correctKin: an optimized method to infer relatedness up to the 4th degree from low-coverage ancient human genomes

Abstract

Kinship analysis from very low-coverage ancient sequences has been possible up to the second degree with large uncertainties. We propose a new, accurate, and fast method, correctKin, to estimate the kinship coefficient and the confidence interval using low-coverage ancient data. We perform simulations and also validate correctKin on experimental modern and ancient data with widely different genome coverages (0.12×–11.9×) using samples with known family relations and known/unknown population structure. Based on our results, correctKin allows for the reliable identification of relatedness up to the 4th degree from variable/low-coverage ancient or badly degraded forensic whole genome sequencing data. Supplementary Information The online version contains supplementary material available at 10.1186/s13059-023-02882-4.