Frameomics for assessing fetal scores in non-invasive prenatal detection

Abstract

The present technology relates, in part, to estimating a fetal score in non-invasive prenatal detection using one or more fragment omics parameters. In some aspects, the technology relates to estimating a fetal fraction from nucleic acid fragment length and sequence motif frequency.

Description

[0001] Related patent applications

[0002] This patent application claims the benefit of U.S. Provisional Patent Application No. 63 / 451,151, filed on March 9, 2023, entitled “FRAGMENTOMICS FORESTIMATING FETAL FRACTION IN NON-INVASIVE PRENATAL TESTING”, inventors are Fan SONG et al., attorney docket number ILM-1001PROV. The entire contents of the foregoing patent application are incorporated herein by reference for all purposes, including all text, tables and drawings. Technical Field

[0003] The present technology relates in part to estimating fetal fraction in non-invasive prenatal testing using one or more fragmentomic parameters. In some aspects, the present technology relates to estimating fetal fraction based on nucleic acid fragment length and sequence motif frequency. Background Art

[0004] The genetic information of living organisms (e.g., animals, plants, and microorganisms) and other forms of replicating genetic information (e.g., viruses) is encoded as deoxyribonucleic acid (DNA) or ribonucleic acid (RNA). Genetic information is a series of nucleotides or modified nucleotides that represent the primary structure of a chemical or hypothetical nucleic acid. In humans, the complete genome contains approximately 30,000 genes located on twenty-three (23) chromosomes. Each gene encodes a specific protein that, after expression through transcription and translation, carries out a specific biochemical function within a living cell.

[0005] Many medical conditions are caused by one or more gene mutations. Some gene mutations cause medical conditions, including, for example, hemophilia, thalassemia, Duchenne muscular dystrophy (DMD), Huntington's disease (HD), Alzheimer's disease and cystic fibrosis (CF). Such genetic diseases may be caused by the addition, substitution or deletion of a single nucleotide in the DNA of a specific gene. Some birth defects are caused by chromosomal abnormalities (also referred to as aneuploidy), such as 21-trisomy (Down syndrome), 13-trisomy (Patau syndrome), 18-trisomy (Edward syndrome), X monomer (Turner syndrome) and some sex chromosome aneuploidy such as Klinefelter syndrome (XXY). Another gene mutation is fetal sex, which can usually be determined according to sex chromosomes X and Y. Some gene mutations may make individuals tend to or cause any of many diseases, such as diabetes, arteriosclerosis, obesity, various autoimmune diseases and cancer (for example, colorectal cancer, breast cancer, ovarian cancer, lung cancer).

[0006] Identifying one or more gene variations or variances can help diagnose specific medical conditions or determine the susceptibility of specific medical conditions. Identifying genetic variances can facilitate medical decision-making and / or adopt useful medical procedures. In certain embodiments, the identification of one or more gene variations or variances involves the analysis of free DNA. Free DNA (cfDNA) is composed of DNA fragments that originate from cell death and circulate in peripheral blood. High concentrations of cfDNA can indicate certain clinical conditions, such as cancer, trauma, burns, myocardial infarction, stroke, sepsis, infection and other diseases. In addition, free fetal DNA (cffDNA) can be detected in maternal bloodstream and used for various non-invasive prenatal diagnosis.

[0007] Due to the presence of fetal nucleic acid in maternal plasma, non-invasive prenatal diagnosis can be performed by analyzing maternal blood samples. For example, quantitative abnormalities in fetal DNA in maternal plasma can be associated with many pregnancy-related conditions, including preeclampsia, preterm birth, antepartum hemorrhage, placenta accreta, fetal Down syndrome, and other fetal chromosomal aneuploidies. Therefore, fetal nucleic acid analysis in maternal plasma can be a useful mechanism for monitoring fetal-maternal health.

[0008] Fetal fraction (FF) is the percentage of maternal plasma free DNA (cfDNA) that is of fetal-placental origin. Accurate measurement of FF is critical for quality control and performance of non-invasive prenatal testing (NIPT). Low FF can lead to “no-call” results due to the limit of detection (LOD). Higher FF enables greater statistical separation of aneuploid and euploid pregnancies and improves detection rates. This article describes methods that exploit nucleic acid fragment length and fragment end motif frequencies in a machine learning framework for FF estimation. Summary of the invention

[0009] In certain aspects, methods are provided for estimating the fraction of fetal nucleic acid in a test sample from a pregnant subject, the method comprising: a) obtaining sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating free (CCF) nucleic acid from a test sample from a pregnant subject; b) measuring fragment lengths of a plurality of circulating free nucleic acid fragments; c) generating one or more fragment length profiles for the test sample; d) determining sequence motifs at ends of a plurality of circulating free nucleic acid fragments; e) determining one or more sequence motif frequencies for the test sample; and f) estimating the fraction of fetal nucleic acid for the test sample based on i) the one or more fragment length profiles for the test sample and ii) the one or more sequence motif frequencies for the test sample.

[0010] Also provided in certain aspects are systems comprising one or more microprocessors and a memory, the memory comprising instructions executable by the one or more microprocessors, and the memory comprising sequence reads mapped to a reference genome, the sequence reads being reads of circulating free (CCF) nucleic acid from a test sample from a pregnant subject, and the instructions executable by the one or more microprocessors being configured to a) measure fragment lengths of a plurality of circulating free nucleic acid fragments; b) generate one or more fragment length profiles of the test sample; c) determine sequence motifs at ends of a plurality of circulating free nucleic acid fragments; d) determine one or more sequence motif frequencies of the test sample; and e) estimate a fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0011] Also provided in certain aspects is a machine comprising one or more microprocessors and a memory, the memory containing instructions executable by the one or more microprocessors, and the memory containing sequence reads mapped to a reference genome, the sequence reads being reads of circulating free nucleic acid from a test sample of a pregnant subject, and the instructions executable by the one or more microprocessors being configured to a) measure fragment lengths of a plurality of circulating free nucleic acid fragments; b) generate one or more fragment length profiles of the test sample; c) determine sequence motifs at ends of a plurality of circulating free nucleic acid fragments; d) determine one or more sequence motif frequencies of the test sample; and e) estimate a fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0012] Also provided in certain aspects is a non-transitory computer-readable storage medium having an executable program stored thereon, wherein the program instructs a microprocessor to perform the following operations: a) access sequence reads mapped to a reference genome, which are reads of circulating free nucleic acid from a test sample from a pregnant subject, b) measure fragment lengths of a plurality of circulating free nucleic acid fragments; c) generate one or more fragment length profiles for the test sample; d) determine sequence motifs at ends of a plurality of circulating free nucleic acid fragments; e) determine one or more sequence motif frequencies for the test sample; and f) estimate a fetal nucleic acid fraction for the test sample based on i) the one or more fragment length profiles for the test sample and ii) the one or more sequence motif frequencies for the test sample.

[0013] Certain embodiments are further described in the following detailed description, examples and claims, and in the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings illustrate certain embodiments of the present technology and are not limiting. For clarity and ease of illustration, the accompanying drawings are not drawn to scale, and in some cases, various aspects may be shown exaggerated or enlarged to facilitate understanding of specific implementations.

[0015] Figure 1 An exemplary workflow of the method described herein is shown. Group 1: Data collection—an internal dataset generated from sequencing cfDNA of approximately 1,500 pregnant women. Group 2: Feature extraction and data processing. The sequencing data was processed with Illumina's DRAGEN platform to obtain fragment spectra of 5Mb genomic bins and sequence motif frequencies at the chromosome level. Group 3: Dimensionality reduction—Features were analyzed using principal component analysis (PCA) to extract the variables most relevant to predicting fetal scores. The dataset was then divided into training and testing in an 80-20 ratio. Group 4: Model training—Three models including linear regression, elastic network, and XGBoost were used in model training. Techniques such as random parameter search and 5-fold cross validation (CV) were used to prevent model overfitting. Group 5: The model was evaluated on the test dataset using two metrics: root mean square error (RMSE) and correlation with the true fetal score.

[0016] Figure 2 The DRAGEN data processing method was shown to be at least two orders of magnitude faster than existing tools.

[0017] Figure 3 Shown are fragment profiles (upper row) and sequence motif frequencies (lower row) from DRAGEN (x-axis) that are highly consistent with existing tools (y-axis) in TSO500 and whole exome sequencing (WES) cfDNA datasets.

[0018] Figure 4 Models trained with fragment size only and elastic net or XGBoost models are shown. Fetal fractions predicted by the models (y-axis) show a high degree of agreement with the true data (x-axis).

[0019] Figure 5 Models trained with only 5' end sequence motif frequencies and elastic net or XGBoost models are shown. Fetal fractions predicted by the models (y-axis) show high consistency with the true data (x-axis).

[0020] Figure 6 A table is provided showing that the sequence motif based method exhibited a similar error rate of 2.5% when compared to the traditional fragment size based method.

[0021] Figure 7An overview of the performance of various models is shown, showing that combining sequence motifs with fragment size or coverage features improves prediction accuracy. RMSE (left) and correlation (right) of different models. The Y-axis, from top to bottom, is the predictions from: 1) Distributed Random Forest (DRF) using only sequence motifs; 2) Gradient Boosting (GBM) using only sequence motifs; 3) XGBoost using only sequence motifs; 4) FF_Size model from the NIPT team; 5) Elastic Net / Generalized Linear Model (GLM) using only sequence motifs; 6) FF_Coverage model, which uses whole genome read coverage to predict fetal fraction; 7) GLM model using both fragment size and sequence motifs; 8) FF_Cov_Size model from the NIPT team; 9) GLM model using fragment coverage and sequence motifs; 10) FF4 model from the NIPT team; 11) XGBoost model using fragment size, coverage, and sequence motifs; 12) FF_Cov_Size_recompute model from the NIPT team; 13) GLM model using fragment size, coverage, and sequence motifs. DETAILED DESCRIPTION

[0022] Provided herein are methods and systems for estimating the fetal nucleic acid fraction in a test sample. The systems and methods herein may include estimating the fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0023] Determination of fetal nucleic acid content

[0024] The methods and systems herein relate to estimating the amount (e.g., concentration, relative amount, absolute amount, copy number, etc.) of fetal nucleic acid in nucleic acid. In certain embodiments, the amount of fetal nucleic acid in a sample (e.g., a test sample) is referred to as the "fetal nucleic acid fraction" or "fetal fraction". In some embodiments, the "fetal fraction" refers to the fetal nucleic acid fraction in the circulating free nucleic acid in a sample (e.g., a blood sample, a serum sample, a plasma sample) obtained from a pregnant subject. The fetal fraction can be estimated according to the fragment length spectrum and sequence motif frequency as described herein. As described herein, the fetal fraction can be estimated by applying one or more model parameters to the fragment length spectrum and sequence motif frequency determined for the test sample. The model parameters can be obtained from a sample training set with a known fetal fraction (e.g., a sample premixed with a known amount of maternal and fetal nucleic acid or a sample in which the fetal fraction is determined according to any suitable method in the art). The fetal fraction of the training sample and / or the test sample can be determined in a suitable manner (e.g., for evaluating the accuracy of the fetal fraction estimation method herein), non-limiting examples of which include the methods described below.

[0025] In certain embodiments, the method is based on markers that are specific to the male fetus (e.g., Y-chromosome STR markers (e.g., DYS19, DYS385, DYS392 markers); RhD markers for RhD-negative females), allele ratios of polymorphic sequences, or one or more markers that are specific to fetal nucleic acid but not maternal nucleic acid (e.g., differential epigenetic biomarkers between mother and fetus (e.g., methylation; described in further detail below), or fetal RNA markers in maternal plasma).

[0026] Sometimes a fetal quantification assay (FQA) is used to determine the fetal nucleic acid content (e.g., fetal fraction). This type of assay allows for the detection and quantification of fetal nucleic acid in a maternal sample based on the methylation status of the nucleic acid in the sample. In certain embodiments, the amount of fetal nucleic acid from a maternal sample can be determined relative to the total amount of nucleic acid present, thereby providing a percentage of fetal nucleic acid in the sample. Methods for distinguishing nucleic acids based on methylation status include, but are not limited to, methylation-sensitive capture, such as using an MBD2-Fc fragment, in which the methyl-binding domain of MBD2 is fused to the Fc fragment of an antibody (MBD-FC); methylation-specific antibodies; bisulfite conversion methods, such as MSP (methylation-sensitive PCR), COBRA, methylation-sensitive single nucleotide primer extension (Ms-SNuPE), or Sequenom MassCLEAVE TM Techniques; and the use of methylation-sensitive restriction enzymes (e.g., using one or more methylation-sensitive restriction enzymes to digest maternal DNA in a maternal sample, thereby enriching fetal DNA). Methyl-sensitive enzymes can also be used to distinguish nucleic acids based on methylation status, such that, for example, when their DNA recognition sequences are not methylated, they can be preferentially or substantially cleaved or digested at the sequence. Therefore, unmethylated DNA samples will be cut into smaller fragments than methylated DNA samples, while highly methylated DNA samples will not be cleaved.

[0027] In certain embodiments, the fetal fraction can be determined based on the allele ratio of a polymorphic sequence (e.g., a single nucleotide polymorphism (SNP)). In such methods, nucleotide sequence reads are obtained for a maternal sample, and the fetal fraction is determined by comparing the total number of nucleotide sequence reads mapped to the first allele at an informative polymorphic site (e.g., SNP) in a reference genome with the total number of nucleotide sequence reads mapped to the second allele. In certain embodiments, fetal alleles are identified by, for example, a mixture of fetal and maternal nucleic acids in a sample, a relatively small contribution of the fetal allele compared to the larger contribution of the maternal nucleic acid to the mixture. Therefore, for each of the two alleles of the polymorphic site, the relative abundance of fetal nucleic acid in the maternal sample can be determined as a parameter of the total number of unique sequence reads of the target nucleic acid sequence mapped to the reference genome.

[0028] sample

[0029] Provided herein are methods and compositions for analyzing nucleic acids. In some embodiments, nucleic acid fragments in a mixture of nucleic acid fragments are analyzed. Nucleic acid mixtures may include two or more nucleic acid fragment species having different nucleotide sequences, different fragment lengths, different sources (e.g., genomic sources, fetal to maternal sources, tumor to host sources, cell or tissue sources, sample sources, subject sources, etc.), or combinations thereof.

[0030] The nucleic acid or nucleic acid mixture used in the methods and devices described herein is usually isolated from a sample obtained from a subject. The subject can be any living or non-living organism, including but not limited to humans, non-human animals, plants, bacteria, fungi or protists. Any human or non-human animal can be selected, including but not limited to mammals, reptiles, birds, amphibians, fish, ungulates, ruminants, bovines (e.g., cattle), equines (e.g., horses), subfamily Ovis and genus Ovis (e.g., sheep, goats), suids (e.g., pigs), camelids (e.g., camels, llamas, alpacas), monkeys, apes (e.g., gorillas, chimpanzees), ursids (e.g., bears), poultry, dogs, cats, mice, rats, fish, dolphins, whales and sharks. The subject can be male, female, hermaphrodite or non-binary. The subject can be of any age (e.g., embryo, fetus, infant, child, adult). The subject can be pregnant or non-pregnant.

[0031] Nucleic acid can be separated from any type of suitable biological specimen or sample (e.g., test sample). Sample or test sample can be any specimen separated or obtained from a subject or part thereof (e.g., human subject, pregnant subject, fetus). Non-limiting examples of specimens include fluids or tissues from a subject, including but not limited to blood or blood products (e.g., serum, plasma, etc.), cord blood, chorionic villi, amniotic fluid, cerebrospinal fluid, spinal fluid, lavage fluid (e.g., bronchoalveolar, stomach, peritoneum, catheter, ear, arthroscopy), biopsy samples (e.g., from preimplantation embryos), celocentesis samples, cells (blood cells, tumor cells, placental cells, embryonic or fetal cells, fetal nucleated cells, fetal cell residues) or parts thereof (e.g., mitochondria, nuclei, extracts, etc.), female reproductive tract washings, urine, feces, sputum, saliva, nasal mucus, prostatic fluid, lavage fluid, semen, lymph, bile, tears, sweat, breast milk, milk, etc., or combinations thereof. In some embodiments, the biological sample is a cervical swab from a subject. In some embodiments, the biological sample is blood. As used herein, the term "blood" refers to a blood sample or preparation from a pregnant subject or a subject who may be pregnant after testing. The term covers whole blood, blood products, or any part of blood, such as serum, plasma, buffy coat, etc., as conventionally defined. Blood or a portion thereof generally contains nucleosomes (e.g., maternal and / or fetal nucleosomes). Nucleosomes contain nucleic acids and are sometimes free or intracellular. Blood also contains buffy coat. Sometimes the buffy coat is separated using a ficoll gradient. The buffy coat may contain leukocytes (e.g., white blood cells, T cells, B cells, platelets, etc.). In certain embodiments, the buffy coat contains maternal and / or fetal nucleic acids. In some embodiments, the biological sample is plasma. Plasma refers to the whole blood portion obtained after centrifugation of blood treated with an anticoagulant. In some embodiments, the biological sample is serum. Serum refers to the watery portion of the fluid remaining after the blood sample coagulates. Fluid or tissue samples are usually collected according to the standard protocols generally followed in hospitals or clinics. For blood, an appropriate amount of peripheral blood (e.g., between 3 ml and 40 ml) is usually collected, and can be stored according to standard procedures before or after preparation. The fluid or tissue sample from which nucleic acid is extracted can be acellular (e.g., free). In some embodiments, fluid or tissue samples can contain cellular components or cell residues. In some embodiments, fetal cells or cancer cells can be included in the sample.

[0032] The sample is usually heterogeneous, which means that there is more than one type of nucleic acid species in the sample. For example, heterogeneous nucleic acids may include, but are not limited to, (i) nucleic acids of fetal origin and maternal origin, (ii) cancer and non-cancer nucleic acids, (iii) pathogen and host nucleic acids, and more generally, (iv) mutant and wild-type nucleic acids. The sample can be heterogeneous because there is more than one cell type, such as fetal cells and maternal cells, cancer cells and non-cancerous cells, or pathogenic cells and host cells. In some embodiments, there are a minority of nucleic acid species and a majority of nucleic acid species.

[0033] For prenatal applications of the technology described herein, fluid or tissue samples can be collected from subjects of a gestational age suitable for testing or subjects being tested for possible pregnancy. Suitable gestational age can vary depending on the prenatal testing performed. Pregnant subjects can be in the first three months of pregnancy, can be in the second three months of pregnancy, or can be in the third trimester of pregnancy. In certain embodiments, fluid or tissue is collected from a pregnant subject between about 1 week to about 45 weeks of fetal gestation (e.g., between 1-4 weeks, 4-8 weeks, 8-12 weeks, 12-16 weeks, 16-20 weeks, 20-24 weeks, 24-28 weeks, 28-32 weeks, 32-36 weeks, 36-40 weeks, or 40-44 weeks of fetal gestation), and sometimes between about 5 weeks to about 28 weeks of fetal gestation (e.g., between 6 weeks, 7 weeks, 8 weeks, 9 weeks, 10 weeks, 11 weeks, 12 weeks, 13 weeks, 14 weeks, 15 weeks, 16 weeks, 17 weeks, 18 weeks, 19 weeks, 20 weeks, 21 weeks, 22 weeks, 23 weeks, 24 weeks, 25 weeks, 26 weeks, or 27 weeks of fetal gestation). In certain embodiments, a fluid or tissue sample is collected from a pregnant subject during or shortly after delivery (eg, vaginal delivery or non-vaginal delivery (eg, surgical delivery)) or shortly after delivery (eg, 0 to 72 hours after delivery).

[0034] Sample acquisition and nucleic acid extraction

[0035] The methods herein may include isolating, enriching and analyzing fetal DNA found in maternal blood as a non-invasive means to detect the presence or absence of maternal and / or fetal genetic variations and / or monitor the health of the fetus and / or pregnant subject during pregnancy and sometimes after pregnancy. Thus, the first step in implementing certain methods herein may include obtaining a blood sample from a pregnant subject and extracting DNA from the sample.

[0036] The method of the present technology can be used to obtain blood samples from pregnant subjects of the gestational age suitable for testing. Suitable gestational age can vary according to the disease tested. Blood is usually collected from subjects according to the standard protocol generally followed in hospitals or clinics. Usually an appropriate amount of peripheral blood (e.g., typically between 5ml-50ml) is collected, and can be stored according to standard procedures before further preparation. The degradation of the nucleic acid present in the sample or the mode of quality minimization can be made to collect, store or transport blood samples.

[0037] For example, whole blood, serum or plasma can be used to analyze the fetal DNA present in maternal blood. Methods for preparing serum or plasma from maternal blood are known. For example, the blood of a pregnant subject can be placed in a tube containing EDTA or a dedicated commercial product such as Vacutainer SST (Becton Dickinson, Franklin Lakes, NJ) to prevent blood clotting, and then plasma can be obtained from the whole blood by centrifugation. Serum can be obtained by centrifugation or without centrifugation after blood clotting. If centrifugation is used, it is usually (although not exclusively) carried out at a suitable speed (e.g., 1,500-3,000 × g). Plasma or serum can be subjected to an additional centrifugation step before being transferred to a new test tube for DNA extraction.

[0038] In addition to the cell-free fraction of whole blood, DNA can also be recovered from the cellular fraction enriched in the buffy coat fraction, which can be obtained after centrifuging a whole blood sample from a female and removing the plasma.

[0039] There are many known methods for extracting DNA from biological samples, including blood. Various commercially available reagents or kits can be used, such as Qiagen's QIAamp Circulating Nucleic Acid Kit, QiaAmp DNA Mini Kit or QiaAmp DNA Blood Mini Kit (Qiagen, Hilden, Germany), GenomicPrep TM Blood DNA Isolation Kit (Promega, Madison, Wis.) or GFX TM The general method of DNA preparation was performed using the Genomic Blood DNA Purification Kit (Amersham, Piscataway, NJ). A combination of more than one of these methods may also be used.

[0040] In certain embodiments, nucleic acid can be provided for carrying out the methods described herein without processing a sample containing the nucleic acid. In some embodiments, after processing a sample containing the nucleic acid, the nucleic acid is provided for carrying out the methods described herein. For example, nucleic acid can be extracted, separated, purified, partially purified or amplified from a sample. As used herein, the term "isolated" refers to a nucleic acid removed from its original environment (e.g., a natural environment if naturally present, or a host cell if exogenously expressed), and then altered from its original environment by human intervention (e.g., "by artificial"). As used herein, the term "isolated nucleic acid" may refer to a nucleic acid removed from a subject (e.g., a human subject). An isolated nucleic acid may have a non-nucleic acid component (e.g., protein, lipid) less than the amount of the component present in the source sample. A composition comprising an isolated nucleic acid may be about 50% to greater than 99% free of non-nucleic acid components. Compositions comprising isolated nucleic acids may be about 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or greater than 99% free of non-nucleic acid components. As used herein, the term "purified" may refer to a provided nucleic acid containing less non-nucleic acid components (e.g., proteins, lipids, carbohydrates) than the amount of non-nucleic acid components present before the nucleic acid is subjected to a purification procedure. Compositions comprising purified nucleic acids may be about 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or greater than 99% free of other non-nucleic acid components. As used herein, the term "purified" may refer to a provided nucleic acid containing fewer nucleic acid species than the sample source from which the nucleic acid is derived. Compositions comprising purified nucleic acids can be about 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or greater than 99% free of other nucleic acid species. For example, fetal nucleic acid can be purified from a mixture comprising maternal and fetal nucleic acid. In some examples, nucleosomes comprising small fragments of fetal nucleic acid can be purified from a mixture of larger nucleosome complexes comprising larger fragments of maternal nucleic acid.

[0041] In some embodiments, nucleic acids are fragmented or cleaved before, during, or after the methods described herein. The fragmented or cleaved nucleic acids can have a nominal, average, or mean length of about 5 to about 10,000 base pairs, about 100 to about 1,000 base pairs, about 100 to about 500 base pairs, or about 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, or 9000 base pairs. Fragments can be generated by suitable methods known in the art, and the average, mean or nominal length of nucleic acid fragments can be controlled by selecting an appropriate fragment generation program.

[0042] In some embodiments, the nucleic acid is fragmented or cleaved by a suitable method, non-limiting examples of which include physical methods (e.g., shearing, such as sonication, French press, heating, ultraviolet irradiation, etc.), enzymatic methods (e.g., enzymatic cleavage agents (e.g., suitable nucleases, suitable restriction enzymes, suitable methylation-sensitive restriction enzymes)), chemical methods (e.g., alkylation, DMS, piperidine, acid hydrolysis, base hydrolysis, heating, etc. or combinations thereof), etc., or combinations thereof.

[0043] Prior to providing the nucleic acid for use in the methods described herein, the nucleic acid may also be exposed to a process for modifying certain nucleotides in the nucleic acid. For example, methods for selectively modifying nucleic acids based on the methylation state of nucleotides in the nucleic acid may be applied to nucleic acids. In addition, conditions such as high temperature, ultraviolet radiation, x-radiation may induce changes in the sequence of nucleic acid molecules. The nucleic acid may be provided in any suitable form for performing suitable sequence analysis.

[0044] In some embodiments, the sample can first be enriched or relatively enriched for fetal nucleic acid by one or more methods. For example, the distinction between fetal and maternal DNA can be performed using certain distinguishing factors. Examples of these factors include, but are not limited to, single nucleotide differences between chromosomes X and Y, chromosome Y-specific sequences, polymorphisms located elsewhere in the genome, size differences between fetal and maternal DNA, and differences in methylation patterns between maternal and fetal tissues. In some applications, maternal nucleic acid is selectively removed (partially, substantially, almost completely, or completely) from the sample.

[0045] Nucleic Acids

[0046] The terms "nucleic acid" and "nucleic acid molecule" are used interchangeably throughout the disclosure. The term refers to nucleic acids of any composition, such as DNA (e.g., complementary DNA (cDNA), genomic DNA (gDNA), etc.), RNA (e.g., messenger RNA (mRNA), short inhibitory RNA (siRNA), ribosomal RNA (rRNA), tRNA, microRNA, RNA highly expressed by the fetus or placenta, etc.) and / or DNA or RNA analogs (e.g., containing base analogs, sugar analogs and / or non-natural backbones, etc.), RNA / DNA hybrids and polyamide nucleic acids (PNA), all of which can be in single-stranded or double-stranded form, and unless otherwise limited, known analogs of natural nucleotides that can function in a manner similar to naturally occurring nucleotides can be encompassed. Nucleic acids can be or can be derived from plasmids, bacteriophages, autonomously replicating sequences (ARS), centromeres, artificial chromosomes, chromosomes, or other nucleic acids that can replicate or be replicated in vitro or in certain cases in host cells, cells, nuclei, or cytoplasm. In some embodiments, the template nucleic acid can be from a single chromosome (e.g., a nucleic acid sample can be from a chromosome of a sample obtained from a diploid organism). Unless otherwise limited, the term encompasses nucleic acids containing known analogs of natural nucleotides, which have binding properties similar to reference nucleic acids, and are metabolized in a manner similar to naturally occurring nucleotides. Unless otherwise specified, specific nucleic acid sequences also implicitly encompass variants (e.g., degenerate codon substitutions), alleles, orthologs, single nucleotide polymorphisms (SNPs) and complementary sequences of its conservative modifications, as well as sequences clearly indicated. Specifically, degenerate codon substitutions can be achieved by producing a sequence in which the third position of one or more selected (or all) codons is replaced by mixed bases and / or deoxyinosine residues. The term nucleic acid can be used interchangeably with locus, gene, cDNA and mRNA encoded by genes. The term can also include RNA or DNA synthesized from nucleotide analogs, single strands ("sense" strands or "antisense" strands, "positive" strands or "negative" strands, "forward" reading frames or "reverse" reading frames) and double-stranded polynucleotides Equivalents, derivatives, variants and analogs. The term "gene" refers to a DNA fragment involved in producing a polypeptide chain; it includes regions before and after the coding region (leader and trailer) that participate in the transcription / translation of the gene product and the regulation of transcription / translation, as well as intervening sequences (introns) between the various coding segments (exons). Deoxyribonucleotides include deoxyadenosine, deoxycytidine, deoxyguanosine, and deoxythymidine. For RNA, the base thymine is replaced by uracil. A template nucleic acid can be prepared using a nucleic acid obtained from a subject as a template.

[0047] In certain embodiments, nucleic acid may include extracellular nucleic acid. As used herein, the term "extracellular nucleic acid" may refer to nucleic acid separated from a source that is substantially free of cells, and is also referred to as "free" nucleic acid, "circulating free nucleic acid" (e.g., CCF fragments) and / or "free circulating nucleic acid". Extracellular nucleic acid may be present in blood and obtained from blood (e.g., obtained from the blood of a pregnant subject). Extracellular nucleic acid does not typically include detectable cells, and may contain cellular components or cell residues. Non-limiting examples of non-cellular sources of extracellular nucleic acid are blood, plasma, serum, and urine. As used herein, the term "obtaining free circulating sample nucleic acid" includes directly obtaining a sample (e.g., collecting a sample, such as a test sample) or obtaining a sample from another person who has collected a sample. Without being limited by theory, extracellular nucleic acid may be a product of apoptosis and cell decomposition, which provides a basis for extracellular nucleic acid that typically has a series of lengths spanning a spectrum (e.g., "ladder").

[0048] Extracellular nucleic acid may include different nucleic acid species and is therefore referred to herein as "heterogeneous" in certain embodiments. For example, serum or plasma from a person with cancer may include nucleic acids from cancer cells and nucleic acids from non-cancerous cells. In another example, serum or plasma from a pregnant subject may include maternal nucleic acids and fetal nucleic acids. In some cases, fetal nucleic acid sometimes comprises between about 5% to about 50% of the total nucleic acid (e.g., about 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, or 49% of the total nucleic acid is fetal nucleic acid). In some embodiments, a majority of the fetal nucleic acid in the nucleic acid has a length of about 500 base pairs or less (e.g., about 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% of the fetal nucleic acid has a length of about 500 base pairs or less). In some embodiments, a majority of the fetal nucleic acid in the nucleic acid has a length of about 250 base pairs or less (e.g., about 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% of the fetal nucleic acid has a length of about 250 base pairs or less). In some embodiments, a majority of the fetal nucleic acid in the nucleic acid has a length of about 200 base pairs or less (e.g., about 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% of the fetal nucleic acid has a length of about 200 base pairs or less). In some embodiments, a majority of the fetal nucleic acid in the nucleic acid has a length of about 150 base pairs or less (e.g., about 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% of the fetal nucleic acid has a length of about 150 base pairs or less). In some embodiments, a majority of the fetal nucleic acid in the nucleic acid has a length of about 100 base pairs or less (e.g., about 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% of the fetal nucleic acid has a length of about 100 base pairs or less).In some embodiments, a majority of the fetal nucleic acid in the nucleic acid has a length of about 50 base pairs or less (e.g., about 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% of the fetal nucleic acid has a length of about 50 base pairs or less). In some embodiments, a majority of the fetal nucleic acid in the nucleic acid has a length of about 25 base pairs or less (e.g., about 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% of the fetal nucleic acid has a length of about 25 base pairs or less).

[0049] Nucleic acid can be single-stranded or double-stranded. For example, single-stranded DNA can be produced by, for example, heating or treating double-stranded DNA with alkali to denature it. In certain embodiments, nucleic acid is in a D-loop structure, formed by oligonucleotides or DNA-like molecules such as peptide nucleic acids (PNA) invading the chains of duplex DNA molecules. D-loop formation can be promoted by adding E. coli RecA protein and / or by changing the salt concentration (for example, using methods known in the art).

[0050] Nucleic acid library

[0051] In some embodiments, nucleic acid libraries are multiple polynucleotide molecules (e.g., samples of nucleic acids) prepared, assembled and / or modified for a particular process, non-limiting examples of which include being fixed on a solid phase (e.g., a solid support, e.g., a flow cell, a bead), enriched, amplified, cloned, detected and / or used for nucleic acid sequencing. In certain embodiments, nucleic acid libraries are prepared before or during a sequencing process. Nucleic acid libraries (e.g., sequencing libraries) can be prepared by suitable methods known in the art. Nucleic acid libraries can be prepared by targeted or non-targeted preparation processes.

[0052] In some embodiments, the library of nucleic acids is modified to include chemical moieties (e.g., functional groups) configured to fix the nucleic acids to a solid support. In some embodiments, the library of nucleic acids is modified to include biomolecules (e.g., functional groups) and / or members of binding pairs configured to fix the library to a solid support, non-limiting examples of which include thyroxine binding globulin, steroid binding proteins, antibodies, antigens, haptens, enzymes, lectins, nucleic acids, repressors, protein A, protein G, avidin, streptavidin, biotin, complement component C1q, nucleic acid binding proteins, receptors, carbohydrates, oligonucleotides, polynucleotides, complementary nucleic acid sequences, etc., and combinations thereof. Some examples of specific binding pairs include, but are not limited to: an avidin portion and a biotin portion; an antigenic epitope and an antibody or an immunologically active fragment thereof; an antibody and a hapten; a digoxigenin portion and an anti-digoxigenin antibody; a fluorescein portion and an anti-fluorescein antibody; an operator and a repressor; a nuclease and a nucleotide; a lectin and a polysaccharide; a steroid and a steroid-binding protein; an active compound and an active compound receptor; a hormone and a hormone receptor; an enzyme and a substrate; an immunoglobulin and protein A; an oligonucleotide or polynucleotide and its corresponding complementary sequence; the like or a combination thereof.

[0053] In some embodiments, the nucleic acid library is modified to include one or more polynucleotides of known composition, non-limiting examples of which include identifiers (e.g., tags, index tags), capture sequences, markers, adapters, restriction enzyme sites, promoters, enhancers, replication origins, stem loops, complementary sequences (e.g., primer binding sites, annealing sites), suitable integration sites (e.g., transposons, viral integration sites), modified nucleotides, etc., or combinations thereof. Polynucleotides of known sequences can be added at suitable positions, such as at the 5' end, the 3' end, or within the nucleic acid sequence. Polynucleotides of known sequences can be identical or different sequences. In some embodiments, polynucleotides of known sequences are configured to hybridize with one or more oligonucleotides fixed on a surface (e.g., a surface in a flow cell). For example, a nucleic acid molecule comprising a 5' known sequence can hybridize with a first plurality of oligonucleotides, and a 3' known sequence can hybridize with a second plurality of oligonucleotides. In some embodiments, the library of nucleic acids can include chromosome-specific tags, capture sequences, markers, and / or adapters. In some embodiments, the library of nucleic acids includes one or more detectable labels. In some embodiments, one or more detectable labels can be incorporated into the nucleic acid library at any nucleotide position in the 5' end, 3' end and / or in the nucleic acid in the library. In some embodiments, the library of nucleic acids comprises hybridized oligonucleotides. In certain embodiments, the hybridized oligonucleotides are label probes. In some embodiments, the nucleic acid library comprises hybridized oligonucleotide probes before being fixed on a solid phase.

[0054] In some embodiments, the polynucleotide of known sequence comprises universal sequence.Universal sequence is a specific nucleotide sequence that is integrated into two or more nucleic acid molecules or two or more subsets of nucleic acid molecules, wherein the universal sequence is identical for all molecules or subsets of molecules to which it is integrated.Universal sequence is usually designed to use a single universal primer complementary to the universal sequence, hybridized with multiple different sequences and / or amplified these multiple different sequences.In some embodiments, two (for example, a pair) or more universal sequences and / or universal primers are used.Universal primers usually comprise universal sequence.In some embodiments, adapter (for example, universal adapter) comprises universal sequence.In some embodiments, one or more universal sequences are used to capture, identify and / or detect multiple types or subsets of nucleic acid.

[0055] In certain embodiments of preparing nucleic acid libraries (e.g., in certain sequencing by synthesis procedures), nucleic acids are size selected and / or fragmented into lengths of a few hundred base pairs or less (e.g., in preparation for library generation). In some embodiments, library preparation is performed without fragmentation (e.g., when ccfDNA is used). For example, certain methods described herein recognize natural end motifs in ccfDNA fragments. Therefore, libraries for such methods are generated using natural fragments from samples and are not subjected to a fragmentation process.

[0056] In certain embodiments, a connection-based library preparation method (e.g., ILLUMINA TRUSEQ, Illumina, San Diego CA) is used. Connection-based library preparation methods are generally designed using adapters (e.g., methylated adapters), which can be incorporated with index sequences in the initial connection step, and are generally useful for preparing samples for single-end sequencing, paired-end sequencing, and / or multiple sequencing. For example, nucleic acids (e.g., fragmented nucleic acids or ccfDNA) can be end-repaired by a filling reaction, an exonuclease reaction, or a combination thereof. In certain configurations, the 5' fragment ends are end-repaired by a filling reaction, and the 3' fragment ends are end-repaired by an exonuclease reaction (e.g., 3' to 5' single-stranded exonuclease). Specifically, the fragment ends with 5' overhangs are end-repaired by a filling reaction, and the fragment ends with 3' overhangs are end-repaired by an exonuclease reaction. In this configuration, the terminal motif sequence is retained at the 5' fragment ends. In some embodiments, the resulting flat-end repaired nucleic acid can then be extended by a single nucleotide that is complementary to the single nucleotide overhang on the 3' end of the adapter / primer. Any nucleotide can be used for extension / protruding nucleotides. In some embodiments, nucleic acid library preparation includes connecting adapter oligonucleotides. Adapter oligonucleotides are usually complementary to flow cell anchors, and can be used to fix nucleic acid libraries to solid carriers such as the inner surface of flow cells. In some embodiments, adapter oligonucleotides include identifiers, one or more sequencing primer hybridization sites (e.g., sequences complementary to universal sequencing primers, single-end sequencing primers, paired end sequencing primers, multiple sequencing primers, etc.) or combinations thereof (e.g., adapter / sequencing, adapter / identifier, adapter / identifier / sequencing).

[0057] The identifier can be a suitable detectable label incorporated into or attached to a nucleic acid (e.g., a polynucleotide) that allows detection and / or identification of the nucleic acid comprising the identifier. In some embodiments, the identifier is incorporated into or attached to a nucleic acid during a sequencing method (e.g., by a polymerase). Non-limiting examples of identifiers include nucleic acid tags, nucleic acid indexes or barcodes, radioactive labels (e.g., isotopes), metal labels, fluorescent labels, chemiluminescent labels, phosphorescent labels, fluorophore quenchers, dyes, proteins (e.g., enzymes, antibodies or parts thereof, joints, members of binding pairs), etc. or combinations thereof. In some embodiments, the identifier (e.g., nucleic acid indexes or barcodes) is a unique, known and / or recognizable sequence of nucleotides or nucleotide analogs. In some embodiments, the identifier is six or more continuous nucleotides. A variety of fluorophores with a variety of different excitation spectra and emission spectra can be obtained. Any suitable type and / or number of fluorophores can be used as identifiers. In some embodiments, one or more, two or more, three or more, four or more, five or more, six or more, seven or more, eight or more, nine or more, ten or more, twenty or more, thirty or more, or fifty or more different identifiers are used in the methods described herein (e.g., nucleic acid detection and / or sequencing methods). In some embodiments, one or two types of identifiers (e.g., fluorescent markers) are connected to each nucleic acid in the library. Detection and / or quantification of identifiers can be performed by suitable methods, devices or machines, non-limiting examples of which include flow cytometry, quantitative polymerase chain reaction (qPCR), gel electrophoresis, photometers, fluorometers, spectrophotometers, suitable gene chips or microarray analysis, protein blotting, mass spectrometry, chromatography, cell fluorescence analysis, fluorescence microscopy, suitable fluorescence or digital imaging methods, confocal laser scanning microscopy, laser scanning cell counting, affinity chromatography, manual batch mode separation, electric field suspension, suitable nucleic acid sequencing methods and / or nucleic acid sequencing devices, and the like, and combinations thereof.

[0058] In some embodiments, a nucleic acid library or a portion thereof is amplified (e.g., by amplification based on a PCR method). In some embodiments, sequencing methods include amplification of a nucleic acid library. A nucleic acid library can be amplified before or after being fixed on a solid support (e.g., a solid support in a flow cell). Nucleic acid amplification includes a process of amplifying or increasing the number of nucleic acid templates and / or their complementary sequences present (e.g., in a nucleic acid library) by generating one or more copies of a template and / or its complementary sequence. Amplification can be implemented by a suitable method. The nucleic acid library can be amplified by thermal cycling or isothermal amplification. In some embodiments, a rolling circle amplification method is used. In some embodiments, amplification occurs on a solid support (e.g., in a flow cell) of an immobilized nucleic acid library or a portion thereof. In some sequencing methods, a nucleic acid library is added to a flow cell and immobilized by hybridizing with an anchor under suitable conditions. This type of nucleic acid amplification is generally referred to as solid phase amplification. In some embodiments of solid phase amplification, all or a portion of the amplified product is synthesized by extension starting from an immobilized primer. Solid phase amplification reactions are similar to standard solution phase amplifications, except that at least one of the amplification oligonucleotides (eg, a primer) is immobilized on a solid support.

[0059] In some embodiments, solid phase amplification includes nucleic acid amplification reaction, and this nucleic acid amplification reaction only includes a kind of oligonucleotide primer fixed to the surface.In certain embodiments, solid phase amplification includes multiple different immobilized oligonucleotide primer species.In some embodiments, solid phase amplification may include nucleic acid amplification reaction, and this nucleic acid amplification reaction includes a kind of oligonucleotide primer fixed on the solid surface and a second different oligonucleotide primer species in the solution.Can use multiple different kinds of immobilized primers or primers based on solution.Non-limiting examples of solid phase nucleic acid amplification reaction include interface amplification, bridge amplification, emulsion PCR, WILDFIRE amplification, etc. or their combination.

[0060] Sequencing

[0061] In some embodiments, nucleic acid (e.g., nucleic acid fragments, sample nucleic acid, test sample nucleic acid, free nucleic acid, circulating free nucleic acid) is sequenced. In some embodiments, a complete or substantially complete sequence is obtained, and sometimes a partial sequence is obtained. In some embodiments, when the method described herein is performed, the nucleic acid is not sequenced, and the sequence of the nucleic acid is not determined by a sequencing method. In some embodiments, a sequencing method is used to determine the fragment length. In some embodiments, a sequencing method is not used to determine the fragment length. In certain embodiments, a non-targeted sequencing method is used, wherein most or all of the nucleic acids in the sample are randomly sequenced, amplified and / or captured. Certain aspects of sequencing and analysis processes are described below.

[0062] In some embodiments, the fragment length is determined using a sequencing method. In some embodiments, the fragment length is determined using a paired end sequencing platform. This platform relates to the sequencing of both ends of the nucleic acid fragment. Generally, the sequence corresponding to the two ends of the fragment can be mapped to a reference genome (for example, with reference to the human genome). In certain embodiments, both ends are sequenced with a read length sufficient to map each fragment end to the reference genome individually. The example description of paired end sequence read length is as follows. In certain embodiments, all or part of the sequence reads can be mapped to the reference genome without mismatching. In some embodiments, each read is mapped independently. In some embodiments, information from two sequence reads (that is, from each end) is considered in the mapping process. The length of the fragment can be determined, for example, by calculating the difference between the genomic coordinates assigned to each mapped paired end read.

[0063] In some embodiments, the fragment length can be determined using a sequencing process to obtain a complete or substantially complete nucleotide sequence of the fragment. Such a sequencing process includes a platform that produces a relatively long read length (e.g., Roche454, Ion Torrent, single molecule (Pacific Biosciences), real-time SMRT technology, etc.).

[0064] In some embodiments, some or all nucleic acids in a sample are enriched and / or amplified before or during sequencing (e.g., non-specifically, e.g., by PCR-based methods). In certain embodiments, specific nucleic acid portions or subsets in a sample are enriched and / or amplified before or during sequencing. In some embodiments, a portion or subset of a pre-selected nucleic acid pool is randomly sequenced. In some embodiments, nucleic acids in a sample are not enriched and / or amplified before or during sequencing.

[0065] As used herein, a "read" (i.e., "read", "sequence read") is a short nucleotide sequence generated by any sequencing process described herein or known in the art. Reads can be generated from one end of a nucleic acid fragment (e.g., a "single-end read"), and sometimes from both ends of a nucleic acid (e.g., paired end reads, double-end reads). A sequence read may include a fragment end sequence associated with the end of a nucleic acid fragment. The fragment end sequence may correspond to the outermost N bases of the nucleic acid fragment, such as 2-30 bases at the end of the nucleic acid fragment. If the sequence read corresponds to the entire nucleic acid fragment, the sequence read may include two fragment end sequences. When paired end sequencing produces two sequence reads corresponding to the ends of the fragment, each sequence read may include one fragment end sequence.

[0066] The length of the sequence read is usually associated with a specific sequencing technology. For example, high-throughput methods provide sequence reads whose size can vary from tens to hundreds of base pairs (bp). For example, Nanopore sequencing can provide sequence reads whose size can vary from tens to hundreds to thousands of base pairs. In some embodiments, the average, median, average or absolute length of the sequence reads is about 15bp to about 900bp long. In certain embodiments, the average, median, average or absolute length of the sequence reads is about 1000bp or longer.

[0067] In some embodiments, the nominal, average, mean, or absolute length of a single-end read is sometimes about 1 nucleotide to about 500 consecutive nucleotides, about 15 consecutive nucleotides to about 50 consecutive nucleotides, about 30 consecutive nucleotides to about 40 consecutive nucleotides, and sometimes about 35 consecutive nucleotides or about 36 consecutive nucleotides. In certain embodiments, the nominal, average, mean, or absolute length of a single-end read is about 20 to about 30 bases, or about 24 to about 28 bases in length. In certain embodiments, the nominal, average, mean, or absolute length of single-end reads is about 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, or 49 bases in length.

[0068] In certain embodiments, the nominal, average, mean, or absolute length of paired-end reads sometimes is about 10 contiguous nucleotides to about 25 contiguous nucleotides (e.g., about 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides in length), about 15 contiguous nucleotides to about 20 contiguous nucleotides, and sometimes is about 17 contiguous nucleotides, about 18 contiguous nucleotides, about 20 contiguous nucleotides, about 25 contiguous nucleotides, about 36 contiguous nucleotides, or about 45 contiguous nucleotides.

[0069] Reads are usually representations of nucleotide sequences in physical nucleic acids. For example, in the reads described by ATGC containing sequences, in physical nucleic acids, "A" represents adenine nucleotides, "T" represents thymine nucleotides, "G" represents guanine nucleotides, and "C" represents cytosine nucleotides. Sequence reads obtained from the blood, plasma, or serum of a pregnant subject may be reads from a mixture of fetal and maternal nucleic acids. A mixture of relatively short reads may be transformed into a representation of genomic nucleic acids present in a pregnant subject and / or a fetus by the process described herein. For example, a mixture of relatively short reads may be transformed into a representation of copy number variation (e.g., maternal and / or fetal copy number variation), gene variation, or aneuploidy. The reads of a mixture of maternal and fetal nucleic acids may be transformed into a representation of a composite chromosome or a segment thereof containing the features of one or both of maternal chromosomes and fetal chromosomes. In certain embodiments, "obtaining" nucleic acid sequence reads from a sample of a subject and / or "obtaining" nucleic acid sequence reads from a biological specimen of one or more reference persons may involve direct sequencing of nucleic acids to obtain sequence information. In some embodiments, "obtaining" may involve receiving sequence information obtained directly from a nucleic acid by another nucleic acid.

[0070] In some embodiments, a representative portion of a genome is sequenced, sometimes referred to as "coverage" or "coverage fold". For example, 1x coverage means that approximately 100% of the nucleotide sequences in the genome are represented by reads. In some cases, the coverage fold is referred to as "sequencing depth" (and is proportional to it). In some embodiments, "coverage fold" refers to a relative term for a previous sequencing run as a reference. For example, a second sequencing run may have a coverage of 1 / 2 of that of the first sequencing run. In some embodiments, a genome is sequenced in a redundant manner, wherein a given region of the genome may be covered by two or more reads or overlapping reads (e.g., "coverage fold" is greater than 1, e.g., 2x coverage). In some embodiments, a genome (e.g., a whole genome) is sequenced at a coverage of about 0.01 to about 100 times, about 0.1 to 20 times, or about 0.1 to about 1 times (e.g., about 0.015, 0.02, 0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, or more). In some embodiments, a specific portion of a genome (e.g., a genome portion from a targeted approach) is sequenced, and the coverage multiple value generally refers to the score of the specific genome portion sequenced (i.e., the coverage multiple value does not refer to the whole genome). In some cases, a specific genome portion is sequenced with 1000 times or higher coverage. For example, a specific genome portion can be sequenced with 2000 times, 5,000 times, 10,000 times, 20,000 times, 30,000 times, 40,000 times or 50,000 times of coverage. In some embodiments, sequencing is carried out with about 1,000 times to about 100,000 times of coverage. In some embodiments, sequencing is carried out with about 10,000 times to about 70,000 times of coverage. In some embodiments, sequencing is carried out with about 20,000 times to about 60,000 times of coverage. In some embodiments, sequencing is carried out with about 30,000 times to about 50,000 times of coverage.

[0071] In certain embodiments, a subset of nucleic acid fragments is selected before sequencing. In certain embodiments, hybridization-based technology (e.g., using oligonucleotide arrays) can be used to first select nucleic acid sequences from certain chromosomes. In some embodiments, nucleic acids can be separated by size fractionation (e.g., by gel electrophoresis, size exclusion chromatography or by a method based on microfluidics), and in some cases, fetal nucleic acids can be enriched by selecting nucleic acids with lower molecular weight (e.g., less than 300 base pairs, less than 200 base pairs, less than 150 base pairs, less than 100 base pairs). In some embodiments, fetal nucleic acids can be enriched by suppressing maternal background nucleic acids, for example, by adding formaldehyde. In some embodiments, random sequencing is performed on a part or a subset of a preselected nucleic acid fragment set. In some embodiments, nucleic acids are amplified before sequencing. In some embodiments, a part or a subset of nucleic acids are amplified before sequencing.

[0072] In some embodiments, a nucleic acid sample from an individual is sequenced. In certain embodiments, nucleic acid from each sample in two or more samples is sequenced, wherein the samples are from one individual or from different individuals. In certain embodiments, nucleic acid samples from two or more biological samples are merged, wherein each biological sample is from one individual or two or more individuals, and the pool is sequenced. In the latter embodiment, nucleic acid samples from each biological sample are usually identified by one or more unique identifiers or identification tags.

[0073] In some embodiments, sequencing methods utilize identifiers that allow multiplexing of sequencing reactions during sequencing. The greater the number of unique identifiers, the greater the number of samples and / or chromosomes used for detection, for example, which can be multiplexed during sequencing. The sequencing process can be performed using any suitable number of unique identifiers (e.g., 4, 8, 12, 24, 48, 96 or more).

[0074] The sequencing process sometimes utilizes a solid phase, and sometimes the solid phase includes a flow cell, on which nucleic acid from a library can be attached, and reagents can flow and contact the attached nucleic acid. A flow cell sometimes includes a flow cell lane, and the use of an identifier can help analyze multiple samples in each lane. A flow cell is typically a solid carrier that can be configured to retain and / or allow reagent solutions to pass in an orderly manner on a bound analyte. A flow cell is typically planar, optically transparent, generally millimeter or submillimeter level, and typically has a channel or a lane in which an analyte / reagent interaction occurs. In some embodiments, the number of samples analyzed in a given flow cell lane depends on the number of unique identifiers used during library preparation and / or probe design. For example, multiplexing using 12 identifiers allows 96 samples (e.g., equal to the number of holes in a 96-well microplate) to be analyzed simultaneously in an 8-lane flow cell. Similarly, for example, multiplexing using 48 identifiers allows 384 samples (e.g., equal to the number of holes in a 384-well microplate) to be analyzed simultaneously in an 8-lane flow cell. Non-limiting examples of commercially available multiplex sequencing kits include Illumina's Multiplex Sample Preparation Oligonucleotide Kit and Multiplex Sequencing Primer and PhiX Control Kit (eg, Illumina's catalog numbers PE-400-1001 and PE-400-1002, respectively).

[0075] Any suitable nucleic acid sequencing method can be used, and its non-limiting example includes Maxim & Gilbert, chain termination method, synthetic sequencing, connection sequencing, mass spectrometry sequencing, microscope-based technology, etc. or their combination.In some embodiments, first generation technology, such as Sanger sequencing method (including automatic Sanger sequencing method, including microfluidic Sanger sequencing) can be used for method provided herein.In some embodiments, sequencing technology including the use of nucleic acid imaging technology (for example, transmission electron microscope (TEM) and atomic force microscope (AFM)) can be used.In some embodiments, high-throughput sequencing method is used.High-throughput sequencing method generally relates to DNA template or single DNA molecule of clone amplification, which is sometimes sequenced in a large-scale parallel manner in a flow cell.The next generation (for example, second generation and third generation) sequencing technology of DNA sequencing can be used for the method described herein in a large-scale parallel manner, and is collectively referred to as "large-scale parallel sequencing" (MPS) in this article.In some embodiments, MPS sequencing method utilizes targeted method, and wherein specific chromosome, gene or region of interest is sequence.In certain embodiments, non-targeted MPS method is used, wherein most or all of nucleic acids in sample are randomly sequenced, amplified and / or captured.

[0076] A suitable MPS method, system or technology platform for performing the methods described herein can be used to obtain nucleic acid sequencing reads. Non-limiting examples of MPS platforms include Illumina / Solex / HiSeq (e.g., Illumina's Genome Analyzer; Genome Analyzer II; HISEQ 2000; HISEQ), SOLiD, Roche / 454, PACBIO and / or SMRT, Helicos true single molecule sequencing, Ion Torrent and ion semiconductor-based sequencing (e.g., as developed by Life Technologies), WildFire, 5500, 5500xl W and / or 5500xl W Genetic Analyzer-based technology (e.g., as developed and sold by Life Technologies, U.S. Patent Publication No. US20130012399); polymerase clone sequencing, pyrosequencing, massively parallel signature sequencing (MPSS), RNA polymerase (RNAP) sequencing, LaserGen systems and methods, nanopore-based platforms, chemical-sensitive field effect transistor (CHEMFET) arrays, electron microscopy-based sequencing (e.g., as developed by ZS Genetics, Halcyon Molecular), and nanoball sequencing.

[0077] MPS sequencing sometimes utilizes synthetic sequencing and certain imaging processes. Nucleic acid sequencing techniques that can be used for the methods described herein are synthetic sequencing and sequencing based on reversible terminators (e.g., Illumina's Genome Analyzer; Genome Analyzer II; HISEQ 2000; HISEQ 2500 (Illumina, San Diego CA)). Using this technology, thousands to millions of nucleic acid (e.g., DNA) fragments can be sequenced in parallel. In an example of this type of sequencing technology, a flow cell comprising an optically transparent slide with 8 separate lanes is used, and oligonucleotide anchors (e.g., adapter primers) are combined on the surface of these 8 separate lanes. The flow cell is typically a solid support that can be configured to retain and / or allow reagent solutions to pass in order on the combined analyte. The flow cell is typically planar, optically transparent, generally millimeter or submillimeter level, and typically has a channel or lane where an analyte / reagent interaction occurs.

[0078] In some embodiments, sequencing by synthesis includes iteratively adding (e.g., by covalent addition) nucleotides to a primer or a pre-existing nucleic acid chain in a template-guided manner. Each iterative addition of nucleotides is detected, and the process is repeated multiple times until the sequence of the nucleic acid chain is obtained. The length of the obtained sequence depends in part on the number of addition and detection steps performed. In some embodiments of sequencing by synthesis, one, two, three or more nucleotides of the same type (e.g., A, G, C or T) are added and detected in a round of nucleotide addition. Nucleotides can be added by any suitable method (e.g., enzymatic or chemical). For example, in some embodiments, a polymerase or a ligase adds nucleotides to a primer or a pre-existing nucleic acid chain in a template-guided manner. In some embodiments of sequencing by synthesis, different types of nucleotides, nucleotide analogs and / or identifiers are used. In some embodiments, a reversible terminator and / or a removable (e.g., cleavable) identifier is used. In some embodiments, fluorescently labeled nucleotides and / or nucleotide analogs are used. In certain embodiments, sequencing by synthesis includes cutting (e.g., cutting and removal of the identifier) ​​and / or washing steps. In some embodiments, the addition of one or more nucleotides is detected by a suitable method described herein or known in the art, non-limiting examples of which include any suitable imaging device or machine, a suitable camera, a digital camera, a CCD (charge coupled device)-based imaging device (e.g., a CCD camera), a CMOS (complementary metal oxide silicon)-based imaging device (e.g., a CMOS camera), a photodiode (e.g., a photomultiplier tube), an electron microscope, a field effect transistor (e.g., a DNA field effect transistor), an ISFET ion sensor (e.g., a CHEMFET sensor), etc. or a combination thereof. Other sequencing methods that can be used to perform the methods herein include digital PCR and hybridization sequencing.

[0079] Other sequencing methods that can be used to perform the methods herein include digital PCR and hybridization sequencing. Digital polymerase chain reaction (digital PCR or dPCR) can be used to directly identify and quantify nucleic acids in samples. In some embodiments, digital PCR can be performed in an emulsion. For example, a single nucleic acid is separated, for example, in a microfluidic chamber device, and each nucleic acid is amplified individually by PCR. The nucleic acid can be separated so that there is no more than one nucleic acid in each hole. In some embodiments, different probes can be used to distinguish various alleles (e.g., fetal alleles and maternal alleles). Alleles can be counted to determine the copy number.

[0080] In certain embodiments, hybrid sequencing can be used. The method involves contacting a plurality of polynucleotide sequences with a plurality of polynucleotide probes, wherein each polynucleotide probe in the plurality of polynucleotide probes can be optionally connected to a substrate. In some embodiments, the substrate can be a flat surface of an array with a known nucleotide sequence. The hybridization pattern with the array can be used to determine the polynucleotide sequences present in the sample. In some embodiments, each probe is connected to beads, such as magnetic beads, etc. The hybridization with beads can be identified and used to identify a plurality of polynucleotide sequences in the sample.

[0081] In some embodiments, nanopore sequencing can be used in the methods described herein.Nanopore sequencing is a single molecule sequencing technology whereby a single nucleic acid molecule (eg, DNA) is directly sequenced as it passes through a nanopore.

[0082] In some embodiments, targeted enrichment, amplification and / or sequencing methods are used. Targeted methods generally separate, select and / or enrich a subset of nucleic acids in a sample for further processing using sequence-specific oligonucleotides. In some embodiments, a library of sequence-specific oligonucleotides is used to target (e.g., hybridize to) one or more groups of nucleic acids in a sample. Sequence-specific oligonucleotides and / or primers are generally selective for specific sequences (e.g., unique nucleic acid sequences) present in one or more chromosomes, genes, exons, introns and / or regulatory regions of interest. Any suitable method or combination of methods can be used for enrichment, amplification and / or sequencing of one or more subsets of targeted nucleic acids. In some embodiments, the targeted sequence is separated and / or enriched by being captured to a solid phase (e.g., a flow cell, beads) using one or more sequence-specific anchors. In some embodiments, sequence-specific primers and / or primer sets are used to enrich and / or amplify the targeted sequence by a polymerase-based method (e.g., a PCR-based method, by any suitable polymerase-based extension). Sequence-specific anchors can generally be used as sequence-specific primers.

[0083] Mapping reads

[0084] In some embodiments, the sequence reads are mapped to a genome (e.g., a reference genome) or a portion thereof. Any suitable mapping method (e.g., a process, an algorithm, a program, software, a module, etc., or a combination thereof) may be used. Certain aspects of the mapping process are described below.

[0085] Mapping nucleotide sequence reads (i.e., sequence information from unknown fragments of physical genomic positions) can be performed in a variety of ways, and typically includes comparing the obtained sequence reads with matching sequences in a reference genome. In such comparisons, sequence reads are typically compared with reference sequences, and the compared sequence reads are designated as "mapped", "mapped sequence reads" or "mapped reads". In certain embodiments, mapped sequence reads are referred to as "hit" or "count". In some embodiments, as described herein, mapped sequence reads are grouped together according to various parameters. In some embodiments, mapped sequence reads are assigned to specific chromosomes or portions thereof.

[0086] As used herein, the terms "aligned," "aligned," or "aligned" refer to two or more nucleic acid sequences that can be identified as a match (e.g., 100% identity) or a partial match. Alignment can be done manually or by a computer (e.g., software, program, module, or algorithm), non-limiting examples of which include the Efficient Local Alignment of Nucleotide Data (ELAND) computer program distributed as part of the Illumina genome analysis pipeline. The alignment of sequence reads can be a 100% sequence match. In some cases, the alignment is less than a 100% sequence match (i.e., non-complete match, partial match, partial alignment). In some embodiments, the comparison is about 99%, 98%, 97%, 96%, 95%, 94%, 93%, 92%, 91%, 90%, 89%, 88%, 87%, 86%, 85%, 84%, 83%, 82%, 81%, 80%, 79%, 78%, 77%, 76% or 75% matching. In some embodiments, the comparison includes mispairing. In some embodiments, the comparison includes 1, 2, 3, 4 or 5 mispairings. Any chain can be used to compare two or more sequences. In certain embodiments, the nucleic acid sequence is compared with the reverse complementary sequence of another nucleic acid sequence.

[0087] Various computational methods can be used for mapping each sequence read to reference genome or its part.Non-limiting examples of computer algorithms that can be used for alignment sequences include but are not limited to BLAST, BLITZ, FASTA, BOWTIE 1, BOWTIE 2, ELAND, MAQ, PROBEMATCH, SOAP or SEQMAP, or their variants or combinations.In some embodiments, sequence reads can be compared with the sequence in the reference genome.In some embodiments, sequence reads can be found and / or compared with the sequence therein in nucleic acid databases known in the art, and these nucleic acid databases include, for example, GenBank, dbEST, dbSTS, EMBL (European Molecular Biology Laboratory) and DDBJ (Japan DNA database).BLAST or similar tools can be used for searching identified sequences for sequence databases.For example, then the identified sequence can be classified into suitable chromosome or genome part using search hits.

[0088] In some embodiments, the read can be mapped to a reference genome or part thereof uniquely or non-uniquely. If the read is aligned with a single sequence in the reference genome, the read is considered to be "uniquely mapped". If the read is aligned with two or more sequences in the reference genome, the read is considered to be "non-uniquely mapped". In some embodiments, the reads of non-unique mappings are eliminated from further analysis (e.g., fragment length measurement, sequence motif quantification). In certain embodiments, a small degree of mismatch (0-1) may be allowed to explain the single nucleotide polymorphisms that may exist between the reference genome and the mapped reads from a single sample. In some embodiments, any degree of mismatch is not allowed for the reads mapped to the reference sequence.

[0089] As used herein, the term "reference genome" may refer to any specific known, sequenced or characterized genome of any organism or virus that can be used to reference the identification sequence from a subject, whether partial or complete. For example, the reference genome for human subjects and many other organisms can be found at the National Center for Biotechnology Information at www.ncbi.nlm.nih.gov. "Genome" refers to the complete genetic information of an organism or virus expressed in a nucleic acid sequence. As used herein, a reference sequence or reference genome is typically an assembled or partially assembled genome sequence from an individual or multiple individuals. In some embodiments, a reference genome is an assembled or partially assembled genome sequence from one or more human individuals. In some embodiments, a reference genome includes a sequence assigned to a chromosome.

[0090] In certain embodiments, when the sample nucleic acid is from a pregnant subject, the reference sequence is sometimes not from the fetus, the mother of the fetus, or the father of the fetus, and is referred to herein as an "external reference". A maternal reference may be prepared and used in some embodiments. When a reference from a pregnant woman (a "maternal reference sequence") is prepared based on an external reference, the reads from the DNA of a pregnant woman that is substantially free of fetal DNA are typically mapped to the external reference sequence and assembled. In certain embodiments, the external reference is from DNA of an individual of substantially the same race as the pregnant woman. The maternal reference sequence may not completely cover the maternal genomic DNA (e.g., it may cover about 50%, 60%, 70%, 80%, 90% or more of the maternal genomic DNA), and the maternal reference may not perfectly match the maternal genomic DNA sequence (e.g., the maternal reference sequence may include multiple mismatches).

[0091] In certain embodiments, mappability is assessed for reference genome or genomic region (e.g., part, genome part, part). Mappability is the ability of nucleotide sequence reads to be clearly compared with a reference genome or a part of a reference genome, usually up to a specified number of mispairings, including, for example, 0, 1, 2 or more mispairings. For a given genomic region, the sliding window method of the preset read length can be used and the resulting read level mappability value averaged to estimate expected mappability. The genomic region comprising a unique nucleotide sequence fragment sometimes has a high mappability value.

[0092] Sequence read quantification

[0093] In some embodiments, the sequence reads mapped or partitioned based on selected features or variables can be quantitatively determined to map to one or more parts, sections, partitions, loci (e.g., parts, sections, partitions, loci with reference to genomes). In certain embodiments, the amount of sequence reads mapped to a portion is referred to as count. Typically, count is associated with a portion. In certain embodiments, the counts of two or more portions (e.g., a group of portions) are mathematically processed (e.g., averaged, added, normalized, etc. or a combination thereof). In some embodiments, counts are determined by mapping to (i.e., being associated with) some or all of the sequence reads of a portion. In certain embodiments, counts are determined from a predefined subset of the sequence reads mapped. Any suitable feature or variable can be used to define or select a predefined subset of mapping sequence reads. In some embodiments, the predefined subset of mapping sequence reads may include 1 to n sequence reads, wherein n represents a number equal to the sum of all sequence reads generated from a test subject or a reference subject sample.

[0094] In certain embodiments, sequence reads are quantitatively or counted from sequence reads processed or operated by suitable methods known in the art, operations or mathematical processes. Quantitative or counting can be determined by suitable methods, operations or mathematical processes. In certain embodiments, quantitative or counting are derived from sequence reads associated with a portion, wherein some or all of the sequence reads are weighted, removed, filtered, normalized, adjusted, averaged, derived as an average, added or subtracted, or processed by a combination thereof. In some embodiments, quantitative or counting are derived from raw sequence reads and or filtered sequence reads. In certain embodiments, quantitative or counting values ​​are determined by mathematical processes. In certain embodiments, quantitative or counting values ​​are average values, averages or sums of sequence reads mapped to a portion. In some embodiments, quantitative or counting are the averages of counts. In some embodiments, quantitative or counting are associated with uncertainty values.

[0095] In some embodiments, the sequence read quantification or counts can be manipulated or transformed (e.g., normalized, combined, added, filtered, selected, averaged, derived as an average, etc., or a combination thereof). In some embodiments, the quantification or counts can be transformed to produce normalized counts. The quantification or counts can be processed (e.g., normalized) by methods known in the art and / or as described herein (e.g., part-by-part normalization, normalization by GC content, linear and nonlinear least squares regression, GC LOESS, LOWESS, RM, GCRM, cQn, and / or combinations thereof).

[0096] Sequence reads can be obtained quantitatively or counted from nucleic acid samples from pregnant subjects with a fetus. The quantitative or count of nucleic acid sequence reads mapped to one or more portions is usually a quantitative or count representing both the fetus and the fetal mother (e.g., a pregnant subject). In certain embodiments, some of the quantitative or counts mapped to the portion are from the fetal genome, and some of the counts mapped to the same portion are from the maternal genome.

[0097] Fragment length measurement

[0098] In some embodiments, the length of one or more nucleic acid fragments is measured (e.g., one or more free nucleic acid fragments; one or more circulating free nucleic acid fragments). In some embodiments, sequence-based fragment length measurement is used. For example, nucleic acid fragment length can be measured using a paired end sequencing platform. This platform involves sequencing at both ends of nucleic acid fragments. Typically, the sequences corresponding to the two ends of the nucleic acid fragment can be mapped to a reference genome (e.g., with reference to the human genome). In certain embodiments, both ends are sequenced with a read length sufficient to map each fragment end to the reference genome alone. In certain embodiments, all or part of the sequence reads can be mapped to the reference genome without mismatching. In some embodiments, each read is mapped independently. In some embodiments, information from two sequence reads (i.e., from each end of the nucleic acid fragment) is considered in the mapping process. The length of the nucleic acid fragment can be measured, for example, by calculating the difference between the genomic coordinates assigned to each mapped paired end read. In other words, the length of the nucleic acid fragment is measured (e.g., inferred or inferred) by mapping two or more reads (e.g., paired end reads) derived from the nucleic acid fragment to the reference genome. For paired-end reads derived from nucleic acid fragments, for example, the reads can be mapped to a reference genome, the length of the genomic sequence between the mapped reads can be determined, and the sum of the two read lengths and the length of the genomic sequence between the reads is equal to the length of the nucleic acid fragment. In some embodiments, the length of the fragment is measured directly from the length of the read (e.g., a single-end read) derived from the nucleic acid fragment.

[0099] In certain applications of the methods described herein, nucleic acid fragment length can be measured using any method suitable for determining nucleic acid fragment length in the art, such as mass sensitive methods (e.g., mass spectrometry (e.g., matrix-assisted laser desorption ionization (MALDI) mass spectrometry and electrospray (ES) mass spectrometry), electrophoresis (e.g., capillary electrophoresis), microscopy (scanning tunneling microscopy, atomic force microscopy), and measuring length using a nanopore. In some applications, the fragment length can be determined by measuring the length of a probe that hybridizes to the fragment.

[0100] Genomic intervals and segments

[0101] In some embodiments, the data (for example, fragment length, sequence motif) obtained from the sequence reads mapped are grouped together according to the part of the genome (for example, with reference to the part of the genome). A part of a genome may also be referred to herein as a chromosome, a genome segment, a genome interval, a genome portion, a genome segment, a box, a region, a partition, a part of a reference genome or a part of a chromosome. In some embodiments, a part of a genome is a whole chromosome, a segment of a chromosome, a segment of a reference genome, a segment across multiple chromosomes, multiple chromosome segments and / or a combination thereof. In some embodiments, a part of a genome is predefined based on a specific parameter (for example, a predefined length). In some embodiments, a part of a genome is defined arbitrarily or non-arbitrarily based on the partition of a genome (for example, by size, GC content, continuous region, partitions such as continuous regions of arbitrary definition size).

[0102] In some embodiments, a portion of a genome (e.g., a genome interval, a genome segment) is depicted based on one or more parameters, the one or more parameters including, for example, the length of one or more specific features of a sequence. A portion of a genome (e.g., a genome interval, a genome segment) can be selected, filtered, and / or removed from consideration using any suitable criteria known in the art or described herein. In some embodiments, a portion of a genome (e.g., a genome interval, a genome segment) is based on a specific length of a genome sequence. Multiple portions of a genome (e.g., multiple genome intervals, multiple genome segments) may have approximately the same length, or multiple portions of a genome (e.g., multiple genome intervals, multiple genome segments) may have different lengths. In some embodiments, multiple portions of a genome (e.g., multiple genome intervals, multiple genome segments) have approximately equal lengths.

[0103] In some embodiments, a portion of a genome (e.g., a genomic interval, a genomic segment) can be a portion of a particular chromosome, a chromosome of interest, for example, a chromosome in which a genetic variation is assessed (e.g., aneuploidy of chromosomes 13, 18, and / or 21, or a sex chromosome). Multiple portions of a genome (e.g., multiple genomic intervals, multiple genomic segments) can be genes, gene fragments, regulatory sequences, introns, exons, etc.

[0104] In some embodiments, the genome (e.g., human genome, reference genome) is partitioned (e.g., partitioned into multiple genome intervals, multiple genome segments) based on the information content of a specific region. In some embodiments, the genome partition can eliminate similar regions (e.g., identical or homologous regions or sequences) across the genome and only retain unique regions. The region removed during the partition can be within a single chromosome or can span multiple chromosomes. In some embodiments, the genome of the partition is pruned and optimized for faster comparison, usually allowing to focus on unique identifiable sequences. In some embodiments, the partition of the genome can be based on other standards, such as, speed / convenience when comparing tags, GC content (e.g., high or low GC content), the uniformity of GC content, other measures of sequence content (e.g., the score of a single nucleotide, the score of pyrimidine or purine, the score of natural nucleic acids and non-natural nucleic acids, the score of methylated nucleotides and CpG content), methylation state, duplex melting temperature, adaptability to sequencing or PCR, uncertainty values ​​assigned to each part of the reference genome and / or targeted searches for specific features.

[0105] In some embodiments, fragment lengths are measured for multiple genomic intervals. Specifically, the fragment lengths obtained by mapping paired end reads to the selected genomic intervals are measured. For each selected interval, a fragment length measurement obtained by mapping paired end reads to the genomic interval can be obtained. As described below, a fragment length ratio can be determined for each selected interval. In some embodiments, the length of the genomic interval is about 10 kilobases (kb) to about 500kb, a length of about 10kb to about 200kb, or a length of about 50kb to about 150kb. For example, the length of the genomic interval may be about 50kb, 60kb, 70kb, 80kb, 90kb, 100kb, 110kb, 120kb, 130kb, 140kb or 150kb. In some embodiments, the genomic interval length is about 100kb. The genomic interval is not limited to the continuous subsequence of the sequence. Therefore, the genomic interval may be composed of continuous sequences and / or non-continuous sequences. The genomic interval is not limited to a single chromosome. In some embodiments, a genomic interval includes all or part of a chromosome or all or part of two or more chromosomes. In some embodiments, a genomic interval may span one, two or more complete chromosomes. In addition, a genomic interval may span a connected or unconnected region of multiple chromosomes.

[0106] In some embodiments, a fragment length spectrum of a genomic segment is generated. In some embodiments, one or more fragment length spectrums of one or more genomic segments are generated. A segment may refer to a segment of a genome and / or a segment of a chromosome. In some embodiments, a segment is a complete chromosome. A segment of a genome or a chromosome generally contains a greater number of nucleotides than a genome interval. Typically, a genomic segment is composed of a plurality of smaller genome intervals. In some embodiments, the length of a genomic segment is about 1 megabase (Mb) to about 10 megabases (Mb). For example, the length of a genomic segment may be about 1Mb, 2Mb, 3Mb, 4Mb, 5Mb, 6Mb, 7Mb, 8Mb, 9Mb or 10Mb. In some embodiments, the length of a genomic segment is about 5Mb.

[0107] Spectrum

[0108] In some implementation methods, the methods herein include generating one or more spectra (e.g., spectra) from various aspects of a data set or a derivation thereof (e.g., a product of one or more mathematical and / or statistical data processing steps known in the art and / or described herein). In some embodiments, the methods herein include generating one or more fragment length spectra. A fragment length spectrum of a genomic segment as described herein can be generated. As described herein, a fragment length spectrum can be generated for a genomic segment partitioned into smaller genomic intervals. A fragment length value can be assigned to each genomic interval of the spectrum. The fragment length value can be the original fragment length or a derived value thereof, such as a fragment length ratio, a median fragment length, an average fragment length, an average fragment length, a normalized fragment length, etc.

[0109] In some embodiments, a fragment length spectrum is generated according to a fragment length ratio (i.e., a fragment length ratio determined for a plurality of genomic intervals). In some embodiments, a fragment length spectrum is generated according to the ratio of X to Y of a plurality of genomic intervals, wherein X is the number of CCF nucleic acid fragments with a length within the first selected fragment length range, and Y is the number of CCF nucleic acid fragments with a length within the second selected fragment length range. In some embodiments, the first selected fragment length range is from about 1 base to about 140 bases, and the second selected fragment length range is from about 141 bases and above. In some embodiments, the first selected fragment length range is from about 1 base to about 150 bases, and the second selected fragment length range is from about 151 bases and above. In some embodiments, the first selected fragment length range is from about 1 base to about 160 bases, and the second selected fragment length range is from about 161 bases and above. In some embodiments, the first selected fragment length range is from about 60 bases to about 170 bases. In some embodiments, the first selected fragment length range is about 70 bases to about 160 bases. In some embodiments, the first selected fragment length range is about 75 bases to about 155 bases. In some embodiments, the first selected fragment length range is about 80 bases to about 150 bases. In some embodiments, the second selected fragment length range is about 131 bases to about 400 bases. In some embodiments, the second selected fragment length range is about 141 bases to about 350 bases. In some embodiments, the second selected fragment length range is about 146 bases to about 325 bases. In some embodiments, the second selected fragment length range is about 151 bases to about 300 bases. In some embodiments, the first selected fragment length range is about 80 bases to about 150 bases, and the second selected fragment length range is about 151 bases to about 300 bases.

[0110] As used herein, the term "spectrum" refers to the product of mathematical and / or statistical operations of data, which can be useful for identifying patterns and / or correlations in large amounts of data. "Spectrum" generally includes values ​​obtained from one or more operations performed on data or data sets based on one or more criteria. Spectrum generally includes multiple data points. Depending on the nature and / or complexity of the data set, any suitable number of data points may be included in the spectrum. In certain embodiments, the spectrum may include 2 or more data points, 3 or more data points, 5 or more data points, 10 or more data points, 24 or more data points, 25 or more data points, 50 or more data points, 100 or more data points, 500 or more data points, 1000 or more data points, 5000 or more data points, 10,000 or more data points, or 100,000 or more data points.

[0111] In some embodiments, the spectrum represents the entirety of a data set, and in certain embodiments, the spectrum represents a portion or subset of a data set. That is, the spectrum sometimes includes data points representing data that have not been filtered to remove any data or is generated from these data points, and the spectrum sometimes includes data points representing data that have been filtered to remove unwanted data or is generated from these data points. In some embodiments, the data points in the spectrum represent the results of data operations (e.g., data operations of fragment lengths of genome intervals) at genomic intervals. In certain embodiments, the data points in the spectrum include the results of data operations of groups at genomic intervals. In some embodiments, groups at genomic intervals may be adjacent to each other, and in certain embodiments, groups at genomic intervals may come from different parts of a chromosome or genome.

[0112] In some embodiments, a spectrum can be generated from data points obtained from another spectrum (e.g., a normalized data spectrum renormalized to a different normalization value to produce a renormalized data spectrum). In certain embodiments, a spectrum generated from data points obtained from another spectrum reduces the number of data points and / or the complexity of the data set. Reducing the number of data points and / or the complexity of the data set often aids in interpreting the data and / or aids in providing results.

[0113] A spectrum can be a collection of normalized or non-normalized values ​​(e.g., fragment length values; fragment length ratios) for two or more genomic intervals. A spectrum typically includes at least one value (e.g., fragment length value; fragment length ratio), and typically includes two or more values ​​(e.g., a spectrum typically has multiple fragment length values ​​or fragment length ratios). In certain embodiments, a spectrum includes one or more genomic intervals that can be weighted, removed, filtered, normalized, adjusted, averaged, derived as an average, added, subtracted, processed or transformed by any combination thereof. A spectrum typically includes fragment length values ​​or fragment length ratios, wherein the fragment length values ​​or fragment length ratios can be further normalized according to a suitable method. In some embodiments, the fragment length values ​​or fragment length ratios of a spectrum are associated with uncertainty values.

[0114] The spectrum can be displayed as a graph. For example, one or more fragment length values ​​for one or more genomic intervals can be plotted and visualized. Non-limiting examples of spectra that can be generated include raw fragment length values, fragment length ratios, average fragment length values, median fragment length values, average fragment length values, normalized fragment length values, z scores, p values, principal components, etc., or combinations thereof. In some embodiments, the spectra allow visualization of the data being manipulated. In some embodiments, a static window process can be used to generate the spectrum, and in certain embodiments, a sliding window process can be used to generate the spectrum.

[0115] Sometimes the spectrum generated for the test subject is compared to the spectrum generated for one or more reference subjects and / or samples in the training set to facilitate the interpretation of mathematical and / or statistical operations of the data set and / or to provide results. In some embodiments, the spectrum is generated based on one or more starting hypotheses (e.g., maternal nucleic acid contribution (e.g., maternal fraction), fetal nucleic acid contribution (e.g., fetal fraction), etc., or a combination thereof).

[0116] sequence motif

[0117] The method herein may include determining the sequence motifs at the ends of nucleic acid fragments. The sequence motif may refer to a short base pattern in a nucleic acid fragment (e.g., a CCF nucleic acid fragment). The sequence motif may appear at the end of a fragment, and is therefore a part of or includes a fragment end sequence. The sequence motif at the end of a fragment may be any length (e.g., 2-30 bases in length) suitable for methods described herein. For example, the length of the fragment end sequence motif may be 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29 or 30 bases. In some embodiments, the length of the fragment end sequence motif is three bases. In some embodiments, the length of the fragment end sequence motif is 4 bases. In some embodiments, the length of the fragment end sequence motif is 5 bases. In some embodiments, the sequence motif is a 5' sequence motif. In some embodiments, the sequence motif is a 3' sequence motif.

[0118] In some embodiments, the sequence motifs of the natural ends of nucleic acid fragments are determined. In some embodiments, the sequence motifs of the natural 5' ends of nucleic acid fragments are determined. In some embodiments, the sequence motifs of the natural 3' ends of nucleic acid fragments are determined. Natural ends generally refer to the unmodified ends of nucleic acid fragments. In some embodiments, the length of the natural ends of nucleic acid fragments is not modified (e.g., shortened). In some embodiments, the length of the 5' ends of nucleic acid fragments is not modified (e.g., shortened). In some embodiments, the length of the 3' ends of nucleic acid fragments is not modified (e.g., shortened). In this context, "unmodified" refers to separating nucleic acid fragments from a sample and performing sequencing treatment (e.g., processing into a sequencing library), without modifying the length of the natural ends of nucleic acid fragments. For example, the ends of nucleic acid fragments are not shortened (e.g., they are not in contact with restriction enzymes or nucleases or are not in contact with physical conditions (e.g., shearing conditions, cutting conditions) that reduce the length to produce non-natural ends). Adding one or more nucleotides, phosphate groups or chemically reactive groups for the purpose of linker connection to the natural ends of nucleic acid fragments is generally not considered to modify the length of nucleic acid fragments.

[0119] Sequence motif frequency

[0120] The method herein may include determining the sequence motif frequency. In some embodiments, the method herein includes determining one or more sequence motif frequencies. In some embodiments, the method herein includes determining the sequence motif frequency of a genome or part thereof. In some embodiments, the method herein includes determining the sequence motif frequency of a chromosome or part thereof. In some embodiments, the method herein includes determining one or more sequence motif frequencies of one or more chromosomes. For example, the sequence motif frequency can be determined according to the frequency of one or more sequence motifs in the mapping sequence reads of one or more chromosomes. In some embodiments, the method herein includes determining one or more sequence motif frequencies of one or more sequence motifs selected from any 256 possible 4bp motifs (four base positions, each position has four base possibilities). In some embodiments, the method herein includes determining one or more sequence motif frequencies of one or more sequence motifs selected from GGAA, AGAA, GTTT, GAAT and GGTT.

[0121] In certain embodiments, motif frequency is processed or operated by suitable methods known in the art, operation or mathematical process.Motif frequency can be determined by suitable methods, operation or mathematical process.In certain embodiments, motif frequency is derived from sequence reads, wherein some or all of the sequence reads are weighted, removed, filtered, normalized, adjusted, averaged, derived as mean, added or subtracted or processed by its combination.In certain embodiments, motif frequency is derived from sequence motifs, wherein some or all of the sequence motifs are weighted, removed, filtered, normalized, adjusted, averaged, derived as mean, added or subtracted or processed by its combination.In some embodiments, motif frequency is derived from original sequence reads and / or filtered sequence reads.In some embodiments, motif frequency is derived from original sequence motifs and / or filtered sequence motifs.In certain embodiments, motif frequency is determined by mathematical process.In certain embodiments, motif frequency is the mean value, average or sum of sequence motifs. In some embodiments, the motif frequency is associated with an uncertainty value (e.g., a calculated variance, error, standard deviation, Z score, p-value, mean absolute deviation, etc.). A deviation value can be used instead of an uncertainty value, and non-limiting examples of deviation measures include standard deviation, mean absolute deviation, median absolute deviation, standard scores (e.g., Z score, Z score, normal score, standardized variable), etc.

[0122] In some embodiments, the motif frequencies may be manipulated or transformed (e.g., normalized, combined, added, filtered, selected, averaged, derived as an average, etc., or a combination thereof). In some embodiments, the motif frequencies may be transformed to produce normalized motif frequencies. In some embodiments, the motif frequencies are normalized. In some embodiments, the motif frequencies are normalized by the total frequency. In some embodiments, a normalized motif diversity score (MDS) is generated. In some embodiments, a normalized motif diversity score (MDS) is generated according to the following equation:

[0123]

[0124] Statistical models and model parameters

[0125] The methods described herein may include estimating the fetal nucleic acid fraction in the test sample based on one or more model parameters obtained from one or more models. The methods described herein may include applying one or more model parameters from one or more models to one or more fragment length spectra of the test sample. The methods described herein may include applying one or more model parameters from one or more models to one or more sequence motif frequencies of the test sample. The methods described herein may include applying one or more model parameters from one or more models to i) one or more fragment length spectra of the test sample, and ii) one or more sequence motif frequencies of the test sample.

[0126] In some embodiments, the methods described herein may include applying one or more model parameters from one or more models to one or more sequence read coverage features of a test sample. For example, model parameters obtained from a model that predicts fetal fraction using whole genome sequence read coverage may be applied to one or more sequence read coverage features of a test sample, such as, for example, in U.S. Pat. No. 10,622,094 and Kim et al., 2015, Prenatal Diagnosis, Vol. 35: pp. 1-6, each of which is incorporated by reference in its entirety. Sequence read coverage features may include mapping sequence read quantification of one or more genomic loci, genomic parts, genomic segments, genomic regions, genomic partitions, etc. In some embodiments, the methods described herein may include applying one or more model parameters from one or more models to i) one or more fragment length spectra of a test sample, and ii) one or more sequence read coverage features of a test sample. In some embodiments, the methods described herein may include applying one or more model parameters from one or more models to i) one or more sequence motif frequencies of a test sample, and ii) one or more sequence read coverage features of a test sample. In some embodiments, the methods described herein may include applying one or more model parameters from one or more models to i) one or more fragment length profiles of a test sample, ii) one or more sequence motif frequencies of the test sample, and iii) one or more sequence read coverage characteristics of the test sample.

[0127] Model parameters can include values ​​of one or more measured features described herein (e.g., fragment length spectrum, motif frequency) and known or measured fetal fraction values ​​for one or more samples (e.g., one or more samples in a training set). In some embodiments, values ​​of one or more measured features described herein (e.g., fragment length spectrum, motif frequency) can be determined from aggregate values ​​as determined for samples of known fetal fraction (e.g., training samples). Model parameters can be defined in a variety of ways, such as as discrete values ​​or as model functions (e.g., model curves). Model functions can be derived from one or more additional mathematical transformations of one or more model parameters.

[0128] In some embodiments, model parameters are coefficients or constants that partially describe and / or define the relationship between fetal fraction and one or more measured features described herein (e.g., fragment length spectrum, motif frequency). In some embodiments, model parameters are determined based on the relationship between multiple fetal fractions and multiple measured features described herein (e.g., multiple fragment length spectrum, multiple motif frequencies). One or more model parameters can define a relationship and one or more model parameters can be determined from the relationship. In some embodiments, model parameters are determined from a fitted relationship based on (i) the fetal nucleic acid fraction of each sample in a plurality of samples and (ii) one or more measured features of a plurality of samples.

[0129] Model parameters can be any suitable coefficients, estimated coefficients or constants derived from a suitable statistical model and / or relationship (e.g., a suitable mathematical relationship, an algebraic relationship, a fitted relationship, a regression, a regression analysis, a regression model). Model parameters can be determined according to a suitable relationship, derived from a suitable relationship or estimated from a suitable relationship. In some embodiments, model parameters are coefficients estimated by the fitted relationship. Fitting a relationship to multiple samples (e.g., multiple samples in a training set) is sometimes referred to as a training model in this article. Any suitable model and / or method for fitting a relationship (e.g., training a model for a training set) can be used. Non-limiting examples of suitable models that can be used include regression models, linear regression models, simple regression models, ordinary least squares regression models, multiple regression models, general multiple regression models, polynomial regression models, general linear models, generalized linear models, discrete choice regression models, logistic regression models, multinomial logit models, mixed logit models, probability unit models, multinomial probability unit models, ordered logit models, ordered probability unit models, Poisson models, multivariate response regression models, multilevel models, fixed effects models, random effects models, mixed models, nonlinear regression models, nonparametric models, semiparametric models, robust models, quantile models, isotonic models, principal component models, minimum angle models, local models, segmented models, variable error models, and combinations thereof. In some embodiments, the fitted relationship is not a regression model. In some embodiments, the fitted relationship is selected from a decision tree model, a support vector machine model, and a neural network model. The result of a training model (e.g., a regression model, a relationship) is typically a relationship that can be described mathematically, wherein the relationship includes one or more coefficients (e.g., model parameters). For example, for a linear least squares model, a general multiple regression model can be trained using fetal fraction values ​​and one or more measured features described herein (e.g., fragment length spectrum, motif frequency) to generate a relationship. More complex multivariate models can determine one, two, three or more model parameters. In some embodiments, a model is trained based on fetal fraction and two or more measured features described herein (e.g., fragment length spectrum, motif frequency) obtained from multiple samples (e.g., such as by fitting relationships to multiple samples through matrix fitting).

[0130] In some embodiments, the relationship is a geometric and / or graphical relationship. As used herein, the terms "relationship" and "contact" are synonymous. In some embodiments, the relationship is a mathematical relationship. In some embodiments, a relationship is drawn. In some embodiments, the relationship is a linear relationship. In certain embodiments, the relationship is a nonlinear relationship. In certain embodiments, the relationship is a regression (e.g., a regression line). Regression can be a linear regression or a nonlinear regression. Relationships can be represented by mathematical equations. Typically, the relationship is defined in part by one or more constants and / or one or more variables. Relationships can be generated by methods known in the art. In certain embodiments, a two-dimensional relationship can be generated for one or more samples, and error proof variables or possible error proof variables can be selected for one or more of these dimensions. For example, a relationship can be generated using drawing software known in the art, which uses the values ​​of two or more variables provided by a user to draw a graph. Relationships can be fitted using methods known in the art (e.g., by performing regression, regression analysis, e.g., by a suitable regression program, such as software). Some relationships can be fitted by linear regression, and linear regression can produce slope values ​​and intercept values. Certain relationships are sometimes not linear and may be fit by a nonlinear function, such as a parabola, hyperbola, or exponential function (eg, quadratic function).

[0131] Model parameters can be derived from a suitable relationship (e.g., a suitable mathematical relationship, an algebraic relationship, a fitted relationship, regression, regression analysis, a regression model) by a suitable method. In some embodiments, the fitted relationship is fitted by estimation, non-limiting examples of which include least squares, ordinary least squares, linear, partial, total, generalized, weighted, nonlinear, iterative reweighting, ridge regression, least absolute deviation, Bayesian, Bayesian multivariate, reduced rank, LASSO, weighted rank selection criterion (WRSC), rank selection criterion (RSC), elastic net estimator (e.g., elastic net regression), and combinations thereof.

[0132] Model parameters can be obtained from any suitable sample or sample group. In some embodiments, model parameters are obtained from a sample training set. In some embodiments, the fetal nucleic acid score of each sample in the sample training set is known. The sample training set may include multiple reference samples. The sample training set may include a euploid sample, an aneuploid sample, or a combination of a euploid sample and an aneuploid sample. The sample training set may include a female sample (from a pregnant subject carrying a female fetus), a male sample (from a pregnant subject carrying a male fetus), or a combination of a female sample and a male sample.

[0133] In some embodiments, model parameters are determined according to one or more samples (e.g., sample training set). In some embodiments, model parameters are determined according to the relationship between the fetal fraction of multiple samples (e.g., sample-specific fetal fraction) and one or more measured features (fragment length spectrum, motif frequency) determined according to multiple samples. Model parameters are usually determined from multiple samples, such as from about 20 to about 100,000 or more samples, from about 100 to about 100,000 or more samples, from about 500 to about 100,000 or more samples, from about 1000 to about 100,000 or more samples, or from about 10,000 to about 100,000 or more samples. In some embodiments, model parameters are determined from about 1,000 samples to about 2,000 samples. Model parameters can be determined from euploid samples (e.g., samples from subjects with euploid fetuses, such as samples without aneuploid chromosomes). In some embodiments, the model parameters are obtained from a sample comprising an aneuploid chromosome (e.g., a sample from a subject with an aneuploid fetus). In some embodiments, the model parameters are determined from multiple samples from a subject with a euploid fetus and from a subject with a triploid fetus. The model parameters can be derived from multiple samples, wherein these samples are from subjects with male fetuses and / or female fetuses.

[0134] The fetal fraction estimate for a sample (e.g., a test sample) can be determined by applying one or more model parameters to one or more measured features (e.g., fragment length spectrum, motif frequency) of the test sample. Applying the one or more model parameters can include adjusting, converting, and / or transforming the one or more measured features (e.g., fragment length spectrum, motif frequency) according to the model parameters by applying any suitable mathematical operation, non-limiting examples of which include multiplication, division, addition, subtraction, integration, symbolic calculations, algebraic calculations, algorithms, trigonometric or geometric functions, transformations (e.g., Fourier transforms), etc. or combinations thereof. Applying the one or more model parameters can include adjusting, converting, and / or transforming the one or more measured features (e.g., fragment length spectrum, motif frequency) according to a suitable mathematical model.

[0135] In some embodiments, one or more model parameters are obtained from the training set based on 1) the fetal nucleic acid fraction of each sample in the sample training set and 2) the fit relationship between one or more fragment length profiles of each sample in the sample training set. In some embodiments, one or more model parameters are obtained from the training set based on 1) the fetal nucleic acid fraction of each sample in the sample training set and 2) the fit relationship between one or more sequence motif frequencies of each sample in the sample training set. In some embodiments, one or more model parameters are obtained from the training set based on 1) the fetal nucleic acid fraction of each sample in the sample training set and 2) the fit relationship between one or more fragment length profiles and one or more sequence motif frequencies of each sample in the sample training set.

[0136] In some embodiments, the model comprises a linear regression (e.g., a linear regression for a sample training set). In some embodiments, estimating the fetal nucleic acid fraction of a test sample comprises applying the regression coefficients from the linear regression to one or more measured features of the test sample. In some embodiments, estimating the fetal nucleic acid fraction of a test sample comprises applying the regression coefficients from the linear regression to one or more fragment length spectra of the test sample. In some embodiments, estimating the fetal nucleic acid fraction of a test sample comprises applying the regression coefficients from the linear regression to one or more sequence motif frequencies of the test sample. In some embodiments, estimating the fetal nucleic acid fraction of a test sample comprises applying the regression coefficients from the linear regression to i) one or more fragment length spectra of the test sample and ii) one or more sequence motif frequencies of the test sample.

[0137] In some embodiments, the model includes an elastic network model (e.g., an elastic network model for a sample training set). Typically, the elastic network is a penalized linear regression model that includes an L1 penalty and an L2 penalty during training. A hyperparameter "alpha" is provided to specify how much weight is assigned to each of the L1 penalty and the L2 penalty. In some embodiments, estimating the fetal nucleic acid score of a test sample includes applying the coefficients from the elastic network model to one or more measured features of the test sample. In some embodiments, estimating the fetal nucleic acid score of a test sample includes applying the coefficients from the elastic network model to one or more fragment length spectra of the test sample. In some embodiments, estimating the fetal nucleic acid score of a test sample includes applying the coefficients from the elastic network model to one or more sequence motif frequencies of the test sample. In some embodiments, estimating the fetal nucleic acid score of a test sample includes applying the coefficients from the elastic network model to i) one or more fragment length spectra of the test sample and ii) one or more sequence motif frequencies of the test sample.

[0138] In some embodiments, the model includes an XGBoost model (e.g., an XGBoost model for a sample training set). Typically, XGBoost is an effective open source implementation of a gradient boosting tree algorithm. Gradient boosting is a supervised learning algorithm that attempts to accurately predict a target variable by combining the estimated values ​​of a set of simpler, weaker models. In some embodiments, estimating the fetal nucleic acid score of a test sample includes applying the coefficients from the XGBoost model to one or more measured features of the test sample. In some embodiments, estimating the fetal nucleic acid score of a test sample includes applying the coefficients from the XGBoost model to one or more fragment length spectra of the test sample. In some embodiments, estimating the fetal nucleic acid score of a test sample includes applying the coefficients from the XGBoost model to one or more sequence motif frequencies of the test sample. In some embodiments, estimating the fetal nucleic acid score of a test sample includes applying the coefficients from the XGBoost model to i) one or more fragment length spectra of the test sample and ii) one or more sequence motif frequencies of the test sample.

[0139] In certain applications, the fetal fraction (e.g., of a test sample) may be determined by any suitable method based on multiple fetal fraction sub-estimates (e.g., of the same test sample). In some embodiments, a method for improving the accuracy of an estimate of a fetal nucleic acid fraction in a test sample from a pregnant female comprises determining one or more fetal fraction sub-estimates, wherein an estimate of a fetal fraction of the sample is determined based on the one or more fetal fraction sub-estimates. In some embodiments, estimating or determining a fetal nucleic acid fraction of a sample (e.g., a test sample) comprises summing one or more fetal fraction sub-estimates. Summing may comprise determining an average, mean, median, AUC, or integral value based on multiple fetal fraction sub-estimates. Fetal fraction sub-estimates may be derived from a subset of measured features of the test sample (e.g., a subset of fragment length spectra and / or a subset of motif frequencies). In some embodiments, a first subset comprises one or more fragment length spectra of the test sample, and a second subset comprises one or more motif frequencies of the test sample.

[0140] Data processing

[0141] The measured features of the test sample and / or the measured features of the samples in the training set may include fragment length, fragment length ratio, fragment length spectrum, motif count and motif frequency, and may be referred to as raw data herein because the data represent the unoperated quantification of such features. In some embodiments, the measured feature data in the data set may be further processed (e.g., mathematically and / or statistically operated) and / or displayed to facilitate providing results. In certain embodiments, the data set (including a larger data set) may benefit from pre-processing to facilitate further analysis. The pre-processing of the data set sometimes involves removing redundant and / or uninformative data. Without being limited by theory, data processing and / or pre-processing may (i) remove noise data, (ii) remove uninformative data, (iii) remove redundant data, (iv) reduce the complexity of a larger data set, and / or (v) promote the transformation of data from one form to one or more other forms. When used with respect to data or data sets, the terms "pre-processing" and "processing" are collectively referred to as "processing" herein. Processing may make data easier to further analyze, and in some embodiments, results may be generated. In some embodiments, one or more or all of the processing methods are performed by a processor, a microprocessor, a computer in conjunction with a memory, and / or a machine controlled by a microprocessor.

[0142] As used herein, the term "noisy data" refers to (a) data that has significant differences between data points when analyzed or plotted, (b) data with significant standard deviations (e.g., greater than 3 standard deviations), (c) data with significant standard errors of the mean, etc., and combinations of the foregoing. Noisy data sometimes arises due to the quantity and / or quality of the starting material (e.g., nucleic acid sample), and sometimes arises as part of the preparation or replication process of the DNA used to generate the sequence reads. In certain embodiments, noise arises from overrepresentation of certain sequences when prepared using PCR-based methods.

[0143] As used herein, the term "uninformative data" refers to data with numerical values ​​that are significantly different from a predetermined threshold or fall outside a predetermined cutoff range of values. Typically, thresholds or value ranges are calculated by mathematically and / or statistically operating data (e.g., from reference and / or subjects). In some embodiments, uncertainty values ​​are determined. Uncertainty values ​​are typically measures of variance or error, and can be any suitable measures of variance or error. In some embodiments, uncertainty values ​​are standard deviation, standard error, calculated variance, p-value, or mean absolute deviation (MAD).

[0144] Any suitable program can be utilized to process the data set described herein. Non-limiting examples of programs suitable for processing data sets include filtering, normalization, weighting, monitoring peak height, monitoring peak area, monitoring peak edge, determining area ratio, mathematical processing of data, statistical processing of data, application of statistical algorithms, analysis with fixed variables, analysis with optimized variables, drawing data to identify patterns or trends for additional processing, etc., and aforesaid combinations. In some embodiments, data sets are processed based on various features (e.g., GC content, redundant mapping reads, centromere regions, telomere regions, etc., and combinations thereof) and / or variables (e.g., fetal sex, maternal age, maternal ploidy, percentage contribution of fetal nucleic acid, etc., and combinations thereof). In certain embodiments, processing data sets as described herein can reduce the complexity and / or dimension of large and / or complex data sets. In some embodiments, data sets may include thousands to millions of sequence reads, fragment lengths, fragment length ratios, fragment length spectra, sequence motifs, motif counts, and motif frequencies for each sample in each test sample and / or training set.

[0145] In certain embodiments, data processing can be carried out in any number of steps. For example, in some embodiments, only a single processing procedure can be used to process data, and in certain embodiments, 1 or more, 5 or more, 10 or more, or 20 or more processing steps (for example, 1 or more processing steps, 2 or more processing steps, 3 or more processing steps, 4 or more processing steps, 5 or more processing steps, 6 or more processing steps, 7 or more processing steps, 8 or more processing steps, 9 or more processing steps, 10 or more processing steps, 11 or more processing steps, 12 or more processing steps, 13 or more processing steps, 14 or more processing steps, 15 or more processing steps, 16 or more processing steps, 17 or more processing steps, 18 or more processing steps, 19 or more processing steps, or 20 or more processing steps) processing data can be used. In some embodiments, a processing step can be the same step repeated twice or more, and in certain embodiments, a processing step can be two or more different processing steps implemented simultaneously or sequentially. In some embodiments, the sequence read data may be processed using any suitable number and / or combination of the same or different processing steps to provide an outcome.

[0146] In certain embodiments, processing a data set by the criteria described herein can reduce the complexity and / or dimension of the data set. For example, principal component analysis (PCA) can reduce the complexity and / or dimension of a data set. PCA can be performed by a suitable PCA method or a variation thereof. Non-limiting examples of PCA methods include canonical correlation analysis (CCA), Karhunen–Loève transformation (KLT), Hotelling transformation, proper orthogonal decomposition (POD), singular value decomposition (SVD) of X, eigenvalue decomposition (EVD) of XTX, factor analysis, Eckart–Young theorem, Schmidt–Mirsky theorem, empirical orthogonal function (EOF), empirical eigenfunction decomposition, empirical component analysis, quasi-harmonic mode, spectrum decomposition, empirical modal analysis, etc., variations or combinations thereof. PCA generally identifies one or more principal components. In some embodiments, PCA identifies the 1st, 2nd, 3rd, 4th, 5th, 6th, 7th, 8th, 9th and 10th or more principal components.

[0147] In some embodiments, one or more processing steps may include one or more normalization steps. Normalization can be performed by suitable methods described herein or known in the art. In certain embodiments, normalization includes adjusting the value measured on different scales to a conceptually common scale. In certain embodiments, normalization includes complex mathematical adjustments to make the probability distribution of the adjusted value consistent. In some embodiments, normalization includes making distribution consistent with a normal distribution. In certain embodiments, normalization includes mathematical adjustments, which allow to compare the corresponding normalized values ​​of different data sets in a manner that eliminates some overall effects (e.g., errors and anomalies). In certain embodiments, normalization includes scaling. Normalization sometimes includes dividing one or more data sets by a predetermined variable or formula.

[0148] Any suitable number of normalizations can be used. In some embodiments, the data set can be normalized 1 or more times, 5 or more times, 10 or more times or even 20 or more times. The data set can be normalized to represent the value (e.g., normalized value) of any suitable feature or variable (e.g., sample data, reference data or both). According to the feature or attribute selected as the predetermined normalization variable, the normalized data set sometimes has the effect of isolating statistical errors. Normalizing the data set sometimes also allows the data characteristics of data with different scales to be compared by placing the data on a common scale (e.g., a predetermined normalization variable). In some embodiments, one or more normalizations to statistically derived values ​​can be utilized to minimize data differences and reduce the importance of abnormal data.

[0149] In some embodiments, the processing step includes weighting. As used herein, the terms "weighted," "weighting," or "weighting function," or grammatical derivatives or equivalents thereof, refer to a mathematical operation sometimes used to change the influence of some data set features or variables relative to other data set features or variables. In some embodiments, a weighting function can be used to increase the influence of data with relatively small measurement variance, and / or to reduce the influence of data with relatively large measurement variance. A non-limiting example of a weighting function is [1 / (standard deviation) 2 ]. The weighting step is sometimes performed in a manner substantially similar to the normalization step. In some embodiments, the data set is divided by a predetermined variable (e.g., a weighting variable). The predetermined variable is typically selected (e.g., to minimize the objective function Phi) to weight different parts of the data set differently (e.g., to increase the impact of certain data types while reducing the impact of other data types).

[0150] In certain embodiments, the processing step may include one or more mathematical and / or statistical operations. Any suitable mathematical and / or statistical operations may be used individually or in combination to analyze and / or operate data sets as described herein. Any suitable number of mathematical and / or statistical operations may be used. In some embodiments, data sets may be mathematically and / or statistically operated 1 or more times, 5 or more times, 10 or more times, or 20 or more times. Non-limiting examples of usable mathematical and statistical operations include addition, subtraction, multiplication, division, algebraic functions, least squares estimators, curve fitting, differential equations, rational polynomials, dual polynomials, orthogonal polynomials, z scores, p values, x values, phi values, analysis of peak levels, determination of peak edge positions, calculation of peak area ratios, analysis of median chromosome levels, calculation of mean absolute deviations, residual sum of squares, mean, standard deviation, standard error, etc., or a combination thereof. Non-limiting examples of data set variables or features that may be statistically operated include fragment length, fragment length ratio, fragment length spectrum, motif count, motif frequency, fetal fraction, etc., or a combination thereof.

[0151] In some embodiments, the processing step may include the use of one or more statistical algorithms. Any suitable statistical algorithm may be used individually or in combination to analyze and / or manipulate the data set described herein. Any suitable number of statistical algorithms may be used. In some embodiments, 1 or more, 5 or more, 10 or more, or 20 or more statistical algorithms may be used to analyze the data set. Non-limiting examples of statistical algorithms suitable for the methods described herein include decision trees, counternulls, multiple comparisons, omnibus tests, Behrens-Fisher problems, bootstrapping, Fisher's method for combining independent significance tests, null hypothesis, type I error, type II error, exact test, one-sample Z test, two-sample Z test, one-sample t test, paired t test, two-sample pooled t test with equal variances, two-sample non-pooled t test with unequal variances, one-proportion z test, two-proportion pooled z test, two-proportion non-pooled z test, one-sample chi-square test, two-sample F test with equal variances, confidence interval, credible interval, significance, meta-analysis, simple linear regression, robust linear regression, etc., or a combination of the foregoing. Non-limiting examples of data set variables or features that can be analyzed using statistical algorithms include fragment length, fragment length ratio, fragment length spectrum, motif count, motif frequency, fetal fraction, etc., or a combination thereof.

[0152] In certain embodiments, a data set can be analyzed by utilizing multiple (e.g., 2 or more) statistical algorithms (e.g., least squares regression, principal component analysis, linear discriminant analysis, quadratic discriminant analysis, bagging, neural networks, support vector machine models, random forests, classification tree models, K nearest neighbors, logistic regression, and / or loss smoothing) and / or mathematical and / or statistical operations (e.g., referred to herein as operations). In some embodiments, the use of multiple operations can generate an N-dimensional space that can be used to provide results. In certain embodiments, the complexity and / or dimensionality of the data set can be reduced by analyzing the data set using multiple operations. For example, the use of multiple operations on the training data set can generate an N-dimensional space (e.g., a probability map) that can be used to represent fetal scores. Analysis of test samples using a substantially similar set of operations can be used to generate N-dimensional points for each test sample. The complexity and / or dimensionality of the test subject data set is sometimes reduced to a single value or N-dimensional point, which can be easily compared to the N-dimensional space generated from the training data.

[0153] In some embodiments, data processing includes cross validation. Cross validation is sometimes referred to as rotation estimation. In some embodiments, a cross validation method is applied to evaluate how accurately a prediction model will be performed in practice using a test sample. In some embodiments, a round of cross validation includes partitioning a data sample into complementary subsets, performing a cross validation analysis on a subset (e.g., sometimes referred to as a training set), and validating the analysis using another subset (e.g., sometimes referred to as a validation set or a test set). In certain embodiments, multiple rounds of cross validation are performed using different partitions and / or different subsets. Non-limiting examples of cross validation methods include leave one out, sliding edge method, K fold method, 2 fold method, repeated random subsampling, etc. or a combination thereof. In some embodiments, cross validation randomly selects a working set of 80% of a sample set with a known fetal score, and uses the subset to train the model. In certain embodiments, random selection is repeated multiple times.

[0154] Machines, software and interfaces

[0155] Certain processes and methods described herein (e.g., obtaining sequence reads, mapping sequence reads, measuring nucleic acid fragment lengths, generating nucleic acid fragment length spectra, identifying sequence motifs, determining sequence motif frequencies, estimating fetal scores, etc.) are generally not accomplished in human thought, and therefore cannot be accomplished without computers, microprocessors, software, modules, or other machines. The methods described herein are generally computer-implemented methods, and one or more parts of the methods are sometimes performed by one or more processors (e.g., microprocessors), computers, or microprocessor-controlled machines. The embodiments related to the methods described in this document are generally applicable to the same or related processes implemented by the instructions in the systems, machines, and computer program products described herein. The embodiments related to the methods described in this document are generally applicable to the same or related processes implemented by non-transitory computer-readable storage media having executable programs stored thereon, wherein the program instructs the microprocessor to perform the method or a portion thereof. In some embodiments, the processes and methods described herein are performed by automated methods. In some embodiments, one or more steps and methods described herein are implemented by a microprocessor and / or a computer, and / or implemented in conjunction with a memory. In some embodiments, the automated method is embodied in software, modules, microprocessors, peripherals, and / or machines comprising the like that determine sequence reads, counts, mapped sequence reads, fragment lengths, levels, profiles, sequence motifs, sequence motif frequencies, fetal fractions, normalization, comparisons, range setting, classification, adjustment, mapping, results, transformations, and identification. As used herein, software refers to computer-readable program instructions that, when executed by a microprocessor, perform computer operations as described herein.

[0156] Sequence reads, fragment lengths, spectra, and / or sequence motif frequencies derived from test subjects (e.g., patients, pregnant subjects) and / or reference subjects may be further analyzed and processed to estimate the fetal nucleic acid fraction in samples from test subjects or reference subjects. Sequence reads, fragment lengths, spectra, and / or sequence motif frequencies are sometimes referred to as "data" or "datasets." In some embodiments, data or data sets may be characterized by one or more features or variables (e.g., sequence-based [e.g., GC content, specific nucleotide sequences, etc.], functional specific [e.g., expressed genes, oncogenes, etc.], position-based [genomic specificity, chromosome specificity, partial or partial specificity], etc., and combinations thereof). In certain embodiments, data or data sets may be organized into a matrix having two or more dimensions based on one or more features or variables. Data organized into a matrix may be organized using any suitable features or variables. Non-limiting examples of data in a matrix include data organized according to maternal age, maternal ploidy, and fetal contribution.

[0157] Machines, software and interfaces can be used to perform the methods described herein. Using machines, software and interfaces, users can input, request, query or determine the option of using specific information, programs or processes (e.g., mapping sequence reads, processing mapping data and / or providing results), which can involve, for example, implementing statistical analysis algorithms, statistical significance algorithms, statistical algorithms, iterative steps, verification algorithms and graphical representations. In some embodiments, data sets can be input by users as input information, users can download one or more data sets through suitable hardware media (e.g., flash drives), and / or users can send data sets from one system to another system for subsequent processing and / or providing results (e.g., sequence read data is sent from a sequencer to a computer system for sequence read mapping; mapped sequence data is sent to a computer system for processing and generating results and / or reports).

[0158] The system typically includes one or more machines. Each machine includes one or more of a memory, one or more microprocessors, and instructions. Where the system includes two or more machines, some or all of the machines may be located at the same location, some or all of the machines may be located at different locations, all of the machines may be located at one location, and / or all of the machines may be located at different locations. Where the system includes two or more machines, some or all of the machines may be located at the same location as a user, some or all of the machines may be located at a different location from the user, all of the machines may be located at the same location as the user, and / or all of the machines may be located at one or more locations different from the user.

[0159] Systems sometimes include a computing machine and a sequencing device or machine, wherein the sequencing device or machine is configured to receive physical nucleic acids and generate sequence reads, and the computing machine is configured to process the reads from the sequencing device or machine. The computing machine is sometimes configured to measure fragment lengths, generate fragment length profiles, determine sequence motifs, determine sequence motif frequencies, and / or estimate the fetal nucleic acid fraction in the test sample from the sequence reads.

[0160] For example, a user may issue a query to the software, which may then access the data set via the Internet, and in certain embodiments, may prompt the programmable microprocessor to obtain a suitable data set based on given parameters. The programmable microprocessor may also prompt the user to select one or more data set options selected by the microprocessor based on given parameters. The programmable microprocessor may prompt the user to select one or more data set options selected by the microprocessor based on information found via the Internet, other internal or external information, etc. Options may be selected to select one or more data feature selections, one or more statistical algorithms, one or more statistical analysis algorithms, one or more statistical significance algorithms, iterative steps, one or more validation algorithms, and one or more graphical representations of methods, machines, devices, computer programs, or non-transient computer-readable storage media on which executable programs are stored.

[0161] The system proposed herein may include general components of a computer system, such as a network server, a laptop system, a desktop system, a handheld system, a personal digital assistant, a computing kiosk, etc. The computer system may include one or more input devices, such as a keyboard, a touch screen, a mouse, voice recognition, or other devices that allow a user to input data to the system. The system may also include one or more outputs, including but not limited to a display screen (e.g., a CRT or LCD), a speaker, a fax machine, a printer (e.g., a laser, inkjet, impact, black and white or color printer), or other outputs for providing visual, auditory and / or hard copy output of information (e.g., results and / or reports).

[0162] In the system, input and output devices may be connected to a central processing unit, which may include, in addition to other components, a microprocessor for executing program instructions and a memory for storing program code and data. In some embodiments, the process may be implemented as a single user system located in a single geographic location. In certain embodiments, the process may be implemented as a multi-user system. In the case of multi-user implementation, multiple central processing units may be connected via a network. The network may be local, including a single department in a part of a building, an entire building, across multiple buildings, across a region, across an entire country or all over the world. The network may be private, owned and controlled by a provider, or it may be implemented as an Internet-based service, where a user accesses a web page to input and retrieve information. Therefore, in certain embodiments, the system includes one or more machines, which may be local or remote relative to the user. A user may access more than one machine in one location or multiple locations, and data may be mapped and / or processed serially and / or in parallel. Therefore, suitable configuration and control may be utilized to use multiple machines to map and / or process data, such as in a local network, a remote network, and / or a "cloud" computing platform.

[0163] In some embodiments, the system may include a communication interface. The communication interface allows software and data to be transmitted between a computer system and one or more external devices. Non-limiting examples of communication interfaces include modems, network interfaces (such as Ethernet cards), communication ports, PCMCIA slots and cards, etc. The software and data transmitted via the communication interface are generally in the form of signals, which may be electrical signals, electromagnetic signals, optical signals, and / or other signals that can be received by the communication interface. The signal is generally provided to the communication interface via a channel. The channel generally carries signals and can be implemented using wires or cables, optical fibers, telephone lines, cellular phone links, RF links, and / or other communication channels. Therefore, in one example, the communication interface can be used to receive signal information that can be detected by a signal detection module.

[0164] Data may be input by suitable devices and / or methods, including but not limited to manual input devices or direct data entry devices (DDE). Non-limiting examples of manual devices include keyboards, conceptual keyboards, touch-sensitive screens, light pens, mice, trackballs, joysticks, graphics tablets, scanners, digital cameras, video digitizers, and voice recognition devices. Non-limiting examples of DDE include barcode readers, magnetic barcodes, smart cards, magnetic ink character recognition, optical character recognition, cursor recognition, and turnaround files.

[0165] In some embodiments, the output from the sequencing device or machine can be used as data that can be input via an input device. In certain embodiments, the mapped sequence reads can be used as data that can be input via an input device. In certain embodiments, the nucleic acid fragment size (e.g., length) can be used as data that can be input via an input device. In certain embodiments, sequence motifs can be used as data that can be input via an input device. In certain embodiments, the combination of nucleic acid fragment size (e.g., length) and sequence motifs can be used as data that can be input via an input device. In certain embodiments, simulated data is generated by a computer simulation process, and the simulated data is used as data that can be input via an input device. The term "computer simulation" refers to research and experiments performed using a computer. The computer simulation process includes, but is not limited to, drawing sequence reads and processing the sequence reads drawn according to the process described herein.

[0166] The system may include software that can be used to perform the processes described herein, and the software may include one or more modules (e.g., sequencing modules, logic processing modules, data display organization modules) for performing such processes. The term "software" refers to computer-readable program instructions that perform computer operations when executed by a computer. Instructions that can be executed by one or more microprocessors are sometimes provided as executable codes, which can enable one or more microprocessors to implement the methods described herein when executed. The modules described herein may exist as software, and the instructions (e.g., processes, routines, subroutines) embodied in the software may be implemented or executed by a microprocessor. For example, a module (e.g., a software module) may be part of a program that performs a specific process or task. The term "module" refers to a self-contained functional unit that can be used in a larger machine or software system. A module may include a set of instructions for implementing the functions of a module. A module may transform data and / or information. Data and / or information may be in an appropriate form. For example, data and / or information may be digital or analog. In certain embodiments, data and / or information may sometimes be packets, bytes, characters, or bits. In some embodiments, data and / or information can be any collected, compiled or available data or information. Non-limiting examples of data and / or information include suitable media, pictures, videos, sounds (e.g., frequency, audible or inaudible), numbers, constants, values, objects, time, functions, instructions, maps, references, sequences, reads, mapped reads, levels, ranges, thresholds, signals, displays, representations, or transformations thereof. A module can accept or receive data and / or information, transform the data and / or information into a second form, and provide or transmit the second form to a machine, a peripheral device, a component, or another module. A module can perform one or more of the following non-limiting functions: for example, obtaining sequence reads, mapping sequence reads, measuring nucleic acid fragment lengths, generating nucleic acid fragment length spectra, identifying sequence motifs, determining sequence motif frequencies, estimating fetal fractions, normalizing, providing normalized spectra, providing normalized sequence motif frequencies, comparing two or more spectra, comparing two or more sequence motif frequencies, providing uncertainty values, providing or determining expected ranges, providing adjustments, classifications, mapping, and / or determining results. In certain embodiments, a microprocessor may implement instructions in a module. In some embodiments, one or more microprocessors are required to implement instructions in a module or group of modules. A module may provide data and / or information to another module, machine, or source, and may receive data and / or information from another module, machine, or source.

[0167] Computer program products are sometimes embodied on tangible computer-readable media, and sometimes tangibly embodied on non-transitory computer-readable media. Modules are sometimes stored on computer-readable media (e.g., disks, drives) or in memory (e.g., random access memory). Modules and microprocessors capable of implementing instructions from modules may be located in one machine or in different machines. Modules and / or microprocessors capable of implementing module instructions may be located in the same location as a user (e.g., a local network), or in a different location from a user (e.g., a remote network, a cloud system). In embodiments in which two or more modules are implemented, modules may be located in the same machine, one or more modules may be located in different machines in the same physical location, and one or more modules may be located in different machines in different physical locations.

[0168] In some embodiments, the machine includes at least one microprocessor for implementing the instructions in the module. The sequence reads mapped to the reference genome are sometimes accessed by a microprocessor, which executes instructions configured to implement the method described herein. The sequence reads accessed by the microprocessor can be in the memory of the system, and these sequence reads can be accessed after obtaining these sequence reads and placed in the memory of the system. In some embodiments, the machine includes a microprocessor (for example, one or more microprocessors), which can execute and / or implement one or more instructions (for example, processes, routines and / or subroutines) from the module. In some embodiments, the machine includes a plurality of microprocessors, such as a microprocessor that coordinates and works in parallel. In some embodiments, the machine runs with one or more external microprocessors (for example, internal or external networks, servers, storage devices and / or storage networks (for example, clouds)). In some embodiments, the machine includes a module. In certain embodiments, the machine includes one or more modules. The machine including a module can generally receive one or more data and / or information and transmit one or more data and / or information to and from other modules. In certain embodiments, the machine includes peripheral devices and / or components. In certain embodiments, the machine may include one or more peripherals or parts, which can transmit data and / or information to and from other modules, peripherals and / or parts. In certain embodiments, the machine interacts with peripherals and / or parts that provide data and / or information. In certain embodiments, peripherals and parts help the machine implement functions or directly interact with modules. The non-limiting examples of peripherals and / or parts include suitable computer peripherals, I / O or storage methods or equipment, including but not limited to scanners, printers, displays (e.g., monitors, LEDs, LCTs or CRTs), cameras, microphones, pads (e.g., ipads, tablet computers), touch screens, smart phones, mobile phones, USB I / O devices, USB mass storage devices, keyboards, computer mice, digital pens, modems, hard drives, jump drives, flash drives, microprocessors, servers, CDs, DVDs, graphics cards, special I / O devices (e.g., sequencers, photocells, photomultiplier tubes, optical readers, sensors, etc.), one or more flow cells, fluid handling components, network interface controllers, ROMs, RAMs, wireless transmission methods and equipment (bluetooth, WiFi, etc.), World Wide Web (www), the Internet, computers and / or another module.

[0169] Software is typically provided on a program product that includes program instructions recorded on a computer-readable medium, including, but not limited to: magnetic media, including floppy disks, hard disks, and tapes; and optical media, including CD-ROMs, DVDs, magneto-optical disks, flash drives, RAMs, floppy disks, and the like, as well as other such media on which program instructions may be recorded. In an online implementation, servers and websites maintained by an organization may be configured to provide software downloads to remote users, or remote users may access a remote system maintained by an organization to remotely access the software. The software may obtain or receive input information. The software may include a module that specifically obtains or receives data (e.g., a data receiving module that receives sequence read data and / or mapping read data), and may include a module that specifically processes data (e.g., a processing module that processes received data). The terms "obtaining" and "receiving" input information refer to receiving data (e.g., sequence reads, mapping reads) from a local or remote site, manual data input, or any other method of receiving data by means of computer communication. The input information may be generated at the same location where it is received, or it may be generated at a different location and transmitted to the receiving location. In some embodiments, input information is modified (eg, placed into a format suitable for processing (eg, tabulated)) before being processed.

[0170] In some embodiments, a computer program product is provided, for example, a computer program product comprising a computer usable medium having computer readable program code embodied therein, the computer readable program code being suitable for being executed to implement a method comprising the following steps: a) obtaining sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating free (CCF) nucleic acid from a test sample from a pregnant subject; b) measuring fragment lengths of a plurality of circulating free nucleic acid fragments; c) generating one or more fragment length profiles for the test sample; d) determining sequence motifs at ends of a plurality of circulating free nucleic acid fragments; e) determining one or more sequence motif frequencies for the test sample; f) estimating a fetal nucleic acid fraction for the test sample based on i) the one or more fragment length profiles for the test sample and ii) the one or more sequence motif frequencies for the test sample.

[0171] In certain embodiments, software may include one or more algorithms. Algorithms may be used to process data and / or provide results or reports according to a finite sequence of instructions. An algorithm is typically a list of defined instructions for completing a task. Starting from an initial state, an instruction may describe a calculation performed by a series of defined continuous states, ultimately terminating at a final end state. The transition from one state to the next state is not necessarily deterministic (e.g., some algorithms include randomness). As an example and not limitation, an algorithm may be a search algorithm, a sorting algorithm, a merging algorithm, a numerical algorithm, a graphic algorithm, a string algorithm, a modeling algorithm, a computational genome algorithm, a combinatorial algorithm, a machine learning algorithm, a cryptographic algorithm, a data compression algorithm, a parsing algorithm, etc. An algorithm may include two or more algorithms that work in combination with an algorithm. An algorithm may have any suitable complexity class and / or parameterized complexity. An algorithm may be used for computing and / or data processing, and in some embodiments, may be used in deterministic or probabilistic / predictive methods. An algorithm may be implemented in a computing environment by using a suitable programming language, and non-limiting examples of programming languages ​​are C, C++, Java, Perl, Python, Fortran, etc. In some embodiments, the algorithm may be configured or modified to include error tolerance, statistical analysis, statistical significance, and / or comparison to other information or data sets (eg, as applicable when using a neural network or clustering algorithm).

[0172] In certain embodiments, several algorithms can be implemented for use in software. In some embodiments, these algorithms can be trained with raw data. For each new raw data sample, the trained algorithm can generate a representative processed data set or result. Compared with the processed parent data set, the processed data set sometimes has reduced complexity. In some embodiments, based on the processed set, the performance of the trained algorithm can be evaluated based on sensitivity and specificity. In certain embodiments, the algorithm with the highest sensitivity and / or specificity can be identified and utilized.

[0173] In certain embodiments, simulation (or emulation) data can assist data processing, for example, by training algorithms or testing algorithms. In some embodiments, simulation data includes various samplings of hypotheses of different sequence read groupings. Simulation data can be based on what is expected from a real population, or may be biased toward testing algorithms and / or assigning correct classifications. Simulation data is also referred to as "virtual" data herein. In certain embodiments, simulation can be performed by a computer program. A possible step using a simulated data set is to assess the confidence of identified results, such as the degree of matching or the best representative degree of random sampling with the original data. One method is to calculate a probability value (p value), which estimates the probability that a random sample has a better score than a selected sample. In some embodiments, an empirical model can be evaluated, wherein it is assumed that at least one sample matches a reference sample (with or without a distinguishable change). In some embodiments, another distribution (such as a Poisson distribution) can be used to define a probability distribution.

[0174] In certain embodiments, the system may include one or more microprocessors. The microprocessor may be connected to a communication bus. The computer system may include a main memory, typically a random access memory (RAM), and may also include a secondary memory. In some embodiments, the memory includes a non-transient computer-readable storage medium. The secondary memory may include, for example, a hard disk drive and / or a removable storage drive, which represents a floppy disk drive, a tape drive, an optical disk drive, a memory card, etc. The removable storage drive is typically read from and / or written to a removable storage unit. Non-limiting examples of removable storage units include floppy disks, tapes, optical disks, etc., which may be read and written to by (for example) a removable storage drive. A removable storage unit may include a computer-usable storage medium having computer software and / or data stored therein.

[0175] The microprocessor can implement the software in the system. In some embodiments, the microprocessor can be programmed to automatically perform the tasks that the user described herein can perform. Therefore, the microprocessor or the algorithm executed by such a microprocessor hardly requires the supervision or input of the user (for example, the software can be programmed to automatically implement the function). In some embodiments, the complexity of the process is so great that a single person or a group of people cannot complete the process in a sufficiently short time to estimate the fetal nucleic acid score.

[0176] In some embodiments, the secondary memory may include other similar devices for allowing computer programs or other instructions to be loaded into the computer system. For example, the system may include a removable storage unit and an interface device. Non-limiting examples of such systems include program cartridges and cartridge interfaces (such as those found in video game devices), removable memory chips (such as EPROM or PROM) and associated sockets, and other removable storage units and interfaces that allow software and data to be transferred from the removable storage unit to the computer system.

[0177] In some embodiments, an entity may generate counts of sequence reads, map the sequence reads to a reference genome, and utilize the mapped reads in the methods, systems, machines, apparatus, or computer program products described herein. In certain embodiments, in the methods, systems, machines, devices, or computer program products described herein, sequence reads mapped to a reference genome are sometimes transferred by one entity to a second entity for use by the second entity.

[0178] In some embodiments, an entity generates sequence reads, and a second entity maps these sequence reads to a reference genome. The second entity sometimes utilizes the mapping reads in the method, system, machine, or computer program product described herein. In certain embodiments, the second entity transfers the mapped reads to a third entity, and the third entity utilizes the mapped reads in the method, system, machine, or computer program product described herein. In embodiments involving a third entity, the third entity is sometimes identical to the first entity. That is, the first entity sometimes transfers the sequence reads to the second entity, and the second entity can map the sequence reads to the reference genome, and the second entity can transfer the mapped reads to the third entity. The third entity can sometimes utilize the mapping reads in the method, system, machine, or computer program product described herein, wherein the third entity is sometimes identical to the first entity, and sometimes the third entity is different from the first or second entity. In some embodiments, an entity obtains blood from a pregnant subject, optionally separates nucleic acids from blood (e.g., from plasma or serum), and transfers blood or nucleic acids to the second entity that generates sequence reads from nucleic acids.

[0179] The systems, methods, and data structures described herein can operate with many other general or special purpose computing system environments or configurations. Examples of known computing systems, environments, and / or configurations that may be suitable include, but are not limited to, personal computers, server computers, thin clients, fat clients, handheld or laptop devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and the like. Any type of computer-readable medium that can store computer-accessible data, such as magnetic tape cassettes, flash memory cards, digital video disks, Bernoulli boxes, random access memory (RAM), read-only memory (ROM), and the like can be used in the operating environment.

[0180] In certain embodiments, provided herein is a system comprising one or more microprocessors and a memory, the memory comprising instructions executable by the one or more microprocessors, and the memory comprising sequence reads mapped to a reference genome, the sequence reads being reads of circulating free (CCF) nucleic acid from a test sample from a pregnant subject, and the instructions executable by the one or more microprocessors being configured to a) measure fragment lengths of a plurality of circulating free nucleic acid fragments; b) generate one or more fragment length profiles for the test sample; c) determine sequence motifs at ends of a plurality of circulating free nucleic acid fragments; d) determine one or more sequence motif frequencies for the test sample; and e) estimate the fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles for the test sample and ii) the one or more sequence motif frequencies for the test sample.

[0181] In certain embodiments, the present invention also provides a machine comprising one or more microprocessors and a memory, the memory comprising instructions capable of being executed by the one or more microprocessors, and the memory comprising sequence reads mapped to a reference genome, the sequence reads being reads of circulating free nucleic acid from a test sample of a pregnant subject, and the instructions capable of being executed by the one or more microprocessors are configured to a) measure fragment lengths of a plurality of circulating free nucleic acid fragments; b) generate one or more fragment length profiles of the test sample; c) determine sequence motifs at ends of a plurality of circulating free nucleic acid fragments; d) determine one or more sequence motif frequencies of the test sample; and e) estimate the fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0182] In certain embodiments, the present invention also provides a non-transitory computer-readable storage medium having an executable program stored thereon, wherein the program instructs a microprocessor to perform the following operations: a) access sequence reads mapped to a reference genome, which are reads of circulating free nucleic acid from a test sample of a pregnant subject, b) measure the fragment lengths of a plurality of circulating free nucleic acid fragments; c) generate one or more fragment length profiles of the test sample; d) determine sequence motifs at the ends of a plurality of circulating free nucleic acid fragments; e) determine one or more sequence motif frequencies of the test sample; and f) estimate the fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0183] Certain Implementations

[0184] The following are non-limiting examples of certain implementations of the technology.

[0185] A1. A method for estimating the fraction of fetal nucleic acid in a test sample from a pregnant subject, the method comprising:

[0186] a) obtaining sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating cell-free (CCF) nucleic acid from a test sample from a pregnant subject;

[0187] b) measuring the fragment lengths of a plurality of circulating free nucleic acid fragments;

[0188] c) generating one or more fragment length profiles of the test sample;

[0189] d) determining sequence motifs at the ends of a plurality of circulating free nucleic acid fragments;

[0190] e) determining one or more sequence motif frequencies of the test sample; and

[0191] f) estimating a fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0192] A2. A method according to embodiment A1, wherein the sequence reads are obtained by a paired-end sequencing process and the sequence reads are paired-end sequence reads.

[0193] A3. A method according to embodiment A2, wherein the fragment length is measured in (b) according to the mapping position of the paired end sequence reads.

[0194] A4. A method according to any one of embodiments A1-A3, wherein the fragment lengths are measured for multiple genomic intervals.

[0195] A5. A method according to embodiment A4, wherein the genomic interval length is about 100 kilobases (kb).

[0196] A6. A method according to any one of embodiments A1-A5, wherein the one or more fragment length profiles are generated in (c) according to the ratio of X to Y for multiple genomic intervals, wherein X is the number of CCF nucleic acid fragments having a length within a first selected fragment length range, and Y is the number of CCF nucleic acid fragments having a length within a second selected fragment length range.

[0197] A7. A method according to embodiment A6, wherein the first selected fragment length ranges from about 80 bases to about 150 bases, and the second selected fragment length ranges from about 151 bases to about 300 bases.

[0198] A8. A method according to embodiment A6 or A7, wherein the genomic interval length is about 100 kilobases (kb).

[0199] A9. The method of any one of embodiments A1-A8, wherein the one or more fragment length profiles are generated for one or more genomic segments in (c).

[0200] A10. A method according to embodiment A9, wherein the length of the genomic segment is 5 megabases (Mb).

[0201] A11. A method according to any one of embodiments A1-A10, wherein the sequence motif in (d) is a 5' sequence motif.

[0202] A12. A method according to any one of embodiments A1-A11, wherein the sequence motif in (d) is a four base pair (bp) sequence motif.

[0203] A13. A method according to any one of embodiments A1-A12, wherein (e) comprises determining one or more sequence motif frequencies of one or more chromosomes.

[0204] A14. A method according to any one of embodiments A1-A13, wherein (e) comprises determining one or more frequencies of one or more sequence motifs selected from GGAA, AGAA, GTTT, GAAT and GGTT.

[0205] A15. A method according to any one of embodiments A1-A14, wherein the one or more sequence motif frequencies are determined based on the frequencies of the one or more sequence motifs in the mapped sequence reads of one or more chromosomes.

[0206] A15. A method according to any one of embodiments A1-A14, wherein estimating the fetal nucleic acid fraction of the test sample in (f) comprises applying one or more model parameters from one or more models to i) the one or more fragment length spectra of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0207] A16. A method according to implementation scheme A15, wherein the model parameters are obtained from a sample training set.

[0208] A17. A method according to embodiment A16, wherein the fetal nucleic acid fraction of each sample in the sample training set is known.

[0209] A18. A method according to any one of embodiments A17, wherein the one or more model parameters are obtained from the training set based on 1) the fetal nucleic acid fraction of each sample in the sample training set and 2) the one or more fragment length spectra of each sample in the sample training set.

[0210] A19. A method according to any one of embodiments A17 or A18, wherein the one or more model parameters are obtained from the training set based on 1) the fetal nucleic acid score of each sample in the sample training set and 2) the one or more sequence motif frequencies of each sample in the sample training set.

[0211] A20. A method according to any one of embodiments A17-A19, wherein the one or more model parameters are obtained from the training set based on 1) the fetal nucleic acid fraction of each sample in the sample training set and 2) the fitting relationship between one or more fragment length spectra and one or more sequence motif frequencies of each sample in the sample training set.

[0212] A20.1 A method according to any one of embodiments A15-A20, wherein the one or more model parameters include coefficients derived from the one or more models.

[0213] A21. A method according to any one of embodiments A15-A20.1, wherein the one or more models include linear regression.

[0214] A22. A method according to embodiment A21, wherein estimating the fetal nucleic acid fraction of the test sample in (f) comprises applying the regression coefficients from the linear regression to i) the one or more fragment length spectra of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0215] A23. A method according to any one of embodiments A15-A22, wherein the one or more models include an elastic network.

[0216] A24. A method according to embodiment A23, wherein estimating the fetal nucleic acid fraction of the test sample in (f) comprises applying coefficients from the elastic net model to i) the one or more fragment length spectra of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0217] A25. The method of any one of embodiments A15-A24, wherein the one or more models comprise XGBoost.

[0218] A26. A method according to embodiment A25, wherein estimating the fetal nucleic acid fraction of the test sample in (f) comprises applying coefficients from the XGBoost model to i) the one or more fragment length spectra of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0219] A27. The method according to any one of embodiments A1-A26, further comprising, before (a), sequencing the circulating free (CCF) nucleic acid from the test sample by a sequencing process.

[0220] A28. A method according to embodiment A27, wherein the sequencing process is a non-targeted sequencing process.

[0221] A29. A method according to embodiment A27, wherein the sequencing process is a massively parallel sequencing process.

[0222] A30. A method according to embodiment A27, wherein the sequencing process is a non-targeted massively parallel sequencing process.

[0223] A31. A method according to any one of embodiments A27-A30, wherein the circulating free (CCF) nucleic acids from the test sample are sequenced at a coverage factor of 1.0 or greater.

[0224] A32. A method according to any one of embodiments A27-A30, wherein the circulating free (CCF) nucleic acids from the test sample are sequenced at a coverage factor of less than 1.0.

[0225] A33. A method according to any one of embodiments A27-A32, wherein the sequencing process generates thousands to millions of sequence reads.

[0226] A34. The method according to any one of embodiments A1-A33 further includes, before (a), mapping the sequence reads to the reference genome.

[0227] A35. The method according to any one of embodiments A1-A33 further includes, before (a), mapping thousands to millions of sequence reads to the reference genome.

[0228] A36. The method according to any one of embodiments A1-A35 also includes obtaining a quantification of the mapped sequence reads of the test sample.

[0229] A37. A method according to embodiment A36, wherein (f) includes estimating the fetal nucleic acid fraction of the test sample based on i) the one or more fragment length spectra of the test sample, ii) the one or more sequence motif frequencies of the test sample, and iii) quantification of the mapped sequence reads of the test sample.

[0230] B1. A system comprising one or more microprocessors and a memory, the memory comprising instructions executable by the one or more microprocessors, and the memory comprising sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating cell-free (CCF) nucleic acid from a test sample from a pregnant subject, and wherein the instructions executable by the one or more microprocessors are configured to:

[0231] a) measuring the fragment lengths of a plurality of circulating free nucleic acid fragments;

[0232] b) generating one or more fragment length profiles of the test sample;

[0233] c) determining sequence motifs at the ends of a plurality of circulating free nucleic acid fragments;

[0234] d) determining one or more sequence motif frequencies of the test sample; and

[0235] e) estimating a fetal nucleic acid fraction for the test sample based on i) the one or more fragment length profiles for the test sample and ii) the one or more sequence motif frequencies for the test sample.

[0236] B2. The system according to embodiment B1 further comprises one or more features described in embodiments A2-A35, and / or is further configured to perform one or more methods described in embodiments A2-A37.

[0237] C1. A machine comprising one or more microprocessors and a memory, the memory comprising instructions executable by the one or more microprocessors, and the memory comprising sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating free nucleic acid from a test sample from a pregnant subject, and wherein the instructions executable by the one or more microprocessors are configured to:

[0238] a) measuring the fragment lengths of a plurality of circulating free nucleic acid fragments;

[0239] b) generating one or more fragment length profiles of the test sample;

[0240] c) determining sequence motifs at the ends of a plurality of circulating free nucleic acid fragments;

[0241] d) determining one or more sequence motif frequencies of the test sample; and

[0242] e) estimating a fetal nucleic acid fraction for the test sample based on i) the one or more fragment length profiles for the test sample and ii) the one or more sequence motif frequencies for the test sample.

[0243] C2. The machine according to embodiment C1 further comprises one or more features described in embodiments A2-A35, and / or is further configured to perform one or more methods described in embodiments A2-A37.

[0244] D1. A non-transitory computer-readable storage medium having an executable program stored thereon, wherein the program instructs a microprocessor to perform the following operations:

[0245] a) accessing sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating free nucleic acid from a test sample from a pregnant subject;

[0246] b) measuring the fragment lengths of a plurality of circulating free nucleic acid fragments;

[0247] c) generating one or more fragment length profiles of the test sample;

[0248] d) determining sequence motifs at the ends of a plurality of circulating free nucleic acid fragments;

[0249] e) determining one or more sequence motif frequencies of the test sample; and

[0250] f) estimating a fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

[0251] D2. The non-transitory computer-readable storage medium according to embodiment D1 further comprises one or more features described in embodiments A2-A35, and / or is further configured to execute one or more methods described in embodiments A2-A37.

[0252] Example

[0253] The examples set forth below illustrate certain embodiments and do not limit the technology.

[0254] Example 1: Predicting fetal fraction using fragmentomics

[0255] In this example, methods are described for predicting fetal fraction in a sample based on fragmentomic parameters (ie, terminal motif frequencies and fragment length profiles).

[0256] Figure 1 An exemplary workflow is provided in . An internal dataset of cfDNA sequencing data from approximately 1500 pregnant women was used for the training and test sets described in this embodiment. Sequencing data was processed using Illumina's DRAGN 4.0 platform to obtain fragment spectra (# short fragments (80bp-150bp) / # long fragments (151bp-300bp)) and chromosome-level motif frequencies (n=256×24) for 5Mb genomic bins (n=503). 256 is the number of possible 4bp motifs, and 24 is 22 autosomes plus chromosome X and chromosome Y. The motif frequencies were normalized to the total frequency, and a normalized motif diversity score (MDS) was generated according to the following equation:

[0257]

[0258] After read alignment and feature extraction, the data was aggregated and summarized into a spreadsheet like the following:

[0259] (Fragment spectrum)

[0260] Sample ID Fetal fraction Fragment spectrum 1 Fragment spectrum 2 … Fragment spectrum 503 1 0.1 0.56 … … 2 0.14 … … … N 0.2 … …

[0261] (Motif frequency)

[0262] Sample ID Fetal fraction Motif 1 Motif 2 … Motif 256 1 0.1 0.35 … … 2 0.14 … … N 0.2 … …

[0263] Principal component analysis (PCA) was applied to the data to extract the most important principal components. Specifically, the features were analyzed with PCA to extract the most relevant variables for predicting fetal fraction. PCA on the fragment spectra and motif frequencies extracted the first principal component that explained at least 99% of the variance, which reduced the total variables to 32.

[0264] The data table is then split into training and test sets at a ratio of 80:20. The following R codes are used to execute three machine learning models (i.e., linear regression, elastic net, and XGBoost) on the training dataset respectively:

[0265]

[0266] The R function lm() was used to train a linear regression model with fetal fraction as the response variable and i) the fragment length spectrum per 5 Mb or ii) the frequency of 256 4-base pair sequence motifs as explanatory variables. As a result of the model, the regression coefficients and residuals for each genomic bin were obtained:

[0267] Fetal fraction in training sample = coefficient × fragment spectrum or motif frequency in training sample + residual

[0268] Elastic net models were trained using the R package "glmnet" with fetal fraction as the response variable and i) fragment length spectra of every 5 Mb or ii) 256 4-base pair sequence motif frequencies as explanatory variables. Five-fold cross validation was used for model training. The model combined L1 and L2 regularization to estimate regression coefficients.

[0269] The XGBoost model was trained using the R package "xgboost" with parameters "nrounds = 100, max_depth = 6, eta = 0.3, gamma = 0, colsample_bytree = 1, min_child_weight = 1, subsample = 1". Fetal fraction was set as the response variable, and i) the fragment length spectrum per 5 Mb or ii) the 256 4-base pair sequence motif frequencies were set as explanatory variables. XGBoost is a decision tree-based method. The training process determines a set of hyperparameters in the tree that best describes the data in the training dataset.

[0270] Techniques such as random parameter search and 5-fold cross validation (CV) were used to prevent the model from overfitting.

[0271] The model trained as described above was then applied to the test dataset to predict the fetal fraction of the test samples using the following R code:

[0272]

[0273] Apply the trained linear regression model to the test sample to estimate the fetal fraction in the test sample using the R predict() function:

[0274] Predicted fetal fraction

[0275] = coefficient from the training model × fragment spectrum or motif frequency in the test sample

[0276] + Residuals from the trained model

[0277] The trained elastic net model is applied to the test sample to estimate the fetal fraction in the test sample using the coefficients and residuals estimated from the training sample using the R predict() function.

[0278] The trained XGBoost model was applied to the test sample to estimate the fetal fraction in the test sample using the hyperparameters estimated from the training sample using the R predict() function.

[0279] The predicted fetal fraction is then compared to the true fetal fraction to estimate model accuracy. Specifically, each model is evaluated on the test dataset using two metrics: root mean square error (RMSE) and correlation with the true fetal fraction. The metric RMSE is used to measure model accuracy according to the following equation:

[0280]

[0281] The method described in this example for estimating fetal fraction in a test sample (DRAGEN) is at least two orders of magnitude faster than existing tools (see Figure 2 ). Figure 3 Shown are fragment profiles (upper row) and sequence motif frequencies (lower row) from DRAGEN (x-axis) that are highly consistent with existing tools (y-axis) in TSO500 and whole exome sequencing (WES) cfDNA datasets.

[0282] Figure 4 Models trained with fragment size only and elastic net or XGBoost models are shown in . The fetal fraction predicted by the model (y-axis) shows a high consistency with the real data (x-axis). Figure 5 Models trained with only 5' end sequence motif frequencies and elastic net or XGBoost models are shown in. The fetal fractions predicted by the models (y-axis) show a high degree of consistency with the real data (x-axis). Figure 6 A table is provided showing that the sequence motif based method exhibited a similar error rate of 2.5% when compared to the traditional fragment size based method.

[0283] Figure 7In shows the performance overview of multiple models, indicating that combining sequence motifs with fragment size or coverage features improves prediction accuracy. Coverage features generally refer to the features of read coverage (i.e., the amount of mapped reads for a given locus or genome portion) derived from the entire genome. RMSE (left) and correlation (right) of different models are provided. The Y axis is from top to bottom the predictions from: 1) Distributed Random Forest (DRF) using only sequence motifs; 2) Gradient Boosting (GBM) using only sequence motifs; 3) XGBoost using only sequence motifs; 4) FF_Size model from NIPT team; 5) Elastic Network / Generalized Linear Model (GLM) using only sequence motifs; 6) FF_Coverage model, which uses whole genome read coverage to predict fetal fraction (e.g., as described in U.S. Patent No. 10,622,094 and Kim et al., 2015, Prenatal Diagnosis, Vol. 35: pp. 1-6, each of which is incorporated by reference in its entirety). 7) GLM model using both fragment size and sequence motifs; 8) FF_Cov_Size model from NIPT team; 9) GLM model using fragment coverage and sequence motifs; 10) FF4 model from NIPT team; 11) XGBoost model using fragment size, coverage and sequence motifs; 12) FF_Cov_Size_recompute model from NIPT team; 13) GLM model using fragment size, coverage and sequence motifs.

[0284] ***

[0285] The entire contents of each patent, patent application, publication and document cited herein are incorporated by reference. The reference to patents, patent applications, publications and documents does not constitute an admission that any of the above is relevant prior art, nor does it constitute an admission of the contents or dates of these publications or documents. Their reference does not represent a search for relevant public content. All statements about the date or content of the document are based on existing information and do not constitute an admission of its accuracy or correctness.

[0286] The present technology has been described with reference to specific embodiments. The terms and expressions used herein to describe the present technology are descriptive and not necessarily restrictive. Certain modifications made to the disclosed embodiments may be considered to be within the scope of the present technology. Certain aspects of the disclosed embodiments may be appropriately practiced with or without certain elements not specifically disclosed herein.

[0287] Each of the terms "comprising", "consisting essentially of", and "consisting of" may be replaced with any of the other two terms. The term "a" or "an" may refer to one or more elements that it modifies (e.g., "agent" may refer to one or more agents), unless the context clearly describes any one of the elements or more than one of the elements. As used herein, the term "about" refers to a value within 10% of the underlying parameter (i.e., plus or minus 10%; for example, a weight of "about 100 grams" may include a weight of 90 grams to 110 grams). The term "about" is used at the beginning of a list of values ​​to modify each value (e.g., "about 1, 2, and 3" refers to "about 1, about 2, and about 3"). When describing a list of values, the list includes all intermediate values ​​and all fractional values ​​thereof (e.g., a list of values ​​"80%, 85%, or 90%" includes an intermediate value of 86% and a fractional value of 86.4%). When a list of values ​​is followed by the term "or more," the term "or more" applies to each value listed (e.g., a list of "80%, 90%, 95%, or more" or "80%, 90%, 95% or more" or "80%, 90% or 95% or more" means "80% or more, 90% or more, or 95% or more"). When describing a list of values, the list includes all ranges between any two listed values ​​(e.g., a list of "80%, 90% or 95%" includes ranges of "80% to 90%," "80% to 95%," and "90% to 95%").

[0288] Certain embodiments of the technology are set forth in the following claims.

Claims

1. A method for estimating the fraction of fetal nucleic acid in a test sample from a pregnant subject, the method comprising: a) obtaining sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating cell-free (CCF) nucleic acid from a test sample from a pregnant subject; b) measuring the fragment lengths of a plurality of circulating free nucleic acid fragments; c) generating one or more fragment length profiles of the test sample; d) determining sequence motifs at the ends of a plurality of circulating free nucleic acid fragments; e) determining one or more sequence motif frequencies of the test sample; as well as f) estimating a fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

2. The method according to claim 1, wherein: i) the sequence reads are obtained by a paired-end sequencing process, and the sequence reads are paired-end sequence reads; ii) measuring the fragment lengths in (b) according to the mapping positions of the paired-end sequence reads; as well as iii) measuring said fragment lengths for multiple genomic intervals.

3. The method of claim 1 or 2, wherein the one or more fragment length profiles are generated in (c) based on a ratio of X to Y for a plurality of genomic intervals, wherein X is the number of CCF nucleic acid fragments having a length within a first selected fragment length range, and Y is the number of CCF nucleic acid fragments having a length within a second selected fragment length range.

4. The method of claim 3, wherein the first selected fragment length ranges from about 80 bases to about 150 bases, and the second selected fragment length ranges from about 151 bases to about 300 bases.

5. The method of any one of claims 1-4, wherein the one or more fragment length profiles are generated for one or more genomic segments in (c).

6. The method according to any one of claims 1 to 5, wherein: i) The sequence motif in (d) is a 5' sequence motif; ii) (d) wherein the sequence motif is a four base pair (bp) sequence motif; or iii) The sequence motif in (d) is a 5' sequence motif and a four base pair (bp) sequence motif.

7. The method of any one of claims 1-6, wherein (e) comprises determining one or more sequence motif frequencies for one or more chromosomes.

8. The method of any one of claims 1-7, wherein (e) comprises determining one or more frequencies of one or more sequence motifs selected from the group consisting of GGAA, AGAA, GTTT, GAAT, and GGTT.

9. The method according to any one of claims 1-8, wherein the one or more sequence motif frequencies are determined according to the frequencies of the one or more sequence motifs in the mapped sequence reads of one or more chromosomes.

10. The method of any one of claims 1-9, wherein estimating the fetal nucleic acid fraction of the test sample in (f) comprises applying one or more model parameters from one or more models to i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

11. The method of claim 10, wherein the model parameters are obtained from a sample training set, and wherein the fetal nucleic acid fraction of each sample in the sample training set is known.

12. The method according to claim 11, wherein: i) obtaining the one or more model parameters from the training set according to a fitted relationship between 1) the fetal nucleic acid fraction of each sample in the sample training set and 2) one or more fragment length spectra of each sample in the sample training set; ii) obtaining the one or more model parameters from the training set according to a fitted relationship between 1) the fetal nucleic acid fraction of each sample in the sample training set and 2) one or more sequence motif frequencies of each sample in the sample training set; or iii) obtaining the one or more model parameters from the training set based on 1) the fetal nucleic acid fraction of each sample in the sample training set and 2) the fitting relationship between one or more fragment length spectra and one or more sequence motif frequencies of each sample in the sample training set.

13. The method of any one of claims 10-12, wherein the one or more model parameters comprise coefficients derived from the one or more models.

14. The method of any one of claims 10-13, wherein the one or more models comprise linear regression, and wherein estimating the fetal nucleic acid fraction for the test sample in (f) comprises applying regression coefficients from the linear regression to i) the one or more fragment length profiles for the test sample and ii) the one or more sequence motif frequencies for the test sample.

15. The method of any one of claims 10-14, wherein the one or more models comprise an elastic net, and wherein estimating the fetal nucleic acid fraction of the test sample in (f) comprises applying coefficients from the elastic net model to i) the one or more fragment length spectra of the test sample and ii) the one or more sequence motif frequencies of the test sample.

16. The method of any one of claims 10-15, wherein the one or more models include XGBoost, and wherein estimating the fetal nucleic acid fraction of the test sample in (f) includes applying coefficients from the XGBoost model to i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

17. The method according to any one of claims 1-16, further comprising, before (a), sequencing the circulating free (CCF) nucleic acid from the test sample by a sequencing process, wherein: i) the sequencing process is a non-targeted sequencing process; ii) the sequencing process is a massively parallel sequencing process; or iii) The sequencing process is a non-targeted massively parallel sequencing process.

18. A system comprising one or more microprocessors and a memory, the memory comprising instructions executable by the one or more microprocessors, and the memory comprising sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating cell-free (CCF) nucleic acid from a test sample from a pregnant subject, and wherein the instructions executable by the one or more microprocessors are configured to: a) measuring the fragment lengths of a plurality of circulating free nucleic acid fragments; b) generating one or more fragment length profiles of the test sample; c) determining sequence motifs at the ends of a plurality of circulating free nucleic acid fragments; d) determining one or more sequence motif frequencies of the test sample; as well as e) estimating a fetal nucleic acid fraction for the test sample based on i) the one or more fragment length profiles for the test sample and ii) the one or more sequence motif frequencies for the test sample.

19. A machine comprising one or more microprocessors and a memory, the memory comprising instructions executable by the one or more microprocessors, and the memory comprising sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating free nucleic acid from a test sample from a pregnant subject, and wherein the instructions executable by the one or more microprocessors are configured to: a) measuring the fragment lengths of a plurality of circulating free nucleic acid fragments; b) generating one or more fragment length profiles of the test sample; c) determining sequence motifs at the ends of a plurality of circulating free nucleic acid fragments; d) determining one or more sequence motif frequencies of the test sample; as well as e) estimating a fetal nucleic acid fraction for the test sample based on i) the one or more fragment length profiles for the test sample and ii) the one or more sequence motif frequencies for the test sample.

20. A non-transitory computer-readable storage medium having an executable program stored thereon, wherein the program instructs a microprocessor to: a) accessing sequence reads mapped to a reference genome, wherein the sequence reads are reads of circulating free nucleic acid from a test sample from a pregnant subject; b) measuring the fragment lengths of a plurality of circulating free nucleic acid fragments; c) generating one or more fragment length profiles of the test sample; d) determining sequence motifs at the ends of a plurality of circulating free nucleic acid fragments; e) determining one or more sequence motif frequencies of the test sample; as well as f) estimating a fetal nucleic acid fraction of the test sample based on i) the one or more fragment length profiles of the test sample and ii) the one or more sequence motif frequencies of the test sample.

Images