Detecting Selection from Patterns of Codon Usage

Jonathan Dushoff

Ecology and Evolutionary Biology, Princeton University

Unequal use of codons encoding the same amino acid (codon bias) is pervasive in living organisms, and has a number of causes, some well understood, some not. I will discuss a method of examining codon usage on a gene-by-gene basis to look for a signal of unusually strong natural selection, either to change, or not to change. A "bootstrap" method which compares a gene with randomly generated codon sequences can control for both the genome-wide codon usage, and the size and composition of each gene, allowing inferences about selection. I will also discuss correcting for local nucleotide biases and an approximation method for rapidly calculating the signal of evolutionary pressure.