abstract
- A permutation method for detecting regional substitution rate heterogeneity in DNA sequences is described. An estimated phylogeny for the sequences is required. The method is likelihood based and searches through all regions of a DNA sequence to find any regions that have an unusually large (or small) number of substitutions in comparison to an optimal substitution rate for the entire sequence. Likelihood ratio tests do not achieve proper statistical significance levels due to the large number of tests performed. Empirical corrections based on permutations are suggested and shown to give correct statistical accuracy. The ability of the method to detect rate heterogeneity is good even when the region is small and with less than twice the rate of evolution in the remainder of the sequence.