Evaluating genetic diversity differences within a likelihood framework

Abstract

The study of genetic diversity has a rich history, as it serves as the foundation upon which evolutionary processes act. Numerous methods have been developed to quantify genetic variation within and among populations. While some methods focus solely on allele frequencies, others assess branching patterns within phylogenetic clades without considering genetic distances. Here we introduce the Proportional Diversity Likelihood Ratio Statistic (PLR), a novel method that integrates genetic distances with phylogenetic tree structure to provide a comprehensive measure of genetic diversity. This method evaluates the likelihood of observed genetic data under models with constrained and unconstrained diversity, offering a robust statistical framework for testing hypotheses about genetic differentiation. The PLR approach is broadly applicable, from assessing immune repertoire diversity, such as B-cell diversity before and after vaccination, to evaluating tumor genetic diversity before and after treatment. By combining genetic distance and tree topology, PLR enables deeper insights into evolutionary dynamics and population structure. We demonstrate the use of this approach on SARS-CoV-2 sequences from two different time frames and found that diversity maybe waning. Moreover, we applied PRL to chronic lymphocytic leukemia (CLL) data, revealing that with temporally order sequences from with treatment we do not see a significant drop in diversity.