Mutation Effect Generalizability under Selection-Drift

TL;DR

This paper investigates the limits of how well mutation effects can be generalized across different populations and evolutionary regimes, using large-scale genomic data and experiments to understand predictability in evolution.

Contribution

It introduces a framework for assessing mutation effect generalizability under various population structures and evolutionary forces, revealing asymptotic behaviors related to natural selection and drift.

Findings

Maximal effect generalization occurs in populations under natural selection and drift.

Distinct limits to fitness and evolution predictability are identified.

Large-scale genomic and barcoding data support the theoretical results.

Abstract

While Neutral Theory famously describes the number of discrete genetic differences in populations, we consider the number of genetic backgrounds under which such differences are observed - setting limits to the generalizability of their effects. This allow us to determine which population structures and diversity rates have maximal effect generalization across (1) environmental and (2) genetic variation, and to demonstrate that they correspond asymptotically to those of populations under (1) natural selection and (2) drift. At the same time, these results suggest distinct limits to the predictability of fitness and evolution across evolutionary regimes. We employ both broad time, large-scale genome sequencing datasets (including whole-genome autocorrelation calculations), and fine time-scale barcoding experiments.