Massive influence of DNA isolation and library preparation approaches on palaeogenomic sequencing data

Abstract

The ability to access genomic information from ancient samples has provided many important biological insights. Generating such palaeogenomic data requires specialised methodologies, and a variety of procedures for all stages of sample preparation have been proposed. However, the specific effects and biases introduced by alternative laboratory procedures is insufficiently understood. Here, we investigate the effects of three DNA isolation and two library preparation protocols on palaeogenomic data obtained from four Pleistocene subfossil bones. We find that alternative methodologies can significantly and substantially affect total DNA yield, the mean length and length distribution of recovered fragments, nucleotide composition, and the total amount of usable data generated. Furthermore, we also detect significant interaction effects between these stages of sample preparation on many of these factors. Effects and biases introduced in the laboratory can be sufficient to confound estimates of DNA degradation, sample authenticity and genomic GC content, and likely also estimates of genetic diversity and population structure. Future palaeogenomic studies need to carefully consider the effects of laboratory procedures during both experimental design and data analysis, particularly when studies involve multiple datasets generated using a mixture of methodologies.