Widespread sequencing efforts are revealing unprecedented amount of genomic variation in populations. Such information is routinely used to derive consensus reference sequences and to infer positions subject to natural selection. Here, we present a new molecular evolutionary method for estimating neutral evolutionary probabilities (EPs) of each amino acid, or nucleotide state at a genomic position without using intraspecific polymorphism data. Because EPs are derived independently of population-level information, they serve as null expectations that can be used to evaluate selective forces on alleles at both polymorphic and monomorphic positions in populations. We applied this method to coding sequences in the human genome and produced a comprehensive evolutionary variome reference for all human proteins. We found that EPs accurately predict neutral and disease- Associated alleles. Through an analysis of discordance between allelic EPs and their observed population frequencies, we discovered thousands of novel candidate sites for nonneutral evolution in human proteins. Many of these were validated in a joint analysis of disease- Associated variants and population data. The EP method is also directly applicable to the analysis of noncoding sequences and genomic analyses of nonmodel species.
ASJC Scopus subject areas
- Ecology, Evolution, Behavior and Systematics
- Molecular Biology