Understanding genetic and environmental complexity in phenotypic resemblance across relatedness in population-scale data

Traits within a population exhibit varying degrees of phenotypic similarity that can be influenced by genetic relatedness. Traditionally, this relationship is believed to be a linear one, with the slope of phenotypic similarity directly proportional to heritability. This concept has been foundational in methods such as heritability estimation from sibling identity-by-descent regression, Haseman-Elston regression, and GREML.

However, empirical evidence suggests that the relationship is not always linear for many traits. With the advent of biobank-scale measurements, researchers are now able to observe phenotypic covariance more finely in relation to genetic relatedness, leading to a push for new theories to explain these deviations. This study delves into three potential sources of non-linearity in this relationship: imperfect tagging of causal variants, assortative mating, and multi-generational shared environments.

Through theoretical modeling and simulation, the study reveals that each of these mechanisms affects the relationship in distinct ways. Imperfect tagging of causal variants reduces the slope of phenotypic similarity among distant relatives. Shared environments, on the other hand, raise the intercept for closer relatives. Assortative mating, where individuals with similar traits mate with each other, influences both the slope and intercept of the relationship.

These findings underscore how complex the relationship between phenotypic similarity and genetic relatedness can be.

By exploring a range of scenarios within their models, the researchers demonstrate that the local slope of the relationship can, in certain cases, be nearly three times higher than the raw heritability value when comparing unrelated pairs versus siblings. This finding suggests that heritability estimates derived from sibling data could be biased, necessitating a more nuanced understanding of the factors at play.

The study's analysis across various traits, using data from the UK Biobank and a supplementary analysis in All of Us, shows that the shape of the relationship with genetic relatedness is not consistent across all traits. Instead, it varies widely, reflecting the influence of each of the theoretical mechanisms discussed. This variation suggests that understanding how phenotypic similarity increases with genetic relatedness in large cohorts requires the disentanglement of these multiple overlaid processes.