Profiling and leveraging relatedness in a precision medicine cohort of 92,455 exomes

Abstract

16Large-scale human genetics studies are ascertaining increasing proportions of 17 populations as they continue growing in both number and scale. As a result, the amount 18 of cryptic relatedness within these study cohorts is growing rapidly and has significant 19 implications on downstream analyses. We demonstrate this growth empirically among 20 the first 92,455 exomes from the DiscovEHR cohort and, via a custom simulation 21 framework we developed called SimProgeny, show that these measures are in-line with 22 expectations given the underlying population and ascertainment approach. For example, 23 we identified ~66,000 close (first-and second-degree) relationships within DiscovEHR 1 involving 55.6% of study participants. Our simulation results project that >70% of the 2 cohort will be involved in these close relationships as DiscovEHR scales to 250,000 3 recruited individuals. We reconstructed 12,574 pedigrees using these relationships 4 (including 2,192 nuclear families) and leveraged them for multiple applications. The 5 pedigrees substantially improved the phasing accuracy of 20,947 rare, deleterious 6 compound heterozygous mutations. Reconstructed nuclear families were critical for 7 identifying 3,415 de novo mutations in ~1,783 genes. Finally, we demonstrate the 8 segregation of known and suspected disease-causing mutations through reconstructed 9 pedigrees, including a tandem duplication in LDLR causing familial hypercholesterolemia. 10In summary, this work highlights the prevalence of cryptic relatedness expected among 11 large healthcare population genomic studies and demonstrates several analyses that are 12 uniquely enabled by large amounts of cryptic relatedness. 13 14 Key words: cryptic relatedness; pedigree reconstruction; relationship inference; identity 15 by decent; compound heterozygous mutation phasing; de novo mutations; precision 16 medicine; healthcare population-based genetic study; exome sequencing; family structure; 17 familial-hypercholesterolemia 18 19 20 21 22

Crossover interference and sex-specific genetic maps shape identical by descent sharing in close relatives

“…Inferring identical by descent (IBD) segments and estimating relatedness are classical problems in human genetics [1], with recent work motivated by the abundance of close relatives in large samples [2][3][4][5][6]. In order to study individuals with a known relationship, many investigators have performed simulations, both to evaluate novel methods [4][5][6][7][8], and to characterize the properties of IBD sharing rates among relatives [9,10].…”

Cox regression increases power to detect genotype-phenotype associations in genomic studies using the electronic health record

“…In the future, more sophisticated models may be able to account for subsequent response to treatment or semi-continuous traits such as lab values. We are especially interested in the potential of Cox mixed models, which, like linear mixed models [14] , use random effects to account for genetic relatedness, an increasingly important factor in EHR-linked samples [15] . Such an approach applied to large-scale datasets such as from the Million Veterans Program or the All of Us Research Program [16,17] , if appropriately adjusted for environmental and societal factors, may enable the creation of clinically useful polygenic hazard scores.…”

Distinguishing pedigree relationships using multi-way identical by descent sharing and sex-specific genetic maps

“…Modern scale genetic datasets contain tens to hundreds of thousands of samples, sample sizes within which numerous close relatives exist 1,2 . Characterizing relatives within these samples is essential to avoid spurious signals and to improve power in genetic association studies [3][4][5] , but standard models consider only kinship estimates while ignoring the potential for different relationship types to vary in their phenotypic correlation 6,7 .…”

“…Characterizing relatives within these samples is essential to avoid spurious signals and to improve power in genetic association studies [3][4][5] , but standard models consider only kinship estimates while ignoring the potential for different relationship types to vary in their phenotypic correlation 6,7 . Moreover, while population genetic studies typically filter close relatives to avoid modeling violations 8 , such an approach will dramatically reduce sample sizes in large datasets 1,2 . One way to enable analyses of full study samples is to directly model the transmission of shared haplotypes-i.e., identical by descent (IBD) segments 9 -using the pedigree structure of each set of relatives, but this requires accurate determination of those pedigrees.…”