Deciphering the relation between genetic relatedness and phenotypic covariance

University Of Chicago, Chicago IL

Investigators

Abstract

PROJECT SUMMARY We take as a starting point exciting new work from biobank-scale analyses of complex disease traits that reveal the degree of phenotypic covariance as a function of genetic relatedness (e.g. Kemper et al, 2021). Such new studies allow phenotypic covariance to finally be interrogated at more than just the two extremes of relatedness (the level of the family or among putative unrelateds), but across a continuum of relatedness. These new studies provoke new questions and offer new opportunities. In our first aim, we seek to understand how subtleties of computing the genetic relatedness matrix may bias heritability estimates to differing degrees at differing levels of genetic relatedness. First, we investigate how, when causal sites are on average at lower frequency than marker sites, the causal sites may be imperfectly captured by the SNP-based GRM due to the incomplete linkage disequilibrium (LD) of markers and rare causal variants. Second, even in the absence of allele frequency differences, biases may arise simply due to the scale of long-range haplotype sharing between individuals at different relatedness scales (“identity-by- descent tracts”). Finally, issues may arise when relatedness measures computed with alternate normalization strategies reflect alternative assumed relationships between SNP frequency and effect size. Work on these topics will help us understand if results such as those observed in Kemper et al might be produced as an intrinsic outcome of how genetic relatedness is measured. For our second aim, we note there are few quantitative genetic models that consider the correlation of environments beyond nuclear families into the domain of distant (2nd-4th) cousin / avuncular relationships that are increasingly relevant to biobank-scale studies of complex disease trait variation. We term such factors as multi-generation shared environments (MGSEs), and propose to study the potential impacts of these factors using combination of theory and simulation. For this work, we will leverage fast forward-in-time population genetic simulators that include pedigree tracking functionality and dimensions of geography that facilitate the simulation of MGSEs. As a second direction of work on this aim, we will develop a modified HE regression approach to infer the contribution of MGSEs to phenotypic variation. Much of the literature of shared environ- ments was established before the recently developed understanding of indirect genetic effects – in which the alleles in a parent or other relative impact the outcome of a phenotype in an offspring (a.k.a. “genetic nurture”). Our work here will focus on shared environments while grappling with the existence of indirect genetic effects. Finally, this project aims to address the recognized need for more work integrating the contribution of complex environmental factors with the study of the genetic architecture of complex traits.