Archives SEEM

Philip Gerrish

Le vendredi 20 mars 2015 - Grande salle de réunion du CEFE - 11h30


Mutations that change fitness have been described as the “raw material” of evolution, because they are the ultimate source of heritable fitness variation upon which natural selection acts. Despite their central role in evolution, however, little is known about them. This fundamental deficiency in evolutionary biology is primarily due to the notorious difficulty of inferring the fitness effects of newly-arising mutations, a difficulty that derives from the unavoidable confounding effects of natural selection. Natural selection imposes a strong sampling bias in favor of higher-fitness genotypes, and as a further complication, this biasing effect itself depends on the dynamic composition of the population. We derive a novel framework for robust inference of fitness effects of mutations – a hierarchy of statistical relations that result from a cumulant expansion of PDE models of evolutionary dynamics. We assess the accuracy with which this framework is able to reconstruct the distribution of fitness effects of newly- arising mutations, using samples taken from simulated evolving populations as well as evolving populations of Escherichia coli. I will discuss our work in the context of other, more mechanistic but perhaps less accurate, attempts to re- construct this fundamental distribution; I will also mention how these methods might be used to decipher general features of fitness landscapes and perhaps even to forecast the near-future evolutionary trajectory or fate of a population.