BetaScan2: Standardized Statistics to Detect Balancing Selection Utilizing Substitution Data

Long-term balancing selection results in a build-up of alleles at similar frequencies and a deficit of substitutions when compared with an outgroup at a locus. The previously published [Formula: see text] statistics detect balancing selection using only polymorphism data. We now propose the [Formula: see text] statistic which detects balancing selection using both polymorphism and substitution data. In addition, we derive the variance of all β statistics, allowing for their standardization and thereby reducing the influence of parameters which can confound other selection tests. The standardized β statistics outperform existing summary statistics in simulations, indicating β is a well-powered and widely applicable approach for detecting balancing selection. We apply the [Formula: see text] statistic to 1000 Genomes data and report two missense mutations with high β scores in the ACSBG2 gene. An implementation of all β statistics and their standardization are available in the BetaScan2 software package at https://github.com/ksiewert/BetaScan.