SNIDE

Human codon context dominated by CpG dinucleotides predisposes certain gene classes to an increased propensity for point mutation, and thus enables the ability to predict which genes will evolve most rapidly. Nonsynonymous mutation probabilities for 12,865 human genes are calculated and contrasted using four mutation models derived from distinct sources: Disease-causing variants, single nucleotide polymorphisms, intronic mutations, and interspecific substitutions from aligned orthologs. We conclude that the most mutable human genes are those that mediate reaction to environmental stimuli, including those involved in immune response (p<0.000879), pathogen response (p<0.00098), and olfaction (p<0.000153). Coupled with observations from studies indicating these classes have experienced positive selection in humans, our results imply that codon usage may shape the size and diversity of the mutation pool on which selection acts. As expected, genes using context inclining low point mutation are those involved in essential processes such as cell proliferation (p<0.000195) and DNA repair (p<0.00543). Developmental (p<1.12E-10) and transcription-related genes (p<1.70E-13) have CpG context biasing them towards point mutation that is conservative with respect to impact on the protein product. We infer that this allows genes to proportionally mutate to effect speciation while avoiding a drastic change in the function of the protein. The methods employed here describe a scoring system by which to compare the relative mutability of gene sequences, which highlights a gene's rigidity or malleability towards point mutation throughout its evolution. .