摘要: The McDonald–Kreitman (MK) test is a simple and widely used of selection in which the numbers nonsilent silent substitutions (Dn Ds) are compared with polymorphisms (Pn Ps). neutrality index (NI 5 DsPn/DnPs), odds ratio (OR) MK table, measures direction degree departure from neutral evolution. mean NI values across genes often taken to summarize patterns species. Here, we show that this leads statistical bias both simulated real data extent species, pattern adaptive evolution, can apparently be subject weak purifying vice versa. We removed by using variant Cochran—Mantel–Haenszel procedure for estimating weighted average OR. also several point estimators statistically biased even when cutoff employed. therefore suggest new statistic study sparse, selection: DoS Dn/(Dn þ � Pn/(Pn