Abstract for presentation at 11th International Congress of Human Genetics

The biological problems to which comparative genomic sequence data can be applied are extremely broad. For a process to be amenable to interrogation by comparative approaches it must affect either the occurrence or fate of mutations. Any process affecting germline stability of DNA influences mutation occurrence while phenotypic effects expose a mutation to the scrutiny of natural selection, influencing its fate. We have sought to devise a software framework (COGENT) that considerably eases the effort required to define novel models of sequence evolution. A probabilistic approach has been employed as it enables the specification of competing hypotheses whose support can be contrasted using likelihood inference. Where possible, we have generalised the probability models such that completely novel models and combinations of models can be readily specified. To illustrate these capabilities, we describe research aimed at evaluating the competing evolutionary forces affecting the so-called 5th base, 5-methyl-cytosine (mC). mC plays both an important functional role in mammals as an epigenetic mark and also increases the spontaneous mutation rate of C. An examination of all possible dinucleotide context substitution rates indicated CpG, the primary context within which mammalian mC occurs, as the strongest individual sequence context. Our analysis further indicates that both transition and transversions are elevated within the CpG context. An elevated rate of substitution affecting CpG dinucleotides is also evident for protein coding sequences. Of the 32 genes sampled for our analysis, 21 exhibited enhanced suppression of nonsynonymous substitutions within CpGs, indicating mC imposes a substantial mutation load on populations. The absence of support for this effect in ~30% of the genes, however, suggests that mutation of mC will have no phenotypic effect for a significant proportion of human protein coding genes.