Predicting the genetic loci of past evolution

Abstract

Repetitions in the mutations found to be responsible for independent evolution of similar phenotypes in various taxa have led some biologists to propose that for certain evolutionary changes the causal mutations are predictable. We examine here the nature of the predictions that have been made and their associated arguments. Predictions about the loci of past evolution are retrodictions, i.e. inferences about events that occurred in the past. They are not based on elaborate models and they derive mainly from the observation of repeated cases of genetic evolution. Predictions at the nucleotide level or at the gene level have a higher inference gain than those for broader categories of genetic changes such as cis-regulatory mutations.