposes only. In addition, the various aspects described in the embodiments below may be used alone or in any combination and are not limited to the combinations explicitly described herein.

[0124] FIG. 1-1 is a schematic block diagram of a trait score facility 100, in accordance with some embodiments of the technology described herein. In the illustrative example of FIG. 1-1, trait score facility 100 includes a genotyping system 110, a trait score system 120, and a remote system 130. It should be appreciated that Trait score facility 100 is illustrative and that a trait score facility 100 may have one or more other components of any suitable type in addition to or instead of the components illustrated in FIG. 1-1. For example, there may be additional remote systems (e.g., two or more) or additional trait score systems (e.g., two or more) within a trait score facility 100.

[0125] As illustrated in FIG.1-1, in some embodiments, one or more of the genotyping system 110, the trait score system 120, and the remote system 130 may be communicatively connected by a network 140. The network 140 may be or include one or more local- and / or wide-area, wired and / or wireless networks, including a local-area or wide-area enterprise network and / or the Internet. Accordingly, the network 140 may be, for example, a hard-wired network (e.g., a local area network within a facility), a wireless network (e.g., connected over Wi-Fi and / or cellular networks), a cloud-based computing network, or any combination thereof. For example, in some embodiments, the genotyping system 110 and the trait score system 120 may be located within a same facility and connected directly to each other or connected to each other via the network 140, while the remote system 130 may be located in a remote facility and connected to the genotyping system 110 and / or the trait score system 120 through the network 140.

[0126] In some embodiments, the genotyping system 110 may be configured to perform genotyping of a biological sample obtained from an individual 112. The biological sample may be any suitable biological material obtained from the individual 112. For example, the biological sample may include one or more of saliva, blood, biopsied tissue, formalin-fixed paraffin-Attorney Docket No.: M0437.70168WO00 embedded (FFPE) tissue, fine-needle aspirate (FNA), core needle biopsies (CNBs), liquid biopsies, urine, feces, hair, epithelial cells, bone marrow, or other biological matter obtained from the individual 112.

[0127] In some embodiments, the genotyping system 110 may be configured to perform one or more types of genotyping. For example, the genotyping system 110 may be configured to perform whole genome sequencing, whole exome sequencing, and / or targeted sequencing (e.g., targeting a particular chromosome or genomic region). In some embodiments, the genotyping system 110 may be configured to perform genotyping using a DNA microarray (e.g., a “gene chip”).

[0128] As illustrated in FIG. 1-1, trait score facility 100 includes trait score system 120 communicatively coupled to the genotyping system 110. It should be appreciated that in some embodiments, the trait score system 120 may not be communicatively coupled directly to the genotyping system 110 but may obtain the genotyping results by retrieving the results from a separate computer-readable memory, including but not limited to a separate database (e.g., as associated with an electronic medical record (EMR)).

[0129] In some embodiments, trait score system 120 may be any suitable electronic device configured to receive information from genotyping system 110 and / or to process obtained genomic data. In some embodiments, trait score system 120 may be a fixed electronic device such as a desktop computer, a rack-mounted computer, or any other suitable fixed electronic device. Alternatively, trait score system 120 may be a portable device such as a laptop computer, a smart phone, a tablet computer, or any other portable device that may be configured to receive information from genotyping system 110 and / or to process obtained genomic data.

[0130] In some embodiments, trait score system 120 includes a computational facility 122. The computational facility 122 may be configured to analyze genomic data obtained by genotyping system 110. The computational facility 122 may be configured to, for example, use a computational model to analyze the obtained genomic data to determine an iPGS and / or a trait score (e.g., using the iPGS) associated with a medical trait for the individual 112, as described herein. The computational facility 122 may be implemented as hardware, software, or any suitable combination of hardware and software, as aspects of the disclosure provided herein are not limited in this respect.Attorney Docket No.: M0437.70168WO00

[0131] As illustrated in FIG. 1-1, the computational facility 122 may be implemented in the trait score system 120, such as being implemented in software (e.g., executable instructions) executed by one or more process of the trait score system 120. However, in other embodiments, the computational facility 122 may be additionally or alternatively implemented at one or more other elements of the Trait score facility 100 of FIG. 1-1. For example, the computational facility 122 may be implemented at the genotyping system 110 and / or the remote system 130 discussed herein. In other embodiments, the computational facility 122 may be implemented at or with another device, such as a device located remote from the trait score facility 100 and receiving data via the network 140.

[0132] As illustrated in FIG.1-1, trait score system 120 also interacts with remote system 130 through network 140, in some embodiments. Remote system 130 may be any suitable electronic device configured to receive information (e.g., from genotyping system 110 and / or trait score system 120) and to display generated graphical user interfaces for viewing. The remote system 130 may be remote from the genotyping system 110 and trait score system 120, such as by being located in a different room, wing, or building of a facility (e.g., a healthcare facility) than the genotyping system 110, or being geographically remote from the genotyping system 110 and trait score system 120, such as being located in another part of a city, another city, another state or country, etc. In some embodiments, remote system 130 may be a fixed electronic device such as a desktop computer, a rack-mounted computer, or any other suitable fixed electronic device. Alternatively, remote system 130 may be a portable device such as a laptop computer, a smart phone, a tablet computer, or any other portable device that may be configured to receive and view generated graphical user interfaces and / or to send instructions and / or information to trait score system 120.

[0133] In some embodiments, remote system 130 may receive information (e.g., information indicative of one or more iPGSes and / or of trait scores, generated graphics describing one or more iPGSes and / or trait scores) from trait score system 120 and / or genotyping system 110 over the network 140. A remote user 132 (e.g., a medical clinician) may use remote system 130 to view the received information on remote system 130. For example, the remote user 132 may view generated graphical user interfaces describing one or more iPGSes and / or trait scores using remote system 130 after the user 124 has completed data analysis using trait score system 120.Attorney Docket No.: M0437.70168WO00

[0134] FIG. 1-2 is a flowchart of a process 200 for predicting a medical trait of an individual by determining one or more trait scores and / or iPGSes, in accordance with some embodiments of the technology described herein. Process 200 may be implemented by a computational facility, such as the computational facility 122 of FIG. 1-1. As such, in some embodiments, the process 200 may be performed by a computing device configured to receive and / or obtain information from a genotyping system (e.g., genotyping system 110 described in connection with FIG. 1-1). As another example, in some embodiments, the process 200 may be performed by one or more processors located remotely (e.g., as part of a cloud computing environment, as connected through a network) from the genotyping system that obtained the input genomic data.

[0135] Process 200 may begin at act 202, where the computational facility obtains genomic data. The genomic data includes information indicative of genetic variants present in the individual’s genome. The genomic data may be obtained by genotyping the individual (e.g., by performing one or more of whole genome sequencing, whole exome sequencing, and / or targeted sequencing). In some embodiments, genotyping the individual may be performed by performing genotyping of a biological sample obtained from an individual 112. The biological sample may be any suitable biological material obtained from the individual 112. For example, the biological sample may include one or more of saliva, blood, biopsied tissue, formalin-fixed paraffin- embedded (FFPE) tissue, fine-needle aspirate (FNA), core needle biopsies (CNBs), liquid biopsies, urine, feces, hair, epithelial cells, bone marrow, or other biological matter obtained from the individual 112.

[0136] In some embodiments, genotyping the individual may be performed prior to and separately from determining the trait score and / or iPGS (e.g., by genotyping a biological sample obtained from the individual to obtain the genomic data prior to determining the trait score and / or iPGS). Alternatively, in some embodiments, genotyping the individual may be performed as a part of determining the trait score and / or iPGS (e.g., by genotyping a biological sample obtained from the individual). In some embodiments, genotyping may be performed using a DNA microarray.

[0137] After act 202, process 200 may proceed to act 204, where the computational facility may determine at least one trait score for the individual based on the information indicative of the genetic variants. The at least one trait score may be calculated using at least one processor (e.g., of a computing device implementing the computational facility) and a computational model trained to determine trait scores and / or iPGSes.Attorney Docket No.: M0437.70168WO00

[0138] In some embodiments, the computational model may be obtained by calculating effect sizes for one or more genetic variants associated with the medical trait. In some embodiments, the computational model may be obtained according to the embodiment of FIG. 1-3. FIG. 1-3 is a flowchart of a process 300 for generating a computational model for determining a trait score and / or an iPGS. Process 300 may be implemented by a computational facility, such as the computational facility 122 of FIG. 1-1. As such, in some embodiments, the process 300 may be performed by a computing device configured to receive and / or obtain information from a genotyping system (e.g., genotyping system 110 described in connection with FIG. 1-1). As another example, in some embodiments, the process 300 may be performed by one or more processors located remotely (e.g., as part of a cloud computing environment, as connected through a network) from the genotyping system that obtained the input genomic data.

[0139] In some embodiments, process 300 may begin at act 302, in which the computational facility may train an initial computational model using supervised learning and data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries. Here, the term “admixed” refers to individuals with recent (<100 generations) admixture events between isolated populations. The supervised learning may be performed using, for example, penalized regression techniques including, but not limited to ^^-penalized regression (LASSO), ^^-penalized regression (Ridge), and / or ^^- and ^^- penalized regression (Elastic Net). Alternatively or additionally, the supervised learning may be performed using, for example, statistical boosting techniques. In some embodiments, batch-based or iterative techniques, including but not limited to batch screening iterative lasso (BASIL) regression, can be used to perform supervised learning using large-scale individual-level data. The computational facility may implement BASIL using the R snpnet package (version 2) on individual-level data.

[0140] In some embodiments, implementing the supervised learning includes fitting an iPGS model by finding the exact solutions for ^^- and ^^-penalized multivariate regression (e.g., Elastic Net regression) using BASIL. Specifically, BASIL directly operates on the individual-level genotype and phenotype data and performs variable selection and effect size estimation simultaneously using penalized generalized linear models (GLMs). Given a phenotype vector ^ = individuals, a covariate matr^ ^^^^^^×^ix # = ^$^⋯ $^^^^^^^ ∈ & of ^^^ !"individuals and ' covariates, and a genotype matrix ( =ofAttorney Docket No.: M0437.70168WO00 ^^^ !"individuals and + variants, the following regularized regression problem may be implemented:where ℒ0^!, ^,!2 is a loss contribution for the Fthindividual, DE0^C2 is the Elastic Net regularization term for the coefficient ^C, 1 is a tuning parameter that controls the sparsity of the solution, >!describes sample weights, and BCare penalty factors for the Ithvariable. The weights, >!, and penalty factors, BC, allow different levels of shrinkage. One may use= 1 and = 1 as the default values.

[0141] Penalty factors less than 1 indicate prioritization of the predictor variables, whereas penalty factors greater than 1 indicate that the effect of the predictor variables is minimized or not preferred in the penalized regression model. One may set penalty factors of 0 for all covariate terms to ensure the covariate terms are unpenalized in the regression.

[0142] In some embodiments, one may assign different degrees of the penalty factor values to prioritize a subset of predictor variables according to biological knowledge (“biological priors”). The source(s) of biological knowledge may include variant pathogenicity (e.g., whether a gene variant is known to be pathogenic, likely pathogenic, likely benign, and / or benign), the predicted consequence of variants (e.g., whether the gene variants are putative loss-of-function variants, non-synonymous variants, etc.), known causal variants, tissue- or cell-type-sample-specific regulatory genomic annotations (e.g., gene variants with specific effects in tissue or cell types), and aggregations of the preceding effects, as non-limiting examples.

[0143] For a continuous phenotype trait (e.g., blood pressure), the loss contribution is squared loss: ℒ0^!, ^,!2 =The Elastic Net loss iswhere L is an Elastic Net parameter that controls the balance between the ^^- (Lasso) and ^^- (Ridge) penalization. The tuning parameter, 1, may be optimized based on the predictive performance on the validation data set. For example, the tuning parameter, λ, may be optimized using, for example, validation set metrics or 10-fold cross-validation, as implemented using the cv.glmnet function in the glmnet package in R. Additional aspects of optimizing the tuning parameter, λ, are described in in the Examples. In some embodiments, L may be set to a value of 0.99, and penalty factorsAttorney Docket No.: M0437.70168WO00 BC^ 1 may be set for non-synonymous coding variants or previously characterized pathogenic or likely-pathogenic variants.

[0144] In some embodiments, the function for the loss contribution may be selected in the generalized linear model (GLM) framework. For example, the loss contribution may be a squared loss for a continuous phenotype (e.g., the Gaussian family in GLM). The loss may be binomial for binary traits (e.g., the binomial family in GLM). In some embodiments, the techniques described herein may be applicable for multiple phenotypes (e.g., the multivariate Gaussian family), time-to-event phenotypes (Cox regression), and / or multiple time-to-event phenotypes. Additional aspects of the loss contribution are described in the Examples.

[0145] The covariate terms may be unpenalized in the regression. When fitting a PGS model from large-scale cohorts, it is not uncommon to have a large number of individuals (P ~ 300,000) and genetic variants (^ ~ 1,000,000). Instead of loading all the large-scale data on memory and fitting a regularized regression, BASIL efficiently solves the exact solution of the penalized regression in an iterative procedure by taking advantage of strong rules that guide the variable selection in each iteration step. Additional aspects of BASIL regression are described in “A fast and scalable framework for large-scale and ultrahigh dimensional sparse regression with application to the UK Biobank,” J. Qian, et al. (2020) PLOS Genetics 16(10): e1009141, which is incorporated herein by reference in its entirety.

[0146] In some embodiments, the computational model may be expanded to introduce genetic variants with linear effects, genetic dominance effects, and / or sex-based effects. Given a phenotype vectorof n individuals and a predictor matrix of n individuals and d variables, the following penalized generalized linear regression may be considered to fit the intercept term and regression coefficient vector as a function of the tuning parameter λ that controls the sparsity of the solution: whereElastic Net penalization term for the coefficient . With a balancing parameter between the L1- (Lasso) and L2- (Ridge) penalty, α, the Elastic Net penalization term is written as follows:Attorney Docket No.: M0437.70168WO00

[0147] The tuning parameter, λ, may be optimized based on the predictive performance on the validation set and by setting α=0.99. As optional parameters, one may specify sample weights, >!, and penalty factor values, BC, that allow different levels of shrinkage to variables. One may use >!= 1 and BC^ 1 as the default values.

[0148] Penalty factors less than 1 indicate prioritization of the predictor variables, whereas penalty factors greater than 1 indicate that the effect of the predictor variables is minimized or not preferred in the penalized regression model. One may set penalty factors of 0 for all covariate terms to ensure the covariate terms are unpenalized in the regression.

[0149] In some embodiments, one may assign different degrees of the penalty factor values to prioritize a subset of predictor variables according to biological knowledge (“biological priors”). The source(s) of biological knowledge may include variant pathogenicity (e.g., whether a gene variant is known to be pathogenic, likely pathogenic, likely benign, and / or benign), the predicted consequence of variants (e.g., whether the gene variants are putative loss-of-function variants, non-synonymous variants, etc.), known causal variants, tissue- or cell-type-sample-specific regulatory genomic annotations (e.g., gene variants with specific effects in tissue or cell types), and aggregations of the preceding effects, as non-limiting examples.

[0150] The loss contribution for the i-th observation,, depends on the types of the exponential family considered in the regression analysis of generalized linear models. For example, it is the squared loss, i.e., for quantitative phenotypes (Gaussian family), and it is logistic loss, i.e., for binary phenotypes (Binomial family). A similar model can be used for time-to-event phenotypes (Cox Proportional Hazards regression), or joint modeling of multiple phenotypes.

[0151] In some embodiments, for modeling linear effects of genetic variants, consider a covariate matrixand dcovcovariates and a genotype matrixof n individuals and dgvariants representing the allelic count of the affected allele. The concatenated vectors may be used as predictors:

[0152] The coefficient vector is also a concatenation of two components, corresponding to covariate-effects and linear effects of genetic variants, respectively:Attorney Docket No.: M0437.70168WO00

[0153] When the covariate terms are set to be unpenalized (i.e.the objective function of the penalized generalized linear regression becomes the following:

[0154] In some embodiments, to incorporate nonlinear genetic dominance effects into the computational model, consider a matrix,is an indicator variable representing whether the i-th individual is homozygous for effect allele for the j-th genetic variant. The concatenated vectors of three components may be used as predictors:The coefficient vector then has three components:One may focus on a smaller subset of variants for genetic dominance effects, instead of considering all of the dggenetic variants.

[0155] In some embodiments, to incorporate sex-based genetic effects, consider two matrices,represent Gene-by-Sex (GxS) interaction terms for male and female individuals, respectively, as follows:and , where is an indicator function. Thehas the original genotype dosage for male individuals and is always set to be zero for female individuals and vice versa for . The concatenated vectors of four components may be used as predictors:The coefficient vector then has four components:Attorney Docket No.: M0437.70168WO00There may also be other encoding of GxS interaction effects, as follows:One may use focus on a smaller subset of variants for GxS interaction effect.

[0156] In some embodiments, it is possible to include both genetic dominance and GxS interaction effects of the linear and genetic dominance effects in the computational model by using concatenated vectors of seven components as predictors:The coefficient vector then has seven components:One may focus on smaller subsets of variants for genetic dominance effects and GxS interaction effects of the linear and genetic dominance effects. Additional aspects related to GenESIS techniques are described in Example 2.

[0157] In some embodiments, after act 302, process 300 may proceed to act 304, in which the computational facility may determine genetic variants with heterogeneous associations between single-ancestry populations (e.g., genetic variants that more strongly indicate an association between the genetic variants and the medical trait in one population versus another population). Genetic variants with heterogeneous associations between single-ancestry populations may be determined using inverse-variance weighted meta-analysis with a heterogeneity test may be applied to GWASes of single-ancestry populations. For example, Cochran’s Q test may be used to determine genetic variants with heterogeneous associations by identifying genetic variants having heterogeneity ^-values smaller than a threshold value (e.g., ^-values < 5 ^ 10^^, as one non- limiting example).

[0158] In some embodiments, after act 304, process 300 may proceed to act 306, in which the computational facility may further train the initial computational model by performing a regression. The regression of act 306 may include, for example, Elastic Net regression or another suitable form of penalized regression. The regression may include a penalty applied to the genetic variants with heterogeneous associations. Additionally, in some embodiments the regression mayAttorney Docket No.: M0437.70168WO00 include covariate effects, representing effects of conventional risk factors that are not genotypic in origin.

[0159] In some embodiments, the regression may proceed in two steps. In a first portion, the regression may be used to fit covariate effects in the training data set. The covariate effects may take into consideration conventional risk factors (e.g., age, sex, weight, or other non-genotypic factors). In some embodiments, the covariate-only model may be determined by fitting an unpenalized regression using individual-level data from the single-ancestry population of interest to characterize the covariate effects of: age, sex, age2, age*sex, the Townsend deprivation index, and / or the genotype principal components, as non-limiting examples. The covariate effects may be fit using an unpenalized regression of the form:where ^^ ^\]^,^^ !"represents the number of individuals in the training data. In some embodiments, ^^ ^\]^,^^ !"may be the same as ^^^ !". In some embodiments, ^^ ^\]^,^^ !"may be different than ^^^ !". For example, in act 306, a subset of the training data may be used or different data may be used to consider genetic variants having heterogeneous effects.

[0160] Thereafter, the act 306 may proceed to further train the computational model, using an estimate obtained from the above regression performed using the covariate effects. In someembodiments, the regression may take the form of an Elastic Net penalized regression:where ℒ(^!, ^^) is a loss for the Fthindividual and DE(^C) is the Elastic Net regularization term forthe coefficient ^, as described above, the subscript m indicates covariate effects, and the subscriptn indicates genetic effects, and the summation for genetic variants W ∈ o is performed over thegenetic variants with heterogeneous effects. The term S / ^+ ^_TU^$!term represents the covariate effects and the term ^`hb^)!represents the predicted phenotype value from the non-covariateAttorney Docket No.: M0437.70168WO00 terms in the iPGS model, described in relation to act 302 herein. The final term represents scores determined for genetic variants with heterogeneous associations between single-ancestry populations. In some embodiments, suitable penalty factors may be assigned for each term in the above expression. For example, a penalty factor of 1.1 may be assigned to the effects due to the genetic variants and a penalty factor of 1.0 may be assigned to the covariate-only and iPGS scores. Additional aspects are described in Example 1 herein.

[0161] Alternatively, in some embodiments, the regression of act 306 may optionally include taking into consideration effects of the individual’s global ancestry on the genetic variants with heterogeneous associations. In some embodiments, including the effects of the global ancestry of the individual comprises including, for each instance of genomic data in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships, an interaction described by a number of copies of each genetic variant with heterogeneous associations in each instance of genomic data and principal components of a genomic matrix generated using each instance of genomic data. These effects may be considered as a measure of how “close” an individual is to a reference single-ancestry population and may be characterized using principal components obtained from genotype matrices generated based on the genomic data associated with each individual included in the training data set and / or in a subset of the training data set. As anexample, the regression performed after fitting the covariate effects may be performed using:where Dm(w)! is the |^}principal component for each individual included in the ^^ ^\]^,^^ !"-size training set and ~ is an integer number of principal components (e.g., one, two, three, five, 10, or any other suitable number). The principal components,, are weighted based on a number ofAttorney Docket No.: M0437.70168WO00 copies of each genetic variant with heterogeneous associations in each instance of genomic data and principal components of a genomic matrix generated using each instance of genomic data. In some embodiments, an additional penalty factor may be assigned to the interaction terms. For example, a penalty factor of 1.2 may be assigned to the interaction terms.

[0162] In some embodiments, alternatively, the computational model may be obtained according to the embodiment of FIG. 1-4. FIG. 1-4 is a flowchart of a process 400 for generating a computational model for determining a trait score and / or an iPGS. Process 400 may be implemented by a computational facility, such as the computational facility 122 of FIG. 1-1. As such, in some embodiments, the process 400 may be performed by a computing device configured to receive and / or obtain information from a genotyping system (e.g., genotyping system 110 described in connection with FIG. 1-1). As another example, in some embodiments, the process 400 may be performed by one or more processors located remotely (e.g., as part of a cloud computing environment, as connected through a network) from the genotyping system that obtained the input genomic data.

[0163] Process 400 is similar to process 300 as described in FIG.1-3 but includes an additional act 403. In act 403, the computational facility may determine if the medical trait is or is not described by an ancestry-dependent genetic effect. The outcome of this determination may guide the form of the computational model that is used in act 204 to calculate the at least one polygenic score.

[0164] In some embodiments, and as in the embodiment of FIG. 1-3, process 400 may begin at act 402, in which the computational facility may perform supervised learning to train the computational model using data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries. Thereafter, process 400 may proceed to act 403, in which it is determined whether the medical trait is or is not described by an ancestry-dependent genetic effect. This determination may be made using, for example, a heterogeneity test (e.g., Cochran’s Q test) applied to meta-analysis of GWASes of single-ancestry populations.

[0165] In some embodiments in which the computational facility determines in act 403 that the medical trait is not described by an ancestry-dependent genetic effect, process 400 may terminate after act 402. In this manner, computational models may be built efficiently for medical traits that are linked to less heterogeneous genetic effects.Attorney Docket No.: M0437.70168WO00

[0166] In some embodiments, after act 403, and if it is determined that the medical trait is described by an ancestry-dependent genetic effect, process 400 may proceed to act 404. Act 404 may include two sub-acts, 404a and 404b. In sub-act 404a, the computational facility may determine genetic variants with heterogeneous associations between single-ancestry populations. In some embodiments, genetic variants with heterogeneous associations between single-ancestry populations may be determined as a result of the determination performed in act 403.

[0167] After sub-act 404a, process 400 may proceed to sub-act 404b, in which the computational facility may perform additional regression, as described in connection with act 306 of FIG. 1-3. The additional regression may include a penalty applied to the determined genetic variants with heterogeneous associations between single-ancestry populations. In some embodiments, the second regression may optionally include taking into consideration effects of the individual’s global ancestry on the genetic variants with heterogeneous associations, also as described in connection with act 306 of FIG. 1-3. After sub-act 404b, process 400 may terminate, and process 200 may proceed to act 206.

[0168] Returning to FIG.1-2, after determining at least one trait score in act 204, process 200 may proceed to act 206, in which the computational facility may generate, using the at least one processor, a graphical user interface in some embodiments. The graphical interface may include a visualization of the calculated at least one trait score and / or iPGS, the calculated at least one trait score being indicative of a prediction of the medical trait in the individual. For example, the graphical interface may include one or more graphs, diagrams, charts, or other graphical illustration of the determined trait scores and / or one or more tables or textual representations of the determined trait scores. In some embodiments, the graphical user interface may be integrated into and / or be an electronic medical record (EMR).

[0169] After act 206, process 200 may proceed to act 208, in which a display device may display the generated graphical user interface. For example, the graphical user interface may be displayed on one or more of a computer monitor, television screen, smartphone display, mobile device display, and / or a tablet display. In this manner, a clinician and / or the individual may review the determined trait scores.

[0170] After act 208, process 200 may optionally proceed to act 210, in which a course of medical treatment related to the medical trait may be altered based on the calculated at least one trait score. For example, the number and / or frequency of medical interventions (e.g., medical tests,Attorney Docket No.: M0437.70168WO00 medical imaging, at-home monitoring, surgeries, vaccinations, and / or other medical treatments) may be changed based on the at least one trait score. In some embodiments, the number and / or frequency of medical interventions in the course of medical treatment may be reduced based on the value of the at least one trait score. If the trait score value is low (e.g., below a threshold value), the individual may be considered “low-risk” such that additional medical interventions may be unnecessary as being invasive and / or cost inefficient. In some embodiments, the number and / or frequency of medical interventions in the course of medical treatment may be increased based on the value of the at least one trait score. If the trait score value is high (e.g., above a threshold value), the individual may be considered “high-risk” such that additional medical interventions may be viewed as necessary to detect a potential disease at an earlier stage and / or to prevent the disease from occurring.

[0171] After, or in place of, act 210, process 200 may optionally proceed to act 212, in which at least one therapeutic agent may be identified and / or administered to the individual based on the calculated at least one trait score. For example, the at least one trait score may indicate a likelihood of a health risk associated with a particular cancer carrying a specific genetic variant. A treatment designed to combat the specific genetic variant of the cancer may then be identified and / or administered to the individual by a medical provider.

[0172] In some embodiments, alternatively or additionally, the at least one trait score may be used to alter a recruitment strategy for a clinical trial related to the medical trait. For example, the individual may be added to or removed from the clinical trial based on the value of the trait score. The trait score may be compared to a threshold value in order to determine whether the individual should be added to or removed from the clinical trial. For example, the trait score may be greater than a threshold value, indicating a likelihood that the individual exhibits or may develop the medical trait such that the individual will be enrolled in the clinical trial. Alternatively, depending on the medical trait being evaluated, the trait score may be less than a threshold value, indicating a likelihood that the individual exhibits or may develop the medical trait such that the individual will be enrolled in the clinical trial.

[0173] FIG. 1-5 is a flowchart of a process 500 for performing a clinical trial, in accordance with some embodiments of the technology described herein. Process 500 may begin at act 502, in which a first trait score associated with a medical trait of a first individual is obtained. The first trait score may be obtained by calculating the first trait score using a trained computational modelAttorney Docket No.: M0437.70168WO00 and first genomic data including information indicative of genetic variants present in a genome of the first individual. The first trait score may be calculated using a computational facility, such as the computational facility 122 of FIG. 1-1. As such, in some embodiments, the process 200 may be performed by a computing device configured to receive and / or obtain information from a genotyping system (e.g., genotyping system 110 described in connection with FIG. 1-1). As another example, in some embodiments, the process 200 may be performed by one or more processors located remotely (e.g., as part of a cloud computing environment, as connected through a network) from the genotyping system that obtained the input genomic data.

[0174] In some embodiments, act 502 may begin at sub-act 502a, in which the trained computational model may be obtained by training a first computational model using data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries. The trained computational model may be obtained, for example, using any techniques described in connection with FIGs. 1-2 through 1-4 herein (e.g., in connection with act 302 of FIG. 1-3, as one non-limiting example).

[0175] After sub-act 502a, process 500 may optionally proceed to sub-act 502b, in which the computational facility may perform additional regression, as described in connection with act 306 of FIG. 1-3. The additional regression may include a penalty applied to genetic variants with heterogeneous associations between single-ancestry populations. In some embodiments, the second regression may optionally include taking into consideration effects of the individual’s global ancestry on the genetic variants with heterogeneous associations, also as described in connection with act 306 of FIG. 1-3. After sub-act 502b, act 502 may terminate, and process 500 may proceed to act 504.

[0176] In some embodiments, after act 502, process 500 may proceed to act 504, in which the first individual is enrolled in the clinical trial based on a value of the first trait score. The trait score may be compared to a threshold value in order to determine whether the individual should be added to the clinical trial. For example, the trait score may be greater than a threshold value, indicating a likelihood that the individual exhibits or may develop the medical trait such that the individual will be enrolled in the clinical trial. Alternatively, depending on the medical trait being evaluated, the trait score may be less than a threshold value, indicating a likelihood that the individual exhibits or may develop the medical trait such that the individual will be enrolled in the clinical trial.Attorney Docket No.: M0437.70168WO00

[0177] In some embodiments, after act 504, process 500 may proceed to act 506, in which a course of medical treatment for the first individual is altered in accordance with the clinical trial. For example, the number and / or frequency of medical interventions (e.g., medical tests, medical imaging, at-home monitoring, surgeries, vaccinations, and / or other medical treatments) may be changed based on the at least one trait score. In some embodiments, the number and / or frequency of medical interventions in the course of medical treatment may be reduced based on the value of the at least one trait score.

[0178] Alternatively or additionally, a therapeutic agent associated with the clinical trial may be administered to the individual based on the calculated at least one trait score. For example, the at least one trait score may indicate a likelihood of a health risk associated with a particular cancer carrying a specific genetic variant. A treatment designed to combat the specific genetic variant of the cancer may then be identified and / or administered to the individual by a medical provider associated with the clinical trial.

[0179] In some embodiments, the process may further include assessing a second trait score determined for a second individual. The second trait score may be calculated using the trained computational model and second genomic data including information indicative of genetic variants present in a genome of the second individual.

[0180] In some embodiments, the process may include declining to enroll the second individual in the clinical trial based on a value of the second trait score. For example, the second trait score may be compared to a threshold value in order to determine whether the individual should be added to the clinical trial. For example, the trait score may be greater than a threshold value, indicating that the second individual is not likely to exhibit or to develop the medical trait such that the individual will not be enrolled in the clinical trial. Alternatively, depending on the medical trait being evaluated, the trait score may be less than a threshold value, indicating that the second individual is not likely to exhibit or to develop the medical trait such that the individual will not be enrolled in the clinical trial.

[0181] Having thus described several aspects and embodiments of the technology set forth in the disclosure, it is to be appreciated that various alterations, modifications, and improvements will readily occur to those skilled in the art. Such alterations, modifications, and improvements are intended to be within the spirit and scope of the technology described herein. For example,Attorney Docket No.: M0437.70168WO00 those of ordinary skill in the art will readily envision a variety of other means and / or structures for performing the function and / or obtaining the data and / or one or more of the advantages described herein, and each of such variations and / or modifications is deemed to be within the scope of the embodiments described herein. Those skilled in the art will recognize or be able to ascertain using no more than routine experimentation many equivalents to the specific embodiments described herein. It is, therefore, to be understood that the foregoing embodiments are presented by way of example only and that, within the scope of the appended claims and equivalents thereto, inventive embodiments may be practiced otherwise than as specifically described. In addition, any combination of two or more features, systems, articles, materials, kits, and / or methods described herein, if such features, systems, articles, materials, kits, and / or methods are not mutually inconsistent, is included within the scope of the present disclosure.

[0182] The terms “program” or “software” are used herein in a generic sense to refer to any type of computer code or set of computer-executable instructions that can be employed to program a computer or other processor to implement various aspects as described above. Additionally, it should be appreciated that according to one aspect, one or more computer programs that when executed perform methods of the present disclosure need not reside on a single computer or processor but may be distributed in a modular fashion among a number of different computers or processors to implement various aspects of the present disclosure.

[0183] Techniques operating according to the principles described herein may be implemented in any suitable manner. Included in the discussion above are a series of flow charts showing the steps and acts of various processes to determine inclusive polygenic risk scores and to use the determined iPGS values to alter an individual’s course of medical treatment. The processing and decision blocks of the flow charts above represent steps and acts that may be included in algorithms that carry out these various processes. Algorithms derived from these processes may be implemented as software integrated with and directing the operation of one or more single- or multi-purpose processors, may be implemented as functionally equivalent circuits such as a Digital Signal Processing (DSP) circuit or an Application-Specific Integrated Circuit (ASIC), or may be implemented in any other suitable manner. It should be appreciated that the flow charts included herein do not depict the syntax or operation of any particular circuit or of any particular programming language or type of programming language. Rather, the flow charts illustrate the functional information one skilled in the art may use to fabricate circuits or toAttorney Docket No.: M0437.70168WO00 implement computer software algorithms to perform the processing of a particular apparatus carrying out the types of techniques described herein. It should also be appreciated that, unless otherwise indicated herein, the particular sequence of steps and / or acts described in each flow chart is merely illustrative of the algorithms that may be implemented and can be varied in implementations and embodiments of the principles described herein.

[0184] Accordingly, in some embodiments, the techniques described herein may be embodied in computer-executable instructions implemented as software, including as application software, system software, firmware, middleware, embedded code, or any other suitable type of computer code. Such computer-executable instructions may be written using any of a number of suitable programming languages and / or programming or scripting tools, and also may be compiled as executable machine language code or intermediate code that is executed on a framework or virtual machine.

[0185] When techniques described herein are embodied as computer-executable instructions, these computer-executable instructions may be implemented in any suitable manner, including as a number of functional facilities, each providing one or more operations to complete execution of algorithms operating according to these techniques. A “functional facility,” however instantiated, is a structural component of a computer system that, when integrated with and executed by one or more computers, causes the one or more computers to perform a specific operational role. A functional facility may be a portion of or an entire software element. For example, a functional facility may be implemented as a function of a process, or as a discrete process, or as any other suitable unit of processing. If techniques described herein are implemented as multiple functional facilities, each functional facility may be implemented in its own way; all need not be implemented the same way. Additionally, these functional facilities may be executed in parallel and / or serially, as appropriate, and may pass information between one another using a shared memory on the computer(s) on which they are executing, using a message passing protocol, or in any other suitable way.

[0186] Generally, functional facilities include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Typically, the functionality of the functional facilities may be combined or distributed as desired in the systems in which they operate. In some implementations, one or more functional facilities carrying out techniques herein may together form a complete software package. These functional facilitiesAttorney Docket No.: M0437.70168WO00 may, in alternative embodiments, be adapted to interact with other, unrelated functional facilities and / or processes, to implement a software program application.

[0187] The above-described embodiments can be implemented in any of numerous ways. One or more aspects and embodiments of the present disclosure involving the performance of processes or methods may utilize program instructions executable by a device (e.g., a computer, a processor, or other device) to perform, or control performance of, the processes or methods. In this respect, various inventive concepts may be embodied as a computer readable storage medium (or multiple computer readable storage media) (e.g., a computer memory, one or more floppy discs, compact discs, optical discs, magnetic tapes, flash memories, circuit configurations in Field Programmable Gate Arrays or other semiconductor devices, or other tangible computer storage medium) encoded with one or more programs that, when executed on one or more computers or other processors, perform methods that implement one or more of the various embodiments described above. The computer readable medium or media can be transportable, such that the program or programs stored thereon can be loaded onto one or more different computers or other processors to implement various ones of the aspects described above. In some embodiments, computer readable media may be tangible (e.g., non-transitory) computer readable media. In some embodiments, the computer readable media may comprise a persistent memory.

[0188] Computer-executable instructions implementing the techniques described herein (when implemented as one or more functional facilities or in any other manner) may, in some embodiments, be encoded on one or more computer-readable media to provide functionality to the media. Such a computer-readable medium may be implemented in any suitable manner, including as computer-readable storage media 606 of FIG. 1-6 described below (i.e., as a portion of a computing device 800) or as a stand-alone, separate storage medium.

[0189] In some, but not all, implementations in which the techniques may be embodied as computer-executable instructions, these instructions may be executed on one or more suitable computing device(s) operating in any suitable computer system, including the exemplary computer system of FIG. 1-6, or one or more computing devices (or one or more processors of one or more computing devices) may be programmed to execute the computer-executable instructions. A computing device or processor may be programmed to execute instructions when the instructions are stored in a manner accessible to the computing device or processor, such as in a data store (e.g., an on-chip cache or instruction register, a computer-readable storage mediumAttorney Docket No.: M0437.70168WO00 accessible via a bus, a computer-readable storage medium accessible via one or more networks and accessible by the device / processor, etc.). Functional facilities comprising these computer- executable instructions may be integrated with and direct the operation of a single multi-purpose programmable digital computing device, a coordinated system of two or more multi-purpose computing device sharing processing power and jointly carrying out the techniques described herein, a single computing device or coordinated system of computing devices (co-located or geographically distributed) dedicated to executing the techniques described herein, one or more Field-Programmable Gate Arrays (FPGAs) for carrying out the techniques described herein, or any other suitable system.

[0190] FIG. 1-6 illustrates one exemplary implementation of a computing device in the form of a computing device 600 that may be used in a system implementing techniques described herein, although others are possible. It should be appreciated that FIG. 1-6 is intended neither to be a depiction of necessary components for a computing device to operate as an iPGS system in accordance with the principles described herein, nor a comprehensive depiction.

[0191] Computing device 600 may comprise at least one processor 602, a network adapter 604, and computer-readable storage media 606. Computing device 600 may be, for example, a desktop or laptop personal computer, a personal digital assistant (PDA), a smart mobile phone, a server, a wireless access point or other networking element, or any other suitable computing device. Network adapter 604 may be any suitable hardware and / or software to enable the computing device 600 to communicate wired and / or wirelessly with any other suitable computing device over any suitable computing network. The computing network may include wireless access points, switches, routers, gateways, and / or other networking equipment as well as any suitable wired and / or wireless communication medium or media for exchanging data between two or more computers, including the Internet. Computer-readable media 606 may be adapted to store data to be processed and / or instructions to be executed by processor 602. Processor 602 enables processing of data and execution of instructions. The data and instructions may be stored on the computer-readable storage media 606.

[0192] The data and instructions stored on computer-readable storage media 606 may comprise computer-executable instructions implementing techniques which operate according to the principles described herein. In the example of FIG.1-6, computer-readable storage media 606Attorney Docket No.: M0437.70168WO00 stores computer-executable instructions implementing various facilities and storing various information as described above. Computer-readable storage media 606 may store a functional

[0193] While not illustrated in FIG. 1-6, a computing device may additionally have one or more components and peripherals, including input and output devices. These devices can be used, among other things, to present a user interface. Examples of output devices that can be used to provide a user interface include printers or display screens for visual presentation of output and speakers or other sound generating devices for audible presentation of output. Examples of input devices that can be used for a user interface include keyboards, and pointing devices, such as mice, touch pads, and digitizing tablets. As another example, a computing device may receive input information through speech recognition or in other audible format.

[0194] Embodiments have been described where the techniques are implemented in circuitry and / or computer-executable instructions. It should be appreciated that some embodiments may be in the form of a method, of which at least one example has been provided. The acts performed as part of the method may be ordered in any suitable way. Accordingly, embodiments may be constructed in which acts are performed in an order different than illustrated, which may include performing some acts simultaneously, even though shown as sequential acts in illustrative embodiments.

[0195] Computer-executable instructions may be in many forms, such as program modules, executed by one or more computers or other devices. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Typically, the functionality of the program modules may be combined or distributed as desired in various embodiments.

[0196] Also, data structures may be stored in computer-readable media in any suitable form. For simplicity of illustration, data structures may be shown to have fields that are related through location in the data structure. Such relationships may likewise be achieved by assigning storage for the fields with locations in a computer-readable medium that convey relationship between the fields. However, any suitable mechanism may be used to establish a relationship between information in fields of a data structure, including through the use of pointers, tags or other mechanisms that establish relationship between data elements.Attorney Docket No.: M0437.70168WO00

[0197] When implemented in software, the software code can be executed on any suitable processor or collection of processors, whether provided in a single computer or distributed among multiple computers.

[0198] Further, it should be appreciated that a computer may be embodied in any of a number of forms, such as a rack-mounted computer, a desktop computer, a laptop computer, or a tablet computer, as non-limiting examples. Additionally, a computer may be embedded in a device not generally regarded as a computer but with suitable processing capabilities, including a Personal Digital Assistant (PDA), a smartphone, or any other suitable portable or fixed electronic device.

[0199] Such computers may be interconnected by one or more networks in any suitable form, including a local area network or a wide area network, such as an enterprise network, and intelligent network (IN) or the Internet. Such networks may be based on any suitable technology and may operate according to any suitable protocol and may include wireless networks, wired networks or fiber optic networks. EXAMPLES Example 1. Enhancing polygenic prediction with admixed individuals

[0200] Polygenic scores (PGSs), used for aggregating genetic effects into the individual-level genetic liability of diseases or non-disease traits, have attracted significant research interest as a result of the recent expansion of genotyped cohort sample sizes, increased recognition of the polygenicity of complex traits, and recent methodological innovations in PGS training. For some traits, the predictive performance indicates the potential clinical relevance of PGS. However, most PGS models suffer from limited transferability across populations, despite the fact that some complex traits manifest substantial trans-ancestry genetic correlation. The limited transferability is partly due to the underrepresentation of non-European individuals in genetic studies and results in delaying the realization of equitable healthcare benefits from advancements in genetic research. Several efforts are underway to improve the transferability of PGS models. On the one hand, active recruitment of non-European individuals in genetic studies along with global partnerships and capacity building are greatly increasing, but most genome-wide association study (GWAS) cohorts have not realized the diversity that proportionally represents global populations. On the other hand, the development of computational methods can complement these efforts and provide immediate benefits to individuals of diverse ancestry groups. Existing efforts include performingAttorney Docket No.: M0437.70168WO00 PGS modeling by prioritizing variants present in diverse populations and cell-type-specific regulatory elements and combining multiple polygenic predictors characterized for multiple ancestry groups.

[0201] Admixed individuals, whose genomes consist of haplotypes from more than one ancestry group and account for one in seven newborns in the U.S., are often excluded in PGS model training given the technical limitations. Most modern PGS methods apply Bayesian multivariate regression by including GWAS summary statistics and ancestry-matched linkage disequilibrium (LD) reference panels. Although methods of applying GWAS analysis to admixed individuals exist, dependencies on the LD reference panels and computational complexities in representing LD for admixed individuals present challenges in the estimation of variant effect sizes in PGS modeling. However, including admixed individuals offers valuable insights into the genomic basis of common complex traits. A recent study indicates that the individual level PGS performance shows linear decay as a function of genomic distance defined as the Euclidean distance on the genotype PCA projection from the PGS training set, highlighting the importance of considering the continuum of genomic ancestry in PGS evaluation. Given the substantial trans- ancestry genetic correlation in some complex traits, one might expect that admixed individuals also offer unique opportunities to train PGS models with improved transferability.

[0202] Here, the technical limitations are overcome and inclusive PGS (iPGS), a PGS training strategy that considers individuals across the continuum of genetic ancestry, is presented. Applying penalized regression directly to individual-level data, iPGS captures genetic effects shared across population groups while avoiding the need for LD reference panels and is applicable to admixed individuals. The improved performance of iPGS is indicated in 33 simulation configurations and systematic application across 60 anthropometric and hematological traits in the UK Biobank. An iPGS+refit strategy is also developed to jointly model the ancestry-shared and ancestry-dependent effects and indicate its utility in improving prediction in a few hematological traits in the African population in the UK Biobank. The present disclosure highlights the benefits of inclusive PGS training in improving predictive performance and transferability. Material and methods Synthetic genotype and phenotype dataAttorney Docket No.: M0437.70168WO00

[0203] Synthetic genotype and phenotype data was prepared to investigate the behavior of iPGS. The recently released simulated genotypes from the INTERVENE consortium were used and their HAPNEST pipeline was used to generate synthetic quantitative phenotypes. The HAPNEST pipeline is capable of generating synthetic genotypes and phenotypes for hundreds of thousands of individuals across multiple continental ancestry groups, thanks to its computational efficiency. The synthetic genotype data preserves the key statistics, such as minor-allele frequency and LD in ancestry-matched reference panels and has lower relatedness with the reference panels. The pipeline allowed us to simulate phenotypes under the specified heritability and polygenicity. The HAPNEST synthetic dataset (BioStudies: S-BSST936; EMBL-EBI's BioStudies repository) was downloaded and synthetic genotype data on chromosome 22 for 168,000 individuals each in African and European ancestry groups was focused on. Each ancestry group was split into training (n = 110,000), validation (n = 20,000), and held-out test sets (n = 38,000) without using phenotypes. The training set was used to fit models, the validation set was used to determine the sparsity of the models, and the held-out test sets were used to evaluate the predictive performance of the models. The same training, validation, and test sets were used for all tested synthetic traits.

[0204] With the phenotype simulation pipeline in HAPNEST, synthetic phenotypes were generated under three polygenicity and heritability parameters: (1) a low polygenicity of 0.01% and equal heritability of 0.1(ℎ^AFR= 0.1 and ℎ^EUR= 0.1); (2) a higher polygenicity of 0.5% and equal heritability of 0.1 in synthetic African and synthetic European samples (ℎ^AFR= 0.1 and ℎE^UR= 0.1); and (3) the higher polygenicity of 0.5% and different heritability between two ancestry groups (ℎ^AFR= 0.03 and ℎE^UR= 0.1). The default value of the trans-ancestry genetic correlation of 1.0 was used, the synthetic phenotypes were assumed to have no covariate effects, and the model was restricted to sample causal variants from chromosome 22 alone. The study population in the UK Biobank

[0205] The UK Biobank is a population-based cohort study with genomic and phenotypic datasets across about 500,000 volunteers collected across multiple sites in the United Kingdom. Sample-level quality control (QC) was performed. P = 406,659 unrelated individuals with genetic data were focused on, based on the following criteria: (1) used to compute principal components (UK Biobank data field 22020); (2) removal of sex mismatch between the sex fieldAttorney Docket No.: M0437.70168WO00 in the genotype dataset and phenotype sex (data field 31); (3) not reported in "outliers for heterozygosity or missing rate" (data field 22027); (4) not reported in "sex chromosome aneuploidy" (data field 22019); and (5) do not have ten or more third-degree relatives (data field 22021).

[0206] A combination of self-reported ethnic background (data field 21000) and genetic principal components (data field 22009) was subsequently used to define population groups. In brief, self-reported European, self-reported African, and self-reported Asian individuals were first identified. Aberrant, a Bayesian-outlier-detection algorithm, was applied to the first six genotype PCs to detect to detect outliers to refine population-group assignment. Unrelated individuals were subsequently focused on. The self-reported ethnic background was used to subdivide European individuals into White British and non-British White. Four population groups were defined as follows: White British (WB), non-British White (NBW), African (Afr), and South Asian (SA). The remaining unrelated individuals were kept as "others."

[0207] Each population group was randomly split into training (70%), validation (10%), and test (20%) sets without using phenotypes. As in the case of the synthetic data, the training, validation, and test sets were used for model fitting, determination of the sparsity hyperparameter, and predictive performance evaluation, respectively. The same training, validation, and test sets were used for all tested traits in the UK Biobank. For iPGS training, four different subsets of the training set were considered (Tables 1 and 2). For MultiPop, NoAdmixed, and InclusiveFixN PGS models, White British individuals in the training set were randomly sub-sampled so that the total number of individuals used in the iPGS training would match that of the WB-only model. A similar procedure was applied to define four subsets of the validation-set individuals for PGS training. Variant annotation and quality control in the UK Biobank

[0208] The directly genotyped dataset (release version 2), imputed genotypes (release version 3), imputed HLA allelotype (release version 2), and the GRCh37 human reference genome were used throughout the study. Variant annotation was performed with Ensembl's Variant Effect Predictor (VEP) (version 101) with the LOFTEE plugin. Using ClinVar, "pathogenic" and "likely pathogenic" variants were annotated. The VEP-predicted consequence of the variants were grouped into six groups: protein-truncating variants (PTVs), proteinaltering variants (PAVs),Attorney Docket No.: M0437.70168WO00 proximal coding variants (PCVs), intronic variants (intronic), genetic variants on untranslated regions (UTRs), and other non-coding variants (others). For the directly genotyped dataset, variants passing the following criteria were focused on: (1) the missingness of the variant is less than 1%, considering that the two genotyping arrays (the UK BiLEVE Axiom array and the UK Biobank Axiom array) cover a slightly different set of variants and (2) Hardy-Weinberg disequilibrium test p value greater than 1.0 × 10^^. For the imputed-genotype dataset, the following criteria was used: (1) the missingness of the variant is less than 1%; (2) the minor-allele frequency (MAF) is greater than 0.01%; (3) the imputation quality score (INFO score) is greater than 0.3; (4) the variant does not present in the directly genotyped dataset; and (5) the variant is present in the HapMap phase 3 dataset. For the HLA allelotype, the imputed allelotype dosage was kept within [0,0.1), (0.9,1.1), or (1.9,2.0] and converted it to a hard call. The HLA allelotype was focused on with (1) missingness no more than 1% and (2) Hardy-Weinberg disequilibrium test p value greater than 1.0 × 10^^. All variants and allelotypes were concatenated into one dataset by using PLINK 2.0 (v. 2.00a3.3LM, 3 Jun 2022). The procedure outlined above resulted in a total of 1,316,181 variants considered in the analysis. Phenotype definition in the UK Biobank

[0209] 60 anthropometric and hematological traits in the UK Biobank were studied (Table 3). Some of those phenotypes are collected at up to four instances, each of which corresponds to (1) the initial assessment visit (2006-2010), (2) the first repeat assessment visit (2012-2013), (3) the imaging visit (2014-present), and (4) first repeat imaging visit (2019- present). Phenotype data was defined by using the median of non-missing values for each individual across the 60 quantitative traits as described elsewhere. Sparse-polygenic-score training from individual-level data

[0210] Sparse-polygenic-score models were fit by using batch screening iterative lasso (BASIL) implemented in the R snpnet package (version 2) on the individual-level data. The additive effects of genetic variants on the phenotypes and fit of a polygenic score model were considered by finding the exact solution for ^^-and ^^-penalized multivariate regression (Elastic Net). Specifically, BASIL directly operates on the individual-level data and performs variable selection and effect-size estimation simultaneously. Given a continuous phenotype ^ ∈ R"of P individuals, a covariate matrix # ∈ R"×^of P individuals and ' covariates, and a genotype matrixAttorney Docket No.: M0437.70168WO00 ( ∈ R"×^of P individuals and ^ variants, the following regularized regression problem is considered:(Equation 1-1) where 1 is a tuning parameter that controls the sparsity of the solution, L is an elastic net parameter that controls the balance between the ^^(Lasso) and ^^(Ridge) penalization, and ^^is the penalty factor for the I-th variable; the penalty factors allow different levels of shrinkage to variables. The tuning parameter, 1, was optimized on the basis of the predictive performance on the validation set and use L = 0.99. Penalty factors of B^= 1 were set for all genetic variants in the application with synthetic data. In contrast, lower penalty factors were assigned for non-synonymous coding variants in the application to the UK Biobank, as described below. Note that the covariate terms are unpenalized in the regression. When fitting a polygenic score model from large-scale cohorts, it is not uncommon to have a large number of individuals (n ≈ 300,000) and genetic variants (^ ≈ 1,000,000). Instead of loading all the large-scale data on memory and fitting a regularized regression, BASIL efficiently solves the exact solution of the penalized regression in an iterative procedure by taking advantage of strong rules that guide the variable selection in each iteration step. A similar model can be used for binary phenotypes (logistic regression), time-to-event phenotypes (Cox proportional hazards regression), or joint modeling of multiple phenotypes, as shown previously. Inclusive polygenic scores with synthetic data

[0211] In the application of iPGS to the synthetic genotype and phenotype data from the HAPNEST pipeline, the impact of the composition of the training-set individuals was assessed on the predictive performance by using the held-out synthetic individuals of African and European ancestry groups.

[0212] Eleven models were fit on ntrain= 110,000 individuals for each of the three synthetic phenotypes. The training set included individuals with synthetic African and synthetic Europeanancestry, each with different ratios. The ratios tested were 100%, 95%, 90%, 75%,60%, 50%, 40%, 25%, 10%, 5%, and 0%. The validation set of P = 40,000 individuals wasAttorney Docket No.: M0437.70168WO00 constructed, matching the ratio in the training set. iPGS was applied to ntrain= 110,000 individuals and ^ = 116,524 variants in chromosome 22. Given that the covariate effects are not modeled in phenotype simulation, covariate terms were not included in the regression. The defaultvalue of 1.0 was used for all variants for penalty factors. The same n = 76,000 individuals (n =38,000 each for African and European) were used in the held-out test set for predictiveperformance evaluation. Inclusive polygenic scores in the UK Biobank

[0213] In the application of iPGS to the UK Biobank, age (UK Biobank data field 34), sex (data field 31), age^, age*sex, Townsend deprivation index (data field 22189), and the first 18 genotype PCs (data field 22009) provided by the UK Biobank were included as unpenalizedcovariates. ^ = 1,316,181 variants were considered for n = 237,055 individuals or n =284,661 individuals, as shown in Tables 1 and 2. The validation set metric was used to select thesparsity of the model. The protein-truncating and protein-altering variants were prioritized as previously described. In brief, a penalty factor was assigned of 0.5 to putative protein-truncating variants and pathogenic variants; 0.75 to putative protein-altering variants, likely pathogenic variants, and HLA allelotypes; 1.2 for genetic variants that are not present in the HapMap phase 3 dataset; and 1.0 for the other remaining variants. The specific values of penalty factors are based on heuristics. Inclusive polygenic score with population-specific refit

[0214] To model the ancestry-dependent genetic effects on top of the ancestry-shared effects captured in iPGS, the iPGS+refit procedure was developed. The individuals in the training and validation sets and also of African ancestry were focused on and unpenalized regression was fit by using the individual-level data to characterize the covariate effects: phenotype ∼ age + sex + age^+ age*sex + Townsend deprivation index + genotype PCs, where genotype PCs represent the first 18 genotype PCs as in the iPGS training. The covariate-only score term was obtained by predicting the phenotype values using the covariate terms alone. For the iPGS+refit model in the training set without interaction effects, the individuals in the training and validation sets and also of African ancestry were focused on and the following elastic-net penalized regression model was fit:Attorney Docket No.: M0437.70168WO00 phenotype ∼ covariate-only score(Equation 1-2) where, the covariate-only score and iPGS represent the predicted phenotype value from the covariate-only model and iPGS model (genotype-only model), respectively, and n represents the set of genetic variants with heterogeneous associations. To nominate the set of genetic variants with heterogeneous associations (n), the heterogeneity test was used in GWAS meta-analysis by using nominal p value = 5 × 10^^as the statistical-significance threshold. The missing values in the genetic variants were imputed with heterogeneous associations in the individual-level data by using the allele frequency computed in the African population in the UK Biobank. A penalty factor of 1.1 was assigned for the genetic variants and 1.0 for the covariate-only score and iPGS.

[0215] For the iPGS+refit model with interaction effects, variant * PC1 and variant * PC2 terms were also considered for genetic variants with heterogeneous associations in the penalized regression model: phenotype ∼ covariate-only score + iPGS(Equation 1-3)

[0216] A penalty factor of 1.2 was assigned for the interaction terms, 1.1 for the genetic variants, and 1.0 for both covariate-only and iPGS scores. For both models, the elastic-net penalized regression was fit by setting elastic-net parameter L to be 0.99 and optimized the tuning parameter by using 10 -fold cross-validation with the cv.glmnet function implemented in the glmnet package in R. Genome-wide association analysis

[0217] Genome-wide association analysis was applied with PLINK (v. 2.00 alpha). Population-specific genotype PCs were first computed for White British, non-British White, South Asian, and African individuals in the UK Biobank by using the randomized algorithm ("approx" modifier) implemented as the "--pca allele-wts 20 approx vzs" command in PLINK2. The GWAS analysis was subsequently applied by using age, sex, Townsend deprivation index, array, and the top ten population-specific genotype PC loadings as covariates and using the approximationAttorney Docket No.: M0437.70168WO00 algorithm ("cc-residualize" modifier) implemented as the "--glm zs omit-ref no-x-sex log 10 hide- covar skip-invalid-pheno cc-residualize firth-fallback" command in PLINK2. The participants from the UK Biobank were genotyped on two different arrays: about 10% of participants were genotyped on the UK BiLEVE Axiom array, and the rest were genotyped on the UK Biobank Axiom array. When genetic variants were directly measured on both arrays, an indicator variable "array" was included in the covariates and whether the UK Biobank Axiom array or UK BiLEVE Axiom array was used in the genotyping was noted.

[0218] For the GWAS meta-analysis and heterogeneity test, quantile normalization was applied by using the "--pheno-quantilenormalize" option in PLINK2. GWAS analysis was conducted by using the individuals in the training set for the following populations: White British, non-British White, South Asian, and African.

[0219] For PGS modeling with PRS-CSx, the same GWAS summary statistics were used for the meta-analysis. GWAS analysis was also conducted by using the union of the training and validation-set individuals in each of the four populations. Those two sets of GWAS summarystatistics were used as the input of the PRS-CSx model (ntrain = 293,301) and the "PRS-CSx(n = 256k)" model (ntrain = 256,637)(Tables 1 and 2).

[0220] For heritability estimation analysis, GWAS analysis was applied by using all the individuals in the White British group in the UK Biobank without applying the quantile normalization. GWAS meta-analysis and heterogeneity test

[0221] Using the GWAS summary statistics for four analyzed populations (White British, non-British White, South Asian, and African), inverse-variance weighted (IVW) meta-analysis was performed by using METAL (version 2020-05-05) and included a heterogeneity-of-effects analysis. Heritability analysis

[0222] Linkage disequilibrium (LD) score regression (LDSC) was applied, and the SNP-based heritability was estimated. The predictive performance of the PRS models and the LDSC-based heritability estimates was compared. Allele frequency and LD pruningAttorney Docket No.: M0437.70168WO00

[0223] The non-reference allele frequency was computed with PLINK2 by using the individuals in the training set and in the following population groups in the UK Biobank: White British, non-British White, South Asian, and African. The cumulative frequency of the minor- allele frequency distribution was computed for all the 1,316,181 genetic variants considered. The analysis was also repeated by focusing on the subset of variants selected in at least one of the PGS models across 60 anthropometric and hematological traits.

[0224] LD pruning was applied with widow size 200 kb and pairwise threshold '^of 0.5 by using the "--indep-pairwise 200 kb 0.5 " command implemented in PLINK2. Protein-truncation, protein-altering, or proximal-coding variants were prioritized by using the "--indep-preferred" command. The procedure was repeated for White British, non-British White, South Asian, and African individuals in UK Biobank. The selected variants were used as approximately LD- independent variants. PGS training with PRS-CSx

[0225] PGS models were fit by using PRS-CSx, a cross-population polygenic prediction method based on Bayesian multivariate regression using continuous shrinkage priors. The precomputed LD reference panels constructed from the UK Biobank data were downloaded from GitHub (github.com / getian107 / PRScsx) and were used for the analysis. Specifically, the European (EUR) reference was used for White British and non-British White populations, the South Asian (SAS) reference was used for the South Asian population, and the African (AFR) reference was used for the African population in the UK Biobank. The Bayesian regression model implemented in PRScsx.py was fit for each chromosome independently. Following the tutorial and recommendations in the GitHub repository, a small-scale grid search was applied for theglobal shrinkage parameter, phi, by fitting four models corresponding to the following phi values:1 × 10^ , 1 × 10^^, 1 × 10^^, and 1. The default values were used for the other parameters andposterior SNP effect-size estimates were obtained for each discovery super-population (i.e., EUR, SAS, and AFR). The optimal linear combination of the three scores was subsequently learned. Specifically, the individuals in the validation set were used, the super-population-specific scores for each individual were computed by using the "--score" command implemented in PLINK2, scaling was applied so that the super-population-specific scores have zero mean and unit variance, and the coefficients of linear combinations of the population-specific scores were learned according to the recommendations provided in the GitHub repository. For each target populationAttorney Docket No.: M0437.70168WO00 (i.e., the White British, non-British White, South Asian, African, and Others groups in the UK Biobank), the global shrinkage parameter was selected on the basis of the predictive performance evaluated in the individuals in the validation set. Two versions of the PRS-CSx models were fit, corresponding to the individuals in the training set (the "PRS-CSx [n = 256k] " model, ntrain=256,637) and the individuals in the union of training and validation set (the PRS-CSx model,ntrain = 293,301), where the optimal phi values and the weights for linear combination ofpopulation-specific scores learned from the "PRS-CSx (n = 256k)" model were used. The similarity of PGS models

[0226] To assess the similarity of PGS models trained for the same phenotype, the Pearson's correlation between the pair of PGS values was computed across the individuals in the held-out test set. PGS performance evaluation

[0227] The held-out test set was used to evaluate the predictive performance (&^) of (1) PGS (genotype-only) models, (2) covariate-only models, and (3) full models that considered both covariates and genotypes (Tables 4 and 5). The 95% confidence interval of predictive performance was evaluated by using the approximate standard error of &^. In the application of iPGS to the UK Biobank, the predictive performance computed for the WB- only model was used as the baseline to evaluate the significance of improvements in predictive performance(&^)in the held-out-test-set individuals. The significance of the difference in &^between the iPGS model and the WB-only model was assessed by using the delta method implemented in the r2redux package in R. The average improvements in PGS models across traits is reported. For each White British, non-British White, South Asian, African, and "others"population in the held-out test set individuals, generalized Deming regression models were fit,&^ ∼ 0 + &^wb-only, accounting for the uncertainties in both predictor and response variables (i.e.,standard errors for &^wB-onlyand &^, respectively) by using the Deming package in R. The slope of the regression model and its 95% confidence interval is reported as the average improvements in the predictive performance.Attorney Docket No.: M0437.70168WO00 Statistics

[0228] For computational and statistical analysis, Jupyter Notebook, R, the R Tidyverse package, and GNU parallel was used. For visualization, ggplot2 with ggrepel and ggrastr packages. The ^ values were computed from two-sided tests unless otherwise specified. Results Overview of inclusive polygenic score (iPGS) methodology

[0229] In iPGS, PGS models were characterized by applying the batch screening iterative lasso (BASIL) algorithm to ancestry-diverse individuals. Unlike most modern PGS methods, which take GWAS summary statistics and LD reference panels as input, BASIL directly operates on the individual-level data and fits a PGS model as the exact solution of penalized multivariate regression via the iterative procedure. Ancestry-shared genetic effects were characterized from large-scale individual-level data of more than one million genetic variants across hundreds of thousands of ancestry-diverse individuals by taking advantage of efficient variable screening rules in BASIL. One may provide genotype principal-component loadings as unpenalized covariates to account for genome-wide admixture fractions and population structure. The individuals were randomly split into training, validation, and held-out test sets and a PGS model was fit on the training-set individuals. The validation set was used to select the sparsity of the penalized regression model and the held-out test set for performance evaluation. Application to synthetic data

[0230] The approach was first tested with a synthetic individual level dataset generated by HAPNEST. Simulated genotypes were used in chromosome 22 and three synthetic phenotypes were created with different polygenicity and heritability for 168,000 individuals each in African and European ancestry groups (material and methods). Each of the ancestry groups was split into a training set (ntrain= 110,000), a validation set (n = 20,000), and a held-out test set (n = 38,000) for evaluation. To systematically assess the impact of the composition of individuals in PGS training on predictive performance, 11 training sets with varying numbers of synthetic African and synthetic European individuals (ntrain= 110,000) were constructed, iPGS models were fit, and their predictive performance was evaluated for each of the three synthetic traits (FIG. 2-1).Attorney Docket No.: M0437.70168WO00

[0231] The lower polygenicity scenario showed the highest predictive performance in both synthetic African and synthetic European individuals, as expected, given that the accuracy of PGS depends on the polygenicity, heritability, and sample size. With the equal heritability between the two ancestry groups, it was found that the difference between synthetic African and synthetic European groups in the test-set predictive performance is the smallest when the iPGS models were trained on 75% of synthetic African and and 25% of synthetic European individuals. The fact that the transferability of PGS models from synthetic European individuals to synthetic African individuals was lower than in the opposite direction possibly reflects that African genomes have greater genetic diversity (e.g., lower degree of LD) than European genomes.

[0232] An increase in the transferability was found with ancestry-diverse training in synthetic African and synthetic European groups when they were the minority samples in the training set. These results indicate that the inclusion of the minority target population, even 5%, in PGS training would help improve the transferability of PGS models. When the heritability was different between the two ancestry groups, the predictive performance for both populations increased as the samples from high heritability was increased. Those results validate the utility of ancestry-diverse training to enhance the transferability of PGS models and motivated the application of iPGS to the UK Bio-bank, one of the largest cohorts with readily available individual-level data. Application to 60 traits in the UK Biobank

[0233] In the application of iPGS to the UK Biobank, unrelated individuals were assigned to four ancestry groups: White British (WB, n = 338,650), non-British White (NBW, n = 14,472), South Asian (SA, n = 7437), and African (Afr, n = 6066) by using the combination of genotype principal-component (PC) loadings and self-reported ethnicity data (material and methods, FIG. 2-2A). The remaining individuals (others, P = 40,034) accounted for nearly 10% of unrelated individuals in the UK Biobank and consisted of many admixed individuals and a smaller number of East Asian individuals (Table 1). Each of the ancestry groups were randomly split into a training set (70%), a validation set (10%), and a held-out test set for evaluation (20%) (Table 1). To assess the effects of ancestry composition of the PGS training sets on predictive performance, four PGS models were fit by using different subsets of the training-set individuals: (1) White British population (WB-only [model i ], ntrain= 237,055), (2) all of the unrelated individuals in the training set (inclusive [model ii], ntrain= 284,661), (3) individuals across the continuum ofAttorney Docket No.: M0437.70168WO00 ancestry, where the number of individuals was kept the same as for the WB-only model (inclusiveFixN [model iii], ntrain= 237,055), and (4) individuals from White British, non-British White, South Asian, and African ancestry groups with population stratification (MultiPop, NoAdmixed [model iv], ntrain= 237,055) (Tables 1 and 2). PGS models were fit across 60 quantitative traits, consisting of anthropometric and hematological measures (material and methods, Table 3).

[0234] The resulting sparse-PGS model contained from 2,827 (inclusive-FixN model for mean corpuscular hemoglobin concentration) to 62,419 (inclusive model for standing height) genetic variants, with a median of 30,787 variants for the inclusive model (FIG. 2-5; Table 4). Given the nature of penalized multivariate regression, shrinkage of effect-size estimates was observed in inclusive PGS models compared GWASs (FIGs. 2-6A to 2-6D). The direction of the effects is more consistent with European ancestry groups than with non-European ancestry groups, as expected, given that most individuals used to train inclusive PGS models are of European ancestry. When the predicted scores in the individuals in the held-out test were compared set across the four models, the scores were largely consistent with substantial variability across ancestry groups. For instance, the median value of Pearson's correlation between the PGSs from the WB-only and the inclusive models was 0.94 for white British individuals and 0.84 for African individuals (FIG. 2-7; Table 5).

[0235] The systematic application of iPGS and evaluation of predictive performance (&^) across 60 quantitative traits indicates that the direct inclusion of diverse-ancestry individuals increases predictive performance in held-out-test set individuals of non-European ancestry (material and methods, FIG. 2-2B, FIGs. 2-3A to 2-3G, FIGs. 2-8A to 2-8B, and FIG. 2-9). Overall, iPGS showed the greatest improvements for individuals of African ancestry; there were an average of 48.9% (95% confidence interval [CI] = [20.2%, 77.7%]) improvements across 60 quantitative traits when the models were trained on the same number of individuals (inclusive- FixN, ntrain= 237,055, material and methods, FIG. 2-2B; Table 6). For some traits (e.g., neutrophil count), the inclusive-FixN PGS model showed substantial improvements(&^= 0.0577 for the genotype-only model, nominal p value = 1.3 × 10^^ ) in the held-out test-set individualsof African ancestry in comparison to the baseline model trained only on White British individuals(&^ = 0.0011, nominal p > 0.05, more than a 50 -fold increase in &^), even though both modelswere trained with the same number of individuals of n = 237,055 (FIG. 2-3F). The difference inAttorney Docket No.: M0437.70168WO00 &^values was statistically significant, with a nominal ^ value of 2.9 × 10^ (material and methods). Similarly, improvements in prediction for leukocyte (red blood cell) count was observed in African individuals(&^= 0.0465, nominal p = 1.2 × 10^^£vs. &^= 0.0044, nominal p = 0.024 for inclu-sive-FixN and WB-only models, respectively) (FIG. 2-3G).

[0236] With the inclusion of all individuals across the continuum of genomic ancestry (inclusive, ntrain= 284,661), it was observed that the inclusive model showed improved or equally competitive performance over the WB-only model across all of the five populations tested in the study (FIG.2-2B; Table 7). Overall, an average improvement was found of 60.8% (95%CI : [ 31.2%, 90.3%]) for individuals of African ancestry, 11.6% (95% CI: [8.9%, 14.4%]) for South Asian, 7.3% (95% CI: [5.0%, 9.5%]) for non-British White, 4.8% (95% CI: [3.7%, 5.9%]) for White British, and 17.8% (95% CI: [ 11.8%, 23.8%]) for the remaining individuals ("others") for the held-out test-set, indicating the power gain in inclusive PGS training (FIG. 2-2B and FIG. 2-10; Table 8). With a fixed number of training individuals (ntrain= 237,055), the inclusive-FixN PGS model (trained on 189,449 WB and 47,606 individuals of the other ancestry groups, instead of 237,055WB) led to a very modest drop in performance (2.5%, 95%CI : [ 2.1%, 2.9%]) in White British, as expected, given the smaller ancestry-matched sample size in the training set(FIG. 2-2B). However, when all the unrelated individuals in the training set (inclusive, ntrain =284,661) were used, there was no drop in the performance across all population groups.Heritability and allele-frequency analysis

[0237] Next, the relationship between heritability and predictive performance of the iPGS model was investigated by focusing on White British individuals because they had the largest sample size in the UK Biobank. Because the additive genetic effects are modeled in the iPGS model, the narrow-sense SNP heritability provides the theoretical upper bound of predictive performance. The heritability was estimated by using LD score regression and it was compared with the predictive performance of the iPGS model. It was found that the predictive performance of the iPGS models for hematological traits was closer to the heritability estimates than it was for the anthropometric traits (FIG. 2-11; Table 8). The observed difference most likely reflects the difference in the power across traits with ntrain= 284,661 individuals in the UK Biobank. The sample size required to achieve predictive performance at the estimated heritability depends on the genetic architecture of each trait, and anthropometric traits might require a larger sample size.Attorney Docket No.: M0437.70168WO00 Indeed, a saturated map of genetic associations for standing height has recently been reported for the European population through meta-analysis of GWAS results from 4 million European individuals.

[0238] The allele frequency of the genetic variants selected in the PGS model were examined across 60 traits and it was found that the PGS models capture common variants (FIG. 2-12). Across the 1,316,181 genetic variants analyzed in the study, some variants were of lower frequency in non-European individuals (FIG. 2-12 and FIG. 2-13). A similar difference between ancestry groups was observed when the analysis was restricted to ∼ 300,000 genetic variants selected for at least one of the 60 traits (FIG. 2-13; Tables 9A-9B). Although the difference between PGS models was much smaller than the difference between ancestry groups, the variants selected in iPGS models were more common in European than in African individuals (FIG. 2-12 and FIG. 2-13). Detecting ancestry-dependent genetic effects with heterogeneity tests

[0239] Although the results across 60 quantitative traits highlight the benefits of modeling genetic effects shared across ancestry groups, in some cases, ancestry-dependent or ancestry- specific genetic effects might result in reduced performance. Previous analyses have revealed a relatively small number of loci with substantial differences in allele frequency between ancestry groups, and some of these loci show associations with complex traits. To identify genetic variants with heterogeneous associations across ancestry groups in the UK Biobank, the heterogeneity test was systematically applied by using GWAS meta-analysis (material and methods, FIG. 2-3B; Table 10). Applying the heterogeneity test across 60 traits, a limited number of approximately LD-independent(&^< 0.5)heterogeneous associations (median of 0) were found with a few exceptions. Neutrophil counts showed the greatest level of heterogeneity; approximately 69 LD- independent loci had statistically significant heterogeneous GWAS associations, all located in chromosome 1 (FIG. 2-3B, FIG. 2-4A, FIGs. 2-14A to 2-14D, and FIG. 2-15). Most of those variants had population-specific effects in individuals of African ancestry (FIG. 2-4B). The neutrophil-count-lowering alleles with heterogeneous associations were of higher allele frequency in individuals of African ancestry in UK Biobank (FIG. 2-4C).

[0240] The lead GWAS-associated variant for neutrophil counts in African ancestry groups is a well-characterized upstream untranslated region (UTR) variant rs2814778 in ACKR1 (atypical chemokine receptor 1, also known as Duffy blood group gene [DARC] [MIM: 613665]) (FIG. 2-Attorney Docket No.: M0437.70168WO00 10), which encodes the subunit of the Duffy receptor and serves as the basis of the Duffy blood group system. The UTR variant rs2814778 disrupts binding sites of the GATA1 transcription factor and shuts down expression of the receptor in erythrocytes; thus, it is considered the nullallele. The null allele is under positive selection in the African population (allele frequency of83% in the African population and 0.3% in the non-Finish European population), given that theDuffy receptor works as the canonical entry point for the malaria parasite, Plasmodium vivax, and the null allele is protective against malaria infection. Beyond its roles in erythrocytes, the null allele is also known as the causal variant for neutrophil-count-lowering associations from admixture mapping studies. A recent GWAS meta-analysis of hematological traits supports the association between rs2814778 and neutrophil counts in individuals of African ancestry. When the variants selected in the PGS models were reviewed, the causal UTR variant rs2814778 was captured in the inclusive PGS model but not in the WBonly model (FIG. 2-12), highlighting the benefits of inclusive PGS training. Joint modeling of ancestry-shared and -dependent effects

[0241] To account for ancestry-shared and ancestry-dependent genetic effects, a series of new models were developed, named iPGS+refit models (material and methods). In these models, a regression model was fit to the individual-level data in a specific ancestry group. An additive combination of covariate effects, iPGS, and genetic variants were considered with ancestry- dependent genetic associations as predictors. An iPGS+refit model that additionally considers the interaction between the first two genotype principal components and the genetic variants with ancestry dependent genetic effects was also explored. These two models were named "iPGS+refit in Afr (wo / interaction)" (model v) and "iPGS+refit in Afr" (model vi), respectively (material and methods). For neutrophil counts, for example, 186 genetic variants with heterogeneous genetic associations were identified from a GWAS meta-analysis in the UK Biobank (FIG. 2-4A). The 4,853 individuals of African ancestry in the training and validation sets were then focused on and the iPGS+refit regression models were applied to their individual-level data. For the "iPGS+refit in Afr (wo / interaction)" model (model v), 188 variables were used as predictors, which included one term each representing the covariate effects and iPGS score as well as the 186 genetic variants. For the "iPGS+refit in Afr" model (model vi), a total of 560 variables were used as predictors, comprising an additional 372 variables that model the interaction between the 186 genetic variantsAttorney Docket No.: M0437.70168WO00 and the first two genotype principal components (material and methods). The predictive performance of these models was evaluated by using the individuals in the heldout test set.

[0242] The iPGS+refit model improved predictive performance when the genetic variants were observed with heterogeneous GWAS associations across ancestry groups. For neutrophil counts, further improvements were seen with population-specific iPGS+refit, even beyond the improvements in the inclusive PGS model without the population-specific refit in the African population (FIG. 2-3F and FIG. 2-4D). The iPGS+refit in Afr (wo / interaction) (model v) showed improvements over the vanilla iPGS model without refit(&^= 0.0737 vs. &^= 0.0567). Furthermore, the iPGS+refit in Afr (model vi), which considers the interaction between genotype PCs and genetic variants with heterogeneous effects, showed the best predictive performance (&^= 0.1148) for African individuals. The performance measure exceeded the best-performing model observed for White British individuals (&^= 0.0902 in iPGS, model ii), even though only 1.49% (4246 / 284,661, Tables 1 and 2) of individuals used in the iPGS training were of African ancestry, highlighting the benefits of population-specific refit using the genetic variants with ancestry-dependent associations. Similar improvements were observed in iPGS+refit models in leukocyte counts(&^= 0.0947, 0.0470, and 0.0441 in iPGS+refit in Afr [model vi], iPGS+refit in Afr [wo / interactions, model v], and iPGS [model ii], respectively) (FIG. 2-3G and FIG. 2-4E). Comparison with summary-statistics-based PGS approach

[0243] Lastly, iPGS was compared with PRS-CSx, a recently developed multi-ancestry-aware PGS approach. Specifically, PRS-CSx fits Bayesian multiple-linear-regression models by using GWAS summary statistics from multiple population groups and a series of ancestry-matched LD reference panels. Its predictive performance was compared across European (White British and non-British White), South Asian, African, and the remaining "other" individuals in the held-outtest set. Two sets of models were fit for each of the 60 traits: PRS-CSx (model vii, trained onntrain = 293,301 individuals) and PRS-CSx (n = 256k) (model viii, trained on ntrain = 256,637individuals), which was used for hyperparameter tuning for both models (Tables 1 and 2, materialand methods). Overall, the iPGS (model ii) outperformed the PRS-CSx model trained on up tontrain = 293,301 individuals in most tested traits and ancestry groups even though iPGS modelswere trained on a smaller number of ntrain= 284,661 individuals (FIGs. 2-3A to 2-3G, FIGs. 2- 8A to 2-8B, and FIG. 2-9; Table 8). The few exceptions were all in the African population andAttorney Docket No.: M0437.70168WO00 when there were genetic variants with ancestry-dependent effects (FIGs.2-3A to 2-3B). Ancestry- dependent genetic effects violate the modeling assumption in iPGS; inclusive PGS training works best to capture ancestry-shared genetic effects. Nonetheless, for neutrophil and leukocyte counts, where PRS-CSx outperformed the vanilla iPGS models without population-specific refit, iPGS+refit showed the best predictive performance for African individuals (FIG. 2-3F, FIG. 2- 3G, FIG. 2-4D, and FIG. 2-4E). Those results highlight the advantage of iPGS and the flexibility of iPGS+refit in jointly modeling ancestry-dependent and ancestry-shared genetic effects. Discussion

[0244] Presented herein is inclusive PGS (iPGS), a PGS training strategy that includes ancestry-diverse individuals. By working directly on the individual-level data, iPGS does not require LD reference panels in PGS fitting and naturally provides a way to include admixed individuals in PGS training. The empirical results disclosed herein, in 33 simulation configurations and 60 anthropometric and hematological traits in the UK Biobank, indicate the power of iPGS training in capturing genetic effects shared across ancestry groups. Across all population groups in the held-out test set, the largest improvements in predictive performance are seen when ancestry-diverse individuals, including admixed individuals, are used in training, highlighting the increase in sample size and power.

[0245] iPGS+refit, a method to model ancestry-dependent effects on top of the shared effects captured in iPGS, was also developed. Its utility was shown when genetic variants have heterogeneous associations, by using neutrophil and leukocyte counts as examples. A systematic benchmarking across the 60 traits revealed the competitive advantage of iPGS and iPGS+refit against the commonly used summary-statistics-based PGS model, PRS-CSx.

[0246] Unlike other existing methods, the iPGS presented herein is directly applicable to admixed individuals without the need for local ancestry inference. In the application of iPGS to the UK Biobank, the predictive performance for the "others" group is reported (FIGs. 2-8A to 2- 8B). The average improvement is 17.8% (95% CI: [11.8%, 23.8%]) for the inclusive model over the WBonly model (FIG. 2-2B). Given the diverse ancestral background, individual-level quantification of the predictive performance would be more appropriate for this group of individuals. A recent study reports a Bayesian approach to evaluate the predictive performance of PGS at the individual level. However, the Bayesian method depends on resampling from Bayesian Markov chain Monte Carlo (MCMC). It is not directly applicable to other statistical models, suchAttorney Docket No.: M0437.70168WO00 as the ones based on the penalized regression on the individual-level data, as is the model presented herein, and fast Bayesian PGS based on variational inference.

[0247] When fitting inclusive PGS models, shared genetic effects were assumed across ancestry groups. The empirical analysis across 60 traits in the UK Biobank shows that iPGS improves the transferability for most situations. However, there is no guarantee as to whether the genetic variants selected in the PGS model have causal roles because iPGS is not designed for fine-mapping analysis. In the application of iPGS to the UK Biobank, it was observed that genetic variants selected in the iPGS model were more common in the White British than in the African population, consistent with the difference in the allele-frequency distribution across 1.3 million genetic variants considered in the study.

[0248] Moreover, some cases were observed where ancestry-dependent or ancestry-specific associations violate the modeling assumptions. In some cases, there are extreme differences in allele frequency across population groups-for example, as a result of the positive selections in some populations, as seen in the UTR variant rs2814778 in ACKR1. For neutrophil count, the inclusive PGS model captures the causal UTR variant, but that is not the case for the WB-only model. An iPGS+refit strategy was developed to jointly model ancestry-shared and ancestry- dependent effects in a specific population when the modeling assumption in the vanilla iPGS did not hold. In the applications of iPGS+refit to hematological traits in the UK Biobank, the candidate genetic variants with ancestry-dependent effects were selected by using the heterogeneity test implemented in a GWAS meta-analysis. It is empirically reported herein that the interaction effects between the genetic variants and genotype PCs help improve the predictive performance in iPGS+refit. However, there is no guarantee that the predictive models constructed from iPGS and iPGS+refit capture the causal effects.

[0249] Overall, the results presented herein highlight the importance and the benefits of inclusive PGS training: it naturally offers a way to include admixed individuals in PGS training and increases the power to model ancestry-shared polygenic effects. It is also indicated that joint modeling of ancestry-shared and -dependent effects, as in the iPGS+refit model disclosed herein, would be beneficial for further improvements of the transferability of PGS models. The models presented herein enable future studies to build more equitable PGS models and apply them to translational research.Attorney Docket No.: M0437.70168WO00 Example 2. GenESIS: enhancing transferability of polygenic scores with gene-by-sex interactions.

[0250] Predicting the disease liability and medically relevant traits from individuals’ genetic, demographic, and environmental factors have substantial implications in epidemiology, medical prevention and intervention strategies, and, more broadly, in realizing precision medicine. Conventional risk factors, for example, age, biological sex, family history, and smoking status, have been extensively studied and used in clinical practice. The incremental utilities of genetic factors have recently attracted substantial research interest.

[0251] Polygenic score (PGS) is a statistical approach to aggregate genetic effects across multiple genetic variants into a single score per individual. Recently developed PGS models show improved predictive performance, highlighting their potential clinical relevance for some traits. Despite the increasing recognition of shared causal genetic effects across genetic ancestry groups, the predictive performance of PGS models trained on a single population does not translate well to other population groups, partly due to the disproportionate representation of non-European samples in existing genetic studies. Several efforts are underway to address the critical gap, including active recruitment of ancestry-diverse individuals in genetic studies and computational method development. Emerging computational approaches integrate genetic data from multiple ancestry groups or individuals across the continuum of genetic ancestry, as exemplified in the inclusive PGS model. PGS can be combined with other factors, such as family history, conventional risk factors, and rare variant genetic effects. Due to technical limitations, most PGS approaches do not consider nonlinear (non-additive) or context-dependent genetic effects. They are mostly limited to considering linear effects captured in summary statistics from genome-wide association studies (GWAS), despite the fact that most common complex traits are influenced by a combination of conventional risk factors, demography, genetic components, and their interactions.

[0252] Recent studies highlight the substantial roles of nonlinear and context-dependent genetic effects on complex traits, thanks to the increased sample size and methodological advancements. On the one hand, nonlinear genetic effects, such as genetic dominance and gene- by-gene (GxG) interactions (also known as epistasis), show smaller but substantial contributions to heritability compared to linear effects. On the other hand, studies on context-dependent genetic effects reveal substantial gene-by-sex (GxS) and gene-by-environment (GxE) interactions. ForAttorney Docket No.: M0437.70168WO00 example, substantial GxS effects have been reported for testosterone, anthropometry, and blood pressure traits. Genetic effects on molecular traits (e.g., expression quantitative trait loci effects) are modulated by cellular contexts, including cellular states, stimulus-response, and exposure to drugs. The examples of studying context-dependent genetic effects also manifest when mapping ancestry-specific or ancestry-dependent genetic effects on heritable traits. The difference in the environment and the resulting GxE interactions in the present and the past and population bottleneck events have contributed to shaping ancestry-specific or ancestry-enriched alleles, some of which show associations with complex traits. Those highlight the potential advantage of integrating linear, nonlinear, and context-dependent genetic effects in a unified framework.

[0253] Indeed, previous studies started to incorporate nonlinear and context-dependent effects into predictive models, one at a time. For nonlinear effects, for example, GenoBoost, a flexible machine-learning-based PGS modeling framework, was recently developed to incorporate nonlinear genetic dominance effects and showed that genetic dominance effects localized in the major histocompatibility complex (MHC) locus improve predictive accuracy for immune-related disorders. For gene-by-demography interactions, it was shown that considering interactions between genotype PCs and genetic variant effects at a relatively small number of genetic loci can improve the accuracy of genetic prediction. For GxS interaction effects, it was revealed that constructing sex-stratified PGS models for testosterone, which shows an exceptionally high degree of sexual dimorphism in its genetic architecture, improved the predictive performance despite the two-fold reduction in the sample size in the PGS training population. Beyond testosterone, there has been limited success in improving the predictive performance of PGS by incorporating GxS or GxE interaction effects. Nonetheless, applying sex-stratified PGS analysis highlights its utility in accurately predicting age-dependent temporal dynamics of hematological traits, and it has been proposed that amplification (e.g., varying magnitude of the causal effects between male and female) plays a substantial role in shaping GxS interaction effects. A recent study proposes a calibration of PGS to account for context-dependent prediction intervals of PGS models. However, prior approaches are limited because they often assess the nuanced genetic effects in aggregate across multiple genetic variants, require stratification of samples (e.g., male- and female-only cohorts in sex-stratified analysis), or focus on a specific type of nonlinear or context-dependent effects at a time.Attorney Docket No.: M0437.70168WO00

[0254] Presented herein is GenESIS (GENe, Environment, and Sex Interaction Score), which overcomes many of the limitations in the previous approaches and is the first unified predictive modeling framework capable of integrating linear and nonlinear genetic effects of millions of genetic variants, demography, environmental factors, and their interactions. It is hypothesized that incorporating nonlinear and context-dependent effects can better capture nuanced and likely causal effects and help improve the transferability of PGS. To that end, GenESIS substantially extends the recently developed inclusive PGS (iPGS). GenESIS and iPGS fit predictive models directly on the individual-level data, thus naturally applicable to individuals across the continuum of genetic ancestry. As an initial application of GenESIS, n=406,659 unrelated ancestry-diverse individuals in the UK Biobank resource were analyze and predictive models for 99 quantitative traits were developed, considering linear and nonlinear effects of genetic variants and GxS interaction effects for each of the genome-wide genetic variants. The approach is validated using sex-stratified GWAS analysis and report substantial improvements in PGS transferability for non- European individuals. Significant gains in predictive performance are reported for individuals of the African ancestry group, most notably in anthropometric traits, including body mass index and hip circumference, which are relevant measures for obesity. Furthermore, it is shown that GenESIS reveals biologically plausible hypotheses. For example, it is demonstrated that genetic variants with GxS effects show pleiotropic associations with sex-specific factors, and genome- wide GxS effects for hip circumference are enriched for the known relevant biological processes in obesity, nominating attractive therapeutic targets for context-dependent intervention. Results GENe, Environment, and Sex Interaction Score (GenESIS) methodology

[0255] In GenESIS, linear and interaction effects of genetic variants, demography, environmental factors, and biological sex are considered by applying supervised learning. The following steps were used as a proof of principle, though a broader set of approaches is possible, as described in the discussion. Here, genetic data was first augmented by constructing variables representing nonlinear genetic effects (Methods). Second, supervised statistical learning was applied directly on the individual-level data while introducing more regularization for nonlinear and context-dependent genetic effects. Third, the incremental utility of nonlinear and context- dependent genetic effects was evaluated in polygenic prediction by benchmarking GenESISAttorney Docket No.: M0437.70168WO00 against linear-only inclusive PGS (iPGS) models. As an illustration of follow-up analysis, biologically relevant variants, genes, and enriched pathways captured in the GenESIS predictive model are nominated. Overview of initial application of GenESIS to UK Biobank

[0256] In the initial application of GenESIS to n=406,659 unrelated individuals across the continuum of genetic ancestry in the UK Biobank resource, a panel of 99 heritable traits was assembled, comprising anthropometry, blood biomarkers, blood pressure, and hematological traits (Tables 13-14). L1- and L2- penalized Elastic Net regression was applied directly on the individual- level data considering linear, nonlinear, and genome-wide gene-by-sex (GxS) interaction effects, represented in 2,630,335 predictor variables across 1,316,147 genetic variants (Methods, Table 13). For the MHC region, the imputed human leukocyte antigens (HLA) allelotypes were used to account for complex linkage disequilibrium (LD) structure and also incorporated genetic dominance effects. Linear-only iPGS models were applied as a baseline using the same set of 1.3 million genetic variants. In both GenESIS and iPGS models, an additional 23 variables were used as unpenalized covariates, including age, sex, age2, age*sex, Townsend deprivation index, and the first 18 genotype PCs. Each population group was split and the training set was used for fitting models, the validation for hyperparameter tuning and model selection, and the held-out test set for predictive performance evaluation (Table 11).

[0257] Of the analyzed individuals, about 53.9% are female (Table 11). The unpenalized covariate term, sex, accounts for the mean differences in trait values between male and female individuals. To allow the magnitude or direction of genetic effects to differ between male and female groups and to capture nuanced genetic effects, about 50.0% of predictor variables were prepared to represent GxS interaction effects (Table 13).

[0258] It was first demonstrated that GenESIS’s GxS effects are highly consistent with sex- stratified genome-wide associations, validating the approach. The predictive performance of GenESIS, linear-only iPGS, and a model that considers covariate terms alone was subsequently evaluated in each of the following population groups: white British (WB), non-British white (NBW), South Asian (SA), African (Afr), and other unrelated individuals (Others). Lastly, the genetic variants selected in the GenESIS model and their enrichment to pathways and biological processes were highlighted to offer an interpretation of context-dependent effects captured in the predictive mode.Attorney Docket No.: M0437.70168WO00 Validating GxS effects in GenESIS by sex-stratified GWAS

[0259] Across the 99 quantitative traits, the GenESIS predictive model was developed. linear- only iPGS models were also prepared using the same sets of individuals and genetic variants. Both approaches use penalized regression (Elastic Net) directly on the individual-level data without requiring users to specify the genetic architecture of the trait. As such, the study design is suitable for systematic comparison across a wide variety of traits, enabling systematic assessment of the benefits of incorporating nonlinear and context-dependent effects in genetic prediction.

[0260] Overall, it was found that GenESIS and linear-only iPGS models have a comparable number of predictor variables (Linear regression fit of y=1.006x, Pearson’s correlation=0.97, 95% confidence intervals, CI: [0.96, 0.98]) with the exceptions of total bilirubin and direct bilirubin (FIG. 3-1A, Table 14). The number of predictor variables increased from 197 to 14,869 for total bilirubin and 278 to 9979 for direct bilirubin in GenESIS. In GenESIS models, a median of 8.0% of predictors (ranging from 0.2% to 27.7%) captures GxS interaction effects. The variants with GxS interaction effects are distributed genome-wide, not necessarily localized in sex chromosomes. On median across 99 UK Biobank traits, the GxS interaction effect size is 24.8% smaller than that of linear effects (FIGs. 3-1A to 3-1B). Genetic dominance effects at HLA allelotypes are captured in two traits (gamma-glutamyl transferase and C-reactive protein), both on the same allele, HLA-DPA1*0103. The high allele frequency of the allelotype (96.0% in White British) works advantageously in capturing genetic dominance effects, possibly reflecting the limited statistical power to detect and capture genetic dominance genetic effects for other imputed allelotypes.

[0261] To validate the GxS interaction effects captured in GenESIS models, sex-stratified GWAS analysis was applied and the presence of GxS effects was replicated. To that end, unrelated individuals in White British, the largest population group in the UK Biobank cohort, were focused on and sex-stratified GWAS was applied for quantile normalized phenotypes (Methods). Effect size estimates were compared and the heterogeneity of genetic associations between male and female individuals were assessed. When the genetic variants selected in GenESIS were investigated, it was found that the GWAS association summary statistics from the sex-stratified analysis are more concordant in variants selected for linear effects alone (FIGs. 3-1B to 3-1D). For hip circumference, for example, Pearson’s correlations of association summary statistics were 0.730 (95% confidence interval, CI: [0.725, 0.735]) for genetic variants with linear effects inAttorney Docket No.: M0437.70168WO00 GenESIS and 0.058 (95% CI: [0.015, 0.101]) for genetic variants with GxS interaction effects (FIGs. 3-1B to 3-1C). The magnitude of GxS effects was further quantified by computing the variant-level test statistics, TDiff, where deviation from TDiff = 0 indicates the presence of sex- biased genetic effects (Methods). Indeed, the distribution of the TDiffstatistics is different between the genetic variant groups with linear-only and GxS effects (Kolmogorov–Smirnov test p-value < 1x10-300) (FIG. 3-1D). A similar pattern was observed for other traits, including the two bilirubin traits (FIGs. 3-5A to 3-5B). Together, those results highlight the unique ability of GenESIS to capture validated GxS effects in predictive models. GenESIS improves PGS transferability, especially for anthropometric traits

[0262] Having validated the GxS effects, the predictive performance of GenESIS models was evaluated and compared against the corresponding linear-only iPGS models. Predictive performance is reported in each population group to minimize the risk of confounding due to population structure. The statistical difference between the two models was subsequently tested in the validation and held-out test set. In the comparison between GenESIS and iPGS, the full models that consider covariate, genetics, and their interactions were studied and their predictive performance was evaluated.

[0263] Across 99 traits, substantial heterogeneity was found across traits in the improvements in predictive performance with GenESIS, as expected, given that the magnitude of GxS effects depends on the genetic architecture of traits (FIGs. 3-6A to 3-6B). Model selection was applied between GenESIS vs. linear-only iPGS for each (population, trait) pair based on the validation set metrics (Methods). It was found that GenESIS was selected for 41 out of 99 traits for at least one population group (FIG. 3-1E, Table 15). For example, for African ancestry groups, GenESIS is selected for 14 out of 31 anthropometric traits considered in the study. The GenESIS PGS models for the two bilirubin traits are selected for European ancestry groups. The selected model shows equivalent or superior predictive performance in the held-out test set for 493 out of 495 (population, trait) pairs, highlighting the improvements in predictive performance with GenESIS for 41 out of 99 analyzed traits (Table 16). For the remaining two cases, the predictive performance was very similar between the two PGS models, with the difference in R2less than 0.0017, indicating the effectiveness of the model selection procedure in determining the most relevant PGS model.Attorney Docket No.: M0437.70168WO00

[0264] Empirically, improved predictive performance of GenESIS models was found for non- European population groups, most notably for the African population and in anthropometric traits, including hip circumference, body mass index, and whole-body fat mass (FIGs. 3-2A to 3-2B). For hip circumference, for example, GenESIS improved trait prediction in the held-out test set (R2=0.067, 95% confidence interval: [0.040, 0.095]) over iPGS (R2=0.018, 95% CI: [0.003, 0.033]) in the full model that considers both covariates, genetics, and their interactions with the statistically significant (p=8.0x10-7) difference (FIG.3-2C, FIG.3-7). Similar improvements were found for body mass index in African individuals (FIG. 3-2C). Moreover, it was found that the predictive performance of the GenESIS model outperforms all publicly available PGS model evaluations in African individuals for hip circumference, despite that only 6% of predictors (2075 / 34,351 variables) capture GxS interaction effects and that only 1.49% of individuals used in the GenESIS model training are of African ancestry (Methods, FIG. 3-2D, Table 17). Similar improvements were found for South Asian ancestry for GenESIS (R2=0.113) over iPGS (R2=0.098), while the predictive performance for white British individuals remained the same (R2=0.1067 for GenESIS and R2=0.1066 for iPGS). The difference in the European group was not significant (nominal p-value=0.62), highlighting the critical benefit of incorporating context- dependent genetic effects in improving PGS transferability.

[0265] To investigate the effects of specific encoding of sex-specific effects in GenESIS, seven phenotypes were selected, additional GenESIS models were fit, and the changes in the predictive performance were assessed. In the primary analysis, 2.6 million predictor variables were considered across 1.3 million unique genetic variants considering sex-shared and male- specific GxS interaction effects, given that the analyzed UK Biobank cohorts consisted of 53.9% female and 46.1% male individuals and that the collinearity of sex-specific GxS effect vectors introduce challenges in numerical stability. In the additional analysis, sex-shared, male-specific, and female-specific effects were considered and the predictive performance was compared (Methods). It found that the two models show highly consistent predictive performance across five population groups, validating the approach (FIG. 3-8). Sparse GenESIS models offer biological interpretations

[0266] To investigate the biology of GxS effects captured in the GenESIS models, it was hypothesized that the genetic variants captured in GenESIS with linear and GxS effects offer interpretation. It was found that the GenESIS model captures genetic effects on non-synonymousAttorney Docket No.: M0437.70168WO00 coding variants of well-known genes, allowing for the interpretation of the biological processes captured in the predictive model. The biological basis of sex-dependent genetic effects was also inferred in the GenESIS model by investigating pleiotropic associations and genome-wide ontology enrichment of genetic variants with GxS effects.

[0267] As an illustrative example, here the GenESIS model for hip circumference was focused on. For linear effects, on the one hand, protein-altering variants in MC4R (rs2229616) and GIPR (rs1800437) were found and selected for their trait-lowering effects (FIG. 3-3A). Melanocortin 4 receptor, encoded in MC4R gene, is known for its role for regulating body weight and energy homeostasis whereas gastric inhibitory polypeptide receptor, encoded in GIPR, is known for glucose metabolism, both known for their relevance in anthropometric traits. For GxS interaction effects, on the other hand, a non-synonymous variant was found in GCKR (rs1260326), where glucokinase regulator, the encoded protein, plays a regulatory role in glucose metabolism and is associated with central fat accumulation (FIG. 3-3B). Moreover, the variant shows pleiotropic associations across blood biochemistry (for example, triglycerides, C-reactive protein, and sex hormone-binding globulin [SHBG] levels), anthropometric (whole-body water mass and trunk fat-free mass), blood pressure (pulse rate and position on pulse wave notch), and sex-specific traits (FIG. 3-3C, Table 18) The associations on the sex-specific traits include genetic effects on age at menopause and “had menopause,” a binary questionnaire-based phenotype. The pleiotropic association of the variant across SHBG and sex-specific traits offers insights into the biology behind the sex-biased effects represented in the GenESIS model.

[0268] To further evaluate the biological processes captured in the genome-wide GxS effects in GenESIS beyond the single-locus analysis, the enrichment to ontology terms, pathways, and biological processes was evaluated using the Genomic Regions Enrichment of Annotations Tools (GREAT) (Methods). For hip circumference, it was found that the genetic variants selected for GxS interaction effects are enriched for negative regulation of interleukin-4 production (hypergeometric fold change=4.5, FDR=3.0x10-6) and phosphofructokinase activity (hypergeometric fold change=5.0, FDR=1.51x10-3) (FIG. 3-3D, Table 19). The PFKP gene encodes phosphofructokinase, a key enzyme in glycolysis, and it has been reported that Interleukin-4 inhibits adipogenesis, regulates lipid metabolism by promoting lipolysis, and influences obesity and type 2 diabetes mellitus. Overall, these results indicate that analysis ofAttorney Docket No.: M0437.70168WO00 genetic variants in GenESIS reveals key pathways with sex-dependent effects, nominating attractive targets for context-dependent interventions. Discussion

[0269] Presented herein is GenESIS, a unified polygenic score (PGS) modeling framework capable of incorporating linear, nonlinear, and context-dependent effects. Herein its advantage in incorporating nuanced effects in genetic prediction is shown, and the biological characterization of GenESIS models is illustrated. As an initial application, 99 quantitative traits were analyzed in UK Biobank using n=406,659 unrelated individuals across the continuum of genetic ancestry. First, the gene-by-sex (GxS) interaction effects were validated, captured in GenESIS by an orthogonal GWAS-based approach. The predictive performance of GenESIS was subsequently compared over the linear-only inclusive polygenic scores (iPGS) trained on the same set of ancestry-diverse individuals. It is demonstrated that modeling GxS effects helps improve the transferability of polygenic scores, especially for anthropometric traits in African individuals. Of note, the GenESIS model for hip circumference, an illustrative example, outperforms all publicly available model evaluations for Africans in the PGS catalog. Lastly, biologically plausible hypotheses for context-dependent effects captured in GenESIS are nominated. For example, phenome-wide associations across menopausal age and sex hormone-binding globulin levels are reported on genetic variants with GxS interaction effects in GenESIS. Moreover, genome-wide enrichment of GxS effects were found in GenESIS to biological processes and pathways, nominating attractive targets for context-dependent interventions.

[0270] Improving the accuracy of predictive models for disease outcomes and medically relevant traits has direct implications for realizing precision medicine. To that end, there has been substantial progress in improving the transferability of PGS. The existing approaches in the field include prioritizing variants, which are present in diverse populations or overlapping with bio- sample-specific regulatory elements, integrating the results of statistical fine mapping, and incorporating genetic data from ancestry-diverse individuals. Here, it is revealed, for the first time to the best of the Applicant’s knowledge, that joint modeling of linear, nonlinear, and context- dependent effects, including GxS interactions, is another effective strategy for developing more equitable PGS models, complementing existing efforts. The GenESIS model does not assume the genetic effects to be linear or ubiquitously shared across everyone and allows genetic effects to be modulated in the presence of demography and environmental variables. The flexible approachAttorney Docket No.: M0437.70168WO00 allowed for capturing more likely causal effects in predictive models and offered improvements in PGS transferability across genetic ancestry groups.

[0271] Methodologically, this study highlights substantial advancements in the predictive modeling of complex traits. For example, to the best of Applicant’s knowledge, GenESIS is the first PGS method that can incorporate linear, nonlinear, and context-dependent genetic effects across millions of genetic variants in a unified framework. Here, penalized regression was focused on as an initial application given its success in PGS modeling, but other types of predictive models, including statistical boosting, would be attractive alternatives. These results, in conjunction with previous PGS modeling efforts, highlight that supervised learning applied directly on large-scale individual-level data would be an effective strategy to incorporate genetic effects that are not captured in the linear association summary statistics from GWAS studies.

[0272] Biologically, it is illustrated that sparse GenESIS predictive models readily offer interpretations. It is shown that pleiotropic associations with sex-specific factors offer plausible hypotheses on sex-dependent genetic effects captured in the GenESIS models. Biological processes in obesity are nominated for hip circumferences via genome-wide enrichment of GxS effects, offering attractive targets for context-dependent interventions.

[0273] Overall, these results underscore the importance of mapping nonlinear and context- dependent effects, highlight the critical benefits of integrating such effects in improving PGS transferability, and, more broadly, pave the way for designing more inclusive and nuanced health interventions. Material and Methods GENe, Environment, and Sex Interaction Score (GenESIS)

[0274] The GENe, Environment, and Sex Interaction Score (GenESIS) methodology is introduced as an application of supervised learning for phenotypic prediction. Herein, L1- and L2- penalized Elastic Net regression was applied directly on the individual-level data using the batch screening iterative lasso (BASIL) algorithm implemented in the R snpnet package (version 2), although other statistical and machine-learning approaches, such as statistical boosting, would be applicable as described in the discussion.

[0275] Given a phenotype vectorof n individuals and a predictor matrix of n individuals and d variables, the following penalizedAttorney Docket No.: M0437.70168WO00 generalized linear regression was considered to fit the intercept term and regression coefficient vector as a function of the tuning parameter λ that controls the sparsity of the solution:(Equation 2-1)

[0276] where is the loss contribution for the i-th individual andis the Elastic Net penalization term for the coefficient. With a balancing parameter between the L1- (Lasso) and L2- (Ridge) penalty, α, the Elastic Net penalization term is written as follows:(Equation 2-2)

[0277] The tuning parameter, λ, was optimized based on the predictive performance on the validation set and use α=0.99. As optional parameters, one may specify sample weights, , and penalty factor values,, that allow different levels of shrinkage to variables. In the application to the UK Biobank dataset, =0 was set for covariate terms to ensure the covariate terms are unpenalized in the regression. Lower penalty factors were assigned for non-synonymous coding variants, as described below. One may use =1 and =1 as the default values.

[0278] The loss contribution for the i-th observation, , depends on the types of the exponential family considered in the regression analysis of generalized linear models. For example, it is the squared loss, i.e., , for quantitative phenotypes (Gaussian family), and it is logistic loss, i.e.,, for binary phenotypes (Binomial family). A similar model can be used for time-to-event phenotypes (Cox Proportional Hazards regression), or joint modeling of multiple phenotypes as shown previously. GenESIS with linear effects alone

[0279] For modeling linear effects of genetic variants, a covariate matrix of n individuals and dcovcovariates and a genotype matrix of n individuals and dgvariants representing the allelic count of effect allele was considered. The concatenated vectors were used as predictors:Attorney Docket No.: M0437.70168WO00(Equation 2-3)

[0280] The coefficient vector is also a concatenation of two components, corresponding to covariate effects and linear effects of genetic variants, respectively:(Equation 2-4)

[0281] Covariate terms were set to be unpenalized (i.e., forthe objective function of the penalized generalized linear regression (Equation 2-1) becomes the following, which is equivalent to the polygenic prediction models considered previously:(Equation 2-5) GenESIS with genetic dominance effects

[0282] To incorporate nonlinear genetic dominance effects, a matrix is considered,, whereis an indicator variable representing whether the i-th individual is homozygous for effect allele for the j-th genetic variant. The concatenated vectors of three components were used as predictors:(Equation 2-6)

[0283] The coefficient vector now has three components:(Equation 2-7)Attorney Docket No.: M0437.70168WO00

[0284] One may focus on a smaller subset of variants for genetic dominance effects instead of considering all of the dggenetic variants. GenESIS with GxS interaction effects

[0285] To incorporate G-by-Sex (GxS) interaction effects, predictor variables were augmented. For example, two matrices may be considered, andand represent GxS interaction terms for male and female individuals, respectively, as follows: , and (Equation 2-8) (Equation 2-9) where is an indicator function. Thehas the original genotype dosage for male individuals and the value is always set to be zero for female individuals and vice versa for . The concatenated vectors of four components were use as predictors:(Equation 2-10)

[0286] The coefficient vector now has four components:(Equation 2-11)

[0287] There would be other encoding of GxS interaction effects, as follows:(Equation 2-12)

[0288] One may focus on a smaller subset of variants for the GxS interaction effect.Attorney Docket No.: M0437.70168WO00 GenESIS with genetic dominance and GxS interaction effects

[0289] It is possible to consider genetic dominance and GxS interaction effects of the linear and genetic dominance effects simultaneously by using concatenated vectors of seven components as predictors:(Equation 2-13)

[0290] The coefficient vector now has seven components:(Equation 2-14)

[0291] One may focus on smaller subsets of variants for genetic dominance effects and GxS interaction effects of the linear and genetic dominance effects. The study population in the UK Biobank resource

[0292] UK Biobank is a population-based cohort study with genomic and phenotypic datasets across about 500,000 volunteers collected across multiple sites in the United Kingdom. Sample- level quality control (QC) was performed. n=406,659 unrelated individuals with genetic data were focused on based on the following criteria: (1) used to compute principal components (UK Biobank Data Field 22020); (2) removal of sex mismatch between the sex field in the genotype dataset and phenotype sex (Data Field 31); (3) not reported in “Outliers for heterozygosity or missing rate” (Data Field 22027); (4) not reported in "Sex chromosome aneuploidy" (Data Field 22019); and (5) do not have ten or more third-degree relatives (Data Field 22021). A combination of self-reported ethnic background (Data Field 21000) and genetic principal components (Data Field 22009) was subsequently used to define four population groups (white British, non-British white, African, and South Asian). The remaining unrelated individuals were kept as Others. Individuals were focused on whose inferred biological sex is either female or male. In each population, 53.9% of unrelated individuals are female and the remaining 46.1% are male (Table 13). Each population group was randomly split into training (70%, n=284,661), validation (10%, n=40,667), and held-out test (20%, n=81,331) sets without using phenotypes. The training set wasAttorney Docket No.: M0437.70168WO00 used for model fitting, the validation set for determining hyperparameters, and the held-out test sets for predictive performance evaluation. The same training, validation, and test sets were used for all tested traits. Variant annotation and quality control in UK Biobank

[0293] The directly genotyped dataset (release version 2), imputed genotypes (release version 3), imputed HLA allelotype (release version 2), and GRCh37 human reference genome were used throughout the study. Variant annotation was performed with Ensembl’s Variant Effect Predictor (VEP) (version 101) with the LOFTEE plugin and ClinVar. The VEP-predicted consequence of the variants were grouped into six groups: protein-truncating variants (PTVs), protein-altering variants (PAVs), proximal coding variants (PCVs), intronic variants (Intronic), genetic variants on untranslated regions (UTR), and other non-coding variants (Others). For the directly genotyped dataset, variants passing the following criteria were focused on: (1) the missingness of the variant is less than 1%, considering that the two genotyping arrays (the UK BiLEVE Axiom array and UK Biobank Axiom array) cover a slightly different set of variants and (2) Hardy-Weinberg disequilibrium test p-value greater than 1.0x10-7. For the imputed genotype dataset, the following criteria were used: (1) the missingness of the variant is less than 1%; (2) minor allele frequency (MAF) greater than 0.01%; (3) imputation quality score (INFO score) greater than 0.3; (4) does not present in the directly genotyped dataset; and (5) present in the HapMap Phase 3 dataset. For the HLA allelotype, the imputed allelotype dosage was kept within [0, 0.1), (0.9, 1.1), or (1.9, 2.0] and converted it to hard call. The HLA allelotype was focused on with (1) missingness no more than 1% and (2) Hardy-Weinberg disequilibrium test p-value greater than 1.0x10-4. All variants and allelotypes were concatenated into one dataset using PLINK 2.0 (v2.00a3.3LM 3 Jun 2022). The quality control procedure resulted in 1,316,181 unique genetic variants and allelotypes considered in the analysis.

[0294] The following variables were defined to consider nonlinear genetic effects. The imputed HLA allelotypes account for complex LD structure in the major histocompatibility complex (MHC) region. The imputed allelotype dosage was kept within (1.9, 2.0] for genetic dominance effects of imputed HLA allelotypes. For GxS interaction effects, sex-specific effects in males were modeled by keeping the original genotype for male individuals and setting zero for female individuals, as in Equation 2-8. Lastly, variables were prepared for GxS interaction effects of genetic dominance terms for the HLA allelotypes. Variables were dropped when none of theAttorney Docket No.: M0437.70168WO00 unrelated individuals considered in the analysis had non-zero values. The procedure above resulted in 2,630,335 variables considered in the analysis (Table 13). Phenotype definition in UK Biobank

[0295] In the UK Biobank resource, 99 anthropometric, blood biochemistry, blood count (hematological), and blood pressure traits were studied (Table 12). Some of those phenotypes are collected at up to four instances, each of which corresponds to (1) the initial assessment visit (2006-2010), (2) the first repeat assessment visit (2012-2013), (3) the imaging visit (2014-), and (4) first repeat imaging visit (2019-). Phenotype data was defined using the median of non-missing values for each individual across the 60 quantitative traits as described elsewhere. Genome-wide association analysis

[0296] Sex-stratified genome-wide association analysis was applied with PLINK (v2.00 alpha). Population-specific genotype PCs were computed for white British individuals in the UK Biobank cohort using the randomized algorithm implemented in PLINK2. The GWAS analysis was subsequently applied using age, Townsend deprivation index, array, and the top ten population-specific genotype PC loadings as covariates, using approximation algorithm implemented as “--glm zs omit-ref no-x-sex log10 hide-covar skip-invalid-pheno cc-residualize firth-fallback” command in PLINK2. The participants of the UK Biobank cohort were genotyped on two different arrays: about 10% of participants were genotyped on the UK BiLEVE Axiom array, whereas the rest were genotyped on the UK Biobank Axiom array. When genetic variants were directly measured on both arrays, an indicator variable “array” was included in the covariates, denoting whether the UK Biobank Axiom array or UK BiLEVE Axiom array was used in the genotyping. Quantile normalization was applied using the “--pheno-quantile-normalize” option in PLINK2. GWAS analysis was conducted using the male and female individuals separately in the white British population group. Sex-stratified GWAS analysis

[0297] To test the sex difference in GWAS associations, the interaction effect-only test described in a recently published study was applied. Specifically, the sex-differential test statistics TDiff, was calculated by applying Equation 2-14 shown below to association summary statistics from the sex-stratified analysis:Attorney Docket No.: M0437.70168WO00(Equation 2-15) Applying GenESIS to the UK Biobank resource

[0298] In the application of GenESIS to UK Biobank, the following variables were included as unpenalized covariates: age (UK Biobank Data Field 34), sex (Data Field 31), age2, age*sex, Townsend deprivation index (Data Field 22189), and the first 18 genotype PCs (Data Field 22009) provided by UK Biobank. The genotype PCs represent genetic ancestry and account for trait mean differences associated with the genome-wide genetic ancestry. Linear, genetic dominance, and sex-specific GxS interaction effects in males represented in p=2,630,335 variables across 1,316,181 unique genetic variants and imputed allelotypes were considered (Table 13). Analysis of both male- and female-specific GxS effect vectors presents technical challenges in numerical stability, partly due to the collinearity of sex-specific GxS effect vectors. In the main analysis, male-specific GxS interaction effects were focused on, given that more female individuals are in the cohort. For select traits, both male- and female-specific GxS effect vectors were incorporated and the difference in predictive performance was investigated. Protein-truncating and protein- altering variants were prioritized using penalty factors shown in Table 13 to improve the interpretation of selected variants. More penalization was imposed for nonlinear genetic effects to reduce the risk of overfitting. The specific values of penalty factors are based on heuristics, and finding the optimal values of penalty factors would be an important direction of follow-up studies as described in the Discussion. Comparison of linear and GxS effect size

[0299] The standard deviation of the phenotype values was computed in the training set for each trait and was used to normalize the GenESIS effect size. The first, median, and third quartile of the absolute value of normalized effect size was computed for linear and GxS effects. They were compared across 99 traits. A linear regression was fit with a fixed intercept at 0 and the slope was reported as an aggregated measure of the difference in the magnitude of effect size between linear and GxS effects across traits. PGS performance evaluation and model selectionAttorney Docket No.: M0437.70168WO00

[0300] The predictive performance (R2) of (1) PGS (non-covariate-only) models, (2) covariate-only models, and (3) full models that considered both covariates and genotypes using the held-out test set were evaluated (Tables 14-15). Unless indicated otherwise, the full models' predictive performance is reported herein. The 95% confidence interval of predictive performance was evaluated using the approximate standard error of the R2metric.

[0301] In the application of GenESIS to UK Biobank, the predictive performance of the linear- only iPGS model was used as the baseline. The significance of the difference in R2between the GenESIS model vs. the iPGS model was assessed using the delta method implemented in the r2redux package in R.

[0302] Model selection between GenESIS vs. linear-only iPGS was performed for each (trait, population group) pair using the predictive performance in the validation set. The GenESIS model was selected if GenESIS showed improved performance with statistical significance of nominal p-values less than 0.05. Otherwise, iPGS models were selected. Predictive performance evaluations reported in the PGS catalog

[0303] To compare the predictive performance of the GenESIS model against that of publicly available PGS evaluation in the PGS catalog, the predictive performance of the GenESIS model was evaluated using partial-R2adjusted for covariate effects using Equation 2-16: (Equation 2-16)

[0304] The 95% confidence intervals of the original R2fullwere used to compute the 95% confidence intervals of partial-R2. Four studies with publicly available PGS evaluations for hip circumference for African-ancestry individuals were found (pgscatalog.org / trait / EFO_0005093 / ). Partial correlation (partial-r) was converted to partial-R2by squaring its value. Phenome-wide association analysis

[0305] The phenome-wide association of genetic variants was investigated with GxS interaction effects in the GenESIS models using the Global Biobank Engine. The association summary statistics for the select genetic variants were obtained based on their coordinates on the GRCh37 reference genome. For example, the association profile for rs1260326 (a missenseAttorney Docket No.: M0437.70168WO00 variant in GCKR) is from the following webpage: biobankengine.stanford.edu / RIVAS_HG19 / variant / 2-27730940-T-C. Associations with nominal p-value < 1x10-10were subsequently focused on across traits in the following categories: anthropometry, arterial stiffness, biomarkers, blood assays, blood pressure, and sex-specific factors. Enrichment to biological processes

[0306] To investigate the relevant biological processes in the nonlinear genetic effects captured in GenESIS models, the Genomic Regions Enrichment of Annotations Tool (GREAT) (version 4.0.4) was applied. To evaluate the enriched biological processes captured in linear and nonlinear effects in GeneESIS models, the top 1000 genetic variants according to the effect size were taken and the GREAT enrichment was applied against the default genome-wide background. To evaluate the enrichment of genetic variants with GxS interaction effects, the variants with non- zero GxS interaction effects were taken as the foreground set and their relative enrichment was evaluated against all the variants included in the GenESIS model (background set). In both analyses, ontology terms were filtered based on the number of genes annotated for the term, focusing on terms with 5 to 500 annotated genes. Statistics

[0307] For computational and statistical analysis, Jupyter Notebook, R, R Tidyverse package, and GNU parallel were used. For visualization, ggplot2 was used with ggrepel and ggrastr packages. The p-values were computed from two-sided tests unless otherwise specified. Example 3. Methods for improving cross-ancestry transferability in predicting polygenic phenotypes from genetic variation Expansion of the predictive performance comparison to 226 traits

[0308] Previously, 60 quantitative traits, consisting of 31 anthropometric traits and 29 blood count traits were focused on. However, the analysis has now been expanded to 226 traits. The 226 traits used in the present analysis consist of 177 quantitative traits, such as blood pressure and clinically relevant blood biomarkers, and 49 binary traits, including disease outcomes. MethodsAttorney Docket No.: M0437.70168WO00

[0309] By including binary traits, such as disease outcomes, in the analysis, the description of the methodology was revised, specifically around Equation 3-1.

[0310] The inclusive polygenic score (iPGS) was applied as described previously. Specifically, the additive effects of genetic variants on the phenotypes was consider and sparse polygenic score models were fit using batch screening iterative lasso (BASIL) implemented in the R snpnet package (version 2), which takes individual-level data and performs variable selection and effect size estimation simultaneously by finding the exact solution for L1- and L2-penalized multivariate regression (Elastic Net).

[0311] Given a phenotype vectorof n individuals and a predictor matrixof n individuals and d variables, the following penalized generalized linear regression was considered to fit the intercept term and regression coefficient vector as a function of the tuning parameter λ that controls the sparsity of the solution:(Equation 3-1) where is the loss contribution for the i-th individual and is the Elastic Net penalization term for the coefficient . With a balancing parameter between the L1- (Lasso) and L2- (Ridge) penalty, α, the Elastic Net penalization term is written as follows:(Equation 3-2)

[0312] The tuning parameter, λ, was optimized based on the predictive performance of the predictive model in the validation set. As optional parameters, one may specify sample weights, , and penalty factor values, , that allow different magnitudes of shrinkage to variables. One may use =1 and =1 as the default values.

[0313] For modeling linear effects of genetic variants with iPGS, a covariate matrixof n individuals and dcovcovariates and a genotype matrix of n individuals and dgvariants representing the allelic count of effect allele were considered. The concatenated vectors were used as predictors:Attorney Docket No.: M0437.70168WO00(Equation 3-3)

[0314] The coefficient vector is also a concatenation of two components, corresponding to covariate effects and linear effects of genetic variants, respectively:(Equation 3-4)

[0315] When covariate terms are set to be unpenalized (i.e.,), the objective function of the penalized generalized linear regression (Equation 3-1) becomes the following, which is equivalent to the polygenic prediction models considered previously herein:(Equation 3-5)

[0316] The loss contribution for the i-th observation,depends on the types of the exponential family considered in the regression analysis of generalized linear models. For example, it is the squared loss, i.e., , for quantitative phenotypes (Gaussian family), and it is logistic loss, i.e.,, for binary phenotypes (Binomial family). A similar model can be used for time-to-event phenotypes (Cox Proportional Hazards regression), or joint modeling of multiple phenotypes as shown previously.

[0317] In the application to the UK Biobank dataset, =0 was set for covariate terms to ensure the covariate terms were unpenalized in the regression. α=0.99 was used as used previously herein. Lower penalty factors were assigned for non-synonymous coding variants, as described previously. Briefly, a penalty factor of 0.5 was assigned to putative protein-truncating variants and pathogenic variants, 0.75 to putative protein-altering variants, likely-pathogenic variants, and HLA allelotypes, 1.2 for variants that are not present in the HapMap Phase 3 dataset, and 1.0 for other variants on remaining variants. The specific values of penalty factors are based on heuristics,Attorney Docket No.: M0437.70168WO00 and finding the optimal values of penalty factors would be an important direction of follow-up studies. Results List of 226 Analyzed Traits

[0318] The analysis was expanded from 60 quantitative traits to 226 traits consisting of 177 quantitative and 49 binary traits, grouped in the following 12 groups as listed in Table 20. The full list of 226 traits analyzed is provided in Table 21. Average Improvements in Predictive Performance

[0319] The average improvements in predictive performance were evaluated. For quantitative traits, improvements were quantified in R2. For binary traits, traits with at least 200 case individuals in the held-out test set were focused on and the improvements in (AUROC - 0.5) were evaluated. For both analyses, Deming regression was applied, which considers standard errors in predictive performance measures in both the baseline model (trained only on white British individuals) and the inclusive PGS models herein. The results are shown in Table 22 and FIGs.4- 1A through 4-5B. Further improvements in PGS transferability with “training & validation refit”

[0320] The participants of the UK Biobank cohort are mostly of European ancestry. In the original application of iPGS to the UK Biobank resource, the substantial difference in the number of individuals across ancestry groups in the training set were not accounted for, raising a possibility that the iPGS models are biased towards statistical patterns observed in European individuals. To reduce the potential biases in the iPGS model trained on mostly European individuals, additional levels of regularization of the statistical model were explored. The difference in the predictive performance was quantified across different ancestry groups in the held-out test set individuals. It is shown that imposing more regularization of iPGS models enhances the transferability of PGS into non-European population groups while keeping the predictive performance for Europeans. Methods

[0321] In the application of iPGS to the UK Biobank dataset, individuals were split into a training set (70% of the unrelated individuals), a validation set (10%), and a held-out test setAttorney Docket No.: M0437.70168WO00 (20%), as described in the initial provisional application. The L1- and L2-regularized regression (Elastic Net) model, a specific class of supervised learning, was considered in iPGS analysis, and it has an objective function specified in Equation 3-1.(Equation 3-1) where, the tuning parameter λ controls the sparsity of the solution, i.e. the number of predictor variables included in the iPGS model. In the original iPGS model, a series of λ values were considered and the predictive performance was used in the validation set to select the optimal tuning parameter to be considered in the final model. With this setup, individuals in the validation set (10%) are used only in determining the tuning parameter λ, and the multivariate effect size estimates (i.e. the vector value across genetic variants) are estimated from the individuals in the training set (70%).

[0322] In “training & validation refit,” the following procedure was applied, and individuals were included in the validation set to maximize the statistical power while keeping the same regularization threshold to reduce the risk of overfitting to the European-biased training set in UK Biobank. 1. Fit the iPGS model by: a. Fitting a series of supervised learning (e.g., penalized regression with Elastic Net) models using individuals in the training set with varying levels of tuning parameter λ; b. Evaluating the predictive performance of each Elastic Net model corresponding to different λ values using the individuals in the validation set; c. Determining the optimal λ value that maximizes the predictive performance in the validation set; and d. Returning the Elastic Net regression model corresponding to the optimal λ value. 2. Record the regularization parameter in the original iPGS model. 3. Fit another iPGS model using a larger number of individuals in the training and the validation set and determine the optimal value of the tuning parameter λ using the regularization criterion in the previous step. This includes the following steps:Attorney Docket No.: M0437.70168WO00 a. Fitting a series of Elastic Net regression models using individuals in the training and validation set with varying levels of tuning parameter λ; b. Evaluating the regularization parameter of each Elastic Net model corresponding to different λ values; c. Returning the Elastic Net regression model corresponding to the optimal regularization parameter.

[0323] The optimal regularization parameter was selected using the initial iPGS models trained only on the individuals in the training set. This indicates that the regularization parameter selected from this procedure has a higher level of regularization when applied to a larger number of individuals in the training and validation sets.

[0324] In some embodiments, obtaining the trained computational model (e.g., using “training validation and refit” techniques) includes techniques to determine an optimal regularization parameter value. The regularization parameter can be the optimal λ value itself, the number of genetic variants and / or predictor variables included in the iPGS model, or information-theoretic metrics, such as Bayesian Information Criterion (BIC). A plurality of computational models may be trained, each having a different value of the regularization parameter, using a subset (e.g., only the training set of data) of the data describing phenotypes, genotypes, and / or phenotype-genotype relationships. Thereafter, a predictive performance of each of the first plurality of trained computational models may be evaluated, and an optimal regularization parameter value may be selected based on the predictive performance metrics. Thereafter, the trained computational model may be obtained by training a computational model using the optimal regularization parameter and a larger data set (e.g., the training set of data and the validation set of data).

[0325] Note that this procedure above does not use the held-out test set individuals in the training; hence, the held-out test set individuals can be used to evaluate the predictive performance of the iPGS models trained with and without the “training & validation refit” strategy.

[0326] To test the new “training & validation refit” strategy, iPGS models were generated with and without “training & validation refit.” Their predictive performance was compared using the individuals in the held-out test set. For the results described in this section, results for 413 traits, consisting of 364 quantitative traits and 49 binary traits, were considered. ResultsAttorney Docket No.: M0437.70168WO00 List of 413 Analyzed Traits

[0327] A total of 413 phenotypes was considered. The number of traits in each trait category is listed in Table 23. The full list of 413 traits analyzed is provided in Table 24. Average Improvements in Predictive Performance

[0328] For each trait, two types of iPGS models were prepared: (A) the original iPGS model without the “training & validation” refit strategy; and (B) A new iPGS model with the “training & validation” refit strategy.

[0329] The average improvements in predictive performance were evaluated. For quantitative traits, improvements were quantified in R2. For binary traits, traits with at least 200 case individuals in the held-out test set were focused on and the improvements in (AUROC - 0.5) were evaluated. For both analyses, Deming regression was applied which considers standard errors in predictive performance measures in both models. The results are shown in Table 25 and FIGs. 4- 6A through 4-10B.

[0330] This application of the “training & validation refit” approach to the UK Biobank resource indicates that this approach is effective in further improving the predictive performance of PGS models in non-European population groups while keeping the predictive performance for European individuals the same. When applying the “train & val refit” strategy, a larger number of individuals are included compared to the original iPGS training while imposing more regularization. These results indicate that the increased regularization in PGS training is helpful in improving the transferability of PGS models. TABLESTables 1 and 2. The number of unrelated individuals in UK Biobank analysis. Table 1. Population assignment. The number of training, validation, and test-set individuals across population groups is shown.Table 2. The number of unrelated individuals used in polygenic score training. The number of individuals used to train PGS models is shown. In the iPGS+refit in Afr models (models v and vi), ^^trainൌ 284,661 individuals were used to train the iPGS model, whereas a subset of ^^ ൌ 4,853 individuals were used in the population-specific refit model (material and methods). Abbreviations are as follows: WB, White British; NBW: non-British White; SA, South Asian; and Afr, African.tit .ar n )te e:nmwnmaogihsndece essa ng r esar ti nonot eec) rcaenxe atn s etaar ncoitO ma fere ttdn tec ss sawiltis tnoiit DIFFN mu fth hg neish re amy ob neopnetaal( lupyagtE(crmnyOF iEc ugt ci iemsrieh hger a gi pumewtmy daf yddooobbateemerm l mu om ceruem m Ktr e eao ufa l elnUhsit oh ecnctene hg y fdydy osbaohpor ehht tni irsBtm n,aedl -iaarnere ie%t obobsdsosatewftfmfmthgth afee nlel abe m m o.orT sia ucuhrpicr gineitt I g0mif nWi M ieydo oh ssaomlh ee ohretlaasetdpyd6an cHicHiS BWBW m WrfW waBarmIobetat :hrto adWeB ecn tx ret-t etacrh iosFt.iec N meerhr,hac ennef gitd ahnifayfwloformehgie ss e dy ybod datecyddaertuos sittiti iaer tf uc g haz,ylemri ni g m)t thagfatn boeerBarlbssoeabeemnaodbeTT sia uc cdrpinnaittyd IMgieyd ecroh sslaomlh eoh sslaasetep lohamheWin a tt cHtSiSoBB(WoBepW m WerfW maBarmIw ans t ti,)irhia wDI 5 102 9 0 1 2 5 6areb:trtt,mBi1afr849 0000010900101010101I4I5I2I21 3 3 3 3 3 3I2I2 2 2 2 2 2o DIunW. TNININININI I I I I II NININININININItstiialratsnoyroyrt yert yert yert y y y y y y y yertrtrtrtrtrtrtrtL t3,oyt it ge e e e e e e e. aiet m m m m m m m m m m m mer elobght vearcotpoopopopopopopopopopopoprororororororororororora etadb inb arhtn hthnthnthnthnthnthnthnthnthnthntnT c aATA A A A A A A A A A A Agel gemrmrR seL seR sa ee L sa eeflofafaf aer% mrffaerf % mrf-% mrf-t t-t.od.o.o %.tmat-t%tmpdpdpdp fagfagfagf ta-t tg fa af tatfafRtaf tatfafL afakfakfkg fgm m mss m mss n n nssmImImImIe e e e e er r r a r rmr a urururaL L L R L L L LA A A m AAA mT T Tm ge e)elmrmrgf aat) s)ttssf sagatfessathsa)telsaof afne oce oeecmnrsseelc(mgirsmfel (smssrsse (sa ee(sa eea emerfncncn epaer)tatm f-epaertm f e-t gatmrtf-ett gatmrtf-t teaf gat-ttafafhdageirp)td)tadfephgep)taff)t taghf)t taf)tafg ghg ghg g)ttaff tagfafng)tfmeaf)taf)tafneafaf kne k k(mIel(mIir(mIel(eLir(eLir(eLir(eLel(eLeLel( rcrmhgmhgmcrm m)tfnucrnunu ssAep rAir( rAir( rAep rArAel(rTeprTrTam 708090011121315 6 7 9 0 1 3 4 5 7 8 91 1 11 1 1 1 2 2 2 2 2 2 2 23 3 3131313131313131313131 1 1 1 1 12 2 2 2 2 2 2 2 23 3 3 3 3 3I I I I I2 2 2 2 2 2 2 2 2 2N N N N NINININININININININININI I III I I I I I I I I I I I I I INININIyrt y y y y y y yerte rte rte rte rte rte rt yert yert yert yrt yrt yrt yrt yrt yrt yrt yrtm m m m m me e e e e e e e eo om m m m m m m m m m m m mp popopopopopopopopopopopo o o o o oo o o o o o o o op p p p p pr r r r ro o o o o o o o o oh h h h hrhrhrhrhrhrhrhrhrhrhrhr r rt t t t t t t t t t t th h hn n n n n n n nt t t t t t tA A A A A A A AnAnAnAnAnAnAnAnAnAnAnAcetetny yi sbtir uspupnsiubpnietnuotierle .tts eityoc celicl uiocccoook rlhg cro . ot ora concol rng obcol yc c cal d congotorhhel te pte ot hhycp hl opo ihey elnlehp t ortnput mc m a.la a ttd tet a.lttmn n uisosLrE e nHoc e e ememyH MovMehMehrE i awlaPle aPMovl dPiwyu oLocMeNoEaB tn rual ral ralhtocll)eetc )et ninuocucucinlld hti tnt t)du t tn tneiocnu uo uo tntt si ydeusu nsiu ni oitec )etw nnu tiele tyw netyoc clcl uocco yc bolarrcgprpool okb r torobporobo ardo ycoitocrc talco oit cetih ih litgonot aatnc e c lgc lgtnol or ub teul telp bmetelubohycpo pohpeldhtn mec mecnamunao nao ecbdht irtetetna ormuetirtpmonrtu niso(e yRreu(oc e nHoc e r e l ememnHepMovMehMehoce yRre( siadlaPlePMhtl(oavlP sidyoLMeNosEaB00001020304050607080900010203040506000000000000000000010101 1 1 1 13 3 3 3 3 3 3 3 30 0 0 0 0I I I I I I I3 3 3 3 3 3 3 3N NININI I I I I I I I I II INININININININININININININIsysysysysysysysysysys s s s s s sa a a a a a a ay y y y y y ysssssss s s s sasasasasasasasasasa a asasasasasasasasasasasasas sdd d d d d d d d d d d a aoo o o o o o o o od d d d dolololololo o o o oooooooooooooooB B B B BlBlBlBlBlBlBlBlBlBlBlBlB040140140140147137137137137137137131575875 5 5 4 5 5 6 6 6 6887 7 7 1 1 1 3 3 3 388 8 8 3 3 3 1 1 1 12 2828 8 4 4 4 4 4 4 43 3 3232323232323232323. %l.c reretne %o attttut % %%ete v derrf acs%ac oscyco etli lhyh i%yty etph licoco yeetchp.eryc tehtpc gyc tehtgycmoo prt on h l loipls loucy noue s sai utcteitRe nn u nouuavta oluil oecitaellmcit hlguil ohluR cMrMc I r icitgHr icitLM NoE Bo e e me eHertnu etlloer r tnec yeccetot ettuldea aorcsccegaett eglih eg liheg legtyc egtycuciteeh etrytehtyesgtehtynyec at po atpoat ih at ol at ol er eps e ut colngilcol atgilcolcorn neeocr rtnec ni necpo necucine uci namnama uocith unce h ucp Me upe rNesporEesparBetpe crRetpeeulRMoeulvMom vmiI tecrargfii Htcerre gpii Hter080 0 0 0 0 0 0 0 0 0 019 0 1 2 4 5 6 7 8 9 001 2 2 2 2 2 2 2 2 2 33030 0 0 0 0 0 0 0 0 0I I3 3 3 3 3 3 3 3 3 3N NINI I I I I I I I II I INININININININININIsysysysysys s s s s s sa a a a ayay y y y y ys s s s s sa a a a a as s s s s ssss s s s sa a a a a a asas s s sdd d d d d d d ad a a aol ol ol ol ol ol ol ol od d do o o o o o o o ol ool ool ooB B B B B B B B B B BlBTable 4. The number of selected genetic variants. The number of genetic variants with non- zero coefficients is shown across four PGS models and the 60 traits analyzed in the study.Table 5. Comparison of predicted PGS in the held-out test set individuals. The Pearson’s correlation (R2) between the two predicted PGS values was evaluated across all unrelated held-out test individuals (“All”) as well as a subset of held-out test set individuals (WB: white British, NBW: non-British white, SA: South Asian, Afr: African, and Others). Reported is the median value across 60 traits.Table 6. Average improvements of PGS models over the WB-only models. The average improvements of PGS models over the WB-only model are reported as the regression coefficients (Estimate), their 95% confidence interval (Estimate_ci_l and Estimate_ci_u), and sigma statistic (scale factor) from Deming regression for each population group in UK Biobank.e- lE-E-E-E-E- - -ua 0v01.16982 4 2 8E4E1E6.1958.163.172.088.263.068.0,6 9 6 5 1 725]15 ,],],],],] , ],98 0845 7438 11 1521905 22976]000238727521649315178 - 00574307.11 321 121 22002213[ 1070 607100[.00.00[.00.00[.00.000. 0[.00[.00.00[.00.00.00.00[.023 8 7 R10. 2 101 0. 7 17 05 0. 7 1 6416 02 0.05 0. 8 23 07 0. 9 02 06 0. 4 1 502 0.00av 0.15 8.11 2.1 6.1 7.2 3.1 0.0 2.1,54],46 ],35 ],29 ],62 ],37 ],87 ],]223302 006 21656 7979476 652546237922 77607 2528 3790 2558 3205 055.000 0[.00.000 0[.00.000 0[.00.0000[.00.1000[.00.0000[.00.0000[.008.000[.024 R07 .0606 9 6 3 6 0407 0. 309 0. 30200. 50700. 6100. 0900. 1700. 97etmse0 0 0 0ifemec pfmecthaa ucnei ggntt tatHucnieith IhtgiMgfeilaeyd%of shydsanWric r ri ecrHiSeh BWoB Wobm n ns ytesriae efrepiefgenidthg t sagnithg mxe)I thtga efga do ssi ytn beaWmuccHmru cna ietiie yd dnM Be d eclmtic th S h i (Woro acric SoBBep h fW D8I49 00 I4NI5I2I 512I 112I 2329320INININI10NI00NI00 INI90 INI01yro opopopo o o o o goer yrt htor yrtor yrptor yrptor yrpor yrpor yrpor yrate hte hte hte ht te ht te hte htec nm Anm Anm m m mtmtm AnAnAnAnAnA- - - - - - - - - - - -,30]59 ,9 ]5 ,3 ]7 ,8 ]4 ,0 ]3 ,8 ]0 ,9 ]5 ,6 ]9 ,6 ]7 ,1 ]5 ,8 ]4 ,2 ]1307 95148 69193 73524 85760 99634 67383 16362 03080 46284 82216 6227903.000 00300 00400 00800 00110 00010 00600 00600 00200 00500 0040 000010[.0 .0[.0 .0[.0 .0[.0 .0[.0 .0[.0 .0[.0 .0[.0 .0[.0 .0[.0 .0[.0 .00[.0200. 22 0600. 23 0700. 86 0100. 09 900. 28 300. 15 704 0. 8001 0. 4904 0. 0203 0. 6001 0. 5400. 21yei0 0 0 0 0 0 0 0 0deselrs lloefo ef f f flo o o R o Ltata R -ta Rtaorbf-tsaaohyd etasaasa ba tar .d o y. R . L . . fh ddgdgdmdmgRfg ssf eerssfgLf moWbw mBte pomcmIwb pmIel pmIel pmIra pmIraeL%eL a gmef aLmeL% y )) )edser osafs-ts b mlcailecnel ecen )c )ntfs oeecebetan )ctnfee tdohydad gel th adl(admr th adl( tagfat ) tnth ah efg eirtrfhegt ga(-tira(f tn )tfaafmelo rhet aaa Btearepwf obepfgirep ge epafgire mr geecgir gesaf s geec el(W w mmI omo (l ImIfo ( pomImIafL roe(p Lsa gme sLa L rmep323 3 3 3 3 3 3 3 3 I1NI012I 2NI012I 5NI012I 6NI012I 7NI012I 8NI012I 9NI012I 0NI112I 1NI112I 23 NI112I 33 NI112I 5NI11opor yop yop yop yop yop yop yop yop yop yop yophrttoerrtomhterrtomhterrtomhterrtomhterrtomhterh rttoerh rttoerh rttoerh rttoerh rttoyerh rtten n n n n nmnmnm m m m m A A A A A A A AnAnAnAnA- - - - - - - - - - - -6.02 7.01 7.14 9.13 3.06 3.0 3.1 2.0 0.1 4.2 5.0 8.2,1 1 5 9 2 3 769]50 ,22 43]5 ,805]1 ,202]1 ,269]9 ,35]4 ,29]4 ,64]2 ,52]6 ,65]5 ,11]2 ,59]30405 42.000 074 23085 74008 516063 423005 247057 41363 64577 76720 51273 05977[.00.000[.00.000[.00.000[.00.000[.00.000[.00.000 0[.00.000 0[.00.000 0[.00.100 0[.00.000 00.00[.00[.0403 0. 1 0406020305020608020 60400 0. 106 0. 405 0. 703 0. 306 0. 407 0. 309 0. 405 0. 901 0. 205 0. 2060. 4tat t0- -fL -tg saf eL a a Rta Rtata Lta Le sLa ges fRf sfe s fLf sfe s kn %k snsa kneerfsy sc tonume rfsamL mr%mArsamA mrerfsamr%mrsamA m A A mrerfsa urt uA mTafrTmt uar-akof Ttafmue cLett)t af eefel r)ft -f te)g tth-t s teg t)tf-t s tg(telafat )ta eg ta -t sll)eth afgirafsa )th afat)t afelafsa)tfat f ss afsaeti ec yctnesLs af(snagsamr ecgrir (mr(ss mm reegir (mrne fcre (l(mr ssammfr eeelk(nunre kcrnu a kr mnm ue h de WoookuuocmeLm AepAaAr mfAepA m Arf Tep TrTr flbel(32I 63 NI112I 73 NI112I 93 NI112I 03 NI212I 13 NI212I 33 NI212I 43 NI212I 53 NI212I 73 NI212I 83 NI212I 90 NI213I 0NI00opor yop yop yop yopopopopopopopdshrttoerrtoerrtoerrtoyerrtoyerrtoyerrtoyerrtoyerrtor yrtor yrt oo yan mhtnmhtnmhtnmhtnmhtnmhtnmhtnmhtemhtemhtemlssA A A A A A A AnAnAnAB a- - - - - - - - -06.0 0 0905000 061 12 7.,1 6.1 3.1 9.3 6.1 7.8 2.8 9.815]30 ,18 52]7 ,875]6 ,337]9 ,797]6 ,119]8 ,515]4 ,272]7 ,37]20404 5.0 1004 102074 002045 000023 10081 40013 848008 34904700[.00.00[.00.00[.00.00[.00.000[.00.000[.00.000[.00.000[.00.000[.0202 3 3 1 1 1 6 5 0. 2 009 0. 1 008 0. 5 004 0. 8 007 0. 5 006 0. 2 000 0. 9 007 0. 6 0304 0.09tlg.couc .l ucolucolc .t tttetynu onocmicebHo ntoa%na su o na su rgo n na su rg.o nconrh sidht eleneleticmetir eMprvr epralmib epralmiboHocalMocehMo eh ctyreditaEtywluPoctal rcPecti oiti eraranir ni oidet nti lylctbeonol tar rtcg loat nue n lubaon lubotaroo lycoit h telt rctcrhtuogo nnntan ae csmu ae cslg ae cslgtnnlb le or ) ubtdet nu elyrce ee)mec mcrMupl uroMpr o um Mpr o emcdechtyirtiwal oc eta(HnoecHepovcocehocenhoc R res(iPdlP03I 00 13I 00 23I 00 33I 00 43I 00 I5 3I 0030 0 I03I 03I 0NI 0NI 0NI 0NI 0N0NI60NI70NI80NI90do soyladso sso yladso soyladso soyladso soyladso soyladso soyladso soyladso soylasB a Bsa Bsa Bsa Bsa Bsa Bsa Bsa Bsa- - - - - - - - - - - -9.11 5.25 4.15 9.13 9.1 1 0 1 0 1 0 01 5.4 2.1 7.8 1.1 1.2 9.1 1.,7,0609,4632]2 ,79]6 ,71]3 ,5] ] 08]6 ,17]0 ,06]9 ,78]2 ,31]4 ,10]0 ,45]1490407 95988 02353 72515 83676 13906 60702 73876 12744 64057 72065 11362.000 0[.00.000 0[.00.000 00.00 00.00 00.00 00.00 00.00 00.00 00.00 00.00 00.00[.00[.00[.00[.00[.00[.00[.00[.00[.00[.060702 35 41 5 3 6 4 1 0. 500 0. 2 004 0. 8 007 0. 80900. 9200. 4600. 7300. 7200. 3300. 6000. 5100. 98telet.tot 0ytpt 0lelo le.t tshthdpet tnuconuor ntu poot0nnl0 0iuiohp tot0 0 0onuhp%yecp o%orpl t%o inhi %pap v al iPd iwmy yco n ocu cscsomtneulis li os% Lc oMe eNlihoE lihaBcyLyc o eM NhoE hat teyentBeli li t eeelieli e elceomteotblit hytco tn tycthp t hptliht yco gattycgathp gat hpgal tihgatalmuloetuabidihp uoonpornuo ortnuoon nuopo nuhp ene on neortne on ne po nehvt() l rt wc o cucis c se P sim dyea occoc u c iscscL MeNoE BmyrLerpMep e rNeporEe apB rep03I 003I 003I 0030 0 0 0 0 0 0 0 I03I 03I 03I 03I 03I 0303030NI01NI11NI21NI31NI41NI51NI61NI81NI91 INI02 INI12 INI22do s s s s s s s s s soyladsoso yladsoso yladsoso yladsoso yladsoso yladsoo yladsoo yladsoo yladsoo yladso soyladso soylasB a B a B a B a B a Bsa Bsa Bsa Bsa Bsa Bsa Bsa- - - - - - -812.02306309101 406406304-1E.5-3E.6-6.9-6.6-6.3-2.2-,152100 36 ,E E E E E],] 2 ]3,7],5 ],5 ],1 ]02 4986 663210623 19442 3178024503.1 200 06[.00.1 200 06- [.00.1[ 70007[.00. 0 400 0850674029.00.100[.00.100[.00.100[.0609 9 0 1 1 1 0. 30700. 10800. 504-00. 50110. 80510. 10410. 307-E-3E- - - - -94 3E7E E E E.029.073.0618.0376.0226050703 1 2 4 5. ., , , , ,1 440] 3 ] 6 ] 7 ] 5 ],1 ],]760404 36609210473087103875 5936102.0 000 082[.00.0 200 0152085105400912045[.00.000[.00.000[.00.0000.000.00[.00[.00[.0101 3 4 1 0. 70500. 20700. 00700. 8 30200. 40100. 93 0300. 109olo uci% lu t ncol .lon de .lrout co.lcahtrceolhtrceoltteetcitety nuae ucvae reh vlame uc rfgi hgitta uc% iegi hgtta ucetynuRycecoRcMiteertyMpli secmI teertyHlcs te tryHilcsite orcegtell e e eaye tcnoelttnn yceemn c e edr tymut cntohigil rteyegthtc atgirteytctncurcueitoecae oluu ae Mclitoeerua ol tv Melhovmumcctol nltolitarhfgaics uceitcr hegaics ucuitocp Rerps IerHer pHer03I 00300000000000NI42 INI523INI623INI723INI823INI923INI03do soy dsy dsy dsdsdsdsl asosol asosol asoso ylasoso ylasoso ylasoso ylasB a B a B a B a B a B a BsaTable 8. Estimated SNP-based heritability in the white British population in UK Biobank. The estimated SNP-based heritability and its standard error (h2 and h2_SE) are shown across 60 anthropometric and hematological traits in UK Biobank.Tables 9A-9B. The number of population-specific variants in UK Biobank in PGS models. Using the empirical minor allele frequency (MAF) in white British and African populationsin UK Biobank, the population-specific genetic variants are counted using MAF ൌ 5% and1% as two MAF thresholds. The "UKB variants" denote all the genetic variants considered inthe study. The four columns under the PGS models correspond to the genetic variants selected for the specified PGS model in at least one of the 60 anthropometric and hematological traits. Table 9A. The minor allele frequency threshold of 5%Table 9B. The minor allele frequency threshold of 1%Table 10. The number of heterogeneous associations from GWAS meta-analysis. The number of genetic variants with heterogeneous GWAS associations is shown across the 60traits analyzed in the study. The number of genetic variants reaching the nominal ^^ value of5 ൈ 10ି଼ in the GWAS heterogeneity tests is shown in the "Without LD pruning,Heterogeneity p < 5e-8" column. The number of approximately LD-independent variants ( ^^ଶ^ 0.5 ) in each population is also shown.Table 11. The number of UK Biobank individuals analyzed in the study. The number of individuals in training, validation, and held-out test sets stratified by biological sex are shown.tia dr tea pk . lxeia f. lxeia f. lxeia f. lxeia f. lalalxeif.x ei .x ei,epaKnoa.h / eso.h / es9o.h / es0o.h / es9o. / es9o. f / eso. f / esmmUboL pd ac201pd ac704pd ac80 pd ac70hpd ac70hpd a hc 8pd ac9 an,)r e iB R Un.kwo = n.kwo = n.kw4o= n. w4o= n. w4o= n. w4o= n. w4o=tiebht n i Knahsdi / ?inahsdi / ?inahsdi / ?kinahsdi / ?kinahsdi / ?kinahsdi / ?kinahsdi / ?iT.G yss diP,tececutsm-)e ron neereenomha hsmerfeftn nti antsm mnihtidn ar PRB P PiuB APacurpicrdwaTePDSMPWicHizslDcy Idlaau e ode e endimrm,ol oollairruc csn navsieta(n etidan t tdaebt,deb,deet sacilert c ert raere erreerfefrtiy tfgoi ia r gresmniotsua gsmni ilotua gsmni nuspestsi mumu9o larT lu otduaaiseortdua ssyeortd a sua e erluacri picri9reotT P aerDp aer S p aerMp PWcHceb nhotitma010404017002708494furnot .r DeI I I IT IN2N97N048N0I 40N0I IN N oca cI I I I9I0 9I Its eihtflruere e e eurur r ryryrL.,eatos2tyr sssussussusst tse e1maneerkiaog ererererermomoelbntmirnarTet p_p_p_p_p_popoa roepboac dodoodood d r roo o hthtThsxeiBlblblobloblbnana979 63737 703 2 3 3 3 3 33 3333333333333333333thgiy y seste%a y sarygtdodomdosacidelthhggth afbesineilsabel ebe elm lfl ob oo.bfeo.tt I gii Meydoohmtorhf-to rhetaas at dp lo dpWfWfW wae hHS BWBa aBmmIwmI Rthg setf fie s )sI y ysMe dodo a ydo sasrocecy odec)tthahggnmBt g( tafatbess bm eeba ile mo naobna hgii yxhgni ye loa lo er lo re la ba de el de ir(e ttido ed e do crm feh tah -t hta sa te poh p gh S BniWB pWfWafW wBmmIwmIel050 1 1 3 3 3 3 3 3 3 I2IN52INI 1I 120 NI 0I 220 NI 0I 920 NI 9I 020 NI 0I 12I 22I 52I 62I 71NI 01NI 01NI 01NI 01NI 01yrt yrt yrt yret yryryryryryryre e etetetetetetetemomomomomomomom m m m pp p p po o o oo o op p p p p pro o o o o o o ohrhrhrhrhrhrhrhr r rt t t t t th h hn n n nt t t t ta a a ananananananana333333333333333 3 3 33 3 3 3 3 3 333333333gs s sem m slr rR sa ee L sa ee Rafaofaf% mrf-% mrf-%tmt.o.o. tatata Rt t tL a adpd df f f sasfafaf s f fsmp p gI LmI RmI LegLegL Rea gLmegLeg amrmrL LeL m A ARfofeofce ) o sna ) tee sa ee) sdt cfn hc) gn tfg rthe sa e sser)t eatatmtf-gigratm f- fet gatmtee ald ir(ad el(afne )taf )t taf(tsafnte) atf) alf(safne )taf )tp (epep gc hgh s c t t s c h hmgm IelmIrm amIraerLegpir (egLigr (ea gLmerLefpegl(efLegl(eamr rLm Aegpirmr g(Air (323232323232323232323 I8 I 9 I 0 I 1 I 2 I 3 I 5 I I I2INI 01NI 01NI 11NI 11NI 11NI 11NI 161NI 171NI 191NI 101NI 21yrt yrt yrt yrt yrt yrt yryryryryre e e e e etetetetetemom pomomomomomomomomomoopopopopopopopop p pr r r r r ro o oh h h h hr r r r rt t th h h h h hn nt t t t t t t ta ananananananananana33 3 3 3 3 3 2 3 3 2e serL seef-t%a eemrf %rt tf-tafRtmsatfa-tfa a a afLf f frss kamrmrmrsankuns knsP TA rur sa usabl L LopA m A ALA mT TmrT m A A A A es sser )t sfa eea eaernirefeesfsto-htg erafigr(tafatm f-)tf te magata-ta nntat e atnarpfafel(f tnf f ini ah en rtoopmsrseamrcr)tef)tef ssknecknknsmsula ps inniilolmrelmr a urreururablkl oh almepA m Ap (A(A mT p T Tm A ApAa sA A 323 3 3 3 3 3 00 01 2 3 I12I 32I 42I 52I 72I 82I 91I601I601I601I6NI 21NI 21NI 21NI 21NI 21NI 21NI 21NI300 0NI30 0N0I30 0N0I30 0yrt yret yryryryryreteteteteteehe e ech h hmom m m m m moc c cpo ioioioioporopoopoopoopoopob_b b bhr r r r r r dy _dy _dy _ ythnthnthnthnththtorotlsiorotlsiorotdlsiorotlsia a a ananana bmbmbmbmesB lC oo T ia P s T cp S BeraCCTrCR yGulA A D UC CG G nie atrfso erfsl npnoee oe C artptialtarrt nrimetnoa ot sivnitnnilppsni cber u t a yesaui elitcirilerclaaieatmma eoao eerto smcatusaulABAmaesDibUChCrC - CrpyC GlgreG 4605 6 01 601 60760860960070170270370473I 03I 01I 01I 02I 01I 01I 01I 01I 01I 00 0NI 0 0NI30 0NI30 0NI30 0NI30 0NI30 0NI30 0NI30 0NI30 0NI30 0ehe e e e e e e e e echchchch h h h h h ho o o ococococococ ci i i i io ob bi i i i i i_ _b_b_b_b_b_b b b bdyr dyr dyr dyr dyr dydy _dy _dy _dy _dyootlsiootsiootsiootsiootsiorotsiorotsiorotsiorotsiorotsiorotsibmlbmlbmlbmlbmlbmlbmlbmlbmlbmlbm731702131 7 4 4 0 92 9 2 2856949224963 2 3 3 2 2 2 3 292e etnoca r1C h etAL1 CApGsob DF Lopso BlitsH HGIDLiLhPHSBTeTPTniAnibtolloc nibeeet urinnoig) retocr1etesrto alie or1idr hpb tspmAL el -po s Gla otleabDo F LhH(HhcGIDpLio B taLhPHSos tTeToT 570 016 7 8 9 1 3 4 5 6 3I701I702I701 701 801 801 801 801 800N0I30 0N0I30 0N0I3I0 0N0I3I 00 0NI3I 00 0NI3I 00 0NI3I 00 0NI3I 00 0NI30 0ehehe e e e e e e ec chchchchchch h ho o o oc c ci io o o o o ob bibibibibibibi i_dy _dy _dy _dy _dy _dy _dy _bdy _ yb_ yorot ort ort ort ort ort ort ortdortd rtlsi olsi olsi o si o si o si o si o si o sioo sbmlbmlbmlibmbmbmlbmlbmlbmlbm33 3 3 3 3 3 2 2 3T Ls etA rD2 cyT Rc ycFRFRF P oktunGA ti SA GuTU V AGeGeGeN AeLocsed T ))-ire DL en Cen)nioitCm otardo)ecnyiAinlo niit ininuiblniniolltbycgietram t ra R ttiTSoaiR atR ta tatFer Fsy Fertsa- ninet mouu ebbotiohllktunuTU V AarGeC(GeC(GeC(yoCNrl lpA G Wecel(oc7808809805600702 00 06 06 0 013I 01I 03I 03I 03I705I703I805I803I0 0NI30 0NI30 0NI30 0NI30 0NI30 0NI30 0NI30 0NI30 0N0I00ehechehehehe e e etoc c c chchchchc nioioioioioioioio ubdybybyb_ ybi o_ _ _ _ yb_ yb_ yb_ yb_ yc_ortdortdortdortd rtd rtd rtd rtd rtdo si o si o si o sioo sioo sioo sioo siosol lmlbmlbmlo i obmb bmlbmlbmlbmlbmlbni % tbtir rol raclaniranietnth uotierle .ttsigoulubotcoluboyctc cal dsc l c l oditete ptem.eca nnmeauepr.nslaugpo nsaugo .rcht w. lele na lehtoe rmpee rmen yr tst t .ialale ltoal diHocH Moc vMoc hMoc hoc E d P PMv Pw ninonllitr r n r noiec )tet ntu te )letb ti e a a i a it d n oir et yc nolagrrtcgoalulubol blu olarto ycoit c cal o obitonett can se cs gcs g nol or ub tetlelpmetelubmecenmecnrauepm runlauepo nrmau o eeprmcbhnd t ir he yt e e nrtsditataaeormul etiart htsdiHocHepMocovMocehMocehoc Re(idwlPlPMht(ovlPidw 030303030 0 0 0 0 0 I N0 I 0 I 0 I 03I3I3I3I3I3II20NI30NI40NI500NI600NI700NI80N0I90N0I01N0I11tntnt t t t t t t tu un n n n n n n no ou u u u u u u uco o o o o o o o_c c c c c c c c cd_ _ _ _ _ _ _ _ _od d d d d d d d dooloooooooooooooooooblblblblblblblblblbetet%%eyll y% lltceotiihlcetiih%lychyhpp tctor tpio tho yhpp th c o piohoplumnoyun ntuu n nisn op o rsnmntu nisoscitLoocMoe ucNocouEocauBocyo eL% M NoEaBeR % tentyceli l uioectyeeeleieli e e tyeothyhcp hplico ga ty ga hp gahpgali ga co gaoooh htct t o thtltp t t r t t pn n orn npnunmn n ntyu o uuenunisn op econectuec niec oec ciecLocMocNocou sEocam ByrLeo rpMee rpNesporEesparBetperRep030 0 0 0 0 0 0 N0303I 03I 030 0 0 II I 03I 03I 03I 03I3I3II21NI31NI41NI51NI61NI81NI91NI02N0I12N0I22N0I42tntntntntntntntntnt tu u u u u un no o o ou u u u uc c c cococococococo_ _ _ _ _ _ _cd d d d_ _ _ _o o o odododododododo o o o o o ool l l l lo o o ob b b b blblblblblblb7373737373778 757 7878755555555527070 0 097 7 7696969696793 2 3 3797979 9313137137131516336363634141 1 12 2424243 3 3 323det er ete eyech eol ps.lrytuchtythtyog cog coucnoivtali ut. .l rhetli ul rhetlu tnelaellmcit cgitacitgitacit urovMecmIerar fHcser% Hcserocdet eryee et tet e tetchmerych ycg h yo ep ul u o ngoatg clsuo t lil rl nil r olcm itunelav a uocithettucechettuctnroellvMemit ccmIerargi afHcit rsere gi apHcit useroc0303030 0 I N0I6I 02NI7I 03I 03I 02NI82NI92NI03tntnt t tun n nououou uc co o_c c cd_d_d_ _ooo ododololo o ob blblblb-enecg setdnehiniaitfeg la 9oton532 9 75653wraot ht3a T 481,-ev nep681580 5 3 2 2376 611914169 403mcim )ye ro6,onte2nnglctfeg eis ens kila cdS afA A A A A A A A (ninx.9pA N N N N N N.ol,sGaate.pde ed maG N N0N rhtnaoDn 67h tcitfeecolafs s sV V) ) ) )nien nde n gnytTV TTs tos s segoiz,ystecu inla ecP P PVnV V V lt aqegisne n( ( ( A ,ests s s,sA A A ,sntntPn( rere P(P(P( reac rnefestn sapf uqaiaiai stnnhtht ststst htio yoesrarara aiO|3O|3 naniainaiO|3afseceleSxdeB se nov v vbv t .nul c gng g raci pcpr r r cpinini v n aiaa a a iagorMnporMpv v v norMpaii tiGcra d didw aeo v dcettcatatanicncncnitrna tna gnigngntnae I|HI|HrieriereI|Hv dana rhro g,sps-ief rid ucrurur tl R niR nitltltl R ni eert-t-t-a-TUdT a a a TteU de- - - U decnitidtcPEanefVsr ps P nin neiteini |sh |shninini |shoteoteotoVctVcteteotetoVcteresfe e ot ehrPrPrPrPCaPmCaPmroPrPrPCaPfepnhtcT m proeoiaf tn.oesrr ytde perr r r r r pe,c l ybsael rdet es azniettirolo o o o o ororo yt roellded d d d d d d ols e s e s e s e e e eelldemu bai ni a)bdp ogira pryt tnpayt tnpayt tnpayt tnpaytstnpst pst pst a pstaytn ytn ytn ytnN .raSnxepidpeo oneir oia ne r oia ne r oia ne r oia ne r oaia ne r oaia ne r oai Aoaia ne raLne ra3 v1rotGr e(uonroaspygvgvgvgvgvgvgvgvHgt yvl de yl de yl de yl de yl de yl dyl dyl d de yl delbcixe gepsten otrainc ttc ttc ttc ttc ttcett ett et tt etades-ree upe upe upe u e u e uce uce u up ce uTrpyb e r rr r r rprprprprprpwocavGiDmiiDmi i i i i iiDmiDmiDmiDmiDmiDmimiIDmiN N N 6 00 0719.50 1.17 0 58 247 83164311774431A A A N N N 2 00 09157.20 1.1 754477163174 117 744 631cinyelgeoki htLaptotno,ns,rs,sererh hethO|c 3 t3 t3ip O|cp O| pnaion a ciarMnporMp orM ta t tpnInana|HIHIHRTn|iRTn|iRTni U| dseVh U deU dec|shc|shcCtaVtaVtaPmCPmCPm rororodedpedytst pey st py stonnaei tnr on ai tnr oairgav eg a nveg ayvltdyl dydceetriu tcepet lriu tcepetriul patDmiDmiDmioTstnrei ociFf.dl0 a25 647 613 8 6 182965 9 5 6 5 28427147537196091515542 3 1 1fe edto 0291918 1 234 3 5 7 0 5 587574232oc u T 1 2 3 6 4 3 4 3 3 4 4 5 4olcrsenitcG_c e0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0z-toeff_Sitcn.s,tncste er ,e)I ssai cnedus ru ereMs araingetsae ss us ceB(eg sa ee mr etealvreg , rpgnier gns nie erc xat mrf- etrohotylrameociop ntiludodap drp ent ea ole dro aoerla eir fet rrusmue hef gitdnehnetataacil wgicsrf fw ey y y ofboderefoiam,e b dlb d e s crm h ies p dododo atecpvi nra tTn a crgil etoacielttarearpic ucrigni ha tgmth afbe be beemnadseutlwtiaesnlid tsmom aon etsttsota s iacpdnnityd gi yl ldo ossloladsepydccoendr uaeiu y ue uli atti oeoh h a ha moli T P rDa S aMPW HS S BWBW W m WB Ie yka bsflneoror 9-r b 79e07404ea hbentDI 297080005011 2 9 0 1 2 5 600000900101010101mui, tllei0a1040401028494050 1 1 3 3 3 3 3 3edorI I I I I I I I2I2I2I2I2I2I2I2I2ITNINININININININ N N N N NIN N N N nheht I I I I I I I I Irm ere e e eururururu yrt yrt yrt yrt yrt yrt yrt yrt yryryryrT.ofSI sssssssssse e e e e e e etetetete4n1SewElon yee rrepreprepreprmomomomomommm m m m m to _po o o o o o o_p p p p p p p p p p p pbhsGia g_ed_d_d d dororororororo o o o o oto o o o o h h h h h hrhrhrhrhrhrhae eTrarht Taocloblobloblobltbntantantant t t tanananant t t tanananana)t )thg)itf hg) ttf h)gth)t)f) t )tth fe )tfr(eli(r(el i(r( )girel( fel )gt irl(el e esgegmetlermrgh(agi sseg )a tf(sesgh(at gi sseag )a t tf(esgsaatssna atah e A B efl of af af tn re(at elsmne (a n r(n l( esmp ssmesmesmcr sa ea neo o o cs e so reinseie ar neibcencecere a eecrresa eecrresa eecrresa eeerp mrtf-thf tpsnotror fsnuril andancana ptmtf-tptmtf-tptmfata-tp mfatat -t tafafafni en e arp p eto o ta artibed d da a a a a af)t f f fafaf k kmi n opipitrotpepepef f f f f fh nkn nulainni loloani cemmmp gegegegegege mr gi mrrmrmrmrmr uA AruTruTr blklalA A Amp ppsaA A AmriI I ImI L L L L L LA(A A AT aD 00010203040 076 6 6 6 656660809 0 112131516 7191021 3 4 5 7 8 9 0 0 0 0 0 0 01 10 11 1 1 111 1 1 12 212 2 2121213030303 3 3 33231313 3 3 3 3 3 3 313 31313 3 3 3 0 0 000000000I2I23I2I2I2I2I2I2I2I2I2I2I2I2I2 2I2I2I1I1I1I1I2I1I1I I ININININININININI IN N NININININININININI NININININININIyrt yryryryryryryryryryryry y y y y y yetetetetetetetetetetetretretretretretretreteehehehehehehehmmmc c c c c c copoopom pommmmmm m mmmmmmmmopopopopopopopopopo o o o o o o iobiobiobiobiobiobibrororororororororororopropropropropropropror_dyr_dyr_dy _dy _dy _dy _dyhthnthnthnthnthnthnthnthnthnthnthnthnthnthnthnthnthnthtootlsiootlsiorotlsiorotlsiorotlsiorotlsiorotlsia a a a a a a a a a a a a a a a anana bmbmbmbmbmbmbmniebnisea oltrgoelf oorlropsnmraeeatsAnietn benmee eutnsivniCritc nt hlit alyes de)tco c1h e icrieted ot urraih liorbetsaei elt aarcltmmao a A1-p ps GloUaoaCheCr erC-sCy matc cuulyl b LDF LoC GlgG GH(HGIDpLioLhBattPHSosTeT0708090 0 0 0 0 0 0 0 0 0 0 0 0 06 6 60717273 4 5 6 7 8 9 1 384 50 0 0 0 0707070707070708 8 83030303003 3030303030303 3030303030 0 0 00 0 0 0 0 0 01 1 2 1010101010 0 0 0 0 0 0 0 0I I I I I I I1I1I1I2I1I1 1I1 1NININININININIININININININIININIINIINIehehehehehehehehehe e e e e e e ec c c c c c c c chchchchchch h ho o o o o o oc c cio o o o o o o o o obibibibibibi i i i i i i i i i i_dy _ y _ y _ y _ y _ yb_ yb_ yb_ yb_ yb_ yb_ yb_ yb_ yb_ yb_b_orotdlsiorotdlsiorotdlsiorotdlsiorotdlsiorotdlsiorotdlsiorotdlsiorotdsiorotdsiorotdsiorotdsiorotdsiorotdsiorotdysiorotdysiorotsibmbmbmbmbmbmbmbmlbmlbmlbmlbmlbmlbmlbmlbmlbmnoi))-ie netilgtearn uiCen on nir b lt in p olle nt tantu uner r rcalalaloll oc reu u unisit aee t it niG c c e) npcscscsntd Torir L aeape Drsye)m otdo)e cet niboittiupupniupni oitnA C CrCCnub niolty do ycl co olar rtcrorobor blo olartayit logietamot( ( ( itaala m be co olb rh gonoeta c e c cnmngng netT RFRFRFts -nub oitihku d ty mcnmaeulaoemaomcnTrr i STU V AGeGeGeyo l tCN AarWel(er eRe(HoecH MovMeehMehoc06070809050002000608 830830830660307307 730308080 0 0 0 0 0 0 00 03 3 3 0 1 2 3 4 5 60 0000000000000000 0 0 0 0 0 0 01 1 1 1 3 3 3 5 3050303030303030I I I I I I I I I I I3N N N N N NI I I I I II I I I I INININININININININININIeheche e e e e e e et t t t t t tochchchchchchchchn n n n n n nioioiocioioio o o o u u u u u u ub b b b b bibibibibococo o o o o_dy _ y _ _ _ _ _ _ _ _ _ _c_c_c_c_c_ortdortdyortdyrtdyrtdyrtdyrtdyrtdyrtdyrtd d d d d d do si o si o sioo sioo sioo sioo sioo sioo sioo siooooooo o o ol lmlbmlbmlbmlbmlbmlo o o obmb bmlbmlbmlblblblblblblbthettlelptode n el hy pppcoor ohphypcppooroh lol enp u u rps u ogiogion ntalu lllliteatlaale etalmntyou neisoosamntyoueisoosci atcitae.lae.lmcih utg cih utg citBeReR MovMovmi iwP PMP LM NE B LM NE IerHerHernoit e eue gba e eig e gaeetgrmtusln ti ooi tnt ne gatatgat etgne tlyactdvu unt tt crn ne ne atcrnu etleolr e r tn)t t)o n net ne bic u u nout yu te on eepcrcr cr nile tyrts ee ee eecpoyccc ueto dcitt cretaeptuaocc o it oyc clicl uioecty p plipl rieepteyt lyuecre ercsctetstetorcrtc ttalcohbdt ceotyhp hpliceotyhp hplicoci lolteh e h y h yeperu ng cog coty hetleelp ee nmleh h ctp oort ohnphpcoort ohnp ucu r scnmnmtaoii rtlhli ulhlueditataaeor(wlPlPMh ttadimnoueisos(lPwyL M NoEamnoueisoByL M Nosi Eati BetRe aeulaeulmcRMovMovmIar gicitgicitfHerHer070 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00809001112131415161819 0 1 2 4 5 6 7 8 9 0030 0 0 0 0 0 0 0 0 01020202020202020202030I3 3I3I3I3 3 3 3I3 3 3I3I3I3 3 3 3I3I3I3 3NIININININIINIINIININIINIININININIINIINIININININIINIINItntuntntntntntntntntntntntntntntntntntntntntnououououououououououou u u u u u u u u u uc_c c c c c c c c c cocococococococococococd_d_d_d_d_d_d_d_d_d_d_ _ _ _ _ _ _ _ _ _ _ooo o o o o o o o o odododododododododododolololololololololololo o o o o o o o o o ob b b b b b b b b b blblblblblblblblblblblbsSISISISISIS S Sd d rentachel t S S S SSI I Ie ESES S S S S SO GnE S S EnS EnS E EnEnS SP GPGPGn nPe e eGPGPeGPeGnPe e eGPGPofrp ge od m p d p en( hn edne a wl e)pntoatce uaool ot duo aoer lariefecrssgitiehbunibenicrfyfofoe itviatliam,taetbc a, lb dreetmu epamhgiri uge li rinlor sseaptdo ececncuprt(n ar gileor gciltaracri tatf)af ni hg btthib etsmafbenadaidoer pti ti,aesnlid tsuss nid otsmona cthdnnit cegilatotydyldo edpy epptar itraaTuPer i e a t epDrperySuaMimHr gAirm(r a t rAtSiS i eD WosTeToBohBWmdIobmInaorth(ca 970 0 0d hesao06 45e crf4 9 0 560 28080 1 9 0 6 7abeD nroo nIt290700800010 9123113 003000003103 0 9 0 0 00000103131313iftc swloihar 1I4NI4NI1NI4NI2NI2 5 2 1 2 1 1 2 2 2 2 2NININININI ININI I I I Iesl e T I I I I I I I I II NII I NININININIedsosi l) yryryremStdS sitsitsioGGPiyr ereMPi r og uresureuru yrsssssss t yert yert yert yertm eehyrctm eem hehyrt yert yert yert yerte. yoetererererm m mmmomcocommm m m 51lnSoISacp_p_p_po_poopo o o iopopopobo i_pobi_bo o o o o_popopopope- ba lEnta d dododorhd d dolr ei or r r r r r r r r raedo e ththththto hto o hththththTni lrololololmG( T b b b bnananananoalbnoaloblbnantanananagir(atsnel(at emn l(a nmec r- csre eef taacil hp sasserniniara crsesa eeecrrsesa er ser epamninboitpt riumfwosrss ucydyd ba oth epfesteotro erfsm p f-p mf-t to arc er o o e np p et ntmatfat tfafaf)ttatata afaf lgtnotrp icbe be menienartopop atartghf)tf f knkno ecaeegLegLe mLr gAistsilosslola lain onii ilrol a onirm(r fAel m(rmu umnml a h a h sk a o pArArTrT eHoc eHuPW W m WaBlAlAma pApAsAma970102030400 6 6 6652 5 6 1 3 461 1 1 2 2 252728202034 1 2 5 0 0 0 0 01 1 1 1 1 1 1 1 1 0 00001010130303 3 33 3 3 3 3 3 3 3 30 0 020 0 0 8 3 3 3 0 0 0 0 0I2I2I2I2I2 2 2 2 3 3 2 4 2 2 2 1 1 1 2 1NINININI I I I I I I I I I I IINININININININININININININII NII NII NII NIyrt y y ysrtrts rt yrtyrt y y y y y y y y y yisi sis ei im m mm m erte rte rte rte rte rte rte rte t t rus rtrt yrt yrt ehehehehehmmm m m mmmmnunuse e e e ec c c c copo o o o o o o o o rm m mmoioioioioiopropopopopopopoopoc_c popopopob b b b bhr r r r r r r r d_d_pdo o o o _d_ _ _ _ththththththththto o orhr r rththth ododododonanananananananoanalobloblt obnanananalobloblobloblbetoef oo ss mreasamninimearni itnilrop nraeatsA m tn e eniseerf T taats ta )meeneiviCrntlh elcied ) oer etetaerf e-reftd-orire-ta DLeArCyeCrCCnauteiseni ttcait a ya mmso etc ha1 c 1id orphpGtatf)tafpl cyfnil kot( ( (itarclloaer ert-sy maatu cuclylAb L- DF Lopso B ghg gatginm uatiT RSFRFRFtsUaChCC C CGlgGGH(HGIDLiLhPHSeLir(eLoTrTrTVAGeGeGeyC070 0 0 0 0 0 0 0 0 0 0 068 9 0 1 2 3 4 5 67 8 901030607090500020006060707070707070707 7 7 8 8 8 8 8 67 7 73 3 3 3 3 3 3 3 3 30303030303317 0 0 9 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11130302130303 3 30 0 0 0 0 0 0 0 0 0 0 00 0 01 1 20 0 0 3 3 0 0 3 0 0 0 0 0I I I1I1I1I1I1I1I1I1 2 1 1 1 2 2 1 1 2 1 3 3 3 5NNNNNN N N N NINININININI I INI I I I I I III I I I I I I I I I I I I I NINIINI NINININININIyrt ysrit ysrit ysrit ysrit ysrit ysryritsit ysryritsit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysryryryritsitsitsitsimememememememememememememememe yryrmmyrmmmm m hchchchchchchchchchchchch h hteteehehteeheheheheho o o o o oc c c c c c c c c ci i i i i ioioioioioioio o om mo omo o o o ob b b b b b b b b b b bibibibo o ibibo ibi i i i_d_ _ _ _ _ _ _ _ _ _ _ _ _ _popo _ _po _b_b_b_b_odododododod d d d d d d d d r r d d r d d d d do o o o o oooooooo o o o o o hthto o hto o o o ol l l l l l l l lolololololololololo o o ob b b b b b b b b b b b b b bnana b bna blblblblborpbonlle nu uliG cm docl oae claulaulu ltldoiv)tu u tc )b oecsc cn nu nut repuscusu no ec )ewtntel etirte utococ oc nu etuoto)blnaiol etd ty pnbyniorprinbpribit d tyno uotir et yc si y eli li oc yo co o o olo olartocoit ct ctal obdt cotyhp hplico-m e cnuob oom liti kb ru uet dhc e ctynamungcao ng nao e olcb rdhutbelelpmel hyirtetetna opcoor ohphpr et htmntu nisosm Nl tAahrWel b(laArUerRee l(Moem v Meem h MenhocerRes(iadla e adoPlPMht(lP iwyLMeNoEaByL060860008083030600830301040506070809000 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 11121314151680 0 0 0 0 0 0 0 0 01 13 5 30 0 0 0 0 0 0 0 0 0I I I1I1I3I3I3I3 3 3 3 3 3 3 3 3 3 3 3N NNNINI N N NINININININI I I I I I IIII I I I I I I I INININININININIyrt ysryritsit ysritsimem mm hech teheh t t t t t t t t t t t t t t toc nioiuc c nooioiunounounoun n n n n n n n n n nouou u u u u u u u u ub b c b b c c c c cocococococo o o o o_d_ _ _ _ _ _ _ _ _ _ _ _ _ _c_c_c_c_c_ododododod d d d d d d d d d d d d d do o o o oooooooooooo o o o o o o o o ol l l l l l l l l lolololololo o o o ob b b b b b b b b b b b b b blblblblblbS GSGSGSGSGSGSGSGSGSGSPP P P P P P PiP PG ii i i i i i i iPi.l.toloenv vtyu etllco r r t%oc yeccluoc e e nittttuo%%%etet lder teacs%acs cetli li%y y ucier eyhet teyt tycp hl oo po ichoco tnrtniploulci uehcrpsru htgilco hi nl gilcolanaa.h uch uoue s sa te t e emcgi itgicitM NoE BR eR M MmIarfHerHereegeag e gta antgaet entt eygaten tcercnegcat encrnu et lylceoclu rneetcr rettnuerpe rpe eep cr pocceetetoldcitae taouer te cs pcs cet li lip y ycieyhcp hoonrplicoco teh r eper thtegyt tchgyctoou n h leipsoulscucr e s e unmnmta noi itl oait hliu l ohluBe itRe aeulaeulmcgR MvMv I f icitgHr icitM NoEo o mareHer0900010204050607080 01 2 2 2 2 2 2 2 29200 0 0 0 0 0 0 0 0 033 3 3 3 3 3 3 3 3 30I I I I I I I I I I3N N N N N N N N N NII I I I I I I I I INItntntntntntntntntntntu u u u u u u u u uno o o o o o o o o ouc c c c c c c c c co_ _ _ _ _ _ _ _ _ _cd d d d d d d d d d_o o o o o o o o o odo o o o o o o o o ool l l l l l l l l lob b b b b b b b b blbrodnSIa.deSSItnES une E eGnoevitcceG idsi h tfoer)”erecp }SGheca hets _e,rci orwo2R“ ( ftehsesiSI ctm naar ttfesSEnnemroroiG ofrebtaderepm ileuah envw vit <eehhtstciid }S }T.narno t er I Sfpoit d orSG et En Pi02 00 0 cesrebte eG{_el abb{ 2esm us_2R ldeetnwoR dcoel ehsmesht e vis dSintatGPicr =iSIeso >deSEtrpns)}}eet”SI SeG t }hteu SSro-GP EG i{n Pi e{7 70_10222moehdl_ G2r f weh 2 {R s s tt eR _2li ht=R uartni >serreo d }hF St.er IaSfode p Emne SytcocG G2 0 9 9P9 9 8 777raeles si{_ imsS2ma RuSwG“( .lePieno61do ylcnni ata W rsre mo-mr lB u WBAS f eAhltb S raeofpoN O a GrTPinilep PSG2ni PeR-lll enba di ,x,xi at o d esn esnliram ed *e oitsC*e oitsCyas , 6 6 6e1,e 1,1a SIulle gaavp S c vP gaa P rr Ct,x t,xet,xaotE nd,Io2i^rp e ,ve py2ire a Peadedadedadedt^ pe pytBpht ni ht ni ht nirec gnpSol aecnee 0aG4aS 2 6 at m2 6 at m2 6 at m2SGPatmr rousoe rru5p s0o0w0aHJ 2giA) 00wo 0aL 2( 20e u 00 FH2( 20e u 00 FH2( 20e u 00 FH2(P.30 gi P. t 0é Jt 0é Jt 0é JtGeacfr S e lcPnl2Pnl enPvienPvenPvenehSe hsi Gaat02 GaateGrmeG irmeG irmel t nGP T d PT| e ( PT| eGPP|A GPP|A GPP|A G b aili d Pavei e el)5htpS9f Sn1,A59f A61f A65f A76f A65s|01, s|14,|34,|78,|34,afitm ma P(ac1i|y la6 1yln u1 |y la6 6u3 | syla9 2u8 | syla9 3u9 | syla9 2u8 | sylauceorSildi f SDI rfrttA|se di 11r 0tse di 70r 0tse di 70r 0tse di 80r 0tse di 70r 0tse dibu erdG eeP SAcvnid SnN AniSaPc cvinid Snr AniSaPc cvinid SnarAniSPc cvinid SnarAn S c cvinidnSnSac cvinidnpe nfw ii P rAi P rAiaosd 4 298- i8- i6- _gnntb.neoo otsi siw ae9 6 4I5ytr SI 83ih_ 5ytih_ 7ytiol)11NI91lig 91lig 12li _)crtadl ohusloacSSE00940 _ 0 bI0 E 0 a otl_)cr 0 bol )i0 at_ cr 0 b 2i0 atdreicapuleaifv nos E e SGGPNSI(GB Sro R c Sro R c Srorp_pPG( )GPp(LP_pGPp(LP_pG p d imv eoe hC.ec Tla P ( l h.siecnD 29235464171na )sd 2R- S a I )7 5 5 562Gmci M818 0 0 2el mro rrtPbofhrtelaiP otrf000101010r ePeM(A MPMPMPM M aTepM(ap PNP P PPPPP251.0:)r-laitrap(noita] 4 2le 380.00.rro1.00 : :Cl,2 2a2R-Rit1la-lar21itrita . arP 0[pap23000200402. .00 0.0:ti:t:tarec iaerec ia ecTn T ne rTndertererr f detef defompe pu ret omu ro muicr pe picr pe picrRHic RHic RHic.).l. 3l6a2t2275 ate7 l5ate0emu 000FH2( 40 erP40 erP Pét0Yvi|vJ e0P g x)0Yvi|)R22 0P g x 22GirmneGni oi0GniRo 0PP|A GPD| b2( PD|ib2(f A7|6f38A|4f84A|379,93| s8y l4 ,2sl48,3sl0 ra1u3 |ya3u7 |yau0tSn se di 0v1rtsd01rtsdi0 neivi 0 neiviSaPc cin d S arAniSPc cin d S arAniSPc cin drAni67 - _g 8-y cr 8-y cr1 yt2ilol35 ac ic35 aic0i _)0ba2cdri 3e_p3 ce_pc 00 d-ti00 d-iStre S h t horp_ n _Sn _GpihGo gP ( l )Go gPp(dlc oPc(ol )264 825203118 801010MPM M PPPPPTable 18. Phenome-wide association across UK Biobank traits for rs1260326. For a non- synonymous variant in GCKR (rs1260326), phenome-wide association summary statistics across UK Biobank traits were obtained from the Global Biobank Engine (Methods). For each phenotype (Code, Group, Name), effects sizes (BETA or log odds ratio), the corresponding 95% confidence intervals (L95OR and U95OR), number of individuals (number of individuals with non-missing values for continuous traits and number of case individuals for binary traits, shown in the n_case column) and statistical significance of the association shown in -log10(p-value) is show. Traits are sorted by the statistical significance of phenome-wide associations.era)r1o. ef0 vitse716 9 d2624723sis< atne eGtatyletua ar p lat onn ay g Totrenid AnTA nuEevol d2 6 51217 cc RiGnu st1 1iSI edaa,tnu-30302020202 2SE o)en oQ- - - - -0-0-nem me rrgepRlE1E6E7E2E8E9E4GSI amh ero y a 5 8 9 4 6 3 3eS N ci HDFV.1.7.2.4.5.4.8ht EmrnFrese etnnerisGe e eTdpy an sia nnoaithpsstecdnlofHko itk al4-soeco3eEsehtff na,(si,eu e cetum urar gneirk fun o4-hp ranni n-6pe gcmGil eSsxtcDIlamre tafNodh yyyeev lroito k oiic it u tces loeos NoInRiti xef mvr- qy mpt siht vobiivta etnuadleu lr udotcbat bir-4l ulqpGf e g roeoh it rit geio get oue euim re T etolfT pcaacca nforpern rrr b osSx,yaryot i p fmp C u cgn 3 0 3 3 2mrGethortilo e oe 4402177604088600ygwtnvodO(c ht siroD8 9 2 2 6 0 0I001 3 3 0 3 800:00 0 0 0 0:0:0:0:0 0oleotild fppmreet s smllr : :eeO O O T O O O O G G G G GG Gnodayafdo n ntntnehsagt osmoi il etnritoi ct sss scescesceecneonoc ws o a n n ororo p pntEntncir.aio t m uFu rrFrPlPlPl momoree odala a a aul ciciciCrCr9a hmo1v yr uycel c Ti. egotgecegogogoalaolololol l ulluioioilllb tenrenw aplluotM MB B BeCeC aTeoghs y g nO O O O O O O HerO G G G G G G Garctr e t.3faofknocuaneresi67791 4 1 3 Irt tsDmerree re2 3 3 353 67 3 elaiScbOyrD eeccnn igeot d r ct mu so t a A A A A A Ade ee actusirectnhmoehocseir la rrpe sssi ( ( ( ( ( (ss s s s s sfnaoeylpocoidop ensem toar eyle e e e e eammmm m m diisa edCsa irmhtb d i id ol o m lo ip nirnaococococococnr-TylesiaF nAoo B olBenbatSUetut t t t touououou uoclacDlBoBehtd d o oed d d dyrim o manienine e e eininininigehtdtim ar ser fefefefefefeco agledid-d d d d dy- - - - -yl Psll yll yll yll yll yllb Anso ycroaci aci aci acacactisa t girtaei i itm ratchm tihm tihm tihm tihm tihtirere6_2 e e e e 2e e ee e6 t riorgororororolglglglglglcncnfp y vitvitvitvitvitvitvitvi 2at2rftA A A A A AaCaC oytstryryryri ti ana a a atinininiiattiattiattiatiatiatiatio epLarB B B B nanantant t t t tananananasiyt_.0 T uQuQuQuQuQuQuQuL. es2Q12 nel eo yryryryryryryryrbalb psearnainainainainainainainiT aT # B B B B B B B Bses ses s s seseseseses s s s smeom memememmmmm memememememct ococt ococococt oct oct ocococococococutut t t t t t t t t touou u u uououououou uououourer re oe e oe oe oe e e e e e oe e e e oecencercercec sasaneas s s s s s s s s s s s s seaeaeaeaea a a a a a a a a aCan n ns CaCaCaCis Dis Dis Dis Dis Dies Dies Dies Dies Dies Dies Dies Dies Dies Dies DiD yryryryryry y y y y y y y y y y y y y y ya a a a ararararar r r r r r r r r r r rn n n n n n n n nanananana a a a a a a ai i i i i i i i i i i i ininininin n n nB B B B B B B B B B B B B B B B BiBiBiBiBses s s s s s s s s sme e e e e e e e e eomomomomomom m m mmy y y y y y y y ctcutctctctct oct oct oct oct octror r r r r r rououououououououououots oitotototo o ohsihsihsihsithsithsithsiese e e e e e e e e ehaseasasasasasasasas s ylylylylylylylyliea a i i i i i i i is siesiesiesiesiesiesieseses m m m m m m m m D D D D D D D DiDiDiDaFaFaFaFaFaFaFaFyryryryryryryry yryryryry y y y y y ya a a ar r r r r r r rn n n nananananana a a a a a a a a ai i i i i i i i inininininin n n n nB B B B B B B B B B B B B BiBiBiBiBiByr yrot yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yryrt yryryrt yrtts ehmeme e e e e e e e e e e e ete etete emei o omomomomomomomomomomommm m mm momylpiop p p p p p p p p p popopopopopopop porohrohrohrohrohrohrohrohrohro o o o o o o o o opohrhrhrhrhrhrhrhrhrhr rma tFntAntAntAntAntAntAntAnt t t t t t t t t t ththtAnAnAnAnAnAnAnAnAnAnAnAnAnA susususususus s s s s s s s s s s s s s so o o o o ououououououououououou u u u uu u u u u u u u u u uo o o o oy n n n n nu u u u u u u u u urainti i i ininininininininininin n n n n nintontont t t t t t t t t t t tititititititononononononononon n n n n n n n n nB C C C C C C C C C C CoCoCoCoCoCoCoCoCoCoCyrt ysrit ysrt ysrt ysrt ysrt ysrt ysrt ysrt ysrt ysrt ysrt ysrts yy y y y yi i i i i i i i i i i irtrtrtrtrtrt yrt yrt yrt yrmmmmmmmmmmmmm e me e e e e e e eteehehehehehehehehehehehehehomm m mm m m m mcpoopopopopopopopopopociocococococococococococororororororororororbidbi i i i i i ibi i i ibd bd bd bd bd bd b b b bhthnthnthnthnthnthnthnth hntntnooo o o o o ododododododolololololololololololo oA A A A A A A A A AB B B B B B B B B B BlBlB susususususus s s s s s s s s s s s s s s s sou u u u u u u u u u u u u u u u uuououououououououououououououo o o o o o o ont nt nt nt nt nit nit nit nit nu u u u u u u ui i i i i it nit nit nit nit nit nit nit nit nit ninininin n n n n n n n n n n n n nt t t to o o o o o o o on n n n n n n n nC C C C C C C C CoCoCoCoCoCoCoCoCoCoCoCoCoCoCyrt ysrit ysrit ysrit ysrit ysrit y ysrits rit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysritsimememememememememememememememmmmmmm hch h h h h h h h he e e e e e et t t toc c c c c c chchc chchc chchchchchchchc nioioioioioioioioioioioioioioioioioioioio unununub b b b b b b b b b b b b bi o o o ocd d d d d d d d d d b b b b b b b c c coo o o od d d d d d d d d d d d d d doo o o o o o o o o o o o o o o o o o o ololololololololololololololololo o o o o o o o oB B B B B B B B B B B B B B B BlBlBlBlBlBlBlBlBlB susususususususususususususususususususususus s so o o o o o o o o o o o o o o o ou u uu u u u u u u u u u u u uo o o o o o o ont nt nt nit nit nit nit nu u u u u u u u u u u ui i i it nit nit nit nit nit nit nit nit nit nit nit ninininininin n n n n n n n n n n n n nt t t t t to o o o o o o o on n n n n n n n n n nC C C C C C C C CoCoCoCoCoCoCoCoCoCoCoCoCoCoCoCoCtntntnt t t t t t t t t t t t t t t tu u unununununun n n n n n n n n n no o o o o o o ououououou uou u u u uc c c c c c c c c c c coc cococococodd d d d d d d d d coo od d d d d d d d do o oooooooooooo o o o o o o o o o ol l l l l l l lolololololo o o o o oB B B B B B B B B B B B BlBlBlBlBlBlB sususus s s s s s s s s s s s s s s so o ouououououou u u u u u u u u u uu u u u u u u uouououououo o o o o on n n n n n n n n n n n nunununununuititititi i i i i i i i i i i i i inin n n ntntntntntnt t t t t t t t t to o o o o o o o onononononon n n n nC C C C C C C C C C C C C CoCoCoCoCoCMM N N nondoel diefliofroprspet sieltiolboat beatem a m a msm al saPlPsusouuonuit nintonCoCM M M M M M M M M M M M M M M N N N N N N N N N N N N N N N nononononononon n non n n n ndd d d d d d od od d od od o o oelieflieflieflieflieflieflielielielieliel diel diel dieli oo o o o ofofofofof f f f fr r r r r r r r r rorororo op p p p p p pr rsp p p p p p p pet sielt siet seseseset set set seset set seseseoliltiltiltililililtilililtiltiltilbotbobo o o o o o o o o o o oae ate at b b b beate ate ate at beat beat b beate at b b b beate ate ate atem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msm m m m lsl sl sl sm lsm lsm lsm m m m m m m aa a a a a a al sal sal sal sal sal sal saP P P P P P P P P P P P P PlPsusususususus s s s s s s s so o o o o ouououou u u u u uu u u u u u u u uouououo o ont nit nit nit ninu u ui inininininininininin n n ntntntntnt t t t t t to o o o o o o onononon n n nC C C C C C C C C C CoCoCoCoCR R R R R R R R R R R R R R R R R R RM M M M M M M M M M M M M M M M M M M N N N N N N N N N N N N N N N N N N N nonononononononononononononononononondl difl difl diedeflifl diefl diefl diefl diefl d oee eiedededededededededeflifliflifliflifliflifliflifli oro o o o o o o o ofprproprprprpr r r r ropro o o o o o opr r r r r r rsetsiesesesespespespespesesespespespesp p p pesesesesltioltiltitiltiltiltiltiltilti tititi tititi tietiboat bolbo obobobobobolbolbolbolbolbol l l lbobobobobeate at beate ate ate ate ate ate ate ate ate ate ate ate ate ate ate atem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msm al sm asm m asasmm m m m mm m m mm m m m asasasas s s s s s s s s s sPlPlPlPlPlPlPlaPlaPlaPlaPlaPlaPla aPlPla a aPlPlPlPsususus s s s s s s s s s s s s s s so o ouououououou u u u u u u u u u uu u u u u u u uouououououo o o o o on n n n n n n n n n n n nunununununuititititi i i i i i i i i i i i i inin n n ntntntntntnt t t t t t t t t to o o o o o o o onononononon n n n nC C C C C C C C C C C C C CoCoCoCoCoCM M M M M M M M M MM M M MM N N N N N N N N N N N N N N N nonononononononon n n n n n ndd d d d d d d od od od od o o oelieflieflieflieflieflieflielielielieliel diel diel dielio o o o o ofofofofof f f f fr r r r r r r r r rorororo op p p p p p pr rsp p p p p p p pet sielt siet set set set set seseseseset seseseolilililililtiltiltiltililtiltiltilbotbobo o o o o o o o o o o oae ate at beat beat beat beat b b b b b b b beate ate ate ate ate ate ate atem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msm m m m ls s s smsmsmmmm msmmm aa a a a a asasasasas s sPlPlPlPlPlPlPlPlPlPlaPlaPla aPlPlPsusususususus s s s s s s s so o o o o ouououou u u u u uu u u u u u u u uouououo o ont nit nit nit ninu u ui inininininininininin n n ntntntntnt t t t t t to o o o o o o onononon n n nC C C C C C C C C C CoCoCoCoCM MM MM M M MM M M M N N N N N N N N N N N N nonon n n n n n n n n ndd od od od od od od od o o oelie eflilielielielielieliel diel diel dieli ofofofofofofofof f f fr r r r r ro o o op p p pr r r r r rsets s spspspspspspspspsieltieltietietietietietie e e eltitititio olololololo ololololbt b b b b b b b boae ate ate ate ate at b b beate ate ate ate ate atey y y y y y y symmmmmrtrtrtrtrtrtrt asa a a a am a m a m a m a m a m a m ae e e e e e esamsmmmmm m mmmm m m aom m m m m m rorororo o oelsasasasasasasaPlPlPlPlPlPlPl sasasasa i i i iriririni PlPlPlPlPpSpSpSpSpSpSpS rU susus s s s s s s s s s s s s s s s s so ou u u u u u u u u u u u u u u u u uuo o o o o o o o o o o o o o o o o onu u u u u u u u u u u u u u u u u u uit nininininininininininininininininininintontntntntntntntntntntntntntntntntntntnCoCoCoCoCoCoCoCoCoCoCoCoCoCoCoCoCoCoCoCfore9 4 1 8 6 0 2 9 5b0.1 1.1 2.1 1.a u1 0.1 0.1 0.1 0.1 0.1der c oetidi adesnr nTifeyr777 7 79 1 058 ocd-sy1717171 4 2 1 3 yrl ets myieola oyrrtmrgecictuoteeshtusim c nustamoho oi oerc htirey po b c pybro estg c snaelird d dsmht o o oilai n ololola ACDaFAB B BWs srt rere31Brfht BWB rfh 4NASA O W NASAtOfoetspiytte e C.iav ve e LritaitvaitvaitaOC C C C32 TROROROROReyry y ytiarararttinttinttintn2 R2R2R2R U U U U Ulb n n nanaA A A A AauauauauTiBiBiBiB Q Q Q Qrreecrcnedna rac osielisysesesesesesefo lam m m m m m rn o o o o o opa cseetcutc c c cututut ttihtolonbid oed od oudoudoane e e edetin n n n ney syder fie efi i i idefdef f fdededed s s s smrtaeassadi -sy- - - - - e e elylylylylylmmmem mm saore nyrlalcalcalcalcalcacoct oct oct octlinPpi SrocUsoi i i i i it giaetm m m m m mur a ht)ouououotc sht)sht)sht)sht)sht)sr r r r re e e e3_i1t rioOirOirOirOirire e e e ereresasasasaDoDoDoDoO DoO Dcncncncncncncnesesese4 arglgftAA( lgAA( lAA g(lAA g(lAA g(lAA(aCaCaCaCaCaCaCiDiDis DiD oets pee eiyvitv vt_ait tiaitttia Lt.i4e2snntantn eopyry y y y y y y y y y y y y yuauaulbsearnararararararararar r r r ryryrininin n n n n n nanananananananQ Q QaTr# B B BiBiBiBiBiBiBiBiBiBiBiBiBiBiBsesmesomesmesmesmesmesmes smemesmesmesmes s s s s s s s s smememememememememememct ocut oct oct oct oct oct oct oct oct oct oct oct ococococococococococyry ory ory oroououououououououououououtoutoutoutout t t t t tououououououotsithsitsitsiese e e e e e e e e e e e eh h haseaseaseaseaseaseaseasease eeaseaseaseas s seseseseseseses yeaeaeaeaeaeaeaeaea a liyliyliylisisisisisisis s s s s s s s s s s s s s seses mmmm D D D D D DiDiDiDiDiDiDiDiDiDiDiDiDiDiDiDiDiDaFaFaFaFyry y y y y y y y y y y y y y y y y y y y yararar r r r r r r r r r r r r r r r r r ryryryryryrn n nanananananananana a a a a a a a a a a a a a a ai i i i i i i i i i ininininininininin n n n n n n nB B B B B B B B B B B B B B B B B B BiBiBiBiBiBiBiBiByry oryryryr yrt yts oths oths oths oths e rt yh me rt yme rt yme rt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrt yrtmememememe e e e e e e e e e e e e e e ei i i i i o o o o o o o omomomomomomomomomomomomomomomomli li li li lp p p p p p p p p p p p p p poyy y y yi orororororororop p p p p p p p prorororororororororororororororomh h h h h h h h h h h h h h hram Fam Fam FamtFaF ntAntAntAntAntAntAntAntAntAntAntAnth h h h h h h h hAntAntAntAntAntAntAntAntAntAntAntAnA evetveveveveveveveveveve e e e e e e e e e e e eiai i i i i i i iv v v v v v v v v v v v vyt tat tat tat tat tat tat tat tait tait tait tait tait tait tait tait tait tait tait tait tait tait tait tatry y y yiarnararara tintintintintintintintintintintintintinti i i i i i i i i intntntntntntntntntntnin n n naBiBiBiBiBuaQuaQuaQuaQuaQuaQuaQuaQuaQuaQuaQuaQuaQua a a a a a a a a aQuQuQuQuQuQuQuQuQuQuQyrt ysrit ysrit ysrit ysrit ysryryryryryryryryryryryryryryryryryritsitsitsitsitsitsitsitststststststststststs yrt yrt yrt yrt yryryi i i i i i i i i i irmmmmmmmmmmmmm m mmmmmmmmm ee e eteteteehehehehehehehehehehehehehehehehehehehehe e emommmmmmcpoopopopopopopoc cio ococococococococococ cocococococochochochcrororororororbi ibibibibibibibibibiobi ibi i i i i i ioioitntntntnthnthntd b b b b b b b b b b bh h h h hnod d d d d d d d d d d d d d d d d d d d d doolooloooooooooooooooooooooooooooooooooooooo oBl l l l l l l l l l lBl l lBlo oA A A A A A AB B B B B B B B B B B B B BlBlBlBlBlBlBeveveveveveveveveveveveveve e e e e e e e e e e e e e e e eitait tiaitaitaitaitaitaitaitaitaitaitaitvaitvaitvaitvaitvaitvaitvaitvaitvaitvaitvaitvaitvaitvaitvaitvaitvaittti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti t t t t t t t tatntntntntntntntntntntntntntnt t t t t t tititititititititita a a a a a a a a an n n n n n n n n n n n n n n nu u u u u u ua a a a a a a a a a a a a a a a a a a aQ Q Q Q Q Q QuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQyrt ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysrit ysritsimem hemememememememememech h h h h h h h h h toc c c c c c c c c c ntntntntntntntntntntntntntnioioioio o o o o o o u u u u u u u u u u u u u ub b b bibibibibibibibococococococo o oco o o o odd d d d d d d d d d d c c c c c c coo o o o o od d d d d d d d d d d d do o o o o o oooooooooooooooooooooooooooooo o o ol l l l l l l l l l l lBlBlBlBlo o o oB B B B B B B B B B B BlBlBlBlBlBlBlBlBlBeve e e e e e e e e e e e e e e e e e e e e e e eitvaiviviviviviviviviviviviviv v v v v v v v v v vttiatt tiatt tiatt tiatt tiatt tiatt tiatt tiat tiat tiat tiat tiat tiait tiait tiait tiait tiait tiait tiait tiait tiait tiait tiait tiatin n n n n n ntntntntntntntntntntntntntntnt t t ta a a a a a a a a a a a a a an n n nu u u u u u u u u ua a a a a a a a a aQ Q Q Q Q Q Q Q Q QuQuQuQuQuQuQuQuQuQuQuQuQuQuQuQytiystinsene e dedtnt t run ntntntntntntntntntntntnt t turesuresuresursuslalrarou u u u u u u u u u u u unununusesesesese ecocococococococococo o o o o o o r r r r r neinid d d d d d d d d d cd cd cd cd cd cd cd pd pd pd pd pm moooloolooloolooloolooloo o o o o o o o o o o o odo elolololololololololololololo nenB B B B B B B B B B B B B B B B B B B B BlBoBoBeveveveveveveveveveveveveveveveveveveve e e e eitait t it it it it i i i i i i i i i i i i iviviviviviiattiattiattiattiatt tiatt tiatt tiatt tiatt tiatt tiatt tiatt tiatt tiatt tiat tiat tiat tiat tiat tiat tiat tiat tiattin n n n n n nt t t t t t t tnta an n n n n n n n n n n n n n n nu uauauauauauauauauauauaua a a a a a a a a a aQ Q Q Q Q Q Q Q Q Q Q Q QuQuQuQuQuQuQuQuQuQuQuQM M M M M MMMMMMMM M MMMMMMMMMMM N N N N N NNNNNNNN N NNNNNNNNNNN nononononononononononononononononononononononon ndl d d d d d d d d d d d d d o oeie e e e ed d d d d d d d d d dflifliflifliefliefliefliefliefliefliefliefliflielielielielielielielielielielieli yo o o o of f f f f f f f f f f ftr r r r rorororororororo oro o o o o o o o o o oispnspeetspspspspspspspspspsprspsprsprsprsprsprsprspr r r r rspspspspsps dieltietietietietietietietietietietietietietietetetetetetetetetet ololololololololololol l lolilililililililililill b b bo o o o o o o o o o o o oar beate at beate ate at beat beat beat beat beat beat beat beat beat beat beat beat beat beat beat beat beat beat beat beatneim a m a m a m a m a mmmmmmmm m mmmmmmmmmmm mea a a a a a a a a a a a a a a a a a a anm osm asm asm asm asm asm asm asm asm asm asm asm asm asm asm asm asm asm asmmmmmm asasasasasasaBlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPlPeveveveveveveveveveveveveveveveveveveveveveve e e eitait tiaitt tiaitt tiaitt tiaitt tiaitt tiaitt tiaitt tiaitt tiaitt t it it it it it it it it it it itvitvitvitvitiattiattiattiattiattiatiatiatiatiatiatiatiatiatiatiatin n n n nt t t t t t t t t t t ta a a a anananananananan n n n n n n n n n n n n nu u u u u u u u u u u uauauauauauauaua a a a a a aQ Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q QuQuQuQuQuQuQuQM MMM M M M M M M M M M MMMMM M MM N NNN N N N N N N N N N NNNNN N NN nonononononononononononononononon n n n ndl d d d d d d d o o o o oed d d d d d d d d d d d dieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieli oo o o o o o o o o of fr r r ro o o o o o o o o opr r rp p prprprprpr rpr r r r r rpr rspspspss s s s s s spsspspsp p psp petieltieltieltieltieltieltieltieltieltieltieltietieltietsietsietsietiet sieltsietio o o o o o o o o o olo olololololo ol lbt beate ate at bet bet bet beat beat beat beate at beate ate ate ate ate at bo oab ba a ab b b b b beat beat beatem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msmsmsm msmsmsmsmsmsmsm msmmmmm msmm alalaPl saPlaPlalalaPlaPlaPlasa asasasasasa asasaP P P PlPlPlPlPlPlPlPlPlPlPlPeveveveveveveveveveveveveveveve e e e e eitait tiaitt tiaitt tiaitt tiaitt tiaitt t it it it it it it it itvitvitvitvitvitvitiattiattiattiattiattiatiatiatiatiatiatiatiatiatiatin n nt t t t t t t t t t ta a ananananananan n n n n n n n n n n nu u u u u u u u uauauauauauaua a a a a aQ Q Q Q Q Q Q Q Q Q Q Q Q Q QuQuQuQuQuQuQMMMMMMM M M M M M M M M M MMM NNNNNNN N N N N N N N N N NNN nonononononononononononononon n n n ndel d d d d d d d o o o o oie ed d d d d d d d d d dfliefliefliefliefliefliflieflieflieflieflieflieflieflieflieflieflieli oro o o o o o o o of fpr r r r r r r r roro o o o o o o ospspspspsp p pspspsprsprsprsprsprspr r rsp p petieltieltietietsietsietietieltieltieltieltieltieltieltielt sieltsietsietio olololololo o o o o o o o o olol lbat beat beat beat beat beat beat beat beat beat beat beat beat beat beat beat boe at boe at beatem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msmsmmmmm msmsmsmsmsmsmsmsmsmmm alaPl saPl saPl saPl saPl saPlaPlala a a a a a asasasaP PlPlPlPlPlPlPlPlPlPlPeveveveveveveveveveveveveveve e e e eitait tiaitt tiaitt tiaitt tiaitt t it it it it it it it itvitvitvitvitvitiattiattiattiattiattiatiatiatiatiatiatiatiatiatin nt t t t t t t t t ta ananananananan n n n n n n n n n nu u u u u u u uauauauauaua a a a a aQ Q Q Q Q Q Q Q Q Q Q Q QuQuQuQuQuQuQRR R R R R R R R R R R R R R R R R R R RM M MMMMMMMMMMMMMMM M M M M N N NNNNNNNNNNNNNNN N N N N nononononononononononon n n n n n n n n ndd d d o o o o o o o o o oeliefliefliel diel diel diel diel diel diel diel diel diel diel diel diel diel diel diel diel dieli oofofofofofofofof f f f f f f f f f f frprspr r r r r r rororo o o o o o ororororetsp p p p p p p p prprprprprprp p p p pielt sietsietsietsietsietsietsietsietsietsietsietsietsietsietsietset siet siet sietiolbol l l l l l l l l l l l l lil lolololoat boe at boe at boe at boe at boe at boe at boe at bo o o o o o o oe at beat beat beat beat beat beat beat b beate at beat beat beatem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msm al smmmmmmmmmmmmmmm m al sal sal sal sal sal sal sal sal sal sal sal sal sal sal sal sm asm asm asaP P P P P P P P P P P P P P P P PlPlPlPlPeveveveveveveveveveve e e e e e e e e e eitait tiaitt t it it it it it it itvitvitvitvitvitvitvitvitvitvitvitiatiatiatiatiatiatiatiatiatiatiatiatiatiatatatatatatnt t t t t t t t t t t t t tititititititananananananan n n n n n n n n n n n n nu u u u u u uauauauauauauaua a a a a a aQ Q Q Q Q Q Q Q Q Q Q Q Q QuQuQuQuQuQuQuQN N N N N N N N N N N NNN N N nononononononon n n n n n n n ndl d d d d d d od od od od o o o o oeie e e e e e e e e ededed dedeflifliflifliflifliflilililililielilili oo o o o o ofofofofof f f f frprprprprprprprprpr rorororororst s sp p p p p p peieltielt sielt sielt sielt sielt s s s s s s s s sieltieltieltietietietietietietio o ol l l l l l lb b bobobobobo o o o o o o o oate ate ate ate ate ate at beat beat beat beat beat beat beat beat beatem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msm al sm al sm al sm asm asm asm asm asm asm asm asm asm m asasm asaP P PlPlPlPlPlPlPlPlPlPlPlPlPlPeveveveveveveveveveveveveveveveitait tiaitt tiaitt tiaitt tiaitt tiaitt tiaitt t it it it it it it itvitiattiattiattiatiatiatiatiatiatin n nt t t t t t ta a ananananan n n n n n n n nu u u u u u uauauauaua a a a aQ Q Q Q Q Q Q Q Q Q QuQuQuQuQuQI I I I I I I I I I I I I I I I I IR R R R R R R R R R R R R R R R R RM M M M M M M M M MMM M M M M M M N N N N N N N N N NNN N N N N N N nononononon n n n n n n n n n n n ndl d d d d od od od od od od o o o o o o oeie e e e e e e e e edededed dededefliflifliflifliflifliflifliflifliflilielililili oropro o o o o o o o o ofofofofofofosprprpr rprprpr r r r r r r r r retsielt siespesesesespesp p p pesesesespesp pesespespeseoltiltilti tiltiltiltiltititi tiltilti tiltiltilboat bo olo o o o olololo o olo o o oe at beat beat b beate at beat beat beat b b b beate ate ate at b b b b beate ate ate ate atem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msm m m m m m m m mmm m m m al sal sal sal sal sal sal sal sal sasasasasasm asm asm asaP P P P P P P P PlPlPlPlPlPlPlPlPlPeveveveve e e e e e e e e e e e e eitait t it itvitvitvitvitvitvitvitvitvitvitvitvitviviiattiattiattiattiattiattiattiattiattiattiattiattiattiattiatt tiatt tiattin n n n n n n n n n nta a a a an n n n n n nu u u u uauauauauauaua a a a a a aQ Q Q Q Q Q Q Q Q Q QuQuQuQuQuQuQuQM M M M M MMMMMM M MMM M M M M M M N N N N N NNNNNN N NNN N N N N N N nonononononononononononononononon n n n ndl d d d d d d d o o o o oed d d d d d d d d d d d dieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieli oo o o o o o o o o of fr r r ro o o o o o o o o opspet spsprspr r r r r r rspspspspspspspr r r rsp p p prspr rsp pr rsp psieltietietietietetetetetetet sietsetsetetiet setet setetolbolololioliolioli i iololololi i iololololioli iol li iol loat beat beat beat beat beat beat beat beat beat beat beat beat beat beat beat boe at beat boe at beat beatem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msm al sm al sm al sm al sm asmsmmmm msmmm msmsm msm msPlaPlaPl saPl saPl saPl sa asasasa a asa asa aP P P PlPlPlPlPlPlPlPlPlPlPlPeveveveveveveveveveveveveveveve e e e e eitait tiaitt tiaitt tiaitt tiaitt tiaitt t it it it it it it it itvitvitvitvitvitvitiattiattiattiattiattiatiatiatiatiatiatiatiatiatiatin n nt t t t t t t t t t ta a ananananananan n n n n n n n n n n nu u u u u u u u uauauauauauaua a a a a aQ Q Q Q Q Q Q Q Q Q Q Q Q Q QuQuQuQuQuQuQI I I I I I I I I I I I I I I I I IR R R R R R R R R R R R R R R R R RM M M M M MMM M M M M M M M M M M N N N N N NNN N N N N N N N N N N nononononon n n n n n n n n n n n ndl d d d d od od od od od od o o o o o o oeie e e e e e e e edededededededeflifliflifliflifliflifliflieflifliflililililili oropro o o o o o o o o ofofofofofofosprpr rpr r r rprpr rpr r r r r retsielt spiesesp p pesesesesespesespespespespespespeseoltilti tiltititi tiltilti tiltiltiltiltiltiltilboat bolo olololo o olo o o o o o o oe at beat b beate at b b b beate ate ate at beat b beate at beat beat b b b beate ate ate atem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msm m m m mmm m m m m m m m al sal sal sal sal sal sal sal sal sasasasasasm asm asm asaP P P P P P P P PlPlPlPlPlPlPlPlPlPeveveveve e e e e e e e e e e e e eitait t it itvitvitvitvitvitvitvitvitvitvitvitvitviviiattiattiattiattiattiattiattiattiattiattiattiattiattiattiatt tiatt tiattin n n n n n n n n n nta a a a an n n n n n nu u u u uauauauauauaua a a a a a aQ Q Q Q Q Q Q Q Q Q QuQuQuQuQuQuQuQM MMM M M M M M M M M M M M MMM M M N N N N N N N N N N N N NNN N N nnN nn n n n nN nnN oo o o o o o o o onononononon n n n ndl d d d d d d d o o o o oed d d d d d d d d d d dieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieflieli oo o o o o o o o o of fr r r ro o o o o o o o opr r rp prpr rpr rprprprpr r r r rspet spspss spsspsps s s sp p p pspsieltietietietietietet sietet sietetietetet setsetsetetetolbololololiololiololiololioliolioli i i iol l l lioloat beat beat beat beat beat beat beat beat beat beat beat beat beat boe at boe at boe at beat beat beatem a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a m a msmmm m al sal sasmsm msm msm msmsmsmsmmm msmsPl saPlaPlal saPlaPl saPlasa a a a asasasa a aP P PlPlPlPlPlPlPlPlPlPlPlPeveveveveveveveveveveveveveve e e e e eitait tiaitt tiaitt tiaitt tiaitt t it it it it it it it itvitvitvitvitvitvitiattiattiattiattiattiatiatiatiatiatiatiatiatiatiatin nt t t t t t t t t t ta ananananananan n n n n n n n n n n nu u u u u u u uauauauauauaua a a a a aQ Q Q Q Q Q Q Q Q Q Q Q Q QuQuQuQuQuQuQRR R R R R R R R RM M M M M M M M M M N N N N N N N N N N nonononon n n n n ndo o o o o oel diefl diel diel diel diel diel diel diel dieli ofofofof f f f f frprsprpropropropropro oprprpetsielt siet set set set set seseseoli i i i i ti ti tibolat bol l l l l l leat boe at boe at boe at bo o o oe at beat beat beat beatem a m a m a m a m a m a m a m a m a m a msm al sm m m m m m m m al sal sal sal sal sal sasasaP P P P P P PlPlPlPeveveveve e e e e eitait t it itvitvitvitvitvitvitiattiatiatiatiatiatiatatatnt t t t t titititananan n n n n n nu u uauauaua a a aQ Q Q Q Q QuQuQuQuQRR R R R R R R R R R R RM M M M M M M M M M M M M N N N N N N N N N N N N N nonononononon n n n n n ndd o o o o o o oeliefl diel diel diel diel diel diel diel diel diel diel dieli ofofofofof f f f f f f fr r r r rororororo o o opsp p p p p p p prprprprpet sielt sielt sielt sielt sielt sielt sielt siet siet siet siet sietio o ol l l l l lbt b bobobobobo o o o o oae ate ate ate ate ate at beat beat beat beat beat beatem a m a m a m a m a m a m a m a m a m a m a m a m a msmsm m m m m m m m m m m alal sal sal sal sal sal sal sal sal sasasaP P P P P P P P P PlPlPlPeveveveveveve e e e e e eitait t it it it itvitvitvitvitvitvitvitiattiatiatiatiatiatiatiatiatatatatnt t t t t t t tititititanananan n n n n n n n nu u u uauauauaua a a a aQ Q Q Q Q Q Q QuQuQuQuQuQN N N N N N N N N N N N N nononononon n n n n n n ndl d d d d od od od o o o o oeie e e e e e edededededflifliflifliflililililililieli oo o o ofofofof f f f fr r r r r r r rororororop p p p p p prst sp p p p p peielt sielt sielt sielt s s s s s s s sieltieltieltietietietietietio ol l l l l lb bobobobo o o o o o o oate ate ate ate at beat beat beat beat beat beat beat beatem a m a m a m a m a m a m a m a m a m a m a m a m a msmsmsmsmsmsmsmsmsmsmsmsm alala a a a a a a a a asaP PlPlPlPlPlPlPlPlPlPlPlPeveveveveveveveveveveveveitait tiaitt tiaitt tiaitt tiaitt t it it it it it itvitiattiattiattiatiatiatiatiatin nt t t t t ta ananananan n n n n n nu u u u u uauauaua a a aQ Q Q Q Q Q Q Q QuQuQuQuQRR R R R R R R R R R R RM M M M M M M M M M M M M N N N N N N N N N N N N N nonononononon n n n n n ndd o o o o o o oeliefl diel diel diel diel diel diel diel diel diel diel dieli ofofofofof f f f f f f fr r r r rororororo o o opsp p p p p p p prprprprpet sielt sielt sielt sielt sielt sielt sielt siet siet siet siet sietio o ol l l l l lbt b bobobobobo o o o o oae ate ate ate ate ate at beat beat beat beat beat beatem a m a m a m a m a m a m a m a m a m a m a m a m a msmsm m m m m m m m m m m alal sal sal sal sal sal sal sal sal sasasaP P P P P P P P P PlPlPlPeveveveveveve e e e e e eitait t it it it itvitvitvitvitvitvitvitiattiatiatiatiatiatiatiatiatatatatnt t t t t t t tititititanananan n n n n n n n nu u u uauauauaua a a a aQ Q Q Q Q Q Q QuQuQuQuQuQRR R R R R R R R R R RM M M M M M M M M M M M N N N N N N N N N N N N nonononononon n n n n ndd o o o o o oeliefl diel diel diel diel diel diel diel diel diel dieli ofofofofof f f f f f fr r r r rorororo o o opsp p p p p p prprprprpet sielt sielt sielt sielt sielt sielt sielt siet siet siet sietio o ol l l l lbt b bobobobobo o o o oae ate ate ate ate ate at beat beat beat beat beatem a m a m a m a m a m a m a m a m a m a m a m a msmsm m m m m m m m m m alal sal sal sal sal sal sal sal sasasaP P P P P P P P PlPlPlPeveveveveve e e e e e eitait t it it itvitvitvitvitvitvitvitiattiatiatiatiatiatiatiatatatatnt t t t t t tititititanananan n n n n n n nu u u uauauauaua a a aQ Q Q Q Q Q Q QuQuQuQuQN N N N N N N N N N nondononononononononoel diefl diefl diefl diefl diefl diefl diefl diefl dieli oo o o o of fr r r ro o o opsp p prprprprprprpet sieseseseses s s sltiltiltiltiltieltieltieltietio o o o o ol lbt b b bo o o oae ate ate at beat beat beat beat beat beateym m m m m m m m m mrt yert yert yert yrtaa a a a a a a a a e em mm m m s m al sm asm asm asm asm asm asm asm asaoriororom ror PlPlPlPlPlPlPlPlPlPpiSpiSpiSpiSpSeveveveveve e e e e e e e e eitait t it it itvitvitviviviviviviviviiattiattiattiattiattiatt tiatt tiatt tiat tiattiatatatat tittitittin n n n n n n ntntn ntnta a a a a a a a ana a ananu u u u u u u u u u u u uauaQ Q Q Q Q Q Q Q Q Q Q Q Q QuQ_1_1 9 3 667553 876 539VEV 3FEFene 797638 195 342 023ien2ri ni2rr44u1149 52062722 565 1 4 3 6092077985749456142363 1 2 3t a80630505050 “03hm242390810 2 50940tit5103834 4062 332827 515 10 880034003 3 i s00000I5I1I1 1 w ee_0e00.000.0 01.0 0. 000.0001.030.0 0.301.000.0N N NININcnsI I I I I a 17961 2 53178585mr 4042501543 44263208ofr et135968591 1ep a5 0 283 71 49453641785 3 13920emi vt009is .9. 00. 80. 0 80.3023.0. 5 10.5 2250.t1 0 1 1 1c e1 1 1 .11id 464 4 4 49 1er 363636363 4 201 583pnists itar nte_ynrt y srtysayssaysasmee esmomasre as vo ne aerpoi ior pininini SpS rmtal srsrUrUri UegupWaroBreh Wrehp WBNAS fAt BO WBNAS fAtO eeve e e e vitv v v vait tiaitaitaita A.ttittit tit tit 52ciC C C C Cnan n n n euauauauaulrbteO2 2 2 2ROUROUROUROURU Q Q Q Q QaTm# R R R R2R A A A A AAttorney Docket No.: M0437.70168WO00 LISTING OF SOME EXAMPLES AND EQUIVALENTS

[0331] Example 1. A method of predicting a medical trait in an individual, the method comprising: obtaining genomic data including information indicative of genetic variants present in a genome of the individual; calculating, using an at least one processor and a trained computational model, at least one trait score based on the information indicative of the genetic variants, wherein the trained computational model was obtained by training a first computational model using data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries; and generating, using the at least one processor, a graphical user interface including a visualization of the at least one trait score, the at least one trait score being indicative of a presence of the medical trait in the individual; and displaying, using a display device, the generated graphical user interface.

[0332] Example 2. The method of example 1, further comprising altering a course of medical treatment related to the medical trait based on the at least one trait score.

[0333] Example 3. The method of example 1 or 2, wherein altering the course of medical treatment comprises reducing a number of medical interventions, reducing a frequency of medical interventions, and / or selecting a different type of medical intervention based on a value of the at least one trait score.

[0334] Example 4. The method of any one of examples 1-3, wherein altering the course of medical treatment is based on the value of the at least one trait score relative to a threshold value.

[0335] Example 5. The method of any one of examples 1-4, wherein altering the course of medical treatment comprises increasing a number of medical interventions, increasing a frequency of medical interventions, and / or selecting a different type of medical intervention based on the at least one trait score.

[0336] Example 6. The method of any one of examples 1-5, wherein altering the course of medical treatment is based on a value of the at least one trait score relative to a threshold value.

[0337] Example 7. The method of any one of examples 1-6, further comprising identifying at least one therapeutic agent based on a value of the at least one trait score.

[0338] Example 8. The method of any one of examples 1-7, further comprising administering the at least one therapeutic agent to the individual based on the value of the at least one trait score.Attorney Docket No.: M0437.70168WO00

[0339] Example 9. The method of any one of examples 1-8, further comprising altering a recruitment strategy for a clinical trial related to the medical trait based on the at least one trait score.

[0340] Example 10. The method of any one of examples 1-9, wherein altering the recruitment strategy for the clinical trial comprises removing the individual from enrollment in the clinical trial based on a value of the at least one trait score.

[0341] Example 11. The method of any one of examples 1-10, wherein altering the recruitment strategy for the clinical trial is based on the value of the at least one trait score relative to a threshold value.

[0342] Example 12. The method of any one of examples 1-11, wherein altering the recruitment strategy for the clinical trial comprises including the individual in enrollment in the clinical trial based on a value of the at least one trait score.

[0343] Example 13. The method of any one of examples 1-12, wherein altering the recruitment strategy for the clinical trial is based on the value of the at least one trait score relative to a threshold value.

[0344] Example 14. The method of any one of examples 1-13, wherein training the first computational model comprises using batch screening iterative lasso (BASIL) regression.

[0345] Example 15. The method of any one of examples 1-14, wherein training the first computational model comprises performing a regression including a loss, the loss comprising a squared loss and / or a binomial loss.

[0346] Example 16. The method of any one of examples 1-15, wherein obtaining the trained computational model further comprises using the trained first computational model and performing a regression including a penalty applied to genetic variants with heterogeneous associations between single-ancestry populations.

[0347] Example 17. The method of any one of examples 1-16, wherein performing the regression further comprises including effects of a global ancestry of the individual on genetic variants with heterogeneous associations.

[0348] Example 18. The method of any one of examples 1-17, wherein including the effects of the global ancestry of the individual comprises including, for each instance of genomic data in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships, an interaction described by a number of copies of each genetic variant with heterogeneousAttorney Docket No.: M0437.70168WO00 associations in each instance of genomic data and principal components of a genomic matrix generated using each instance of genomic data.

[0349] Example 19. The method of any one of examples 1-18, wherein performing the regression comprises performing Elastic Net regression.

[0350] Example 20. The method of any one of examples 1-19, wherein training the first computational model comprises performing a regression including at least one penalty factor determined based on one or more biological priors.

[0351] Example 21. The method of any one of examples 1-20, wherein the one or more biological priors include one or more of the following: variant pathogenicity, predicted variant consequences, known causal variants, tissue-specific regulatory genomic annotations, cell-type- specific regulatory genomic annotations, and / or aggregation of one or more of the preceding effects.

[0352] Example 22. The method of any one of examples 1-21, further comprising determining the genetic variants with heterogeneous associations between single-ancestry populations by determining genetic variants associated with the medical trait for two or more single-ancestry populations using inverse-variance weighted meta-analysis of genome-wide association studies (GWAS) for each of the two or more single-ancestry populations.

[0353] Example 23. The method of any one of examples 1-22, wherein determining the genetic variants with heterogeneous associations further comprises using Cochran’s Q test to determine genetic variants associated with ^-values smaller than a threshold value.

[0354] Example 24. The method of any one of examples 1-23, wherein calculating the at least one trait score further comprises using information indicative of one or more conventional risk factors and / or genetic variants associated with the individual.

[0355] Example 25. The method of any one of examples 1-24, wherein obtaining the trained computational model further comprises performing a regression including effects on phenotypes in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships of the one or more conventional risk factors and / or the genetic variants associated with the individual.

[0356] Example 26. The method of any one of examples 1-25, wherein obtaining the genomic data comprises obtaining genotyping data previously obtained by genotyping a biological sample obtained from the individual.Attorney Docket No.: M0437.70168WO00

[0357] Example 27. The method of any one of examples 1-26, wherein obtaining the genomic data comprises obtaining genotyping data by sequencing a biological sample obtained from the individual.

[0358] Example 28. The method of any one of examples 1-27, wherein obtaining the genotyping data comprises obtaining microarray data, whole-genome sequencing data, whole- exome sequencing data, and / or genotype imputation from partially observed data.

[0359] Example 29. The method of any one of examples 1-28, wherein training the first computational model further comprises using one or more of a linear genetic effect, a genetic dominance effect, and / or a sex-based genetic effect associated with the medical trait.

[0360] Example 30. The method of any one of examples 1-29, wherein obtaining the trained computational model further comprises: determining an optimal regularization parameter value by: training a first plurality of computational models using a plurality of different regularization parameter values and a subset of the data describing phenotypes, genotypes, and / or phenotype- genotype relationships; evaluating a predictive performance of each of the first plurality of trained computational models; and selecting the optimal regularization parameter value based on the evaluated predictive performance of each of the first plurality of trained computational models.

[0361] Example 31. The method of any one of examples 1-30, wherein obtaining the trained computational model further comprises training the first computational model using the optimal regularization parameter value.

[0362] Example 32. A system, comprising: at least one computer hardware processor; and at least one non-transitory computer readable medium storing processor-executable instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of any one of examples 1 or 14 to 31.

[0363] Example 33. At least one non-transitory computer readable medium storing processor- executable instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of any one of examples 1 or 14 to 31.

[0364] Example 34. A method of performing a clinical trial, comprising: obtaining, for a first individual, a first trait score associated with a medical trait by: calculating the first trait score using an at least one processor, a trained computational model, and fi...

Claims

Attorney Docket No.: M0437.70168WO00 CLAIMS 1. A method of predicting a medical trait in an individual, the method comprising: obtaining genomic data including information indicative of genetic variants present in a genome of the individual; calculating, using an at least one processor and a trained computational model, at least one trait score based on the information indicative of the genetic variants, wherein the trained computational model was obtained by training a first computational model using data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries; and generating, using the at least one processor, a graphical user interface including a visualization of the at least one trait score, the at least one trait score being indicative of a presence of the medical trait in the individual; and displaying, using a display device, the generated graphical user interface.

2. The method of claim 1, further comprising altering a course of medical treatment related to the medical trait based on the at least one trait score.

3. The method of claim 2, wherein altering the course of medical treatment comprises reducing a number of medical interventions, reducing a frequency of medical interventions, and / or selecting a different type of medical intervention based on a value of the at least one trait score.

4. The method of claim 3, wherein altering the course of medical treatment is based on the value of the at least one trait score relative to a threshold value.

5. The method of claim 2, wherein altering the course of medical treatment comprises increasing a number of medical interventions, increasing a frequency of medical interventions, and / or selecting a different type of medical intervention based on the at least one trait score.Attorney Docket No.: M0437.70168WO00 6. The method of claim 5, wherein altering the course of medical treatment is based on a value of the at least one trait score relative to a threshold value.

7. The method of claim 1, further comprising identifying at least one therapeutic agent based on a value of the at least one trait score.

8. The method of claim 7, further comprising administering the at least one therapeutic agent to the individual based on the value of the at least one trait score.

9. The method of claim 1, further comprising altering a recruitment strategy for a clinical trial related to the medical trait based on the at least one trait score.

10. The method of claim 9, wherein altering the recruitment strategy for the clinical trial comprises removing the individual from enrollment in the clinical trial based on a value of the at least one trait score.

11. The method of claim 10, wherein altering the recruitment strategy for the clinical trial is based on the value of the at least one trait score relative to a threshold value.

12. The method of claim 9, wherein altering the recruitment strategy for the clinical trial comprises including the individual in enrollment in the clinical trial based on a value of the at least one trait score.

13. The method of claim 12, wherein altering the recruitment strategy for the clinical trial is based on the value of the at least one trait score relative to a threshold value.

14. The method of claim 1, wherein training the first computational model comprises using batch screening iterative lasso (BASIL) regression.Attorney Docket No.: M0437.70168WO00 15. The method of claim 1, wherein training the first computational model comprises performing a regression including a loss, the loss comprising a squared loss and / or a binomial loss.

16. The method of claim 1, wherein obtaining the trained computational model further comprises using the trained first computational model and performing a regression including a penalty applied to genetic variants with heterogeneous associations between single-ancestry populations.

17. The method of claim 16, wherein performing the regression further comprises including effects of a global ancestry of the individual on genetic variants with heterogeneous associations.

18. The method of claim 17, wherein including the effects of the global ancestry of the individual comprises including, for each instance of genomic data in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships, an interaction described by a number of copies of each genetic variant with heterogeneous associations in each instance of genomic data and principal components of a genomic matrix generated using each instance of genomic data.

19. The method of claim 16, wherein performing the regression comprises performing Elastic Net regression.

20. The method of claim 1, wherein training the first computational model comprises performing a regression including at least one penalty factor determined based on one or more biological priors.

21. The method of claim 20, wherein the one or more biological priors include one or more of the following: variant pathogenicity, predicted variant consequences, known causal variants, tissue-specific regulatory genomic annotations, cell-type-specific regulatory genomic annotations, and / or aggregation of one or more of the preceding effects.Attorney Docket No.: M0437.70168WO00 22. The method of claim 1, further comprising determining the genetic variants with heterogeneous associations between single-ancestry populations by: determining genetic variants associated with the medical trait for two or more single- ancestry populations using inverse-variance weighted meta-analysis of genome-wide association studies (GWAS) for each of the two or more single-ancestry populations.

23. The method of claim 22, wherein determining the genetic variants with heterogeneous associations further comprises using Cochran’s Q test to determine genetic variants associated with ^-values smaller than a threshold value.

24. The method of claim 1, wherein calculating the at least one trait score further comprises using information indicative of one or more conventional risk factors and / or genetic variants associated with the individual.

25. The method of claim 24, wherein obtaining the trained computational model further comprises performing a regression including effects on phenotypes in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships of the one or more conventional risk factors and / or the genetic variants associated with the individual.

26. The method of claim 1, wherein obtaining the genomic data comprises obtaining genotyping data previously obtained by genotyping a biological sample obtained from the individual.

27. The method of claim 1, wherein obtaining the genomic data comprises obtaining genotyping data by sequencing a biological sample obtained from the individual.

28. The method of claims 26 or 27, wherein obtaining the genotyping data comprises obtaining microarray data, whole-genome sequencing data, whole-exome sequencing data, and / or genotype imputation from partially observed data.Attorney Docket No.: M0437.70168WO00 29. The method of claim 1, wherein training the first computational model further comprises using one or more of a linear genetic effect, a genetic dominance effect, and / or a sex-based genetic effect associated with the medical trait.

30. The method of claim 1, wherein obtaining the trained computational model further comprises: determining an optimal regularization parameter value by: training a first plurality of computational models using a plurality of different regularization parameter values and a subset of the data describing phenotypes, genotypes, and / or phenotype-genotype relationships; evaluating a predictive performance of each of the first plurality of trained computational models; and selecting the optimal regularization parameter value based on the evaluated predictive performance of each of the first plurality of trained computational models.

31. The method of claim 30, wherein obtaining the trained computational model further comprises training the first computational model using the optimal regularization parameter value.

32. A system, comprising: at least one computer hardware processor; and at least one non-transitory computer readable medium storing processor-executable instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of any one of claims 1 or 14 to 31.

33. At least one non-transitory computer readable medium storing processor-executable instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of any one of claims 1 or 14 to 31.

34. A method of performing a clinical trial, comprising:Attorney Docket No.: M0437.70168WO00 obtaining, for a first individual, a first trait score associated with a medical trait by: calculating the first trait score using an at least one processor, a trained computational model, and first genomic data including information indicative of genetic variants present in a genome of the first individual, wherein the trained computational model was obtained by training a first computational model using data describing phenotypes, genotypes, and / or phenotype-genotype relationships for a population including admixed individuals of multiple ancestries; enrolling the first individual in the clinical trial based on a value of the first trait score; and altering a course of medical treatment for the first individual in accordance with the clinical trial.

35. The method of claim 34, wherein enrolling the first individual in the clinical trial based on the value of the first trait score comprises enrolling the first individual in the clinical trial based on the value of the first trait score relative to a threshold value.

36. The method of claim 34, further comprising: obtaining, for a second individual, a second trait score associated with the medical trait by calculating the second trait score using an at least one processor, the trained computational model, and second genomic data including information indicative of genetic variants present in a genome of the second individual; and declining to enroll the second individual in the clinical trial based on a value of the second trait score.

37. The method of claim 36, wherein declining to enroll the second individual in the clinical trial based on the value of the second trait score comprises declining to enroll the second individual in the clinical trial based on the value of the second trait score relative to a threshold value.

38. The method of claim 34, wherein training the first computational model comprises using batch screening iterative lasso (BASIL) regression.Attorney Docket No.: M0437.70168WO00 39. The method of claim 34, wherein training the first computational model comprises performing a regression including a loss, the loss comprising a squared loss and / or a binomial loss.

40. The method of claim 34, wherein obtaining the trained computational model further comprises using the trained first computational model and performing a regression including a penalty applied to genetic variants with heterogeneous associations between single-ancestry populations.

41. The method of claim 40, wherein performing the regression further comprises including effects of a global ancestry of the first individual on genetic variants with heterogeneous associations.

42. The method of claim 41, wherein including the effects of the global ancestry of the first individual comprises including, for each instance of genomic data in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships, an interaction described by a number of copies of each genetic variant with heterogeneous associations in each instance of genomic data and principal components of a genomic matrix generated using each instance of genomic data.

43. The method of claim 40, wherein performing the regression comprises performing Elastic Net regression.

44. The method of claim 34, wherein training the first computational model comprises performing a regression including at least one penalty factor determined based on one or more biological priors.

45. The method of claim 44, wherein the one or more biological priors include one or more of the following: variant pathogenicity, predicted variant consequences, known causal variants, tissue-specific regulatory genomic annotations, cell-type-specific regulatory genomic annotations, and / or aggregation of one or more of the preceding effects.Attorney Docket No.: M0437.70168WO00 46. The method of claim 34, further comprising determining the genetic variants with heterogeneous associations between single-ancestry populations by: determining genetic variants associated with the medical trait for two or more single- ancestry populations using inverse-variance weighted meta-analysis of genome-wide association studies (GWAS) for each of the two or more single-ancestry populations.

47. The method of claim 46, wherein determining the genetic variants with heterogeneous associations further comprises using Cochran’s Q test to determine genetic variants associated with ^-values smaller than a threshold value.

48. The method of claim 34, wherein calculating the at least one trait score further comprises using information indicative of one or more conventional risk factors and / or genetic variants associated with the first individual.

49. The method of claim 48, wherein obtaining the trained computational model further comprises performing a regression including effects on phenotypes in the data describing phenotypes, genotypes, and / or phenotype-genotype relationships of the one or more conventional risk factors and / or the genetic variants associated with the first individual.

50. The method of claim 34, wherein obtaining the genomic data comprises obtaining genotyping data previously obtained by genotyping a biological sample obtained from the first individual.

51. The method of claim 34, wherein obtaining the first genomic data comprises obtaining genotyping data by sequencing a biological sample obtained from the first individual.

52. The method of claims 50 or 51, wherein obtaining the genotyping data comprises obtaining microarray data, whole-genome sequencing data, whole-exome sequencing data, and / or genotype imputation from partially observed data.Attorney Docket No.: M0437.70168WO00 53. The method of claim 34, wherein training the first computational model further comprises using one or more of a linear genetic effect, a genetic dominance effect, and / or a sex- based genetic effect associated with the medical trait.

54. The method of claim 34, wherein obtaining the trained computational model further comprises: determining an optimal regularization parameter value by: training a first plurality of computational models using a plurality of different regularization parameter values and a subset of the data describing phenotypes, genotypes, and / or phenotype-genotype relationships; evaluating a predictive performance of each of the first plurality of trained computational models; and selecting the optimal regularization parameter value based on the evaluated predictive performance of each of the first plurality of trained computational models.

55. The method of claim 54, wherein obtaining the trained computational model further comprises training the first computational model using the optimal regularization parameter value.

56. A system, comprising: at least one computer hardware processor; and at least one non-transitory computer readable medium storing processor-executable instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of any one of claims 34 to 55.

57. At least one non-transitory computer readable medium storing processor-executable instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform the method of any one of claims 34 to 55.

Images