Systems and methods for predicting variant pathogenicity based on amino acid signal-to-noise ratios

Abstract

Tools and related methods for analyzing the pathogenicity of gene variants are provided, the analysis being based on signal-to-noise ratios of the amino acids for the gene. The tools allow a user to input a gene, along with an amino acid position, and then the tool outputs information related to the susceptibility to a particular disease based on the inputted data as compared to information in a database about the gene, amino acid position, and / or other information. The information in the database can include data from others in the population and / or predicted information about the public, for example using artificial intelligence and other related predictive measures. The result is a tool, and related methodologies, that more accurately predict the true susceptibility for a patient to have or eventually get a particular disease than current tools, that tend to over-estimate the likelihood of having or getting such disease.

Description

Systems and Methods for Predicting Variant Pathogenicity Based on Amino Acid Signal- to-Noise RatiosFEDERAL FUNDING LEGEND

[0001] This invention was made with Government support at least under Federal Grant nos. K08-HL136839 and R01-HL160654 awarded by the National Institutes of Health. The Federal Government has certain rights to this invention.INCORPORATION BY REFERENCE

[0002] The present disclosure claims priority to and the benefit of U.S. Provisional Application No. 63 / 51 1 ,798, entitled “Systems and Methods for Predicting Variant Pathogenicity Based on Amino Acid Signal-to-Noise Ratios,” filed on July 3, 2023, the content of which is incorporated by reference herein in its entirety.FIELD

[0003] The present disclosure relates to systems and methods that enable accurate predictions of variant pathogenicity for genes, often rare genes, in helping to determine disease susceptibility for a patient in view of the gene(s), and more particularly relates to a web-based tool, and related methods, that provide these accurate predictions relying upon, for example amino acid signal-to-noise ratios.BACKGROUND

[0004] With genetic testing advancements, the burden of incidentally-identified cardiac disease-associated gene variants is rising. These variants may carry a risk of sudden cardiac death (SCD), highlighting the need for accurate diagnostic interpretation.

[0005] Evaluating cardiomyopathy- and channelopathy-associated genetic variants is increasingly challenging, though a population-based approach to determine probability of pathogenicity can help to whether a given variant may be associated with disease development. Advances in genetic analysis platforms and bioinformatics infrastructure significantly decreased the cost and time burden of genetic sequencing. Consequently, clinical genetic testing has expanded beyond disease-specific gene panels to increasing exome and genome sequencing. With recent expansions in clinical exome sequencing (ES), more variants incardiac genes are being incidentally identified in patients without suspicion of cardiac disease. Though ES is valuable diagnostically, incidental variants (also known as secondary variants) are challenging to interpret and evaluate.

[0006] To address this issue, the American College of Medical Genetics and Genomics (ACMG) established guidelines for reporting secondary findings and defined medically actionable genes. Of 78 genes currently included, 42 are associated with cardiac diseases carrying a heightened risk of SCD. Variants can be classified into 5 categories: pathogenic (P), likely pathogenic (LP), variant of uncertain significance (VUS), likely benign (LB), and benign (B). Incidental variants are often VUSs, and while many likely reflect background variation, discerning between a LP / P variant and VUS at time of genotyping is vital, as only LP / P variants are recommended for reporting and considered “actionable.” Thus, evaluating variant pathogenicity is crucial to understand their clinical actionability, especially in genes associated with SCD.

[0007] Conditions presenting with SCD include arrhythmogenic cardiomyopathy (ACM), catecholaminergic polymorphic ventricular tachycardia (CPVT), long QT syndrome (LQTS), hypertrophic cardiomyopathy (HCM) and dilated cardiomyopathy (DCM). These have pediatric diagnostic yield of 60-80%. Desmosomal genes are a major cause of ACM (PKP2, DSC2, DSG2, DSP), where fibrofatty tissue replaces cardiac myocytes, increasing risk of fatal arrhythmias. Defects in ion channels (RYR2, KCNQ1, KCNH2, SCN5A) are primarily responsible for channelopathies, including CPVT and LQTS. CPVT classically results in syncope or SCD secondary to bidirectional ventricular tachycardia, particularly in heightened adrenergic states. The cardiac ryanodine receptor 2 (RyR2) implicated in CPVT maintains Ca2+release channel complexes in the sarcoplasmic reticulum of cardiac myocytes. LQTS is marked by delayed cardiac repolarization, with the implicated ion channels crucial for cardiac action potential regulation and in which defects contribute to life-threatening torsades de pointes. Finally, MYH7, MYBPC3, and TTN are the major sarcomeric genes that cause cardiomyopathy. HCM is defined by unexplained thickening of the left ventricular wall without dilation.Truncating variants in TTN (TTNtvs) resulting in premature translation termination cause familial cardiomyopathies (CM), including nonischemic DCM and left ventricular noncompaction cardiomyopathy (LVNC). TTN encodes the largest human protein, contributing to a high population variant prevalence and complicating evaluation of incidental TTNtvs. In all these SCD-predisposing diseases, phenotypic heterogeneity highlights the utility of genetic testing in confirming diagnosis.

[0008] Accordingly, there is a need for improved methodologies for identifying pathogenic hotspots in SCD-associated genes, methodologies that more accurately predict whether a person having those genes is at risk for disease, thus avoiding false-positive predictions. There is also a need for providing an accessible tool to conduct such an analysis.SUMMARY

[0009] The present disclosure is directed to systems, methods, and techniques for producing identifying pathogenic hotspots in SCD-associated genes using amino acid- level signal-to-noise analysis, as well as a web-based precision medicine tool, referred to as DiscoVari or the DiscoVari tool herein, to improve variant evaluation and make it more accessible. The disclosed technology may also leverage artificial intelligence (Al) models, algorithms, and techniques to predict disease risk in various genomic variants on a population-based level. More particularly, the Al model(s) described herein can be trained and used to predict disease association and also disease penetrance, which includes likelihood that a particular individual will manifest a particular associated disease phenotype.

[0010] More particularly, variants in the primary genes implicated in ACM, CPVT, LQTS, HCM, and familial CM are considered reportable, medically actionable secondary findings. With existing challenges in variant interpretation, a population-based approach to determining the likelihood a given protein region is associated with disease development, so-called amino acid level signal-to-noise (S:N), can predict disease risk in incidental variants. The present disclosure provides for a bioinformatics tool (e.g., web-based, mobile application, software) to refine variant evaluation in cardiomyopathy-associated genes, channelopathy-associated genes, and / or other genes. The disclosures contained herein use amino acid-level S:N analysis to identify disease-associated genetic hotspots.

[0011] The minor allele frequency (MAF) of putatively pathogenic variants can be derived from cohort-based cardiomyopathy and channelopathy studies. In conjunction with the present disclosure, disease-associated MAFs were normalized to rare variants in an ostensibly healthy population (gnomAD) to calculate amino acid-level S:N. Amino acids with S:N above the gene-specific threshold were defined as hotspots. Accordingly, the disclosed DiscoVari tool can be used to identify pathogenic variants using variants from databases that provide information about genetic variants and their relationship to human health, like ClinVar, and individuals clinically evaluated with cardiac genetic testing.

[0012] The DiscoVari tool can be used as an internet-based tool for S:N-based variant hotspots. In an illustrative example of validation, a higher proportion of ClinVar likely pathogenic / pathogenic (LP / P) variants localized to DiscoVari hotspots (43.1%) than likely benign / benign (LB / B) variants (17.8%, P < 0.0001). Further, 75.3% of ClinVar variants reclassified to LP / P were in hotspots, compared to 41.3% of those reclassified as variants of uncertain significance (P < 0.0001) and 23.4% of those reclassified as LB / B ( < 0.0001). Of the clinical cohort variants, 73.1% of LP / P were in hotspots, compared to 0.0% of LB / B (P < 0.01).

[0013] The DiscoVari tool reliably can be used to identify disease-susceptible amino acid residues to evaluate variants by searching amino acid-specific S:N ratios.

[0014] One embodiment of a method of analyzing a gene to determine susceptibility to a particular disease includes receiving a gene selection from a user, receiving an amino acid position selection from that user, and outputting information. The gene selection is one gene of a plurality of genes stored in a database, with the plurality of genes being linked to one or more associated diseases. The outputted information includes at least one of: a signal-to-noise ratio of the selected gene-amino acid combination, a relative risk level of the one or more associated diseases, an indication of whether the signal-to-noise ratio corresponds to a statistical mutation hotspot, or a disease susceptibility determination. The outputted information is based on the received gene selection and amino acid position selection.

[0015] In at least some instances, the outputted information can further include information about a functional domain for the selected gene-amino acid combination. Alternatively, or additionally, the outputted information can include data generated from population predictions for one or more of the plurality of genes stored in the database. The population predictions can be determined, for example, based on providing the selected gene-amino acid combination as input to an artificial intelligence (Al) model. In at least some embodiments, the outputted information can include data generated by an Al model. The Al model can have been trained to generate the data based on receiving, as model input, the selected gene-amino acid combination.

[0016] An embodiment of a gene analysis tool includes a database of a plurality of genes and a processor. The database including information for each gene of the plurality of genes that includes: one or more associated diseases for the gene, a threshold signal-to-noise value for each disease of the one or more associated diseases, and a signal-to-noise value for each amino acid position of the gene. The processor is configured to receive user input that includes a geneselection from the database of a plurality of genes and an amino acid position selection, as well as output information about disease susceptibility in view of the gene selection and the amino acid position selection based on the information in the database.

[0017] The information for each gene of the plurality of genes can include a signal- to-noise ratio of the selected gene-amino acid combination and / or a signal-to-noise threshold value of the selected gene and the one or more associated diseases for the gene. Alternatively, or additionally, the information for each gene of the plurality of genes can include a functional domain for the selected gene-amino acid combination. The processor can be further configured to compare the signal-to-noise ratio of the selected gene-amino acid combination and the signal-to-noise threshold value of the selected gene and the one or more associated diseases for the gene, as well as determine, based on the comparison, whether the signal-to-noise threshold value may be exceeded by the signal-to-noise ratio of the selected gene-amino acid combination. Still further, the processor can be configured to output the determination about the signal-to-noise ratio as compared to the signal-to-noise threshold value. In at least some instances, the information for each gene of the plurality of genes can include a functional domain for the selected gene-amino acid combination, and the processor can be further configured to identify the functional domain for the selected gene-amino acid combination and output the identified functional domain.

[0018] In at least some implementations, the tool can be web-based. The processor can be further configured to present a graphical user interface (GUI) in a display of a computing device of a user. The GUI can include, for example, a first graphical element that can correspond to the gene selection and a second graphical element that can correspond to the amino acid position selection. In at least some such instances, at least one of the first and second graphical elements can include dropdown menus, and the user input can include selection of an option in a respective dropdown menu. In at least some embodiments, the information about disease susceptibility can be presented in a graph.

[0019] An embodiment of a computer-readable medium storing instructions is also provided. When that computer-readable medium storing instructions is executed by a processor, it causes the processor to receive a gene selection from a user, receive an amino acid position selection from the user, and output information. The gene selection is one gene of a plurality of genes stored in a database, with the plurality of genes being linked to one or more associated diseases. The outputted information includes at least one of: a signal-to-noise ratio of the selected gene- amino acid combination, a relative risk level of the one or more associated diseases, anindication of whether the signal-to-noise ratio corresponds to a statistical mutation hotspot, or a disease susceptibility determination. The outputted information is based on the received gene selection and amino acid position selection.

[0020] The outputted information can include data generated by an Al model. The Al model can have been trained to generate the data based on receiving, as model input, the selected gene-amino acid combination. The processor can be further configured to present a graphical user interface (GUI) in a display of a computing device of the user. The GUI can include, for example, a first graphical element that can correspond to the gene selection and a second graphical element that can correspond to the amino acid position selection. The information for each gene of the plurality of genes further can include a functional domain for the selected gene-amino acid combination. In at least some such embodiments, the processor can be further configured to identify the functional domain for the selected gene-amino acid combination and output the identified functional domain in a GUI at a computing device of the user.

[0021] One embodiment of a method of analyzing a gene to determine susceptibility to a particular disease includes receiving a gene selection from a user, receiving an amino acid position selection from the user that made the gene selection, and outputting information in view of the same. The gene selection is one gene of a plurality of genes stored in a database, with the plurality of genes being linked to one or more associated disease. The information that is output includes at least one of: a signal-to-noise ratio of the selected gene-amino acid combination, a relative risk level of the one or more associated diseases, an indication of whether the signal-to-noise ratio corresponds to a statistical mutation hotspot, and / or a disease susceptibility determination.

[0022] The outputted information can include information about a functional domain for the selected gene-amino acid combination. Alternatively, or additionally, the outputted information can include data generated from population predictions for one or more of the plurality of genes stored in the database. In at least some such embodiments, the population predictions can be determined based on providing the selected gene-amino acid combination as input to an artificial intelligence (Al) model. In at least some embodiments, the outputted information can include data generated by an Al model. The Al model can have been trained to generate the data based on receiving, as model input, the selected gene-amino acid combination.

[0023] One embodiment of a gene analysis tool includes a database of a plurality of genes and a processor. The database includes information for each gene of the plurality of genes.Such information includes: one or more associated diseases for the genes; a threshold signal-to- noise value for each disease of the one or more associated diseases; and a signal-to-noise value for each amino acid position of the gene. The processor is configured to receive user input that includes a gene selection from the database of a plurality of genes and an amino acid position selection. The processor is also configured to output information about disease susceptibility in view of the gene selection and the amino acid position selection based on the information in the database.

[0024] The information for each gene of the plurality of genes can include a signal-to-noise ratio of the selected gene-amino acid combination, as well as a signal-to-noise threshold value of the selected gene and the one or more associated diseases for the gene. The information for each gene of the plurality of genes can additionally, or alternatively, include a functional domain for the selected gene-amino acid combination.

[0025] The processor can be configured to perform other functions as well. For example, the processor can be configured to compare the signal-to-noise ratio of the selected gene-amino acid combination and the signal-to-noise threshold value of the selected gene and the one or more associated diseases for the gene, as well as determine, based on the comparison, whether the signal-to-noise threshold value is exceeded by the signal-to-noise ratio of the selected gene- amino acid combination. Further, the processor can be configured to output the determination about the signal-to-noise ratio as compared to the signal-to-noise threshold value. The information about disease susceptibility can be presented in a graph.

[0026] The information for each gene of the plurality of genes can include a functional domain for the selected gene-amino acid combination. In some such embodiments, the processor can be further configured to identify the functional domain for the selected gene- amino acid combination, and output the identified functional domain.

[0027] The tool can be web-based. In at least some embodiments, the processor can be further configured to present a graphical user interface (GUI) in a display of a computing device of a user. The GUI can include, for example, a first graphical element that corresponds to the gene selection and a second graphical element that corresponds to the amino acid position selection. In at least some embodiments, at least one of the first and second graphical elements can include dropdown menus and the user input can include selection of an option in a respective dropdown menu.

[0028] One embodiment of a computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform a variety of functions. These include receive a gene selection from a user, receive an amino acid position selection from the user that made the gene selection, and output information. The gene selection is one gene of a plurality of genes stored in a database, with the plurality of genes being linked to one or more associated diseases. The output information includes at least one of: a signal-to-noise ratio of the selected gene-amino acid combination; a relative risk level of the one or more associated diseases; an indication of whether the signal-to-noise ratio corresponds to a statistical mutation hotspot; and / or a disease susceptibility determination.

[0029] The outputted information can include data generated by an Al model. The Al model can have been trained to generate the data based on receiving, as model input, the selected gene-amino acid combination.

[0030] In some embodiments, the processor can be further configured to present a graphical user interface (GUI) in a display of a computing device of the user. In at least some such embodiments, the GUI can include a first graphical element that corresponds to the gene selection and a second graphical element that corresponds to the amino acid position selection.

[0031] The information for each gene of the plurality of genes can include a functional domain for the selected gene-amino acid combination. In at least some such embodiments, the processor can be further configured to identify the functional domain for the selected gene- amino acid combination and output the identified functional domain in a GUI at a computing device of the user.

[0032] The disclosed technology can provide one or more of the following advantages. For example, the disclosed technology can provide for accurately predicting and identifying pathogenic variants for various diseases over a population-level. The disclosed techniques further provide that S:N can be useful in identifying disease-causative variants and may also distinguish variants correlated with sub-clinical evidence of cardiac disease. Improved identification of disease-causing variants can allow for preemptive evaluation and treatment of patients and their family members at increased risk for developing cardiac disease. The disclosed techniques’ usefulness in downgrading variants can also reduce patient anxiety and healthcare expenditure in patients without disease found to have variants in CM-associated genes. In a research setting, this tool may allow for more accurate prediction of whether currently unidentified or uncertain genetic variants will lead to cardiac disease. This can beused for refining current functional domains and for determining new functional domains on proteins encoded by the genes, and / or identifying new areas of the protein that can result in disease when changed. This can also be used for targeting new therapies directly to the protein based on where disease causative genetic variants are more likely to localize.

[0033] To provide robust risk predictions and classifications, the disclosed technology can use a complex collection of algorithms, Al, and / or machine learning techniques to analyze data related to different genes for a population of users and to inform the relevant users about potential disease risks derived from genomic variances. This complex collection of algorithms, Al, and / or machine learning techniques can provide an unconventional solution to the problem of trying to accurately detect various disease risks for populations of users (and / or individual users). This unconventional solution can be rooted in technology and provides information that was not available in conventional systems. This unconventional solution also represents an improvement in the subject technical field otherwise unrealized by conventional systems. Specifically, unlike conventional systems, the disclosed technology may predict, with high accuracy and by using minimal processing time and compute resources, disease risks of a population level.

[0034] Advantageously, the disclosed technology may also display relevant information and data using a graphical user interface (GUI) on a display of computing devices of the relevant users in a unique and easy way to understand format. Conventional systems may not provide the disclosed solutions (e.g. , information about predicted disease risks) for at least the following reasons: (i) the significant processing power required to predict the disease risks of a population level real-time or near real-time; (ii) the considerable data storage requirements for maintaining information collected and determined by the disclosed technology; (hi) a large enough pool of parameter data to provide accurate thresholds for the disclosed algorithms, Al, and / or machine learning techniques; and / or (iv) algorithms, Al, and / or machine learning techniques that allow for the thresholds to be self-updated in light of additional data that can be added to the pool of relevant parameter data. The GUI(s) can display results of the execution of these complex algorithms, Al, and / or machine learning techniques in a manner that can be easily understandable by a human user. For example, translation of outcomes from these complex algorithms, Al, and / or machine learning techniques through the GUI onto images or other information displayed for a user can improve comprehension of considerable quantities of highly processed and complex data. As an illustrative example, an exemplary algorithm from this complex collection of algorithms can require: receiving genomic data from a variety of computing sources; selecting some data provided by the computing sources; ignoring some ofthe data that was provided by the computing sources; performing multiple calculations on a selected subset of the data; combining the data from these multiple calculations; and then outputting that data within a short amount of time (e.g., preferably less than a minute), all for multiple relevant users. Likewise, the disclosed technology may require analyzing millions of data points to accurately predict and quantify risk of different types of disease on a population level, generating and outputting information based on the analysis and predictions, and then repeating the above operations over a relatively short time period e.g., every day, every half day, every hour, every 10 minutes, every 5 minutes, every 1 minute, etc.) and for many different users and / or input data (e.g. , genes, diseases).

[0035] The details of one or more implementations are set forth in the accompanying drawings and the description below. Other features and advantages will be apparent from the description and drawings, and from the claims.BRIEF DESCRIPTION OF THE DRAWINGS

[0036] This disclosure will be more fully understood from the following detailed description, taken in conjunction with the accompanying drawings, in which:

[0037] FIG. 1A is a schematic illustration of derivation of disease-associated variants;

[0038] FIG. IB is a diagram of a methodology for population variant inclusion;

[0039] FIG. 1C depicts a schematic illustration of an S:N modeling design;

[0040] FIG. 2 A is an S:N validation workflow using variants in ClinVar;

[0041] FIG. 2B is a bar graph of percent of variants located within S:N hotspots between ClinVar likely benign or benign (LB / B) and likely pathogenic and pathogenic (LP / P) variants;

[0042] FIG. 2C is a bar graph of the S:N values for LB / B and LP / P variants in hotspots;

[0043] FIG. 2D is a workflow illustrating CM and channelopathy variants with a classification change in ClinVar from initial variant of uncertain / unknown significance (VUS) to LP / P or LB / B (or vice versa);

[0044] FIG. 2E is a bar graph depicting percent of variants located within S:N hotspots of variants that were reclassified from VUS to LB / B, from LB / B or LP / P to VUS, or from VUS to LP / P;

[0045] FIG. 2F is a bar graph illustrating S:N values for reclassified variants;

[0046] FIG. 3A is a bar graph of percent of variants in a clinical validation cohort located in S:N hotspots stratified by variants classified as LB / B, VUS, and LP / P at time of genetic testing;

[0047] FIG. 3B is a bar graph of S:N values for VUSs and LP / P variants in the clinical cohort;

[0048] FIG. 3C is a bar graph of percent of VUSs located in S:N hotspots stratified by those considered disease-causative, of unknown significance, or suspected benign;10049] FIG. 4 A is an example architecture of an application for performing the disclosed techniques;

[0050] FIG. 4B is one embodiment of a graphical user interface (GUI) based on using the application described in reference to FIG. 4A;

[0051] FIG. 4C is the GUI of FIG. 4B illustrating a gene selection being made;

[0052] FIG. 4D is the GUI of FIG. 4C illustrating both a gene selection and an amino acid position having been selected;

[0053] FIG. 4E is the GUI of FIG. 4D illustrating a GUI output after the gene selection and the amino acid position of FIG. 4D were searched;

[0054] FIG. 4F is the GUI of FIG. 4C illustrating both the same gene selection as FIG. 4D but a different amino position having been selected;

[0055] FIG. 4G is the GUI of FIG. 4F illustrating a GUI output after the gene selection and the amino acid position of FIG. 4F were searched;

[0056] FIG. 5 is a flowchart of an example process for incorporating S:N analysis into ACMG criteria when evaluating incidental variants;

[0057] FIG. 6A illustrates a workflow for derivation of UK Biobank (UKBB) variants for prediction of variants that can cause disease at a population level;

[0058] FIG. 6B is a bar graph of the variant prevalence in the UKBB, stratified by DiscoVari genes, HCM-associated variants, and TTN variants;

[0059] FIG. 6C is a bar graph of a proportion of cohort variants where classification between LB / B, VUS, and / or LP / P, as described by ACMG criteria, hinges upon whether the variant does, or does not, satisfy ACMG PM1 Criteria, the PM1 Criteria asking whether the variant does, or does not, lie in a genetic “hotspot,” which increases chances that the variant will cause disease, stratified by DiscoVari genes, HCM-associated variants, and TTN variants;

[0060] FIG. 7 A is a workflow of missense variants identified in HCM-associated genes among UKBB participants and their classification change to DiscoVari classification;

[0061] FIG. 7B is a bar graph of prevalence of HCM by ICD-10 code in UKBB participants included in an example cohort;

[0062] FIG. 7C is a bar graph of a proportion of variants identified in S:N hotspots stratified by variants in participants with no evidence of HCM or with HCM;

[0063] FIG. 7D is a bar graph of signal-to-noise values for variants in HCM negative participants compared to variants in participants with HCM;

[0064] FIG. 7E is a bar graph of a proportion of variants hosted in individuals with HCM where ACMG PM1 criteria was met or not by InterVar and using DiscoVari S:N analysis;

[0065] FIG. 7F is a bar graph of LP / P variants in HCM negative individuals and those with HCM using InterVar and DiscoVari S:N for applying PM1 criteria;

[0066] FIG. 8A is a workflow of identification of HCM negative individuals and phenotypic analysis for sub-clinical evidence of disease among UKBB participants;

[0067] FIG. 8B is a bar graph of a proportion of missense HCM-associated variants hosted in individuals with evidence of an HCM-associated phenotype, including atrial fibrillation / flutter, cardiomyopathy, cardiac murmur, and chest pain, stratified by variants outside S:N hotspots or in S:N hotspots;

[0068] FIG. 8C is a bar graph of a proportion of variants hosted in HCM-negative individuals with Afib / flutter in downgraded or upgraded variants with DiscoVari incorporated for classification;

[0069] FIG. 8D is a bar graph of a proportion of variants hosted in HCM-negative individuals with chest pain in downgraded or upgraded variants with DiscoVari incorporated for classification;

[0070] FIG. 9 A is a workflow of identification of CM negative individuals and phenotypic analysis for sub-clinical evidence of disease;

[0071] FIG. 9B is a bar graph of a proportion of variants identified in S:N hotspots stratified by variants in participants with no evidence of CM or with CM;

[0072] FIG. 9C is a bar graph of signal-to-noise values for variants localizing to hotspots in CM negative participants compared to variants in participants with CM;

[0073] FIG. 9D is a bar graph of a proportion of variants hosted in CM negative individuals with dizziness and a murmur outside S:N hotspots compared to in S:N hotspots;

[0074] FIG. 9E is a bar graph of a proportion of variants hosted in individuals with syncope in variants with no classification change versus upgraded with DiscoVari incorporated for classification;

[0075] FIG. 10 illustrates a conceptual diagram of a system for predicting risk of disease on a population-level using the disclosed techniques and artificial intelligence (Al) techniques; and

[0076] FIG. 11 is a schematic diagram that shows an example of a computing device and a mobile computing device for use in conjunction with the present disclosures.DESCRIPTION

[0077] Certain exemplary embodiments will now be described to provide an overall understanding of the principles of the structure, function, manufacture, and use of the devices and methods disclosed herein. One or more examples of these embodiments are illustrated in the accompanying drawings and / or are described herein. Those skilled in the art will understand that the devices and methods specifically described herein and illustrated in the accompanying drawings are non-limiting exemplary embodiments and that the scope of the present disclosure is defined solely by the claims. The features illustrated or described in connection with one exemplary embodiment may be combined with the features of other embodiments. Such modifications and variations are intended to be included within the scope of the present disclosure. Terms commonly known to those skilled in the art may be used interchangeably herein.

[0078] The following is a non-exhaustive list of abbreviations and acronyms that may be used herein: ACM, arrhythmogenic cardiomyopathy; ACMG, American College of MedicalGenetics and Genomics; B, benign; CM, cardiomyopathy; CPVT, catecholaminergic polymorphic ventricular tachycardia; DCM, dilated cardiomyopathy; ES, exome sequencing; gnomAD, Genome Aggregation Database; GS, genome sequencing; HCM, hypertrophic cardiomyopathy; LB, likely benign; LP, likely pathogenic; LQTS, long QT syndrome; MAF, minor allele frequency; P, pathogenic; SCD, sudden cardiac death; S:N, signal-to-noise; TTN, titin; TTNtvs, truncating variants in TTN', VUS, variant of uncertain / unknown significance.

[0079] Articles “a’- and “an’- are used herein to refer to one or to more than one (z.e. , at least one) of the grammatical object of the article. By way of example, “an element” means at least one element and can include more than one element.

[0080] “About” is used to provide flexibility to a numerical range endpoint by providing that a given value may be “slightly above” or “slightly below” the endpoint without affecting the desired result.

[0081] The use herein of the terms "including," "comprising," or "having," and variations thereof, is meant to encompass the elements listed thereafter and equivalents thereof as well as additional elements. As used herein, “and / or” refers to and encompasses any and all possible combinations of one or more of the associated listed items, as well as the lack of combinations where interpreted in the alternative (“or”).

[0082] Recitation of ranges of values herein are merely intended to serve as a shorthand method of referring individually to each separate value falling within the range, unless otherwise indicated herein, and each separate value is incorporated into the specification as if it were individually recited herein. For example, if a concentration range is stated as 1% to 50%, it is intended that values such as 2% to 40%, 10% to 30%, or 1 % to 3%, etc., are expressly enumerated in this specification. These are only examples of what is specifically intended, and all possible combinations of numerical values between and including the lowest value and the highest value enumerated are to be considered to be expressly stated in this disclosure.

[0083] As used herein, the term "subject" and "patient" are used interchangeably herein and refer to both human and nonhuman animals. In some embodiments, the subject comprises a human who is undergoing a medical procedure using a system or method as prescribed herein.

[0084] Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.

[0085] The present disclosure provides for improved performance in predictive medicine. Recent improvements in genetic testing technologies have facilitated the rapid rise of cardiovascular genomics and its expanding use in clinical settings. The understanding of genetic underpinnings of inherited cardiovascular disease can further be expanded particularly in rare diseases. Implementing ES and GS in critically ill infants can, for example, improve outcomes and reduces healthcare costs over time by avoiding the common diagnostic odyssey for families of patients with rare diseases. Additionally, predictive genetic testing can provide for identifying at-risk individuals before disease manifestation. This is both in the form of cascade testing, which can help in risk stratification in kindred of affected probands, and predictive genetic testing in individuals with no evidence of cardiovascular disease. Predictive genetic testing thereby offers the opportunity to reduce the morbidity and mortality of SCD- associated diseases and to lower healthcare costs to patients and the overall system.

[0086] The disclosed techniques provide for improved specificity in putatively pathogenic variant identification by downgrading most LP / P variants in HCM negative individuals to VUS. The importance of this finding is highlighted by the markedly high general population prevalence of 1: 149 for LP / P variants in HCM-associated genes, despite a notably low disease penetrance approximately in the range of about 1.3% to about 2.1%. Similar work in dilated CM also found markedly low penetrance in the UK biobank (UKBB), as an illustrative example, the UKBB being a large-scale biomedical database and research resource. Further, variant classification changes in individuals with an HCM diagnosis can be correlated with ClinVar and in silico predictions.

[0087] The illustrative tool described, referred to herein as DiscoVari of the DiscoVari tool, enables a more accurate prediction of how likely a gene(s), often a rare gene(s), is going to result in a patient developing a particular disease. While in the past the detection of such rare genes often led to an over-inclusive determination that a particular disease was likely to occur in view of the detection of that gene, the disclosures herein provide a much more refined approach, improving false determinations that a disease is likely to occur in view of a particular gene by at least one-third (1 / 3). This is achieved, at least in part, on improved calculations and data, yielding more useful thresholds that serve as the measuring stick between determining whether the detected gene(s) is likely to lead to a particular disease or not. This is further achieved through the application of Al to the calculations and data used in predicting disease likelihood for a particular gene(s). A person skilled in the art, in view of the totality of what is disclosed herein, will understand how to apply the artificial intelligence and other related predictive measures to the calculations and data described herein for purposes of providing ahighly accurate (approximately 90% accurate or higher, or approximately 95% accurate or higher) determination about disease in view of detected genes.L0088 J More particularly, Al algorithms, models, and / or techniques and other related predictive measures can be applied to populations to enable population predictions that further supplement the model and data of the underlying tool. The thresholds in making determinations can be refined, and further, variants can even be predicted, at least in some instances, on data alone, or at least predominantly on data. The result is improved predictions for sudden cardiac death (SCD), hypertrophic cardiomyopathy (HCM), and other diseases. This is because, prior to the present disclosures, a major hurdle to accurate predictive testing is the markedly higher prevalence of variants associated with the disease than variants that actually lead to the disease (e.g., SCD, HCM). However, when viewed in the context of the rest of the present disclosure, the population-identified, disease-associated variants enable highly accurate predictions about disease susceptibility when analyzing a gene(s) using the tools and methodologies disclosed.L0089 J In one embodiment, a platform provides a user the ability to input a gene and amino acid of input, and the user then obtains a signal-to-noise (S:N) value for a given amino acid residue. For example, in at least some iterations, a user can select a gene from a given list of genes, such as provided by a dropdown menu. The user can also select the amino acid position, which can be any whole number (e.g., 1, 2, 3. ..135, 136, 137. ..etc.). After both the gene and amino acid position have been selected, the tool can search the database to identify the S:N value for that gene’s amino acid position in conjunction with a disease(s) for which the genedisease pair has been validated. The result can be compared to an S:N threshold value for that gene-disease pair. The tool can flag when the S:N ratio has or has not exceeded the S:N threshold for that gene-disease pair, among other information. Refer to FIGs. 10B and 10C for further discussion.

[0090] RESULTS

[0091] Establishing hotspots in DiscoVari genes

[0092] Referring to the figures, FIGs. 1 A, IB, and 1C depict an illustrative summary of cohorts and methodology for performing the disclosed techniques.

[0093] To establish disease-associated genetic hotspots, rare variant prevalence in literature- derived disease cohorts were compared to rare variants in controls from gnomAD, as shown byFIGs. 1 A and IB. For example, FIG. 1A is a schematic of derivation of disease-associated variants. In this illustrative example, only TTN truncating variants were included in a CM variant (e.g., 3722 Probands TTN). ACM, CPVT, LQTS, and HCM variants (e.g., 1421 Probands PKP2, DSC2, DSG2, DSP; 155 Probands RYR2; 4356 Probands KCNQ1, KCNH2, SCN5A; and 5520 Probands MYH7, MYBPC3, TNNT2, TNNI3, ACTC1, TMP1, MYL2, MYL3, respectively) were restricted in this example to missense and single amino acid indels. Variant pathogenicity was assessed in ClinVar and by in silico algorithms. FIG. IB is a diagram of methodology for gnomAD variant inclusion. Variants with an MAF greater than the highest pathologic MAF for that gene in the disease cohort can be excluded. Variants from disease cohorts were compiled from summarized data in previously published work applying S:N analysis to ACM, CPVT, LQTS, HCM, and TTNtvs. Pathogenic variant MAF approximately ranged from about 0.15% (ACTC7) to about 40.3% (RYR2). The control cohort consisted of rare genetic variants derived from gnomAD with genotyping of 141,456 individuals, as shown by FIG. IB. Aggregate gnomAD MAF approximately ranged from about 0.87% (TTNtvs) to about 5.38% (ACM-associated genes), highlighting intragenic heterogeneity in MAF.

[0094] More particularly, FIG. 1C shows a schematic of an S:N modeling design. Variant prevalence and gene-level S:N were evaluated by averaging the S:N across all amino acid positions to compare variant frequencies in the case cohorts and gnomAD, as shown by FIG. 1A. Disease-associated MAF was normalized to gnomAD MAF to establish S:N ratios at the amino acid-level. Pathogenic hotspots were areas with S:N above calculated gene-specific thresholds. Amino acid residues were considered a “hotspot” if the S:N ratio exceeded respective gene-specific thresholds. In topology maps overlaying S:N across domains, high S:N was found in the A-band of TTN and pore regions of LQTS genes, which are known genetic hotspots. Topology maps for ACM-, CM-, CPVT-, LQTS-, and HCM-associated genes may also demonstrate the S:N across functional domains. Overall, these findings highlight a potential for S:N analysis to identify functional domains of cardiovascular disease proteins to determine putatively pathogenic hotspots with greater specificity.

[0095] Validation of S:N hotspots using ClinVar.

[0096] FIGs. 2A, 2B, 2C, 2D, 2E, and 2F illustrate workflows and bar graphs of LP / P ClinVar variants as being commonly found in S:N hotspots compared to LB / B variants.

[0097] Whether the S:N analysis aligned with ClinVar calls of variant pathogenicity was assessed to validate whether LP / P variants could be discriminated from VUSs. S:N was applied to all ClinVar LB / B and LP / P variants in DiscoVari genes, excluding any variants from the provided disease cohorts used to identify hotspots, as shown by FIG. 2A, which represents an S;N validation workflow. In FIG. 2A, CM and channelopathy variants that were not found in the provided disease cohorts were included. S:N was applied to all LB / B and LP / P variants.

[0098] FIG. 2B illustrates a bar graph of the percent of variants located within S:N hotspots for all LB / B and all LP / P variants. The proportion of LP / P variants in hotspots (about 43.1 % [40.4-45.9]) was higher than LB / B variants (about 17.8% [13.4-23.2], P < 0.0001), as shown by block B.

[0099] Additionally, for hotspot variants, the mean S:N ratio for LB / B variants (about 13.8 [4.9-22.7]) was lower than LP / P variants (about 31.0 [26.8-35.3], P < 0.0001), as shown in FIG. 2C, which provides a plotting of the S:N values for LB / B and LP / P variants in hotspots. There was variability across genes analyzed, with the highest ratio of LP / P and LB / B hotspot variants in KCNH2 (P < 0.0001), compared to MYPBC3, where there was no significant difference in the proportion of LP / P versus LB / B in hotspots.

[0100] FIG. 2D illustrates that CM and channelopathy variants with a classification change in ClinVar from initial VUS to LP / P or LB / B (or vice versa) were included. S:N was next applied to ClinVar variants reclassified over time to determine whether hotspots could predict those reassessed from LB / B or VUS to LP / P, as shown by FIG. 2D. Variants in the provided disease cohorts were excluded. Variants were assessed using S:N analysis.

[0101] FIG. 2E illustrates a bar graph depicting the percent of variants located within S:N hotspots of variants that were reclassified from VUS to LB / B, from LB / B or LP / P to VUS, or from VUS to LP / P. The lowest proportion of variants found in hotspots was among ClinVar variants reevaluated as LB / B (about 23.4% [14.7-35.2]), as shown in FIG. 2E. A large proportion of variants reclassified to LP / P localized to S:N hotspots (about 75.3% [64.3-83.9]), which was higher than in LB / B variants and VUSs (P < 0.0001). The percentage of variants reinterpreted as VUSs within a S:N peak was also higher than LB / B variants (P < 0.01).

[0102] FIG. 2F illustrates plotted S:N values for reclassified variants. Moreover, the mean S:N for hotspot variants reclassified to LP / P (about 37.2, [24.0-50.4]) was greater than those reassessed to LB / B (about 5.10, [2.00-8.20], P < 0.0001), as shown by FIG. 2F. Labels indicate most recent variant classification. **, P < 0.01. ****, P < 0.0001. LB / B, likelybenign / benign; LP / P, likely pathogenic / pathogenic; S:N, signal-to-noise; VUS, variant of uncertain significance.LOO 103 J Overall, these findings suggest that amino acid-level S:N analysis can reliably identify putatively pathogenic hotspots in cardiomyopathy- and channelopathy-associated genes and offers value for clarifying the diagnostic weight of VUSs.|00104| S:N Validation using clinical cohort.

[0105] FIGs. 3A, 3B, and 3C illustrate bar graphs showing that clinically re-evaluated LP / P variants can more likely fall within S:N hotspots compared to LB / B variants and VUSs.

[0106] An illustrative clinical cohort of patients with cardiovascular genetic disease evaluated using the disclosed techniques can include 152 variants in 103 individuals meeting inclusion criteria, including 47 (45.6%) male subjects. The illustrative mean age of genetic testing can be about 26.5 years [21.7-31.3], though the date of genetic testing may be unavailable for some. Of variants evaluated, about 51.3 % [43.4-59.1] were classified as LP / P by ACMG before incorporating PM1 criteria. PM1 criteria represents one of the criteria present in the 2015 ACMG criteria for variant interpretation. This criterion is satisfied if the variant is located within a genetic “hotspot” that is more likely to cause disease or a functional domain contributing to disease when disrupted. PM1 was applied to variants with a S:N ratio above the respective gene-specific threshold.

[0107] FIG. 3 A illustrates a bar graph of the percent of variants in the clinical cohort located in S:N hotspots stratified by variants classified as LB / B, VUS, and LP / P at time of genetic testing. Upon variant re-assessment, about 59.9% [51.9-67.3] were LP / P with PM1 incorporated. After re-evaluation, about 71.8% [61.0-80.6] of the variants called as LP / P at time of genetic testing localized to S:N hotspots, while only about 27.0% [18.2-38.1] of VUSs did (P < 0.0001), as shown by FIG. 3 A.

[0108] FIG. 3B illustrates the plotted S:N values for VUSs and LP / P variants in the clinical cohort. The mean S:N ratio for LP / P variants in hotspots was about 37.9 [25.1-50.7] compared to about 10.6 [4.5-16.8] for VUSs (P < 0.0001), as shown by FIG. 3B. The application of S:N in this cohort suggests that the analysis is clinically valid and aligns with pathogenicity assignments in individuals with diagnostic gene testing.

[0109] To further determine whether DiscoVari informs the diagnostic weight of identified VUSs, variants interpreted as VUSs at the time of genetic test reporting were categorized basedon suspected clinical re-evaluation of the variant in a clinical cohort. Of all VUSs identified in the clinical cohort, about 32.9% [23.7-43.7] were located in hotspots. Upon re-evaluation of the variant, variants disease-associated were considered if there was a phenotypic match to the disease associated with the gene in question, and if that variant was suspected to be pathogenic due to significant family history or proven co-segregation of the variant with affected family members.

[0110] FIG. 3C illustrates a bar graph of the percent of VUSs located in S:N hotspots stratified by those considered disease-causative, of unknown significance, or suspected benign based on clinical presentation, family history, and / or co-segregation of variants with disease. In suspected disease-associated VUSs, about 80.0% [49.0-96.4] localized to S:N hotspots, compared to only about 23.3% [13.2-37.7] of suspected benign variants (P < 0.001), as shown by block FIG. 3C. Similar proportions hotspot variants were found when considering variants called VUSs by ACMG criteria before PM1 incorporation. These results support the ability of S:N analysis to help discriminate potentially disease-associated VUSs from ones that are likely benign.

[0111] Creation of a novel variant analysis web-based interface: DiscoVari

[0112] FIG. 4A is an example architecture 400 of an application for performing the disclosed techniques. A frontend computing system 402 can provide a user interface for end users 406. A hackend computing system 404 can be configured to process information with programming logic. 256-bit Advanced Encryption Standard, or other similar standards, may be used in the backend 404 to store data, and each encryption key can be encoded with a rotating set of master keys. As an illustrative example, JAVASCRIPT ES6 programming language can be used to build the frontend 402 and the backend 404 of the web application described herein. The application can be accessed using any standard web browser (e.g., SAFARI, GOOGLE CHROME, FIREFOX).

[0011] As an illustrative example of the architecture 400, the user(s) 406 can provide / collect data using the frontend 402 (408). The frontend 402, as described above, can be configured to provide user interfaces at the user(s) 406 computing device (e.g., computer, laptop, tablet, mobile device, smartphone) through which the user(s) 406 interacts. Providing / collecting data can include providing a gene and / or amino acid position to the frontend 402 using GUI features presented in a display of the user(s) 406 computing device. The frontend 402 can then make a request of the backend system 404 to generate and obtain a S:N ratio for the user-inputtedposition (410). The backend system 404 can access one or more application logic, file systems, databases, and / or web servers to determine the S:N ratio and / or generate results / information to be presented to the user(s) 406 (412). Accordingly, the backend system 404 can transmit a response back to the frontend 402 and the frontend 402 can display the results at the user(s) 406 computing device (414).

[0114] With an increasing burden of incidental variants in cardiomyopathy and channelopathy genes, the disclosed tool can be built and used to improve variant analysis based on S:N analysis. An example of the tool is provided for at https: / / discovarilab.duke.edu / , the content at the provided web link of which is incorporated by reference herein in its entirety, including all links and sublinks associated with the provided web link. At the web-interface provided by the frontend 402, the user(s) 406 can input a gene and amino acid position to obtain a S:N ratio for the position. The results include the disease in which the disclosed tool can be validated for that gene, the gene-specific S:N threshold, a gene-level S:N plot, and / or the S:N and applicable functional domain(s) for that amino acid residue. Illustrations of an example graphical user interface (GUI) based on using the application described in reference to FIG. 4A are provided in FIGs. 4B-4G, the GUI serving as the frontend 402.

[0115] FIG. 4B illustrates a GUI 420, which can be presented for the user(s) 406 at their respective computing device as the frontend 402. The GUI 420 can include, for example, selectable options or fields 422 and 424, also referred to as graphical elements or graphical input elements. In the illustrated embodiment, the selectable option 422 represents a selection of genes that can be listed in a dropdown menu 422d, as shown in FIG. 4C, while the selectable option 424 represents an amino acid (AA) position that can be manually entered, for instance by entering a number. A person skilled in the art will appreciate that other options or fields can be provided, and that any such option or field can include a dropdown menu, a text field, or other way by which data can be inputted by a user. In the illustrated embodiment, once a GUI input is elected by the user, for example the gene KCNH2 (NM_000238.4) illustrated as gene selection 428 and AA position 200 illustrated as AA position selection 429, as shown in FIG. 4D, the user can then click on or otherwise choose the select button 426 to cause his or her selections to be processed using the architecture 400. Results from such processing can be presented at the user’s device, for example as shown by the GUI output 480 in FIG. 4E.

[0116] The example GUI output 480 of FIG. 4E includes an identified associated disease 486, as shown long QT syndrome, a visual result 488, and a tabular result 490. In some implementations, the output 480 may additionally display overlap between diseases associatedwith particular genes. The visual result 488 and the tabular result 490 can be considered graphical elements or graphical output elements. The visual result 488 includes a graph 488g comparing AA position of the gene and the LQTS / gnomAD variant frequency, as well as a bar line 488b illustrating various aspects of the result: PAS, Pore, SF, cNBD, Hinge, and STK. That is, the visual result 488 may include a functional domains (e.g., the bar line 488b) mapped to a plot (e.g., the graph 488g). In the illustrated embodiment, PAS, Pore, SF, cNBD, Hinge, and STK are indications that can represent the relevant functional domains of the hERG potassium channel encoded by the KCNH2 gene. These are relevant as mutations overlapping with these areas would potentially disrupt crucial elements of the resulting protein, thereby impacting its normal function and contributing to disease mechanism. The tabular result 490 includes relevant results for the selected AA position, including an S:N threshold, a Signal-to- Noise ratio, a determination as to whether the S:N is above the threshold, and the functional domain. Being above the threshold denotes a variant that would be considered to fulfill the PM1 criteria described herein and therefore predict to be more likely to cause disease, while being below the threshold denotes that the variant is more likely to be a benign population genetic variation. Accordingly, the visual and / or tabular results 488 and / or 490, respectively, can be used to indicate whether a variant of the gene may exist at the user-indicated position and / or whether a mutation exists (or does not exist) within a particular hotspot. In the illustrated embodiment, the S:N ratio at position 200 is 0.00, and thus that value can be flagged as not being above a threshold value, which for the illustrated gene-disease pair may be 1.56. There is thus not a functional domain that is applicable, as represented by the “NA” in the tabular results 490.

[0117] FIGs. 4F and 4G illustrate a second example using the GUI input 420 having a different resulting GUI output 480'. More particularly, as shown in FIG. 4F, a gene selection 428 made in the selection portion 422 is again the gene KCNH2 (NM_000238.4), but now an AA position selection 429' made in the selectable option 424 is position 600. Again, after clicking on or otherwise choosing the select button 426, using the techniques described herein, the disclosed tool processes the selections and outputs the GUI output 480' using the architecture 400, the output 480' being shown in FIG. 4G. In the illustrated embodiment, the output 480' again includes an associated diseases 486', a visual result 488', and a tabular result 490'. The associated disease 486' is again long QT syndrome. Further, the visual result 488' again includes a graph 488g' and a bar line 488b', the graph and chart conveying similar information as for the output 480. Both the graphs 488g and 488g' are and the bar lines 488b and 488b' are the same. Likewise, the tabular result 490' includes the same relevant results forthe selected AA position, including an S:N threshold, a Signal-to-Noise ratio, a determination as to whether the S:N is above the statistical threshold for being labeled as a genetic “hotspot,” and the corresponding functional domain of the protein if present. In this instance, however, the S:N ratio at position 600 is 73.10, and thus that value is flagged as being above the threshold value of 1.56, meaning that the variant localizes in an area of the gene and protein that is a genetic “hotspot.” Further, the resulting functional domain is identified as “Pore.” As shown by bar-line 488b', an AA position of 775 would yield a functional domain of “cNBD.”

[0118] By way of non-limiting examples, other genes for which the disclosed tools and methods have been validated for missense variants in long QT syndrome may include but are not limited to KCNQ1 and SCN5A. Also by way of non-limiting examples, the disclosed tools and methods can be used for missense variants in: arrhythmogenic cardiomyopathy for genes PKP2, DSC2, DSG2, and DSP; catecholaminergic polymorphic ventricular tachycardia for gene RYR2; and hypertrophic cardiomyopathy for genes MYH7, MYBPC3, TNNT2, TNNI3, ACTC1, TPM1, MYL2, and MYL3. Still further, the disclosed tools and methods can be used for cardiomyopathy-associated truncating variants in gene TTN.

[0119] FIG. 5 is a flowchart of an example process 500 for incorporating S:N analysis into ACMG criteria when evaluating incidental variants. The process 500 can be performed by a computing system as described herein (e.g., refer to the backend computing system 404 of FIG. 4A). The process 500 can be performed by any other type of computing system and / or network of computing systems configured to perform the disclosed techniques. For illustrative purposes, the process 500 is described from the perspective of a computing system.

[0120] Individuals with variants incidentally identified in CM- or channelopathy-associated genes should be referred to a multi-disciplinary center specializing in cardiovascular genetic testing. The disclosed tool may be used and deployed by the computing system as a component of a comprehensive clinical evaluation by providing a correlate for ACMG PM1 criteria (where FHx corresponds to family history and PHx corresponds to personal history).

[0121] Referring to the process 500 of FIG. 5, the computing system can identify an incidental variant in a CM and / or channelopathy associated gene (block 502). The computing system can identify the incidental variant using input provided by a relevant user, as described in reference to FIGs. 4A-4G. The incidental variant can be identified, for example, using a variety of genetic testing techniques, including but not limited to exome sequencing, genome sequencing, and / or targeted panel sequencing.

[0122] Based on the incidental variant identified in block 502, the computing system can generate a referral recommendation for an associated user or patient to a multi-disciplinary cardiovascular genetics team (block 504).

[0123] In response to the referral recommendation, a clinical evaluation of the associated user can be performed, as indicated in block 506. Results for the clinical evaluation can be documented, recorded, and / or processed. The results may include, but are not limited to, PHx, physical examination, and / or FHx.

[0124] The results of the clinical examination can be compared against one or more rules and / or criteria by the computing system to perform the clinical evaluation. Performing the clinical evaluation may include determine whether there is a low suspicion of disease risk (block 508) or a high suspicion of disease risk (block 510). In some implementations, the computing system may apply one or more ML models and / or Al models in block 506 to evaluate the clinical results and determine the associate user’s level of disease risk.

[0125] Regardless of whether the computing system identifies and tags the associate user as having a low suspicion of disease risk (block 508) or a high suspicion of disease risk (block 510), the computing system can proceed to block 512. In block 512, the computing system may search for one or more related variants using the disclosed tool (e.g., the DiscoVari tool).

[0126] Depending on what variant is found for the low suspicion of disease risk, the computing system can identify a low S:N area (block 514) or an S:N hotspot PM1 being met (block 516). For example, the computer system can match the inputted variant value and inputted gene value to data (e.g. , in a database, data repository, data store) that correlates to the same gene position. The low S:N area in block 514 can indicate a variant that does not meet a predetermined threshold level / value and therefore does not meet the PM1 criteria described herein. Accordingly, the variant is predicted as more likely to be benign. If the S:N hotspot PM1 is met in block 516, then the S:N of the variant is predicted to be above the predetermined threshold, and therefore is predicted to be more likely disease causative.

[0127] Depending on what variant is found for the high suspicion of disease risk, the computing system can identify a low S:N area (block 518) or an S:N hotspot PM1 being met (block 520). Refer to blocks 514 and 516 for further discussion.

[0128] In response to blocks 514, 516, 518, and / or 520, the computing system can apply one or more ACMG criteria (block 522). In some implementations, a relevant user can provideinput indicating that the one or more criteria should be applied. Sometimes, the computing system may apply the one or more criteria without first receiving user input indicating such action to be taken.

[0129] Where there was an initial clinical evaluation of there being a low suspicion of disease, as indicated by block 508, and the variant is found in a low S:N area, as indicated by block 514, applying the ACMG criteria can lead to the computing system generating a determination of LB / B for the associated user in block 524 that indicates the the user’s CM- or channelopathy-gene is less likely to be disease associated (block 526). Such information can be presented in GUI displays at a computing device of the associated relevant user. Refer to at least FIGs. 4A-4G for further discussion.

[0130] Where there was an initial clinical evaluation of there being a low suspicion of disease, as indicated by block 508, but the variant is found in an S:N hotspot PM1 met, or alternatively, where there was an initial clinical evaluation of there being a high suspicion of disease, as indicated by block 510, but the variant is found in a low S:N area, as indicated by block 518, applying the ACMG criteria can lead to the computing system generating a VUS indication (variant of uncertain significance) for the associated user (block 528). Accordingly, the computing system may generate a recommendation that additional evaluation and / or followup may be required (block 530). The computing system may additionally or alternatively recommend VUS resolution testing, if possible (block 532). Such information can be presented in GUI displays at a computing device of the associated relevant user. Refer to at least FIGs. 4A-4G for further discussion.

[0131] Still further, where there was an initial clinical evaluation of there being a high suspicion of disease, as indicated by block 510, and the variant is found in an S:N hotspot PM1 met, applying the ACMG criteria can lead to the computing system generating an LP / P indication in block 534. The computing system can generate a recommendation for continued follow-up in block 536, and can subsequently recommend cascade testing of relatives (e.g., first degree relatives). Such information can be presented in GUI displays at a computing device of the associated relevant user. Refer to at least FIGs. 4A-4G for further discussion.

[0132] A person skilled in the art will appreciate that the process 500 is one example of an implementation of the techniques and systems provided for herein. It is intended to illustrate an example of how the disclosed DiscoVari tool can be used as a first-in-kind tool for assisting users in identifying disease variant hotspots (e.g. cardiovascular disease variant hotspots), or alack of hotspots, so there can be a better identification of those most likely to be susceptible to medical issues. A person skilled in the art, in view of the present disclosures, will appreciate other workflows that can be followed in view of the present disclosures and / or result from the present disclosures.

[0133] Using the disclosed technology, identifying individuals at-risk of developing disease before symptom onset can reduce morbidity and mortality. Moreover, cascade testing of a LP / P variant in a proband can identify family members at risk of disease development (see, e.g., block 538 in the process 500 of FIG. 5). This can be essential as it relates to heart disease, for example, as SCD can be the first presentation of inherited cardiac disorders. The disclosed technology can make accurate determinations of whether a variant may, or may not, be disease- associated critical. To aid in variant interpretation, a population-based amino acid-level S:N analysis can be used by the computing system to identify genetic hotspots where variants may be more likely to be disease-causative. S:N analysis can be incorporated into ACMG criteria by using S:N as a correlate for PM1, which can define variants located in mutational hotspots or functional domains with no benign variation. PM 1 implementation can sometimes be inconsistent across laboratories; thus, S:N can be used to standardize application. The disclosed tool (e.g., DiscoVari) can help determine whether variants fulfill this criterion, offering an easy, precise search to pinpoint hotspots with a higher probability of causing disease and low background variation. As an illustrative example, this user-friendly method for incorporating S:N into ACMG criteria was validated in patients clinically evaluated across ClinVar variants. A higher proportion of ClinVar LP / P variants were found in areas of high signal than VUSs and LB / B variants, which also extended to variants with an increase in pathogenicity. In the provided clinical cohort, LP / P variants were more commonly found in hotspots and with a higher mean S:N than VUSs. Incorporating S:N into variant classification may help determine whether individuals with genetic variants need to be followed or evaluated to ensure early identification and treatment of SCD-predisposing diseases.

[0134] The high risk of morbidity and mortality, which may be reduced with proper clinical management, underlies the recommendation for reporting LP / P variants in DiscoVari genes associated with ACM, HCM, CPVT, LQTS, DCM, and / or LVNC. Expanding indications for diagnostic ES and GS, as well as ACMG guidelines suggesting reporting variants in clinically actionable genes, can increase a number of incidentally identified variants being reported. These variants may trigger an individualized cardiac evaluation. Notably, the burden of these incidental findings can be greater than disease prevalence, indicating that most variants, even those deemed LP / P, may be unlikely to cause penetrant disease. As a merely illustrativeexample, work in ACM found a background variant rate of ~16% in controls. Moreover, a TTNtv frequency of -3% in the general population was previously described. In LQTS1-3, literature estimates of variant frequency range from 3-8%. Finally, CPVT (RYR2) has an estimated variant frequency of -11%, and others have found a background rate of pathogenic variants in RYR2, KCNQ1, KCNH2. and SCN5A of -0.5%. The challenges inherent in genetic variant interpretation extends beyond a clinical evaluation. For example, there are significant psychosocial implications to genetic variant identification which may, or may not, be associated with disease. This can include, but is not limited to, anxiety for patients and families, uncertainty around variant pathogenicity and disease risk, additional healthcare costs, higher insurance premiums, and / or unnecessary clinical testing burdening the healthcare system and insurers. This high background rate of variants, and the challenge implicit in determining disease-risk, highlights the need for the disclosed tools to aid in variant evaluation.

[0135] A compounding factor among incidental variants is the often limited clinical history, phenotyping, and family history available. Such phenotyping can be critical to variant interpretation but may be absent when variants are found incidentally rather than through diagnostic testing. Moreover, variant calls can be fluid and change over time, adding additional burden on various entities. Overcalling pathogenic variants can cause significant stress for patients and their families, potentially resulting in misdiagnoses and inappropriate clinical management. Pre-genetic sequencing counseling can remain key to discussing potential findings with patients. Multi-disciplinary interpretation of genetic testing results can be important before reporting to appropriately determine variant pathogenicity and risk.

[0136] Variant interpretation should incorporate a probabilistic, Bayesian approach to determine diagnostic weight of a genetic test before evaluation. Here, DiscoVari can help distinguish between VUS and LP / P variant by assessing variants when clinical data is unavailable and there is low pre-genetic test suspicion of cardiac disease. S:N can provide clarity when determining variant pathogenicity, evidenced by retrospectively evaluated clinical cohorts. DiscoVari represents one part of the return of results by providing a basis for ACMG PM1 criteria, particularly for diagnostic VUSs and for incidental variants. It can be part of a full clinical evaluation at a multi-disciplinary center specializing in cardiovascular genetic testing. Importantly, decisions regarding clinical patient management should not be based on any single evaluation component, including DiscoVari. The variant should be clinically reevaluated in the context of a complete clinical evaluation (personal history, physical exam for relevant disease, family history). Amino acid-level analysis can also offer an additional tool todefine the diagnostic weight of variants by identifying those in genetic loci, and therefore, greater diagnostic relevance.LOO 137 J As provided for herein, amino-acid level signal-to- noise analysis can be a more precise way to determine variant disease-risk and may also impact ACMG criteria. For example, the PM1 “hotspot” criteria can represent a binary instrument for incorporating continuous relative risk variable that the disclosed S:N analysis provides. There can be significant variation in S:N values across a gene. For example, in MYBPC3, residues 5 (S:N 1.97) and 502 (42.56) can both be considered to be located in hotspots based on the provided gene S:N threshold of 1.42. The region around residue 5 may not be considered a hotspot based on previous studies, and there may be two variants at this position listed in ClinVar without evidence of pathogenicity). Conversely, a more common HCM-associated variants in MYBPC3 can be p.Arg502Trp. These case examples highlight that applying PM1 as a binary “present” or “absent” may not reflect heterogeneity within current “hotspots,” nor does it allow for a quantitative assessment of disease association. The disclosed DiscoVari tool can provide a more nuanced consideration of genetic hotspots and S:N analysis for various frameworks of variant interpretation.

[0138] Future validation using large, diverse, prospective, and longitudinally followed clinical cohorts can facilitate evaluation of genotype -phenotype correlations and captures individuals with later disease presentation. Including both individuals with and without clinical evidence of disease can be crucial to clarify the diagnostic relevance of DiscoVari and can provide clinical information absent from literature-derived cohorts in conjunction with the present disclosure. Confirming the present results with these studies can underscore the importance of incorporating S:N into clinical frameworks and guidelines, such as those published by the ACMG.

[0139] Though not a clinical validation tool, variants with a clinically significant change (VUS to LB / B or LP / P and vice versa) in their ClinVar classification can be used as another method of validation. Despite lack of available phenotypic data for variants listed in ClinVar, listed variants and their classifications can be curated and regularly reviewed by expert groups or other relevant users. While this strategy may not replicate the strength of an individualized clinical validation in a large, prospective cohort, it can support utility of S:N analysis given the stringent criteria used by ClinVar working groups for variant classification. S:N can also be a relative measure of effect size, such that the strength of the provided results can be confined by the number of disease-associated variants incorporated into the present analysis. Further,analysis can sometimes be limited to a single disease per gene (except for TTNtvs), though many of the genes that were included can be associated with a range of phenotypes. In genes such as SCN5A, comparing S:N hotspots across phenotypes can be a next step. SCN5A can be associated with LQTS, Brugada syndrome, and / or DCM, which have different mechanisms of disease (with gain of function variants associated with LQTS, compared to loss of function variants in Brugada). Using S:N across the spectrum of cardiomyopathies for individual genes can be an important future direction to expand DiscoVari' s utility and understand differences and similarities in phenotype-specific genetic hotspots.

[0140] FIGs. 6A, 6B, and 6C illustrate example variants and methodology for performing the disclosed techniques. To determine whether the disclosed tool DiscoVari can accurately identify variants associated with development of cardiovascular disease, a large populationbased cohort can be used, such as UKBB, as a merely illustrative example. FIG. 6 A provides a schematic of derivation of UKBB variants for analysis. The UKBB can be accessed for participants with exome sequencing (n = 200,619 in the illustrated example) and with variants in DiscoVari genes (n = 162,037 in the illustrated example), as shown by FIG. 6A. Only participants with ES and variants in DiscoVari genes may be included in this illustrative example. Variants can be categorized into variants in HCM-associated genes, or TTN variants.

[0141] FIG. 6B illustrates a bar graph of the variant prevalence in the UKBB, stratified by DiscoVari genes 600, HCM-associated variants 602, and TTN variants 604. Notably, in this example, about 80.8% [80.6-80.9] of individuals in the UKBB hosted a variant in a DiscoVari gene, as shown by FIG. 6B. Prevalence can be shown across all variants, or those classified as LBB, VUS, or LPP. In this illustrative example, an overall prevalence of about 26.0% [25.8- 26.1] was found, and about a 21.8% [21.7-22.0] prevalence of variants of uncertain significance (VUSs) specifically (n = 43,799) among participants with HCM-associated gene variants (n = 52,064). An overall prevalence of about 56.3% [55.8-56.3] and about 44.7% [44.5-44.9] prevalence of VUSs (n = 89,675) was found in participants with variants in TTN (n = 112,470). Together, this suggests that VUSs can be markedly common among UKBB participants.

[0142] Next, the disclosed technology can be used to determine whether PM1 criteria, ACMG criteria refined by using DiscoVari, is important for interpretation of variants within the UKBB. Specifically, variants within the UKBB that had a clinically meaningful change in variant classification based upon PM 1 criteria can be identified. For example, change from VUS to LB / B or LP / P, and vice versa, depending on whether PM1 criteria was either met or not. Reversing the InterVar PM1 call changed interpretation in a notable about 26.5% [25.6-27.4] of HCM-associated missense variants. FIG. 6C illustrates a bar graph of the proportion of cohort variants hinging on PM1 stratified by DiscoVari genes 600, HCM-associated variants 602, and TTN variants 604. Variants hinging on PM1 were defined as those that changed in classification from VUS to LPP / LBB and vice versa. When PM1 was added, about 54.1% [52.4-55.7] of TTNtvs changed classification, as shown by FIG. 6C.

[0143] The results shown by FIG. 6C suggest that PM1 plays a critical role in variant interpretation. They further highlight the need for accurate variant interpretation and the high impact of PM1 criteria in determining variant classification. PM1 was noted as one source of discrepancy in ACMG variant interpretation across laboratories evaluating genetic variants, with others recommending a quantitative approach to this criteria to address this issue. The disclosed tool DiscoVari can therefore offer a precise and replicable way of evaluating PM1 criteria.

[0144] The disclosed DiscoVari S:N can be interpreted by using it as a correlate for ACMG PM1 criteria. DiscoVari may represent one component of the return of results, which can be done in a multi-disciplinary fashion at a specialized cardiovascular genetics center incorporating evidence-based guidelines to appropriately determine variant pathogenicity and risk. Clinical patient management generally should not be decided upon any individual aspect of the evaluation. The disclosed techniques can be implemented in the context of a full clinical evaluation, to include a personal and family history, as well as a physical exam and testing for the disease in question. Guidelines recommend a probabilistic, Bayesian approach for variant interpretation by incorporating the pre-test suspicion of cardiac disease in addition to the diagnostic weight of a genetic test. The amino acid-level S:N analysis provided by DiscoVari can provide an additional tool for quantifying diagnostic relevance of variants in S:N-predicted genetic hotspots.

[0145] FIGs. 7A, 7B, 7C, 7D, 7E, and 7F illustrate an example Signal-to-Noise implementation using the disclosed techniques to enrich for HCM-associated genes. The disclosed / WvcoVuri-based S:N analysis can be validated to accurately identify putatively pathogenic variants in individuals with disease. FIG. 7A illustrates a schematic of missense variants identified in HCM-associated genes and their classification change to DiscoVari classification. As an illustrative example a total of 9,378 missense variants were identified in HCM-associated genes, with 36 of these variants being identified in participants with an ICD- 10 code-based diagnosis of HCM, as shown by FIG. 7A. Notably, 1,371 variants were downgraded in HCM negative participants when incorporating S:N, representing a downgradeof about 56.2% [54.2-58.1] of the original LP / P variants. Only 47 variants were upgraded from VUS to LP / P in HCM negative participants using DiscoVari.

[0146] FIG. 7B illustrates a bar graph of the prevalence of HCM by 1CD-10 code in UKBB participants included in the example cohort. A total of 215 participants were identified in this illustrative example with an ICD-10 code-based diagnosis of HCM among the UKBB participants included for analysis (n = 162,037), representing a prevalence of about 0.1% [0.1- 0.2], as shown by FIG. 7B.

[0147] FIG. 7C illustrates a bar graph of the proportion of variants identified in S:N hotspots (defined as those with a S:N ratio above the gene-specific threshold) stratified by variants in participants with no evidence of HCM (HCM negative) or with HCM. When analyzed based on S:N values, there was a greater proportion of variants localizing to S:N hotspots in those with HCM (about 58.3% [42.2-72.9]) compared to in HCM negative participants (about 37.8% [36.8-38.8], p < 0.05), as shown by FIG. 7C.

[0148] FIG. 7D illustrates plotted signal-to-noise values for variants in HCM negative participants compared to variants in participants with HCM. There was a higher median S:N value for variants found in people with HCM (p < 0.05), shown in FIG. 7D. Such results support that S:N hotspots can correlate with evidence of HCM in the UKBB.

[0149] S:N can then be used to correlate for the PM1 criteria, and re-evaluated variants based on current ACMG guidelines using this new refined PM1 criteria. Application of DiscoVari- based PM1 criteria showed enrichment for individuals with a diagnosis of HCM when using S:N. FIG. 7E illustrates a bar graph of the proportion of variants hosted in individuals with HCM where ACMG PM1 criteria was met or not by Interpretation and Validation of Genomic Variants (InterVar) (700) (which is a non-limiting example of a tool that can be used to assist in interpreting genetic variants identified through sequencing technologies) and using DiscoVari signal-to-noise analysis (702). Specifically, with PM1 criteria not met, about 0.26% [0. 16-0.42] of variants may be hosted in individuals with HCM, compared to about 0.59% [0.39-0.90] with PM1 applied by S:N (p < 0.01), as shown by FIG. 7E. Conversely, InterVar, which is currently used in the UKBB to predict variant pathogenicity, did not show enrichment for HCM by their PM1 assignment (about 0.54% [0.23-1.26] when PM1 was not met and about 0.37% [0.26- 0.52] when PM1 was met).

[0150] FIG. 7F illustrates a proportion of LP / P variants in HCM negative individuals and those with HCM using InterVar (700) and DiscoVari signal-to-noise (702) for applying PM1criteria. *, p< 0.05. ****, p <0.001. Using DiscoVari to evaluate these variants reduced the burden of LP / P variants in individuals without an HCM diagnosis (about 12.0% [11.3-12.6]) compared to InterVar (about 26.1% [25.2-27.0], p < 0.0001), as shown in FIG. 7F. Further, DiscoVari showed a significant enrichment for LP / P variants in those with HCM (about 36.1% [22.5-52.4]) compared to HCM negative individuals (about 12.0% [11.3-12.6], p < 0.0001). Conversely, this may not be seen when using InterVar for variant interpretation (about 38.9% LP / P in HCM [24.8-55.1] and about 26.1% in HCM negative [25.2-27.0]; p = 0.08). Together, such findings underline the ability of the disclosed technology to enrich for an HCM diagnosis when variants are re-evaluated with ACMG criteria informed by S:N for PM1.

[0151] FIGs. 8A, 8B, 8C, and 8D illustrate that sub-clinical HCM phenotypes can be more prevalent in variants upgraded in classification by using the disclosed techniques. Given the relatively rare prevalence of HCM in the UKBB described herein, the disclosed techniques can be used to determine whether S:N predicted hotspot variants correlated with sub-clinical evidence of disease, as shown by FIG. 8A. FIG. 8A illustrates a schematic of identification of HCM negative individuals and phenotypic analysis for sub-clinical evidence of disease.

[0152] A higher proportion of variants hosted by individuals with any type of cardiomyopathy among S:N hotspot variants (about 7.5% [6.7-8.4]) was found compared to those outside hotspots (about 7.2% [6.6-7.9], p < 0.01), as shown in FIG. 8B. FIG. 8B illustrates a bar graph of the proportion of missense HCM-associated variants hosted in individuals with evidence of an HCM-associated phenotype (<?.g., Afib / flutter, cardiomyopathy, murmur, chest pain) stratified by variants outside S:N hotspots 800 or in S:N hotspots 802. A similar trend was found using the disclosed technology when looking at variants in individuals with atrial fibrillation / flutter (p = 0.05) and with a reported murmur (p = 0.06). The disclosed technology can also be used to focus in on individuals with missense HCM-associated variants with no evidence of HCM (n = 9,117). In these individuals, 1,943 (about 21.4% [20.6-22.2]) were identified as having some evidence of a sub-clinical phenotype, meaning one or more of the following: atrial fibrillation / flutter, pacemaker / ICD, ventricular arrhythmia, sudden death, heart failure, cardiomyopathy, syncope, dizziness, tachycardia, palpitations, murmur, chest pain, family history, and / or cardioversion.

[0153] In variants in HCM negative individuals, a higher proportion can be found in participants with atrial fibrillation / flutter and chest pain in variants upgraded by DiscoVari compared to downgraded (p < 0.05 for both) as shown in both FIGs. 8C and 8D. Taken together, these results suggest S:N is not only helpful for identifying disease-causative variants,but may also help distinguish those that are correlated with sub-clinical evidence of HCM. FIGs. 8C and 8D provide example bar graphs of the proportion of variants hosted in HCM- negative individuals with Afib / flutter (refer to FIG. 8C) or chest pain (refer to FIG. 8D) in downgraded or upgraded variants with DiscoVari incorporated for classification.

[0154] FIGs. 9A, 9B, 9C, 9D, and 9E provide examples of cardiomyopathy in TTNtvs localizing to S:N hotspots. As described herein, the disclosed technology can provide for the utility of S:N analysis in TTNtv evaluation, which can be found as a common source of population based variants in the UKBB described above. Participants with TTNtvs can be categorized into those with a CM-associated ICD10 and CM negative participants, as shown by FIG. 9A.

[0155] FIG. 9A illustrates a schematic 900 of identification of CM negative individuals and phenotypic analysis for sub-clinical evidence of disease. Patients can be evaluated with truncating variants in TTN (block 902). After review and analysis of the patient’s medical records, ECG(s), and / or cardiac MRI (block 904), the patients can be evaluated as CM negative (block 906) or CM positive individuals (block 910) to distinguish variants found in individuals with or without disease. In individuals with no clear diagnosis (block 906), further analysis and evaluation can be performed for sub-clinical evidence of disease that is related to cardiomyopathies (block 908). Block 908 may not warrant a full diagnosis, in some implementations.

[0156] FIG. 9B is a bar graph of the proportion of variants identified in S:N hotspots (defined as those with a S:N ratio above the gene-specific threshold) stratified by variants in participants with no evidence of CM (CM negative) or with CM. There was a trend towards a higher proportion of hotspot variants in those with CM (about 25.0% [16.4-36.1]) compared to CM negative individuals (about 16.7% [15.5-17.9], p = 0.06), as shown in FIG. 9B.

[0157] FIG. 9C demonstrates plotted signal-to-noise values for variants localizing to hotspots in CM negative participants compared to variants in participants with CM. Similar to in HCM, there was a higher median S:N of hotspot variants in CM (p < 0.01), shown by FIG. 9C.

[0158] FIG. 9D illustrates a bar graph of the proportion of variants hosted in CM negative individuals with dizziness and a murmur outside S:N hotspots 900 compared to in S:N hotspots 902. The disclosed techniques and technology can also be used to focus on variants specifically in CM negative participants, where it was found there were a higher proportion of individuals with reported dizziness in variants localizing to S:N hotspots (about 3.6% [2.3-5.4]) comparedto outside hotspots (about 2.2% [1.7-2.8], p < 0.05), and a similar trend with participants with a reported murmur (p = 0.09), as shown by FIG. 9D.[00159J FIG 9E illustrates a proportion of variants hosted in individuals with syncope in variants with no classification change versus upgraded with DiscoVari incorporated for classification. In CM negative individuals, a higher proportion of variants was in individuals with reported syncope in upgraded variants (about 6.5% [4.5-9.4]) compared to those that did not change (about 3.8% [3.2-4.5]) classification (p < 0.01), as shown by FIG. 9E. Overall, such results suggest that S:N hotspots correlated with evidence of CM in TTNtvs, and correlated with sub-clinical symptoms in CM negative UKBB participants.

[0160] FIG. 10 illustrates a conceptual diagram of a system 1000 for predicting risk of disease on a population-level using the disclosed techniques and artificial intelligence (Al) techniques. The system 1000 can include a backend computing system 1002, a user computing device 1004, and / or a data store 1008 in communication (e.g., wired, wireless) via network(s) 1006. The system components 1002, 1004, and / or 1008 are similar to other computer components described herein.

[0161] The backend computing system 1002 can be configured to generate predictions of disease risks based on inputs such as particular genes and / or amino acid (AA) positions. The backend computing system 1002 can generate such predictions using the disclosed techniques and technology. Moreover, the backend computing system 1002 can generate the predictions using Al techniques, models, and / or algorithms. The user computing device 1004 can be configured to present the disclosed tool in one or more web applications and / or mobile applications. Refer to at least FIGs. 4A-4G for further discussion.

[0162] Referring to the system 1000 of FIG. 10, the backend computing system 1002 can transmit code for launching an application for predicting disease risk at the user computing device 1004 (block A, 1010). The code can be transmitted in response to a user providing input at their device 1004 to launch a web-based application in a web browser (e.g., by typing in a URL or searching for the application using a search engine in their browser).

[0163] In response to receiving the code, the user computing device 1004 can execute the code so that the application can be launched and displayed in one or more GUIs at the device 1004 (block B, 1012). Refer to at least FIGs. 4B-4G for further discussion about the application presented in the GUIs.

[0164] The user computing device 1004 can receive user input indicating one or more genes and / or amino acid positions in block C (1014). The user input can be provided using GUI features presented as part of launching the application at the device 1004. Refer to FIGs. 4B- 4G for further discussion. The user input can be provided for a particular user / patient and / or a population of users / patients. As a result, the disclosed techniques can be used as a diagnostic variant classifier to predict disease risk on a population level.

[0165] The user computing device 1004 can transmit the user input to the backend computing system 1002 in block D (1016).

[0166] The backend computing system 1002 can retrieve an Al model from the data store 1008 in block E (1018). Sometimes, the Al model can be locally stored at the backend computing system 1002 and / or the data store 1008 can be part of the backend computing system 1002. In some implementations, the backend computing system 1002 can train genespecific machine learning / Al models. Then, the backend computing system 1002 can retrieve the model(s) that corresponds to the particular gene(s) provided as part of the user input. The gene-specific model(s) can provide for post-genetic testing diagnostic analyses and high performance prediction of variant pathogenicity. In some implementations, the backend computing system 1002 can also train and use a model that is trained to automatically compare different model configurations with ensembling and cross-validation to identify high- performing models for a particular gene of interest. The identified model(s) can then be retrieved and used by the backend computing system 1002 to perform the disclosed techniques.

[0167] To train the model(s) described herein, the backend computing system 1002 can generate datasets for one or more gene variants. Each of the gene variants can be associated with a ground truth pathogenic label or benign label. A model for each gene can be applied to a model deployment dataset of VUS for that particular gene to reclassify those VUS as likely benign, still uncertain, or likely pathogenic, and perform functional and clinical evaluation of model predictions on VUS. Model input features for each gene variant can include consensus amino acid, mutant amino acid, amino acid position, domain, evolutionary conservation, rate of evolution, signal-to-noise ratio, position-specific scoring matrix score, and / or any combination thereof. For each gene, the backend computing system 1002 can perform cross-validation of each model configuration on the training dataset (e.g., the dataset of pathogenic and benign variants) to calculate an area under a receiver operating characteristic, an average precision, and / or a calibration curve slope. The backend computing system 1002 may then automatically select a preferred model configuration for each gene based on the calibration curve slope e.g.,being close to 1) and average precision (e.g., as high as possible). The backend computing system 1002 can then train a model with the selected model configuration using the benign and pathogenic variants in the training dataset mentioned above. This trained model can then be applied to the model deployment dataset of VUS. Based on the final trained model for each gene, the backend computing system 1002 can also calculate thresholds for VUS reclassification based on negative predictive values and positive predictive values.

[0168] In some implementations, an Al model can be trained using one or more of the following: (i) known disease-associated variants (or benign variants) through databases (e.g., ClinVar); (ii) whether each (or one or more) of the variants lands in a functional domain of a resultant protein; (iii) how evolutionarily-conserved an amino acid residue hosting the variant is across species; and / or (iv) a signal-to-noise analysis at the amino acid level. Accordingly, the model can be trained to identify and correlate user inputs with different types of diseases based, at least in part, on validated gene-disease associations. Various types of machine learning models may be used with the disclosed technology, including but not limited to Random Forest models and / or Gradient Boosting models.

[0169] In block F (1020), the backend computing system 1002 can provide the user input as inputs to the Al model. The Al model can be used to improve calculations described herein on the user input, thereby yielding accurate thresholds that can serve for measuring whether one or more detected genes are more likely than not to leave to particular diseases.

[0170] The backend computing system 1002 can receive, as output from the Al model, indications of disease risk for the gene(s) and / or the AA position(s) (block G, 1022). The model output can include, for example, a probabilistic value of a likelihood that the particular variant will result in a penetrant disease. This can be an improvement over existing techniques, which may be designed around how likely it is that the variant is disease-associated. Using existing techniques, a variant can be predicted to be 100% disease-associated, but an individual hosting the variant may never develop any signs of disease. The Al model described herein, on the other hand, may not only predict disease association, but more specifically, disease penetrance, which includes the likelihood that the individual will actually manifest the particular disease phenotype.

[0171] In block H (1024), the backend computing system 1002 may then generate disease risk results and / or recommendations based on the model output and / or one or more criteria. Refer to the process 500 of FIG. 5 for further discussion about determining how the disease riskcan factor into treatments, additional testing, and / or other recommendations for the relevant user(s).

[0172] The backend computing system 1002 can return the results and / or recommendations to the user computing device 1004 and / or the data store 1008 (for storage and / or later retrieval) in block I (1026). The results and / or recommendations can be presented at the user computing device 1004, as described at least in reference to FIGs. 4B-4G.

[0173] FIG. 1 1 is a schematic diagram that shows an example of a computing system 1 100. The computing system 1100 includes one or more computing devices (e.g., computing device 1110), which can be in wired and / or wireless communication with various peripheral device(s) 1180, data source(s) 1190, and / or other computing devices (e.g. , over network(s) 1170). The computing device 1110 can represent various forms of stationary computers 1112 (e.g., workstations, kiosks, servers, mainframes, edge computing devices, quantum computers, etc.) and mobile computers 1114 (e.g., laptops, tablets, mobile phones, personal digital assistants, wearable devices, etc.). In some implementations, the computing device 1110 can be included in (and / or in communication with) various other sorts of devices, such as data collection devices (e.g., devices that are configured to collect data from a physical environment, such as microphones, cameras, scanners, sensors, etc.), robotic devices (e.g., devices that are configured to physically interact with objects in a physical environment, such as manufacturing devices, maintenance devices, object handling devices, etc.), vehicles (e.g., devices that are configured to move throughout a physical environment, such as automated guided vehicles, manually operated vehicles, etc.), or other such devices. Each of the devices (e.g., stationary computers, mobile computers, and / or other devices) can include components of the computing device 1110, and an entire system can be made up of multiple devices communicating with each other. For example, the computing device 1110 can be part of a computing system that includes a network of computing devices, such as a cloud-based computing system, a computing system in an internal network, or a computing system in another sort of shared network. Processors of the computing device (1110) and other computing devices of a computing system can be optimized for different types of operations, secure computing tasks, etc. The components shown herein, and their functions, are meant to be examples, and are not meant to limit implementations of the technology described and / or claimed in this document.

[0174] The computing device 1110 includes processor(s) 1120, memory device(s) 1130, storage device(s) 1140, and interface(s) 1150. Each of the processor(s) 1120, the memory device(s) 1130, the storage device(s) 1140, and the interface(s) 1150 are interconnected using asystem bus 1160. The processor(s) 1120 are capable of processing instructions for execution within the computing device 1110, and can include one or more single-threaded and / or multithreaded processors. The processor(s) 1120 are capable of processing instructions stored in the memory device(s) 1130 and / or on the storage device(s) 1140. The memory device(s) 1130 can store data within the computing device 1110, and can include one or more computer-readable media, volatile memory units, and / or non-volatile memory units. The storage device(s) 1140 can provide mass storage for the computing device 1110, can include various computer- readable media (e.g., a floppy disk device, a hard disk device, a tape device, an optical disk device, a flash memory or other similar solid state memory device, or an array of devices, including devices in a storage area network or other configurations), and can provide date security / encryption capabilities.

[0175] The interface(s) 1150 can include various communications interfaces e.g., USB, Near-Field Communication (NFC), Bluetooth, WiFi, Ethernet, wireless Ethernet, etc.) that can be coupled to the network(s) 1170, peripheral device(s) 1180, and / or data source(s) 1190 (e. g. , through a communications port, a network adapter, etc.). Communication can be provided under various modes or protocols for wired and / or wireless communication. Such communication can occur, for example, through a transceiver using a radio-frequency. As another example, communication can occur using light e.g., laser, infrared, etc.) to transmit data. As another example, short-range communication can occur, such as using Bluetooth, WiFi, or other such transceiver. In addition, a GPS (Global Positioning System) receiver module can provide location-related wireless data, which can be used as appropriate by device applications. The interface(s) 1150 can include a control interface that receives commands from an input device (e.g., operated by a user) and converts the commands for submission to the processors 1120. The interface(s) 1150 can include a display interface that includes circuitry for driving a display to present visual information to a user. The interface(s) 1150 can include an audio codec which can receive sound signals (e.g., spoken information from a user) and convert it to usable digital data. The audio codec can likewise generate audible sound, such as through an audio speaker. Such sound can include real-time voice communications, recorded sound (e.g., voice messages, music files, etc.), and / or sound generated by device applications.

[0176] The network(s) 1170 can include one or more wired and / or wireless communications networks, including various public and / or private networks. Examples of communication networks include a LAN (local area network), a WAN (wide area network), and / or the Internet. The communication networks can include a group of nodes (e.g. , computing devices) that areconfigured to exchange data (e.g., analog messages, digital messages, etc.), through telecommunications links. The telecommunications links can use various techniques e.g., circuit switching, message switching, packet switching, etc.) to send the data and other signals from an originating node to a destination node. In some implementations, the computing device 1110 can communicate with the peripheral device(s) 1180, the data source(s) 1190, and / or other computing devices over the network(s) 1170. In some implementations, the computing device 1110 can directly communicate with the peripheral device(s) 1180, the data source(s), and / or other computing devices.

[0177] The peripheral device(s) 1180 can provide input / output operations for the computing device 1110. Input devices e.g., keyboards, pointing devices, touchscreens, microphones, cameras, scanners, sensors, etc.) can provide input to the computing device 1110 (e.g., user input and / or other input from a physical environment). Output devices (e.g. , display units such as display screens or projection devices for displaying graphical user interfaces (GUIs)), audio speakers for generating sound, tactile feedback devices, printers, motors, hardware control devices, etc.) can provide output from the computing device 1110 (e.g. , user-directed output and / or other output that results in actions being performed in a physical environment). Other kinds of devices can be used to provide for interactions between users and devices. For example, input from a user can be received in any form, including visual, auditory, or tactile input, and feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback).

[0178] The data source(s) 1190 can provide data for use by the computing device 1110, and / or can maintain data that has been generated by the computing device 1110 and / or other devices (e.g., data collected from sensor devices, data aggregated from various different data repositories, etc.). In some implementations, one or more data sources can be hosted by the computing device 1110 (e.g., using the storage device(s) 1140). In some implementations, one or more data sources can be hosted by a different computing device. Data can be provided by the data source(s) 1190 in response to a request for data from the computing device 1110 and / or can be provided without such a request. For example, a pull technology can be used in which the provision of data is driven by device requests, and / or a push technology can be used in which the provision of data occurs as the data becomes available (e.g., real-time data streaming and / or notifications). Various sorts of data sources can be used to implement the techniques described herein, alone or in combination.

[0179] In some implementations, a data source can include one or more data store(s) 1190a. The database(s) can be provided by a single computing device or network (e.g., on a file system of a server device) or provided by multiple distributed computing devices or networks e.g., hosted by a computer cluster, hosted in cloud storage, etc.). In some implementations, a database management system (DBMS) can be included to provide access to data contained in the database(s) e.g., through the use of a query language and / or application programming interfaces (APIs)). The database(s), for example, can include relational databases, object databases, structured document databases, unstructured document databases, graph databases, and other appropriate types of databases.

[0180] In some implementations, a data source can include one or more blockchains 1190b. A blockchain can be a distributed ledger that includes blocks of records that are securely linked by cryptographic hashes. Each block of records includes a cryptographic hash of the previous block, and transaction data for transactions that occurred during a time period. The blockchain can be hosted by a peer-to-peer computer network that includes a group of nodes (e.g., computing devices) that collectively implement a consensus algorithm protocol to validate new transaction blocks and to add the validated transaction blocks to the blockchain. By storing data across the peer-to-peer computer network, for example, the blockchain can maintain data quality (e.g. , through data replication) and can improve data trust (e.g. , by reducing or eliminating central data control).

[0181] In some implementations, a data source can include one or more machine learning systems 1190c. The machine learning system(s) 1190c, for example, can be used to analyze data from various sources (e.g., data provided by the computing device 1110, data from the data store(s) 1190a, data from the blockchain(s) 1190b, and / or data from other data sources), to identify patterns in the data, and to draw inferences from the data patterns. In general, training data 1192 can be provided to one or more machine learning algorithms 1 194, and the machine learning algorithm(s) can generate a machine learning model 1196. Execution of the machine learning algorithm(s) can be performed by the computing device 1110, or another appropriate device. Various machine learning approaches can be used to generate machine learning models, such as supervised learning (e.g., in which a model is generated from training data that includes both the inputs and the desired outputs), unsupervised learning (e.g., in which a model is generated from training data that includes only the inputs), reinforcement learning (e.g., in which the machine learning algorithm(s) interact with a dynamic environment and are provided with feedback during a training process), or another appropriate approach. A variety of different types of machine learning techniques can be employed, including but not limited toconvolutional neural networks (CNNs), deep neural networks (DNNs), recurrent neural networks (RNNs), and other types of multi-layer neural networks.LOO 182 J Various implementations of the systems and techniques described herein can be realized in digital electronic circuitry, integrated circuitry, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. A computer program product can be tangibly embodied in an information carrier (e.g. , in a machine-readable storage device), for execution by a programmable processor. Various computer operations (e.g. , methods described in this document) can be performed by a programmable processor executing a program of instructions to perform functions of the described implementations by operating on input data and generating output. The described features can be implemented in one or more computer programs that are executable on a programmable system including at least one programmable processor coupled to receive data and instructions from, and to transmit data and instructions to, a data storage system, at least one input device, and at least one output device. A computer program is a set of instructions that can be used, directly or indirectly, by a computer to perform a certain activity or bring about a certain result. A computer program can be written in any form of programming language, including compiled or interpreted languages, and can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program product can be a computer- or machine-readable medium, such as a storage device or memory device. As used herein, the terms machine-readable medium and computer-readable medium refer to any computer program product, apparatus and / or device (e.g., magnetic discs, optical disks, memory, etc.) used to provide machine instructions and / or data to a programmable processor, including a machine -readable medium that receives machine instructions as a machine-readable signal. The term machine -readable signal refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0183] Suitable processors for the execution of a program of instructions include, by way of example, both general and special purpose microprocessors, and can be a single processor or one of multiple processors of any kind of computer. Generally, a processor will receive instructions and data from a read-only memory or a random access memory or both. The elements of a computer are a processor for executing instructions and one or more memory devices for storing instructions and data. Generally, a computer can also include, or can be operatively coupled to communicate with, one or more mass storage devices for storing data files. Such devices can include magnetic disks (e.g. , internal hard disks and / or removabledisks), magneto-optical disks, and optical disks. Storage devices suitable for tangibly embodying computer program instructions and data can include all forms of non-volatile memory, including by way of example semiconductor memory devices, flash memory devices, magnetic disks (e.g., internal hard disks and removable disks), magneto-optical disks, and optical disks. The processor and the memory can be supplemented by, or incorporated in, ASICs (application-specific integrated circuits).

[0184] The systems and techniques described herein can be implemented in a computing system that includes a back end component e.g., a data server), or that includes a middleware component e.g., an application server), or that includes a front end component (e.g., a client computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). The computer system can include clients and servers, which can be generally remote from each other and typically interact through a network, such as the described one. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

[0185] While this specification contains many specific implementation details, these should not be construed as limitations on the scope of the disclosed technology or of what may be claimed, but rather as descriptions of features that may be specific to particular embodiments of particular disclosed technologies. Certain features that are described in this specification in the context of separate embodiments can also be implemented in combination in a single embodiment in part or in whole. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination. Moreover, although features may be described herein as acting in certain combinations and / or initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a subcombination or variation of a subcombination. Similarly, while operations may be described in a particular order, this should not be understood as requiring that such operations be performed in the particular order or in sequential order, or that all operations be performed, to achieve desirable results. Particular embodiments of the subject matter have been described. Other embodiments are within the scope of the following claims.

[0186] One skilled in the art will appreciate further features and advantages of the present disclosure based on the above-described embodiments. Accordingly, the disclosure is not to be limited by what has been particularly shown and described. Further, a person skilled in the art, in view of the present disclosures, will understand how to implement the disclosed systems and methods provided for herein in conjunction with at least vat polymerization printers, including SLA-style and DLP-style additive manufacturing printers. All publications and references cited herein are expressly incorporated herein by reference in their entireties.

[0187] Examples of the above-described embodiments can include the following:1. A method of analyzing a gene to determine susceptibility to a particular disease, comprising: receiving a gene selection from a user, the gene selection being one gene of a plurality of genes stored in a database, the plurality of genes being linked to one or more associated diseases; receiving an amino acid position selection from the user that made the gene selection; and outputting information that includes at least one of: a signal-to-noise ratio of the selected gene-amino acid combination, a relative risk level of the one or more associated diseases, an indication of whether the signal-to-noise ratio corresponds to a statistical mutation hotspot, or a disease susceptibility determination.2. The method of example 1 , wherein the outputted information further comprises information about a functional domain for the selected gene-amino acid combination.3. The method of example 1 or 2, wherein the outputted information comprises data generated from population predictions for one or more of the plurality of genes stored in the database.4. The method of example 3, wherein the population predictions are determined based on providing the selected gene-amino acid combination as input to an artificial intelligence (Al) model.5. The method of any of examples 1 to 4, wherein the outputted information comprises data generated by an Al model, the Al model having been trained to generate the data based on receiving, as model input, the selected gene-amino acid combination.6. A gene analysis tool, comprising: a database of a plurality of genes, the database including information for each gene of the plurality of genes that includes: one or more associated diseases for the gene; a threshold signal-to-noise value for each disease of the one or more associated diseases; and a signal-to-noise value for each amino acid position of the gene; a processor configured to: receive user input that includes a gene selection from the database of a plurality of genes and an amino acid position selection; and output information about disease susceptibility in view of the gene selection and the amino acid position selection based on the information in the database.7. The tool of example 6, wherein the information for each gene of the plurality of genes comprises: a signal-to-noise ratio of the selected gene-amino acid combination; and a signal-to-noise threshold value of the selected gene and the one or more associated diseases for the gene.8. The tool of example 6 or 7, wherein the information for each gene of the plurality of genes further comprises a functional domain for the selected gene-amino acid combination.9. The tool of any one of examples 6 to 8, wherein the processor is further configured to: compare the signal-to-noise ratio of the selected gene-amino acid combination and the signal-to-noise threshold value of the selected gene and the one or more associated diseases for the gene; determine, based on the comparison, whether the signal-to-noise threshold value is exceeded by the signal-to-noise ratio of the selected gene-amino acid combination; and output the determination about the signal-to-noise ratio as compared to the signal-to- noise threshold value.10. The tool of any one of examples 6 to 9, wherein the information for each gene of the plurality of genes further comprises a functional domain for the selected gene-amino acid combination, and wherein the processor is further configured to: identify the functional domain for the selected gene-amino acid combination;and output the identified functional domain.1 1. The tool of any of examples 6 to 10, wherein the tool is web-based.12. The tool of any one of examples 6 to 11, wherein the processor is further configured to present a graphical user interface (GUI) in a display of a computing device of a user, and wherein the GUT includes a first graphical element that corresponds to the gene selection and a second graphical element that corresponds to the amino acid position selection.13. The tool of example 12, wherein at least one of the first and second graphical elements comprises dropdown menus and the user input comprises selection of an option in a respective dropdown menu.14. The tool of any one of example 6 to 13, wherein the information about disease susceptibility is presented in a graph.15. A computer-readable medium storing instructions that, when executed by a processor, cause the processor to: receive a gene selection from a user, the gene selection being one gene of a plurality of genes stored in a database, the plurality of genes being linked to one or more associated diseases; receive an amino acid position selection from the user that made the gene selection; and output information that includes at least one of: a signal-to-noise ratio of the selected gene-amino acid combination, a relative risk level of the one or more associated diseases, an indication of whether the signal-to-noise ratio corresponds to a statistical mutation hotspot, or a disease susceptibility determination.16. The computer-readable medium of example 15, wherein the outputted information comprises data generated by an Al model, the Al model having been trained to generate the data based on receiving, as model input, the selected gene-amino acid combination.17. The computer-readable medium of example 15 or 16, wherein the processor is further configured to present a graphical user interface (GUI) in a display of a computing device of the user, and wherein the GUI includes a first graphical element that corresponds to the gene selection and a second graphical element that corresponds to the amino acid position selection.18. The computer-readable medium of any one of examples 15 to 17, wherein the information for each gene of the plurality of genes further comprises a functional domain for the selected gene- amino acid combination, and wherein the processor is further configured to: identify the functional domain for the selected gene-amino acid combination; and output the identified functional domain in a GUI at a computing device of the user.

Claims

What is claimed is:

1. A method of analyzing a gene to determine susceptibility to a particular disease, comprising: receiving a gene selection from a user, the gene selection being one gene of a plurality of genes stored in a database, the plurality of genes being linked to one or more associated diseases; receiving an amino acid position selection from the user that made the gene selection; and outputting information that includes at least one of: a signal-to-noise ratio of the selected gene-amino acid combination, a relative risk level of the one or more associated diseases, an indication of whether the signal-to-noise ratio corresponds to a statistical mutation hotspot, or a disease susceptibility determination.

2. The method of claim 1 , wherein the outputted information further comprises information about a functional domain for the selected gene-amino acid combination.

3. The method of claim 1, wherein the outputted information comprises data generated from population predictions for one or more of the plurality of genes stored in the database.

4. The method of claim 3, wherein the population predictions are determined based on providing the selected gene-amino acid combination as input to an artificial intelligence (Al) model.

5. The method of claim 1, wherein the outputted information comprises data generated by an Al model, the Al model having been trained to generate the data based on receiving, as model input, the selected gene-amino acid combination.

6. A gene analysis tool, comprising: a database of a plurality of genes, the database including information for each gene of the plurality of genes that includes: one or more associated diseases for the gene; a threshold signal-to-noise value for each disease of the one or more associated diseases; and a signal-to-noise value for each amino acid position of the gene; a processor configured to:receive user input that includes a gene selection from the database of a plurality of genes and an amino acid position selection; and output information about disease susceptibility in view of the gene selection and the amino acid position selection based on the information in the database.

7. The tool of claim 6, wherein the information for each gene of the plurality of genes comprises: a signal-to-noise ratio of the selected gene-amino acid combination; and a signal-to-noise threshold value of the selected gene and the one or more associated diseases for the gene.

8. The tool of claim 6, wherein the information for each gene of the plurality of genes further comprises a functional domain for the selected gene-amino acid combination.

9. The tool of claim 6, wherein the processor is further configured to: compare the signal-to-noise ratio of the selected gene-amino acid combination and the signal-to-noise threshold value of the selected gene and the one or more associated diseases for the gene; determine, based on the comparison, whether the signal-to-noise threshold value is exceeded by the signal-to-noise ratio of the selected gene-amino acid combination; and output the determination about the signal-to-noise ratio as compared to the signal-to- noise threshold value.

10. The tool of claim 6, wherein the information for each gene of the plurality of genes further comprises a functional domain for the selected gene-amino acid combination, and wherein the processor is further configured to: identify the functional domain for the selected gene-amino acid combination; and output the identified functional domain.

11. The tool of claim 6, wherein the tool is web-based.

12. The tool of claim 6, wherein the processor is further configured to present a graphical user interface (GUI) in a display of a computing device of a user, andwherein the GUI includes a first graphical element that corresponds to the gene selection and a second graphical element that corresponds to the amino acid position selection.

13. The tool of claim 12, wherein at least one of the first and second graphical elements comprises dropdown menus and the user input comprises selection of an option in a respective dropdown menu.

14. The tool of claim 6, wherein the information about disease susceptibility is presented in a graph.

15. A computer-readable medium storing instructions that, when executed by a processor, cause the processor to: receive a gene selection from a user, the gene selection being one gene of a plurality of genes stored in a database, the plurality of genes being linked to one or more associated diseases; receive an amino acid position selection from the user that made the gene selection; and output information that includes at least one of: a signal-to-noise ratio of the selected gene-amino acid combination, a relative risk level of the one or more associated diseases, an indication of whether the signal-to-noise ratio corresponds to a statistical mutation hotspot, or a disease susceptibility determination.

16. The computer-readable medium of claim 15, wherein the outputted information comprises data generated by an Al model, the Al model having been trained to generate the data based on receiving, as model input, the selected gene-amino acid combination.

17. The computer-readable medium of claim 15, wherein the processor is further configured to present a graphical user interface (GUI) in a display of a computing device of the user, and wherein the GUI includes a first graphical element that corresponds to the gene selection and a second graphical element that corresponds to the amino acid position selection.

18. The computer-readable medium of claim 15, wherein the information for each gene of the plurality of genes further comprises afunctional domain for the selected gene-amino acid combination, and wherein the processor is further configured to: identify the functional domain for the selected gene-amino acid combination; and output the identified functional domain in a GUI at a computing device of the user.

Images