Improving the performance of true single molecule sequencing for ancient DNA

Abstract

Background: Second-generation sequencing technologies have revolutionized our ability to recover genetic information from the past, allowing the characterization of the first complete genomes from past individuals and extinct species. Recently, third generation Helicos sequencing platforms, which perform true Single-Molecule DNA Sequencing (tSMS), have shown great potential for sequencing DNA molecules from Pleistocene fossils. Here, we aim at improving even further the performance of tSMS for ancient DNA by testing two novel tSMS template preparation methods for Pleistocene bone fossils, namely oligonucleotide spiking and treatment with DNA phosphatase.

Results: We found that a significantly larger fraction of the horse genome could be covered following oligonucleotide spiking however not reproducibly and at the cost of extra post-sequencing filtering procedures and skewed %GC content. In contrast, we showed that treating ancient DNA extracts with DNA phosphatase improved the amount of endogenous sequence information recovered per sequencing channel by up to 3.3-fold, while still providing molecular signatures of endogenous ancient DNA damage, including cytosine deamination and fragmentation by depurination. Additionally, we confirmed the existence of molecular preservation niches in large bone crystals from which DNA could be preferentially extracted.

Conclusions: We propose DNA phosphatase treatment as a mechanism to increase sequence coverage of ancient genomes when using Helicos tSMS as a sequencing platform. Together with mild denaturation temperatures that favor access to endogenous ancient templates over modern DNA contaminants, this simple preparation procedure can improve overall Helicos tSMS performance when damaged DNA templates are targeted.

Figure 1

Sequence features of the population of tSMS reads recovered from sample CA (>50,300 BP). Sample CA was extracted in duplicate and both extracts were tSMS sequenced on a Helicos 1100 FOV channel following different template preparation procedures as described in the methods section. Top: Read length distribution (spiked sequences are reported as vertical bars). Middle: Read %GC contents. Right: Cumulative guanine to adenine misincorporation rates as a function of the distance from sequencing start. This class of mismatch derives from the post-mortem deamination of cytosine residues and can be taken as a proxy for post-mortem DNA damage.

Figure 2

Sequence features of the population of tSMS reads recovered from sample TP (13,389 ± 52 BP). Sample TP was extracted in duplicate and identical volumes of the extracts were tSMS sequenced on a Helicos 110 FOV channels following different template preparation procedures as described in the methods section. The analyses are restricted to the fraction of reads identified as originating from the horse genome. Top: Read length distribution (spiked sequences are reported as vertical bars). Middle: Read GC contents. Right: Cumulative guanine to adenine misincorporation rates as a function of the distance from sequencing start. This class of mismatch derives from the post-mortem deamination of cytosine residues and can be taken as a proxy for post-mortem DNA damage.

Figure 3

Nucleotide misincorporation patterns observed on extracts CA1 and CA2. The frequencies of all possible mismatches and indels observed between the horse genome and the reads are reported in grey as a function of the position on sequencing reads, except for C→T, G→A, insertions, and deletions that are reported in blue, red, pink, and green respectively. Only the first 25 nucleotides sequenced are considered. Left: CA1 extract. Right: CA2 extract. Template preparation procedures for tSMS sequencing are reported on the left hand side of the graphs. All graphs correspond to the analyses of horse reads (80°C, 95°C, and 80°C + Spiking) except for the ones labelled 80°C olig. that refer to the analysis of reads mapping against the 30 oligonucleotides used for spiking.

Figure 4

Base composition at the 5’-end of tSMS reads and preceding genomic regions. Read base composition is reported for the first position sequenced (position 1, bottom). In addition, the base composition of the genomic region is indicated for the two positions preceding the beginning of tSMS reads. The first (position −1, middle) corresponds to the locking site while the second (position −2, top) corresponds to the genomic position following the last nucleotide preserved in the ancient DNA template and located on the complementary strand (see Additional file 1, Figure S2).

Figure 5

Phylogenetic relationships among ancient and modern horses. All sequences were downloaded from GenBank, except for sample TP that has been analyzed for the first time in this study. Accession numbers are reported together with available information about the different samples considered. The analysis was restricted to sites showing a minimal coverage depth of 2 and >50% of sequence identity among reads (11,316 sites). Phylogenetic inference is performed with Maximum Likelihood using a HKY + Γ 8 model of molecular evolution including a gamma correction for rate heterogeneity among sites. Node supports are estimated from a total number of 1,000 bootstrap pseudo-replicates and are reported for those above 75%.