Predicting Molecular Evolutionary Trajectories in Principle and in Practice

Abstract

AbstractThe discrete, digital nature of DNA (deoxyribonucleic acid) defines a large but countable number of mutational trajectories between any two alleles. Even though some set of mutations jointly confer an improvement in an organism's reproductive success, it need not follow that all mutational trajectories to that allele are equivalent in the eyes of evolution by natural selection. To a good approximation, the key question is whether all mutations improve reproductive success in all combinations; if they do not, then some subset (even, possibly all) trajectories will be selectively inaccessible. Mutations that are only conditionally beneficial are said to exhibit sign epistasis, which several recent empirical studies demonstrate is pervasive. Biochemical and biophysical considerations of protein biology are beginning to shed light on the underlying mechanisms of sign epistasis and current theoretical attention is focused on its influence on selectively accessible trajectories in sexual population.Key concepts:Phenotypic change over evolutionary time must reflect underlying changes in genetic constitution of the population.Under reasonable assumptions, it is possible to enumerate all mutational trajectories a population might follow in the course of evolving from one genetic variant to another.Recent experimental work demonstrates that for several proteins, only a very small fraction of mutational trajectories are accessible to populations evolving under the action of natural selection.The identity of selectively accessible mutational trajectories reflects the underlying biochemistry and biophysics of the protein in question.