Ligation bias in illumina next-generation DNA libraries: implications for sequencing ancient genomes.

Ancient DNA extracts consist of a mixture of endogenous molecules and contaminant DNA templates, often originating from environmental microbes. These two populations of templates exhibit different chemical characteristics, with the former showing depurination and cytosine deamination by-products, resulting from post-mortem DNA damage. Such chemical modifications can interfere with the molecular tools used for building second-generation DNA libraries, and limit our ability to fully characterize the true complexity of ancient DNA extracts. In this study, we first use fresh DNA extracts to demonstrate that library preparation based on adapter ligation at AT-overhangs are biased against DNA templates starting with thymine residues, contrarily to blunt-end adapter ligation. We observe the same bias on fresh DNA extracts sheared on Bioruptor, Covaris and nebulizers. This contradicts previous reports suggesting that this bias could originate from the methods used for shearing DNA. This also suggests that AT-overhang adapter ligation efficiency is affected in a sequence-dependent manner and results in an uneven representation of different genomic contexts. We then show how this bias could affect the base composition of ancient DNA libraries prepared following AT-overhang ligation, mainly by limiting the ability to ligate DNA templates starting with thymines and therefore deaminated cytosines. This results in particular nucleotide misincorporation damage patterns, deviating from the signature generally expected for authenticating ancient sequence data. Consequently, we show that models adequate for estimating post-mortem DNA damage levels must be robust to the molecular tools used for building ancient DNA libraries.

Mitochondrial phylogenomics of modern and ancient equids.

The genus Equus is richly represented in the fossil record, yet our understanding of taxonomic relationships within this genus remains limited. To estimate the phylogenetic relationships among modern horses, zebras, asses and donkeys, we generated the first data set including complete mitochondrial sequences from all seven extant lineages within the genus Equus. Bayesian and Maximum Likelihood phylogenetic inference confirms that zebras are monophyletic within the genus, and the Plains and Grevy's zebras form a well-supported monophyletic group. Using ancient DNA techniques, we further characterize the complete mitochondrial genomes of three extinct equid lineages (the New World stilt-legged horses, NWSLH; the subgenus Sussemionus; and the Quagga, Equus quagga quagga). Comparisons with extant taxa confirm the NWSLH as being part of the caballines, and the Quagga and Plains zebras as being conspecific. However, the evolutionary relationships among the non-caballine lineages, including the now-extinct subgenus Sussemionus, remain unresolved, most likely due to extremely rapid radiation within this group. The closest living outgroups (rhinos and tapirs) were found to be too phylogenetically distant to calibrate reliable molecular clocks. Additional mitochondrial genome sequence data, including radiocarbon dated ancient equids, will be required before revisiting the exact timing of the lineage radiation leading up to modern equids, which for now were found to have possibly shared a common ancestor as far as up to 4 Million years ago (Mya).