Hybrid RY-mers increase aDNA alignment sensitivity

Abstract

One of the key steps in ancient DNA data analysis is the accurate identification of endogenous short reads that originate from the ancient genome. Currently, this step is performed using general-purpose sequence aligners, such as BWA-aln, with parameter settings optimized for sensitivity. This approach results in increased runtime, even relative to standard DNA alignment workflows, and is not sufficient for aligning samples that are highly divergent from a reference database. We introduce hybrid RY-mers, a seed construct designed to incorporate aDNA-specific sequence damage. We adapt hybrid RY-mers to the existing concept of multi-context seeds and implement them in the read aligner strobealign. On simulated genomic and metagenomic datasets, we show that hybrid RY-mers result in higher mapping accuracy, while also reducing runtime by 4-6 times compared to the state-of-the-art aDNA alignment method BWA-aln. Furthermore, substituting aDNA aligner MALT in the metagenomic tool-suite aMeta with our new alignment approach demonstrated that aMeta gave a faster and less memory consuming detection of ancient taxa present in the samples.