Integrating Bottleneck Size into Selection Tests for Biological Diversity Data

Abstract

Population bottlenecks profoundly shape genetic diversity, but distinguishing stochastic drift from selective pressure requires precise estimation and accounting for bottleneck size. While deep-sequencing data enable inference via frameworks like beta-binomial modeling, integrating these estimates directly into selection tests remains a critical challenge. In this study, based on existing computational approaches, we propose a new method that explicitly incorporates bottleneck size estimates into neutrality tests for biological diversity data. Designed for variant frequency data, our framework accounts for sequencing errors and sampling biases to improve the precision and interpretability of selection signature detection. We validate this framework using previously published Streptococcus pneumoniae in vivo experimental data, successfully replicating established fitness results, while uncovering novel genes relevant to infection and pathogenesis. This integrated new model with explicit bottleneck effects narrows down the set of candidate genes under selection and provides a robust, generalizable tool for disentangling drift from selection across a wide range of biological systems.