This week's edition of Nature has a brief paper (doi:10.1038/nature07390) reporting on the identification of an HIV positive tissue sample collected in Leopoldville (now Kinshasa) in what was then the Belgian Congo, and now known as the Democratic Republic of the Congo. Sequence data derived from the tissue was used to investigate the chronology of the appearance of HIV from its likely simian origin.
This is a piece of research which hit the news services (see for example this page at the BBC news website). The research has a number of features which earmark it for media interest: an important virus, a serious disease with a global spread, and a simple take-home message as to the origin of the virus. This raised my interest and I looked at the paper. Incidentally, the paper raises issues to do with complexity of statistical analysis: I imagine many readers such as I, and the journos who wrote articles in the press, have little or no chance of understanding what an "unconstrained Bayesian Markov chain Monte Carlo method" is, and are similarly limited in one's real critical analysis of conclusions reached by that means! I am forced to assume that all is above board in the statistical and computational aspects of this paper, and that the referees have done their job! In addition, it's always interesting in studies of ancient DNA (and cases where sample preservation was not originally intended to preserve nucleic acids) to know what measures were taken to ensure that contamination with modern DNA did not happen.