Systems and methods for leveraging relatedness in genomic data analysis

Abstract

Methods, non-transitory computer-implemented methods and systems for identifying compound heterozygous mutations (CHMs) and de novo mutations (DNMs) in populations are provided. Also provided are methods for phasing genetic variants in a population by leveraging the population's relatedness. Further provided is a prediction model of relatedness in a human population.

What is claimed: 
     
         1 . A method for phasing heterozygous loss-of-function mutations (hLoF), comprising:
 establishing an ancestral superclass designation for each of one or more of the samples,   generating first identity-by-descent estimates of subjects within an ancestral superclass,   generating second identity-by-descent estimates of subjects independent from subjects' ancestral superclass;   clustering subjects into primary first-degree family networks based on one or more of the second identity-by-descent estimates;   generating third identity-by-descent estimates of subjects within a primary first-degree family network;   merging first and third identity-by-descent estimates to obtain merged identity-by-descent estimates;   constructing secondary first-degree family networks of samples based on merged identity-by-descent estimates,   phasing variants in accordance with merged identity-by-descent estimates and secondary first-degree family networks as either being or not being a compound heterozygous mutation (CHM);   phasing variants that are identified as CHMs as putative, missense, or non-putative; and   phasing variants identified as CHMs and as putative or missense as hLoFs.   
     
     
         2 . The method according to  claim 1 , wherein phasing putative variants as hLoFs further comprises:
 classifying variants identified as CHMs as hLoFs if said variants result in a frameshift, stop codon gain, stop codon loss, start codon gain, start codon loss, splicing acceptor altering, or splicing donor alter variant.   
     
     
         3 . The method according to  claim 1 , wherein phasing missense variants as hLoFs comprises:
 classifying CHMs that result in missense mutations as disruptive or non-disruptive, and   classifying CHMs that result in disruptive missense mutations as hLoFs.   
     
     
         4 . The method according to  claim 1 , wherein phasing variants to either be or not be CHMs comprises:
 phasing variants according to population allele frequencies;   removing variants outside of Hardy-Weinberg equilibrium (HWE) or within 10 base pairs of another variant in the same sample or both; and   removing SNPs with a quality by depth (QD) of about 2 or less, or a read depth (DP) of less than about 5, or an alternate allele balance (AB) of about 10% or less, or a combination thereof, and removing insertions or deletions (INDELS) with a QD of about 2 or less, or a DP of less than about 5, or an AB of about 10% or less, or a combination thereof, selecting remaining variants as potential compound heterozygous mutations (pCHMs) where there are one or more pairs of variants in the same sample and in the same gene, phasing pCHMs as either cis or trans pCHMs, and then classifying the pCHM phased as trans pCHM as CHMs.   
     
     
         5 . The method according to  claim 4 , wherein phasing variants to either be or not be CHMs comprises:
 removing variants outside of Hardy-Weinberg equilibrium (HWE) or within 10 base pairs of another variant in the same sample or both; and   removing SNPs with a quality by depth (QD) of about 3 or less, or a read depth (DP) of less than about 7, or an alternate allele balance (AB) of about 15% or less, or a combination thereof, and removing insertions or deletions (INDELS) with a QD of about 5 or less, or a DP of less than about 10, or an AB of about 20% or less, or a combination thereof.