PLOS Genetics paper cautions against uncritical rescaling in population genetics simulations

A methodological paper by Parul Johri, Fanny Pouyet, and Brian Charlesworth, published in PLOS Genetics, examines the rights and wrongs of rescaling in population genetics computer simulations — a topic of direct relevance to researchers who routinely use coalescent or forward-time simulations to interpret large genomic datasets.

Rescaling is the practice of simulating populations much smaller than the real biological population of interest, whilst adjusting parameters such as mutation rates and selection coefficients so that their products with population size remain constant. This reduces computational cost substantially, but Johri and colleagues demonstrate that the approach introduces non-trivial distortions under certain conditions — particularly when complex demographic histories, strong selection, or interactions between selection and drift are involved.

The paper provides analytical and simulation-based guidance on when rescaling is and is not appropriate, making it a practically useful reference for researchers designing studies that depend on simulated null distributions or demographic inference.

For statistical geneticists and population geneticists, this is a methods paper with direct implications for study design and interpretation. Educators teaching computational genomics will find it a useful reference for discussing the assumptions that underpin widely used tools.