Non-Random Patterns of Genome-Wide Sequence Polymorphisms

Base composition of DNA molecules is a fundamental genome feature. Here, we report two common patterns of base composition, individual-strand base equality (i.e., [A] ≈ [T] and [G] ≈ [C] on the same strand) and AT-increase. While further validating these two patterns across 2,210 species at the genome level, we revealed the validity of these patterns at the broad population level by examining genome-wide nucleotide polymorphisms across a diverse set of species (human, dog, mouse, silkworm, rice, maize, soybean, and grape). Our analysis discovered a consistent AT-increase divergence across polymorphic sites between samples from basal ancestral groups and samples from derived groups. Further analysis of a series of mutation tracking experiments showed that these two base composition patterns can be directly generated by spontaneous mutations or induced mutations. With the human 1000 Genomes data, genome-wide scans by treating base composition across polymorphic sites as a quantitative trait identified strong signals underlying the observed base composition variation. A series genes involved in different biological processes are significantly enriched with the identified genome regions, including many DNA repair genes.