ReSkmer: modeling repeats allows k-mer-based alignment-free methods to calculate population genomic distances

Understanding ecological change requires robust, inexpensive methods for monitoring populations. One approach is using low-coverage genome sequencing (genome skimming) to compute genetic diversity via assembly-free, alignment-free, k-mer-based methods to compute genomic distances from shotgun sequences using the intersection of k-mer sets. Skmer extended this approach to genome skims by modeling coverage and error. However, these methods ignore genome repetitiveness, hampering population genetic distance calculation. Here, we mathematically derive the expected intersection size between k-mers sampled from two repetitive genomes, accounting for repeats, coverage, and errors, leading to the method ReSkmer. Our experiments show highly accurate distances despite sampling highly repetitive genomes.