The genetic code is highly redundant with multiple codons coding for a particular amino acid. Even before the first genome was sequenced, it became apparent that synonymous codons of an amino acid are used unequally in coding sequences. Ever since, the relative importance of adaptive and non-adaptive factors affecting such codon usage bias (CUB) have been actively debated. The lack of a coherent framework to test alternate hypotheses, and a focus on heuristic indices have hindered our ability to disentangle the effects of different evolutionary pressures on codon bias.
We are developing a framework that integrates mechanistic models of protein translation with population-genetics models to understand and distinguish the effects of various adaptive and non-adaptive forces on the evolution of codon biases. Such models will allow us to make quantitative predictions on how codon usage would change with varying mutation and selection regimes.