Post date: Feb 28, 2016 12:44:13 PM
A news article describes how a colleague discovered the error in the study entitled ‘Ancient Ethiopian genome reveals extensive Eurasian admixture throughout the African continent’ which completely invalidated the claim of African-wide effects.
[side note: kudos to Nature for highlighting this event -- although it may not be a coincidence that the original paper was published in Science -- the research community needs to hear more about these behind-the-scenes happenings so they can be avoided in the future.]
The paper is still interesting and publishable based on the genome it presents. In the erratum (the not explaining the error), the authors write:
"[the results] were affected by a bioinformatics error. A script necessary to convert [from one software program to another] was not run"
Of course, it happens, but I think the authors should take more responsibility. Part of bioinformatic training is learning to run sanity checks, question and replicate pipelines, and consider the implications of surprising results.