Three Reagents for in-Solution Enrichment of Ancient Human DNA at More than a Million SNPs

Abstract

In-solution enrichment for hundreds of thousands of single nucleotide polymorphisms (SNPs) has been the source of >70% of all genome-scale ancient human DNA data published to date. This approach has made it possible to generate data for one to two orders of magnitude lower cost than random shotgun sequencing, making it economical to study ancient samples with low proportions of human DNA, and increasing the rate of conversion of sampled remains into working data thereby facilitating ethical stewardship of human remains. So far, nearly all ancient DNA data obtained using in-solution enrichment has been generated using a set of bait sequences targeting about 1.24 million SNPs (the ‘1240k reagent’). These sequences were published in 2015, but synthesis of the reagent has been cost-effective for only a few laboratories. In 2021, two companies made available reagents that target the same core set of SNPs along with supplementary content. Here, we test the properties of the three reagents on a common set of 27 ancient DNA libraries across a range of richness of DNA content and percentages of human molecules. All three reagents are highly effective at enriching many hundreds of thousands of SNPs. For all three reagents and a wide range of conditions, one round of enrichment produces data that is as useful as two rounds when tens of millions of sequences are read out as is typical for such experiments. In our testing, the “Twist Ancient DNA” reagent produces the highest coverages, greatest uniformity on targeted positions, and almost no bias toward enriching one allele more than another relative to shotgun sequencing. Allelic bias in 1240k enrichment has made it challenging to carry out joint analysis of these data with shotgun data, creating a situation where the ancient DNA community has been publishing two important bodes of data that cannot easily be co-analyzed by population genetic methods. To address this challenge, we introduce a subset of hundreds of thousands of SNPs for which 1240k data can be effectively co-analyzed with all other major data types.