Purifying Selection Influences the Comparison of Heterozygosities between Populations

Heterozygosity is a fundamental measure routinely used to compare between populations to infer the level of genetic variation and their relative effective population sizes. However, such comparison is highly influenced by the magnitude of selection pressure on the genomic regions used. Using over 2 million Single Nucleotide Variants (SNVs) from chimpanzee and mouse populations, this study shows that the heterozygosities estimated using neutrally evolving sites of large populations were two times higher than those of small populations. However, this difference was only ~1.6 times for the heterozygosities estimated using nonsynonymous sites. This suggests an excess in the nonsynonymous heterozygosities due to the segregation of deleterious variants in small populations. This excess in the nonsynonymous heterozygosities of the small populations was estimated to be 23-31%. Further analysis revealed that the magnitude of the excess is modulated by effective population size (N) and selection intensity (s). Using chimpanzee populations, this investigation found that the excess in nonsynonymous diversity in the small population was little (6%) when the difference between the N values of large and small populations was small (2.4 times). Conversely, this was high (23%) when the difference in N was large (5.9 times). Analysis using mouse populations showed that the excess in the nonsynonymous diversity of highly constrained genes of the small population was much higher (38%) than that observed for the genes under relaxed selective constraints (21%). Similar results were observed when the expression levels of genes were used as a proxy for selection intensity. These results emphasize the use of neutral regions, less constrained genes, or lowly expressed genes when comparing the heterozygosities between populations.