Quantifying and reducing cross-contamination in single- and multiplex hybridization capture of ancient DNA

The use of hybridization capture has enabled a massive upscaling in sample sizes for ancient DNA studies, allowing the analysis of hundreds of skeletal remains (Mathieson et al., 2015; Narasimhan et al., 2019) or sediments (Vernot et al., 2021; Wang et al., 2021; Zavala et al., 2021) in single studies. Yet demands in throughput continue to grow, and hybridization capture has become a limiting step in sample preparation due to the large consumption of reagents, consumables and time. Here we explore the possibility of improving the economics of sample preparation via multiplex capture, i.e. the hybridization capture of pools of double-indexed ancient DNA libraries. We demonstrate that this strategy is feasible for small genomic targets, such as mitochondrial DNA, if the annealing temperature is increased and PCR cycles are limited in post-capture amplification to avoid index swapping by jumping PCR, which manifests as cross-contamination in resulting sequence data. We also show that the re-amplification of double-indexed libraries to PCR plateau before or after hybridization capture can sporadically lead to small, but detectable cross-contamination even if libraries are amplified in separate reactions. We provide protocols for both manual capture and automated capture in 384-well format that are compatible with single- and multiplex capture and effectively suppress cross-contamination and artefact formation. Last, we provide a simple computational method for quantifying cross-contamination due to index swapping in double-indexed libraries, which we recommend using for routine quality checks in studies that are sensitive to cross-contamination.