Year: 2026
Type: preprint
Abstract: The rapidly expanding scale of ancient DNA data creates an increasing need for high-performance software methods. One area for such improvement is the inference of pairwise biological relatedness, a prominent aspect of many ancient DNA studies. We present fastpmr, a fast and memory-efficient method for biological relatedness inference based on the computation of pairwise mismatch rates. We demonstrate that fastpmr can achieve over 1000x speed-up over READv2, a widely used relatedness-inference method. We leverage this performance to investigate relatedness at the scale of the global ancient DNA record. Analyzing biological relatedness across all pairs in the Allen Ancient DNA Resource (n = 19,118 ancient and 23,089 total samples) recapitulates human evolutionary history and identifies previously unknown genetic duplicates and likely sample mix-ups, demonstrating fastpmr's utility for ancient DNA analysis and quality control at scale.
Authors: Edward C. Huang, Harald Ringbauer
Institution: Max Planck Institute for Evolutionary Anthropology
Language: English
Subfield: Genetics
Domain: Life Sciences
Open Access status: green
Funder: Max-Planck-Gesellschaft
OpenAlex ID: W7237254265
