More info about heritability and genetic correlation

Geneticists seek to understand the role that genetic variation plays in the expression of inherited traits and, of interest for improving health care, have explored the genetic architecture of complex human diseases and traits. There is evidence that suggests some common diseases are the result of variation in many common alleles acting in concert, which each contributes relatively small individual effects. In order to detangle the myriad ways that different genetic variations may interact to produce observed phenotypes, it is useful to also understand how multiple different traits are affected by each variant, i.e. pleiotropy.

Genetic correlation is a result of pleiotropy and is an estimate of the additive genetic effect that is common to pairs of traits. Recent studies have demonstrated that many complex traits are genetically correlated and unearthing these relationships may aid in understanding the etiology of diseases via shared pathways. For example, recognition of the overlap between loci that contribute to genetic susceptibility for the development of multiple tumor types can help identify novel and important mechanisms of carcinogenesis. In addition to an improved understanding of the genetic and molecular basis of complex traits, studying genetic correlation can boost statistical power in association studies by effectively increasing the sample size. Observed pleiotropy is also not necessarily indicative of genetic correlation, and studying the components of this correlation can determine if genetics or other factors - e.g. shared environment, assortative mating - are more explanatory for phenotypic correlation.

Estimates of genetic correlation can be derived from studies in structured populations. They require much information on the investigators’ part, including genotype data, phenotype data for multiple traits of interest, and often relatedness data for the subjects in the cohort. Issues that traditionally plague heritability estimates - e.g. assumptions about interaction effects, and the difficulty of estimating parameters based on data with potentially large sampling errors - apply to calculations of genetic correlation as well. Confounding variables including, but not limited to, environmental factors can also impact heritability, and thus, genetic correlation estimates. One strategy for controlling aspects of experimental design so that fewer biases seep into the ultimate analysis is to use animal models instead of data from humans.

Diversity Outbred (DO) mice were developed as a large, heterogeneous population derived from eight inbred founder strains bred randomly over many generations. The resulting mice contain extensive genetic diversity from random matings that ensure more allelic recombination across the genome, breaking up long spans of linkage disequilibrium that have previously hampered the use of inbred strains for mapping genetic loci associated with traits of interest. This finer mapping resolution and diversity make the DO mice more analogous to a human model compared to other resources. Due to their recombination density, DO mice need to be genotyped on high-density arrays, but the generated data is expected to provide a more precise picture of genetic variation and a greater range of phenotypic measurements. As a result, the use of more accurate kinship matrices minimizes biases due to mis-estimation of relatedness and the increased phenotypic variation facilitates exploration of correlations between different traits.