Estimation of kinship coefficient in structured and admixed populations using sparse sequencing data

Abstract

Knowledge of biological relatedness between samples is important for many genetic studies. In large-scale human genetic association studies, the estimated kinship is used to remove cryptic relatedness, control for family structure, and estimate trait heritability. However, estimation of kinship is challenging for sparse sequencing data, such as those from off-target regions in target sequencing studies, where genotypes are largely uncertain or missing. Existing methods often assume accurate genotypes at a large number of markers across the genome. We show that these methods, without accounting for the genotype uncertainty in sparse sequencing data, can yield a strong downward bias in kinship estimation. We develop a computationally efficient method called SEEKIN to estimate kinship for both homogeneous samples and heterogeneous samples with population structure and admixture. Our method models genotype uncertainty and leverages linkage disequilibrium through imputation. We test SEEKIN on a whole exome sequencing dataset (WES) of Singapore Chinese and Malays, which involves substantial population structure and admixture. We show that SEEKIN can accurately estimate kinship coefficient and classify genetic relatedness using off-target sequencing data down sampled to ~0.15X depth. In application to the full WES dataset without down sampling, SEEKIN also outperforms existing methods by properly analyzing shallow off-target data (~0.75X). Using both simulated and real phenotypes, we further illustrate how our method improves estimation of trait heritability for WES studies.

Large-scale whole-genome sequencing of three diverse Asian populations in Singapore

“…We also applied the SEEKIN software (using GT mode and SGVP as reference) to estimate kinship coefficients without pruning SNPs, and obtained similar results. 60 We classified as k-degree related pairs if 2 -k-1.5 < φ< 2 -k-0.5 . 61 Zero degree means monozygotic twins (MZ) or duplicates.…”

“…The inbreeding coefficient for each sample was estimated as (2φ ii -1), where φ ii is the self-kinship coefficient for sample i output by PC-Relate. 31,60 Evaluation of genotyping accuracy and sensitivity. We have 1,263 samples from the SERI cohort that were previously genotyped by Illumina Quad610 arrays.…”

Linkage disequilibrium connects genetic records of relatives typed with disjoint genomic marker sets

“…This study contributes to a growing body of work on inference of genetic relationships in scenarios more challenging than when relatives are typed for the same markers [22][23][24][25][26]. In the setting of ancient DNA, Vohr et al [22] focused on the scenario in which DNA sequence is generated for different DNA samples possibly representing the same or related individuals-from a burial site, for example-but sequence is sufficiently sparse that reads do not necessarily overlap between samples.…”

Allele frequency‐free inference of close familial relationships from genotypes or low‐depth sequencing data

“…Numerous pairwise relatedness inference methods have been developed, for example, Thompson (), Lee (), Purcell et al (), Albrechtsen et al (), Manichaikul et al (), Stevens et al (), Korneliussen and Moltke (), Conomos, Reiner, Weir, and Thornton (), Dou et al (), and many are available in popular software packages, like PLINK (Purcell et al, ), and KING (Manichaikul et al, ). Most of these methods estimate either the three relatedness coefficients k 0 , k 1 and k 2 , or the kinship coefficient for each pair of diploid individuals, where k 0 , k 1 and k 2 are the proportions of the genome where a pair of individuals share 0, 1 or 2 alleles identical by descent (IBD) (Thompson, ).…”

“…This makes it infeasible to call genotypes accurately (Nielsen, Korneliussen, Albrechtsen, Li, & Wang, ), precluding the use of these methods. There are a few methods that estimate relatedness from low‐depth sequencing data by utilizing genotype likelihoods (e.g., Korneliussen & Moltke, ), or by using imputed genotype dosages (Dou et al, ). However, these methods function by leveraging access to many samples to estimate allele frequencies or perform accurate genotype imputation and are therefore not designed to apply to data sets with a low number of samples.…”