How common are intragene windows with K-A > K-S owing to purifying selection on synonymous mutations?

One method for diagnosing the mode of sequence evolution considers the ratio of nonsynonymous substitutions per nonsynonymous site (K-A) to the corresponding figure for synonymous substitutions (K-S). A ratio (K-A/K-S) greater than unity is taken as evidence for positive selection. This, however, need not necessarily be the case. Notably, there is one instance of a high intragenic K-A/K-S peak, revealed by sliding window analysis and observed in two pairwise comparisons, better accounted for by localised purifying selection on synonymous mutations that affect splicing. Is this example exceptional? To address this we isolate intragenic domains with K-A/K-S > 1 from more than 1000 long mouse-rat orthologues. Approximately one K-A/K-S > 1 peak is found per 12-15 kb of coding sequence. Surprisingly, low synonymous substitution rates underpin more incidences than do high nonsynonymous rates. Several reasons, however, prevent us from supposing that the low synonymous rates reflect purifying selection on synonymous mutations. First, for many peaks, the null that the peak is no higher than expected given the underlying rates of evolution, cannot be rejected. Second, of 18 statistically significant incidences with unusually low K-S values, only 3 are repeatable across independent comparisons. At least two of these are within alternatively spliced exons. We conclude that repeatable statistically significant intragenic domains of low intragenic K-S are rare. As so few K-A/K-S peaks reflect increased rates of protein evolution and so few hold statistical support, we additionally conclude that sliding window analysis to infer domains of positive selection is highly error-prone.