Ultrarare variants drive substantial cis heritability of human gene expressio

These GEITP pages have often described genome-wide association studies (GWAS), because they provide information about what genes are correlated with any multifactorial trait (phenotype) — these can include complex diseases (e.g. obesity, type-2 diabetes, coronary arter disease), quantitative traits (e.g. height, serum cholesterol levels, blood pressure), and responses to drugs as well as environmental toxicants. And it is well known that hundreds or thousands of “small-effect” genes often contribute to these multifactorial traits. The topic today [see attached article] concerns single-nucleotide variants (SNVs) having minor allele frequencies (MAFs) of less than 1% (recall that each gene has two alleles, one on each of a chromosome pair, and one derived from each parent). “Common variants” in any population might exhibit a frequency of 10% or 70%, compared with “rare variants” with frequencies of less than 1% or “ultra-rare variants” (singletons) found in only one individual in a large population study. Whereas many rare variants and ultra-rare variants underlying Mendelian diseases (e.g. phenylketonuria, sickle cell disease, cystic fibrosis), have been found, their role in complex disease is unknown.

Recent methodological breakthroughs have enabled researchers to estimate the independent contributions of low- versus high-frequency alleles to multifactorial traits — often demonstrating a large-effect contribution by a rare variant, probably driven by

natural selection [process whereby any organism that adapts better to its environment — will improve its survival (find food, avoid predators, and produce more offspring)]. However, these studies excluded the rarest variants; this is a problematic limitation — given that some plausible evolutionary models predict that the largest contributions to phenotypic variance could come from the rarest variants. Directly querying the role of all variants with large-scale sequencing and sensitive statistical tests — has the potential to reveal important sources of missing heritability, inform strategies to increase the success rate of association studies, and clarify how natural selection has shaped human phenotypes.

Authors [see attached article] developed, validated, and applied an approach for inferring the relative phenotypic contributions of all variants — from the ultra-rare to the high-frequency variants. Authors focused on the narrow-sense heritability (h2) of gene expression (because a growing body of literature suggests that genetic variants primarily affect disease by modifying gene regulatory programs, and recent examinations have identified significant rare-variant effects on transcription). To characterize genetic architecture (i.e. underlying genetic basis of a phenotypic trait and its properties of variability) of gene expression, authors analyzed 360 unrelated individuals of European ancestry with paired whole-genome DNA and RNA-sequencing (RNA-seq) of immortal lymphoblastoid cell lines.

Authors conservatively estimate that singletons (i.e. rare variants often seen only once, within a large human population study) contribute ~25% of cis heritability across genes — which dwarfs the contributions of other allele frequencies). [Cis-regulatory elements (CREs) are regions of non-coding DNA that regulate transcription of nearby gene(s).] The majority (~76%) of singleton heritability is derived from ultra-rare variants — that are almost always absent from thousands of additional samples..!! Authors developed an inference procedure to demonstrate that their results are consistent with pervasive purifying selection which shapes the regulatory architecture of most human genes. These findings have enormous implications to gene-environment interactions in molecular and environmental genetic toxicology. 😊



· Nature Genetics Sept 2019; 51: 1349–1355

This entry was posted in Center for Environmental Genetics. Bookmark the permalink.