Concordance between male- and female-specific GWAS results helps define underlying genetic architecture of complex traits

Abstract

A better understanding of genetic architecture will help translate genetic data into improved precision-based medicine and clinical care. Towards this end, we explored the use of sex-stratified analyses for several traits in the Hybrid Mouse Diversity Panel (HMDP) and UK Biobank to better determine trait polygenicity and identify contributing loci. This was accomplished by comparing the direction of allelic effects in males and females in sex-stratified analyses under the hypothesis that loci that are not associated with a trait should have equal chances of trending in the same direction of effect. Instead, we found that even for most alleles that do not meet nominal levels of statistical significance, the direction of effect in the two sexes was highly concordant. Results were consistent with hundreds of loci regulating each mouse trait and thousands of loci regulating each human trait, including traits for which no statistically significant loci were identified using conventional approaches. We also found evidence of likely spurious widespread epistasis. Collectively, these findings demonstrate the importance of stratifying by sex to discover novel associating loci, suggest a new method for identifying biologically rather than statistically significant associations, and caution that widespread marginal effects can lead to phantom epistasis.