MODELING SELECTION BIAS ON RECOMBINATION RATES INFERRED THROUGH LINKAGE DISEQUILIBRIUM

Abstract

Recombination is a key part of evolutionary theory, and understanding the ways selection can bias inferred rates in a population can help us investigate better models for inference. This experiment models the bias in recombination rate inferences on a simulated genome with selection. Using SLiM forward genetic simulation, this experiment creates two basic genomic structures, one with and one without a hotspot. Then, using Pyrho a fine-scaled linkage disequilibrium-based inference model, the experiments reveal how selection biases the linkage disequilibrium model. Notably, with increasing nonsynonymous distribution of fitness effects (DFE), the inferences worsen and show a decreasing trend. This is most notable in the hotspot region of the second genomic structure (with a hotspot). The results show that the assumption of neutral selection in popular population-based inference methods is extremely important and should be addressed. In particular, among organisms with less compact genomes, the issue of selection would become more extreme and disruptive. In future models, this could be taken into consideration to improve inference ability among a diverse set of organisms with different levels of selection in their genome. Understanding how recombination rates differ across the tree of life can also reveal interesting molecular structures which further motivates accuracy in these inference methods.