Extreme purifying selection against point mutations in the human genome

Abstract

Genome sequencing of tens of thousands of human individuals has recently enabled the measurement of large selective effects for mutations to protein-coding genes. Here we describe a new method, called ExtRaINSIGHT, for measuring similar selective effects at individual sites in noncoding as well as in coding regions of the human genome. ExtRaINSIGHT estimates the prevalance of strong purifying selection, or "ultraselection" (λs), as the fractional depletion of rare single-nucleotide variants (minor allele frequency <0.1%) in a target set of genomic sites relative to matched sites that are putatively neutrally evolving, in a manner that controls for local variation and neighbor-dependence in mutation rate. We show using simulations that, above an appropriate threshold, λs is closely related to the average site-specific selection coefficient against heterozygous point mutations, as predicted at mutation-selection balance. Applying ExtRaINSIGHT to 71,702 whole genome sequences from gnomAD v3, we find particularly strong evidence of ultraselection in evolutionarily ancient miRNAs and neuronal protein-coding genes, as well as at splice sites. Moreover, our estimated selection coefficient against heterozygous amino-acid replacements across the genome (at 1.4%) is substantially larger than previous estimates based on smaller sample sizes. By contrast, we find weak evidence of ultraselection in other noncoding RNAs and transcription factor binding sites, and only modest evidence in ultraconserved elements and human accelerated regions. We estimate that ~0.3-0.5% of the human genome is ultraselected, with one third to one half of ultraselected sites falling in coding regions. These estimates suggest ~0.3-0.4 lethal or nearly lethal de novo mutations per potential human zygote, together with ~2 de novo mutations that are more weakly deleterious. Overall, our study sheds new light on the genome-wide distribution of fitness effects for new point mutations by combining deep new sequencing data sets and classical theory from population genetics.

Mutation saturation for fitness effects at human CpG sites

“…An analogous strategy to overcome sample size limitations at other types of sites is to infer selection in bins of sites ( Dukler et al, 2021 ); however, if sites within a bin vary in their fitness effects, inferences based on these bins are not straight-forward. Indeed, the mutation frequency in a bin reflects the harmonic mean of hs across sites in the bin weighted by (unknown) mutation rates across sites (see Materials and methods).…”

The sequences of 150,119 genomes in the UK biobank

“…Regions under strong negative selection are expected to have a greater fraction of rare variants (FRV, defined here as variants carried by at most 4 WGS individuals) than the rest of the genome 36 . We observe a greater FRV in the most depleted regions (DR<5) than in the least depleted regions (DR>95) 74.8% vs 69.1% (Fig.…”

“…However, such statistics are only partially informative about sequence conservation specific to humans 34 . Sequence variation in humans 35, 36 can be used to characterize human specific conservation, but large sample sizes are required for accurate inference, as much fewer mutations separate pairs of humans than different species.…”

“…Inter species conservation 33 have been used to characterize selection beyond the exome, using the extensive accumulation of mutations over millions of years, but these methods fail to identify sequence conservation specific to humans. Sequence variation within human populations 34,35 has been used to characterize human specific conservation, however, this requires many human genomes to make accurate inference as a much smaller number of mutations separate two humans than two species.…”

“…After removing coding exons, among the 1% of regions with lowest and highest DR score we see a 3.4 and 0.4-fold overrepresentation of GWAS variants, respectively (Table 2). Regions under strong negative selection are also expected to have a greater fraction of rare variants (FRV) than the rest of the genome 35 . As most variants are carried only by a few individuals, we define FRV as the fraction of variants carried by at most 4 WGS individuals and find FRV of 74.7% in windows with DR less than 5 compared to FRV of 69.6% in regions with DR above 95 (Fig.…”