Nonsynonymous substitution rate (Ka) is a relatively consistent parameter for defining fast-evolving and slow-evolving protein-coding genes.

BACKGROUND: Mammalian genome sequence data are being acquired in large quantities and at enormous speeds. We now have a tremendous opportunity to better understand which genes are the most variable or conserved, and what their particular functions and evolutionary dynamics are, through comparative genomics.

RESULTS: We chose human and eleven other high-coverage mammalian genome data-as well as an avian genome as an outgroup-to analyze orthologous protein-coding genes using nonsynonymous (Ka) and synonymous (Ks) substitution rates. After evaluating eight commonly-used methods of Ka and Ks calculation, we observed that these methods yielded a nearly uniform result when estimating Ka, but not Ks (or Ka/Ks). When sorting genes based on Ka, we noticed that fast-evolving and slow-evolving genes often belonged to different functional classes, with respect to species-specificity and lineage-specificity. In particular, we identified two functional classes of genes in the acquired immune system. Fast-evolving genes coded for signal-transducing proteins, such as receptors, ligands, cytokines, and CDs (cluster of differentiation, mostly surface proteins), whereas the slow-evolving genes were for function-modulating proteins, such as kinases and adaptor proteins. In addition, among slow-evolving genes that had functions related to the central nervous system, neurodegenerative disease-related pathways were enriched significantly in most mammalian species. We also confirmed that gene expression was negatively correlated with evolution rate, i.e. slow-evolving genes were expressed at higher levels than fast-evolving genes. Our results indicated that the functional specializations of the three major mammalian clades were: sensory perception and oncogenesis in primates, reproduction and hormone regulation in large mammals, and immunity and angiotensin in rodents.

CONCLUSION: Our study suggests that Ka calculation, which is less biased compared to Ks and Ka/Ks, can be used as a parameter to sort genes by evolution rate and can also provide a way to categorize common protein functions and define their interaction networks, either pair-wise or in defined lineages or subgroups. Evaluating gene evolution based on Ka and Ks calculations can be done with large datasets, such as mammalian genomes.

REVIEWERS: This article has been reviewed by Drs. Anamaria Necsulea (nominated by Nicolas Galtier), Subhajyoti De (nominated by Sarah Teichmann) and Claus O. Wilke.