Correspondence Between Genomic- and Genealogical/Coalescent-Based Inference of Homozygosity by Descent in Large French-Canadian Genealogies

Research on the genetics of complex traits overwhelmingly focuses on the additive effects of genes. Yet, animal studies have shown that non-additive effects, in particular homozygosity effects, can shape complex traits. Recent investigations in human studies found some significant homozygosity effects. However, most human populations display restricted ranges of homozygosity by descent (HBD), making the identification of homozygosity effects challenging. Founder populations give rise to higher HBD levels. When deep genealogical data are available in a founder population, it is possible to gain information on the time to the most recent common ancestor (MRCA) from whom a chromosomal segment has been transmitted to both parents of an individual and in turn to that individual. This information on the time to MRCA can be combined with the time to MRCA inferred from coalescent models of gene genealogies. HBD can also be estimated from genomic data. The extent to which the genomic HBD measures correspond to the genealogical/coalescent measures has not been documented in founder populations with extensive genealogical data. In this study, we used simulations to relate genomic and genealogical/coalescent HBD measures. We based our simulations on genealogical data from two ongoing studies from the French-Canadian founder population displaying different levels of inbreeding. We simulated single-nucleotide polymorphisms (SNPs) in a 1-Mb genomic segment from a coalescent model in conjunction with the observed genealogical data. We compared genealogical/coalescent HBD to two genomic methods of HBD estimation based on hidden Markov models (HMMs). We found that genomic estimates of HBD correlated well with genealogical/coalescent HBD measures in both study genealogies. We described generation time to coalescence in terms of genomic HBD estimates and found a large variability in generation time captured by genomic HBD when considering each SNP. However, SNPs in longer segments were more likely to capture recent time to coalescence, as expected. Our study suggests that estimating the coalescent gene genealogy from the genomic data to use in conjunction with observed genealogical data could provide valuable information on HBD.