Copy number allelic strand determination
AssigneeGuardant Health, Inc.
InventorsSean GORDON, Andrew M. GROSS, Geethika YALAMANCHILI
Life of this patent
Estimating the term…
Abstract
Methods, systems, and computer-readable media for determining allele-specific copy number (ASCN) and mutant allele dosage in tumor cells using Next-generation sequencing (NGS) genomic data, and broad methylation patterns (promoter regions and selected DMRs), and for integrating both modalities into a robust estimate of mutant allele dosage (the ratio of mutant allele copies to total tumor allele copies), with clinical reporting and quality control.
Claims17
A method for determining allele-specific copy number and mutant allele dosage in a biological sample, comprising:
- receiving a plurality of sequencing reads, each comprising at least one germline
- heterozygous single nucleotide polymorphism (SNP) and one or more somatic variant calls for a genomic locus;
- obtaining a tumor fraction estimate;
- generating a matrix comprising each candidate allele-specific configuration and each candidate purity and ploidy setting;
- computing an expected SNP minor allele fraction (MAF) and coverage likelihood;
- computing a combined likelihood for each of a candidate configuration;
- selecting the candidate allele-specific configuration based on maximizing the combined likelihood;
- determining at least one tumor specific total copy number and tumor specific minor allele copy number;
- and
- estimating mutant allele dosage based on one or more of:
- the selected allele specific configuration, the tumor fraction estimate, SNP MAF, and coverage likelihood.
A method for predicting allele-specific copy number and mutant allele dosage
- using methylation data, comprising:
- receiving tumor-adjusted methylation measurements across a plurality of regions comprising genomic promoters and differentially methylated regions;
- generating features normalized by tumor methylation fraction and/or control gene signals;
- applying a predictive model, optionally a trained predictive model, to the features to estimate allele-specific copy number, optionally a probability of allele-specific states;
- estimating mutant allele dosage by combining the estimate allele-specific copy number, and optionally observed somatic variant allele fraction;
- and
- generating outputting a dosage value.
A method, comprising:
- determining one or more of:
- genomic allelic strand copy number (ASCN), methylation predictions, somatic variant allele fraction (VAF), and tumor fraction for a plurality of sequence reads;
- computing mutant allele dosage based on the one or more determinations for the plurality of sequence reads;
- assigning one or more mutations in the plurality of sequence reads as amplified, non-amplified, or indeterminate allele scenario;
- and
- generating a dosage estimate.
The method of any preceding claim, comprising:
- computation of a genomic likelihood ASCN and/or methylation, optionally for to resolving ambiguity or to evaluating otherwise non-evaluable tumor (TF) cases.
The method of any preceding claim, comprising a methylation predictor to yield candidate ASCN and dosage, optionally used to inform genomic likelihood search.
The method of any preceding claim, comprising determination of a full Bayesian posterior estimate, optionally based on genomic, epigenomic methylation and/or priors by tumor type.
The method of any preceding claim, comprising):
- wherein the method interrogates a locus for one or more candidate genes of interest, optionally including one or more of:
- KRAS, EGFR, ERBB2, and PIK3CA.
The method of any preceding claim, comprising reporting a dosage matrix for one or more candidate genes of interest.
The method of any preceding claim, wherein the sequence reads are derived from tissue DNA and/or plasma cfDNA.
The method of any preceding claim, wherein the method comprises use of data trained on matched tissue/plasma pairs.
The method of any preceding claim, wherein the method comprises training sequence reads derived from tissue, and/or liquid biopsy.
The method of any preceding claim, wherein the sequence reads are generated from sequencing nucleic acid molecules derived from a sample obtained from a subject and generating testing sequence data.
The method of any preceding claim, wherein the nucleic acid molecules comprise cell-free DNA.
The method of any preceding claim, wherein the sample is at least one of blood, serum, plasma or tissue.
The method of any preceding claim, comprising determination of treatment for the subject.
A system configured to perform the method of any of the preceding claims.
A computer readable medium comprising the method of any of the preceding claims.
Dependencies are inferred from the claim text.