Assessing DNA sequence alignment methods for characterizing ancient genomes and methylomes

Abstract

Applying high-throughput DNA sequencing technologies to the ancient DNA molecules preserved in subfossil material can provide genetic information from past individuals, populations and communities at the genomic scale. The combination of dedicated statistical techniques and specific molecular tools aimed at reducing the impact of post-mortem DNA damage can also help recover epigenetic data from ancient individuals. However, the capacity of different sequence aligners to identify ultrashort and deaminated ancient DNA templates, and their impact on the characterization of ancient methylomes, remain overlooked. In this study, we use both simulated and real ancient DNA sequence data to benchmark the performance of the read alignment tools most commonly used in ancient DNA research. We identify a read alignment strategy making use of the Bowtie2 aligner that substantially reduce computational times but shows increased sensitivity relative to previous recommendations based on the BWA aligner. This strategy significantly improves the genome coverage especially when DNA templates are shorter than 90bp, as is typically the case for ancient DNA. It also impacts on ancient DNA methylation estimates as it maximizes coverage improvement within CpG dinucleotide contexts, which hold the vast majority of DNA methylation marks in mammals. Our work contributes to improve the accuracy of DNA methylation maps and to maximize the amount of recoverable genetic information from archaeological and subfossil material. As the molecular complexity of ancient DNA libraries is generally limited, the mapping strategy recommended here is essential to both limit sequencing costs and sample destruction.