Evaluating Natural Selection In Schizophrenia Using Data From Genome-Wide Association Studies

Abstract

Schizophrenia is a debilitating psychiatric condition that manifests itself in the early adult life and is often associated with poor quality of life and decreased life expectancy. It is both highly heritable and polygenic, with risk variants spanning the full spectrum of population frequencies. There has been a long-standing debate, from an evolutionary perspective, as to how risk alleles persist in the human gene pool, particularly given the decreased fecundity associated with schizophrenia. It has been repeatedly show that rare risk alleles of strong effect persist due to mutation-selection balance, a mechanism which implies that they are removed from the population by natural selection at the same rate that they appear by mutation. However, an explanation for the persistence of common risk alleles of small effect has not been explicitly given, though this is often attributed to various evolutionary scenarios. One of the most popular hypotheses links this persistence to positive selection, arguing that current common risk alleles for schizophrenia might have been beneficial at some point during human evolution, and these benefits resulted in the allele frequencies we observe today.While detecting selection signals in genomic data has been traditionally challenging, recent advances in the statistical and population genetics communities have made possible to systematically test for the presence of selected alleles in GWAS summary statistics. To demonstrate the possibilities of such a procedure, this work uses a large-scale meta-analysis of schizophrenia GWAS data (40,675 cases and 64,643 controls) to test four different metrics of positive selection, each tailored to a particular timeframe of human evolution. The LD-Score framework shows that none of these metrics are significantly associated with common genetic variation, raising doubts into the validity of positive selection as a driver for the persistence of schizophrenia risk alleles. Furthermore, strong association is shown for a metric of negative selection, background selection (h2 enrichment = 1.801; p=0.001), which recurrently affects regions of low recombination and reduces genetic diversity. An examination of the mechanism of action of background selection, coupled with a simulation analysis, allows us to establish that regions with reduced genetic diversity are more likely to harbour common risk alleles in schizophrenia, and that this is compatible with our current knowledge about the epidemiology of the disorder. This also suggests a genetic model for the action of natural selection in schizophrenia, which relates our findings to the nearly-neutral theory of molecular evolution (Ohta 1973) and could be extended to other psychiatric complex traits.