The functional effects and fitness consequences of GWAS loci

Abstract

In the past 20 years, genome-wide association studies have identified hundreds of thousands of genetic loci linked to human traits. Examining these genetic associations can reveal both the molecular mechanisms contributing to traits and the evolutionary pressures constraining them. However, efforts to study the mechanisms of genetic loci or their evolution are hindered by technical challenges and conceptual gaps. Our analyses used targets designed to reduce the technical challenges in order to answer the conceptual questions---including genes selected for their known functions, variants grouped by their effects, or species chosen because of their unusual evolutionary histories.

First, we performed colocalization analysis---a strategy for connecting genetic loci, molecular mechanisms, and traits---to a set of genes selected for their known roles in relevant Mendelian traits. Colocalization has a low rate of success; the reasons for this are unclear. By selecting genes and loci where our \textit{a priori} knowledge strongly suggests a connection, we showed that some challenges have higher potential to explain the limitation of colocalizations while others are not fully consistent with it.

Next, we evaluated the consistency of genome-wide association study results with various forms of natural selection. Though the existence of selection has been widely reported, by designing our tests around trait-increasing alleles, we were able to show that the form of selection was stabilizing. This was true not just for quantitative traits---where stabilizing selection has been expected and observed---but also disease risk, which is more often assumed to be under directional selection, not stabilizing.

Finally, we evaluated whether the selection on traits may explain our limited ability to identify biological mechanisms. We examined the high success of colocalization analysis in cattle---a species with similar biology to humans, but a very different set of evolutionary pressures. The observations in cattle are consistent with an understanding in which, though the colocalization model correctly describes human genetics, selection on traits limits our statistical power to identify it.

This work demonstrates that the challenges of statistically identifying the functions of genetic loci, though intractable, are not insurmountable. It also expands our understanding of evolution's effects on human traits, and demonstrates that an evolutionary understanding may be crucial for addressing the problems of functional analysis.