Recommendations for improving statistical inference in population genomics

Abstract

The field of population genomics has grown rapidly with the recent advent of affordable, large-scale sequencing technologies. As opposed to the situation during the majority of the 20th century, in which the development of theoretical and statistical population-genetic insights out-paced the generation of data to which they could be applied, genomic data are now being produced at a far greater rate than they can be meaningfully analyzed and interpreted. With this wealth of data has come a tendency to focus on fitting specific (and often rather idiosyncratic) models to data, at the expense of a careful exploration of the range of possible underlying evolutionary processes. For example, the approach of directly investigating models of adaptive evolution in each newly sequenced population or species often neglects the fact that a thorough characterization of ubiquitous non-adaptive processes is a prerequisite for accurate inference. We here describe the perils of these tendencies, present our views on current best practices in population genomic data analysis, and highlight areas of statistical inference and theory that are in need of further attention. Thereby, we argue for the importance of defining a biologically relevant baseline model tuned to the details of each new analysis, of skepticism and scrutiny in interpreting model-fitting results, and of carefully defining addressable hypotheses and underlying uncertainties.

Efficient ancestry and mutation simulation with msprime 1.0

“…The efficiency of coalescent simulations depends crucially on the assumption of neutrality, and it is important to note that there are many situations in which this will be a poor approximation of biological reality ( Johri et al 2021 ). In particular, background selection has been shown to affect genome-wide sequence variation in a wide range of species ( Charlesworth et al 1993 , 1995 ; Charlesworth and Jensen 2021 ).…”

Background selection from unlinked sites causes non-independent evolution of deleterious mutations

“…What these are can be subject to dispute, e.g. whether adaptive mutations should be excluded from "baseline models" (Comeron 2017;Johri et al 2021) vs. whether independent lines of evidence so conclusively support widespread adaptation (Pennings et al 2014;Enard et al 2016) such that adaptive mutations must be considered until sufficient power for their exclusion has been proved. For our own somewhat different purposes, we take the empirical evidence for high deleterious mutation rates and widespread linkage disequilibrium to be broadly accepted, and hence their consideration to be required.…”

The recombination landscapes of spiny lizards (genusSceloporus)

“…, pyrho; Spence and Song 2019 ), our analyses are limited by the scarce data available for lizards in the genus Sceloporus. Namely, it is challenging to infer the demographic history of the two species in the first place without any prior knowledge of, for example, mutation rates, effective population sizes, or even which genomic regions to use for such inference as there are no genome annotations available that could be leveraged to select regions unaffected (or at least less affected) by selection [see Johri et al (BioRxiv) for a discussion regarding statistical inference in population genomics]. This highlights the importance of developing further genomic resources for these important model organisms to improve our understanding of recombination rate evolution in squamates.…”