Compression for population genetic data through finite-state entropy

Abstract

We improve the efficiency of population genetic file formats and GWAS computation by leveraging the distribution of sample ordering in population-level genetic data. We identify conditional exchangeability of these data, recommending finite state entropy algorithms as an arithmetic code naturally suited to population genetic data. We show between 10% and 40% speed and size improvements over dictionary compression methods for population genetic data such as Zstd and Zlib in computation and and decompression tasks. We provide a prototype for genome-wide association study with finite state entropy compression demonstrating significant space saving and speed comparable to the state-of-the-art.