Statistical evidence for ancestral correlation patterns

Statistical correlations in DNA sequences are an important source of information for processes of genome evolution. As a special case of such correlations and building up on our previous work, here we study, how short-range correlations in Eukaryotic genomes change under elimination of various classes of repetitive DNA. Our main result is that a residual correlation pattern, common to most mammalian species, emerges under elimination of all repetitive DNA, suggesting features of an ancestral correlation signature. Furthermore, using this general framework, we find classes of repeats, which upon deletion move the correlation pattern towards this residual pattern (simple repeats and SINEs) or away from this residual pattern (LINEs). These findings suggest that the common correlation pattern visible in the mammalian species after repeat elimination can be associated with a common mammalian ancestor.