Identification of evolutionary hotspots in the rodent genomes

We describe a whole-genome comparative analysis of the human, mouse, and rat genomes to describe the average substitution patterns of four genomic regions: ancient repeats, rodent-specific DNA, exons, and conserved (coding and noncoding) regions, and to identify rodent evolutionary hotspots. In all types of regions, except the rodent-specific DNA, the rat branch is slightly longer than the mouse branch. Moreover, the mouse-rat distance is longer in the rodent-specific DNA than in the ancient repeats. Analysis of individual conserved regions with different substitution models yielded the conclusion that the Jukes-Cantor model is inadequate, and the Hasegawa-Kishino-Yano model is almost as good as the REV model. Using human as an outgroup, we identified 5055 evolutionary hotspots, which are highly conserved subalignment blocks (each consisting of at least 100 aligned sites and a small fraction of gaps) with a large and statistically significant difference in the branch lengths of the rodent species. The cutoffs used to identify the hotspots are partially based on estimates of the average rates of substitution. The fractions of hotspots overlapping with the rodent RefSeq genes, RefSeq exons, and ESTs are all higher than expected. Still, more than half of the hotspots lie in noncoding regions of the mouse genome. We believe that the hotspots represent biologically interesting regions in the rodent genomes.