Title: Calling Genotypes and Estimating Heterozygosity from ancient DNA

Abstract:

The analysis of ancient genomes is complicated by the degraded nature of ancient DNA: ancient DNA fragments are often short and carry base modifications that manifest as substitutions in sequence reads. In principle, the errors that arise from base-modifications could be overcome by calling genotypes from higher coverage data. I present snpAD, an iterative method that estimates error rates and genotype frequencies from diploid ancient genomes. These parameters can be used to call posterior probabilities of genotypes. I show that the method can be used to reliably estimate heterozygosity from high to medium-coverage genomes (>10x). However, approximate estimates of heterozygosity can be gained from lower-coverage genomes (>1x) and I present heterozygosity estimates from the low-coverage genomes of 5 Neandertals, a Denisovan-Neandertal first generation offspring and several ancient modern human genomes. Further extensions to estimate contamination, regions of inbreeding and reference bias are discussed.