I have to give kudos to Keith Baggerly, PhD, and Kevin Coombes, PhD. Not only are they the biostatisticians at MD Anderson Cancer Center who initially uncovered problems with Anil Potti’s data, but they continue to work for solutions in the system so that such problems won’t happen again.
In an editorial published online March 1 in Clinical Chemistry (subscription required), they argue, once again, that investigators should be required to make the raw data and computer code used to develop ‘omic signatures freely available.
Although the events at Duke are an extreme case, simple errors are not, according to Drs. Baggerly and Coombes. And even simple errors can lead to big problems in a genomic signature. Having the code and raw data available won’t make those problems disappear, but they will make it easier for researchers to understand what has happened if and when they try to reproduce the original results.
In my graduate laboratory, we used to (affectionately) call my advisor’s administrative assistant a “bull dog” because of her tenacity and ability to get things done. Drs. Baggerly and Coombes are showing a similar determination and I think the community will be better off for it.
Dr. Baggerly is scheduled to speak at the second meeting of the Institute of Medicine committee, which is charged with reviewing ‘omics-based tests, later this month.
More retractions in the works?
I asked Doug Stokke, Associate Vice President of Duke Medicine News & Communications, if we are likely to see more retractions. His response was short, but too the point -- and good news for the community: “A team of researchers is continuing to carefully evaluate other related publications.”