Molecular Biology and Evolution, doi:10.1093/molbev/msn095
Bayesian Inference of Errors in Ancient DNA Caused by Post Mortem Degradation
Ligia M. Mateiu and Bruce Rannala
Methods for extracting and amplifying sequences using ancient DNA (aDNA) can be prone to errors caused by post mortem modifications of the DNA strand. A new statistical method is developed for predicting errors in aDNA sequences caused by such processes. In addition to the canonical DNA substitution model parameters, a discrete Markov chain is used to describe nucleotide substitutions occurring via post mortem degradation of the aDNA sequences. A computer program, BYPASSR-degr, was developed implementing the method and was used in subsequent analyses of simulated datasets under the new model. Simulation studies show that the new method can be powerful and accurate in identifying damaged sites. The method is applied to analyze aDNA sequences of Etruscans, Adélie penguins and horses. No significant signals of degradation were observed at any sites of the aDNA sequences we analyzed.