Placing ancient DNA sequences into reference phylogenies

Abstract

During the last decade, large volumes of ancient DNA (aDNA) data have been generated as part of whole-genome shotgun and target capture sequencing studies. This includes sequences from non-recombining loci such as the mitochondrial or Y chromosomes. However, given the highly degraded nature of aDNA data, post-mortem deamination and often low genomic coverage, combining ancient and modern samples for phylogenetic analyses remains difficult. Without care, these factors can lead to incorrect placement. For the Y chromosomes, current standard methods focus on curated markers, but these contain only a subset of the total variation. Examining all polymorphic markers is particularly important for low coverage aDNA data because it substantially increases the number of overlapping sites between present-day and ancient individuals which may lead to higher resolution phylogenetic placement. We provide an automated workflow for jointly analysing ancient and present-day sequence data in a phylogenetic context. For each ancient sample, we effectively evaluate the number of ancestral and derived alleles present on each branch and use this information to place ancient lineages to their most likely position in the phylogeny. We provide both a parsimony approach and a highly optimised likelihood-based approach that assigns a posterior probability to each branch. To illustrate the application of this method, we have compiled and make available the largest public Y-chromosomal dataset to date (2,014 samples) which we used as a reference for phylogenetic placement. We process publicly available African ancient DNA Y-chromosome sequences and examine how patterns of Y-chromosomal diversity change across time and the relationship between ancient and present-day lineages. The same software can be used to place samples with large amounts of missing data into other large non-recombining phylogenies such as the mitochondrial tree.