IMPROVED TESTS FOR HETEROGENEITY ACROSS A REGION OF DNA-SEQUENCE IN THE RATIO OF POLYMORPHISM TO DIVERGENCE

The neutral theory of molecular evolution predicts that the ratio of p olymorphisms to fixed differences should be fairly uniform across a re gion of DNA sequence. Significant heterogeneity in this ratio can indi cate the effects of balancing selection, selective sweeps, mildly dele terious mutations, or background selection. Comparing an observed hete rogeneity statistic with simulations of the heterogeneity resulting fr om random phylogenetic and sampling variation provides a test of the s tatistical significance of the observed pattern. When simulated data s ets containing heterogeneity in the polymorphism-to-divergence ratio a re examined, different statistics are most powerful for detecting diff erent patterns of heterogeneity. The number of runs is most powerful f or detecting patterns containing several peaks of polymorphism; the Ko lmogorov-Smimov statistic is most powerful for detecting patterns in w hich one end of the gene has high polymorphism and the other end has l ow polymorphism; and a newly developed statistic, the mean sliding G s tatistic, is most powerful for detecting patterns containing one or tw o peaks of polymorphism with reduced polymorphism on either side. Nine out of 27 genes from the Drosophila melanogaster subgroup exhibit het erogeneity that is significant under at least one of these three tests , with five of the nine remaining significant after a correction for m ultiple comparisons, suggesting that detectable evidence for the effec ts of some kind of selection is fairly common.