Small Insertions Are More Deleterious than Small Deletions in Human Genomes

Abstract

Although lines of evidence suggest that small insertions and deletions differ in their mechanisms of formation, there remains the debate on whether natural selection acts differently on the two indel types. Currently available personal genomes and the 1000 Genomes Project permit population level and genome scale comparison of the selection regimes on the two indel types. We first developed a statistical model to evaluate the indel frequency biases of the 1000 Genomes Project phase 1 data. We then identified four independent lines of evidence demonstrating that human small (1-4bp) insertions are on average more deleterious than deletions. This genome-wide selection pattern is not affected by methodology, demography, and regional differences including indel density, introns versus exons, repeats versus nonrepeats, recombination rates, and the timing of DNA replication. This selection pattern has a profound effect on indel frequency spectra, deletional bias, and local single-nucleotide mutation rates. Finally, we observed that small insertions appear to be more actively implicated in shaping fast-evolving genomic sequences (or nonconserved regions).