Understanding the "flattening" of gene contributions to human complex trait habitability

点击翻译按钮获取中文摘要

Genome-wide association studies (GWAS) indicate that heritable variation for most human complex traits is widely distributed across the genome. This is surprising since we expect each trait to have specific mechanisms and pathways underlying its biology and yet, for most traits, such mechanisms and pathways are not evident from GWAS results. Recently, it has been suggested that this phenomenon might be explained by the natural selection’s suppression of heritability. If variants at genes directly underlying trait biology have large effect sizes on the trait which cause increased selection, they will only appear at low frequencies and therefore only make a limited contribution to heritability. This intuition has recently been dubbed “flattening” because, by keeping large effect alleles at low frequencies, natural selection flattens the distribution of heritability. In this work, I will take the concept of “flattening” from basic intuition to fully fleshed out theory and from theory to the first genome- wide measure of flattening per gene per trait. First, I will build a series of models to describe the evolutionary and genomic factors that determine flattening, including pleiotropy and human demographic history. Then, I will validate my models by directly quantifying the effects of flattening on 38 blood and urine markers using UK Biobank data. These biomarkers provide a unique and exciting opportunity to test my predictions because we know the regulatory networks underlying the synthesis of these biomarkers. Finally, I will estimate the heritability deficit of each gene for each trait by comparing the gene’s contribution to heritability to a neutral expectation. The heritability deficit will be the first ever genome-wide measure of flattening per gene per trait and I will use it to characterize the extent of flattening on available UK Biobank traits.