Abstract
Recently, the finding of 8,000 year old wheat DNA from submerged marine sediments (1) was challenged on the basis of a lack of signal of cytosine deamination relative to three other data sets generated from young samples of herbarium and museum specimens, and a 7,000 year old human skeleton preserved in a cave environment (2). The study used a new approach for low coverage data sets to which tools such as mapDamage cannot be applied to infer chemical damage patterns. Here we show from the analysis of 148 palaeogenomic data sets that the rate of cytosine deamination is a thermally correlated process, and that organellar generally shows higher rates of deamination than nuclear DNA in comparable environments. We categorize four clusters of deamination rates (alpha,beta,gamma,epsilon) that are associated with cold stable environments, cool but thermally fluctuating environments, and progressively warmer environments. These correlations show that the expected level of deamination in the sedaDNA would be extremely low. The low coverage approach to detect DNA damage by Weiss et al. (2) fails to identify damage samples from the cold class of deamination rates. Finally, different enzymes used in library preparation processes exhibit varying capability in reporting cytosine deamination damage in the 5 prime region of fragments. The PCR enzyme used in the sedaDNA study would not have had the capability to report 5 prime cytosine deamination, as they do not read over uracil residues, and signatures of damage would have better been sought at the 3 prime end. The 8,000 year old sedaDNA matches both the thermal age prediction of fragmentation, and the expected level of cytosine deamination for the preservation environment. Given these facts and the use of rigorous controls these data meet the criteria of authentic ancient DNA to an extremely stringent level.
