My original page on the Kundu case is here. Current Science has published a letter I wrote on this affair; it is followed by a response by G. Padmanaban.
He reiterates that the original autoradiograms are genuine. In the June 10 issue, he had displayed very poor-quality images of what purport to be the original autoradiograms for figures 7c in paper I and 6a (and 8a) in paper II -- so poor that I chose not to comment. But I chose the very same figures for illustration; it turns out that these are top halves of figures, whose independent bottom halfs line up too, without any further modification other than scaling. Moreover, the black borders, too, align perfectly (as is the case for nearly all the offending figures).
Padmanaban ignores these points, supplying once again the same autoradiograms that purportedly prove the authenticity of the top halves of these figures. This time the image quality is better, but is still very grainy and full of artifacts of heavy image compression. The published images are poor too, but one would expect someone with access to the original autoradiograms to make high-quality reproductions.
Rather than recycle these images from his previous response, it would have been interesting to see the original autoradiograms corresponding to the bottom halves of these figures, which display a variety of features -- streaks, smears, and so on -- that one really would not expect to be duplicated by chance. It would have also been interesting to hear Padmanaban's explanation of the fact that the black borders overlap.
They indeed look very similar; I tried overlaying one on the other, and on this occasion no scaling was required, but I needed to rotate one of them. They align up very well -- so well that one is tempted to say, again, that they are the same.
But there are differences -- the one that purports to represent figure 6a of paper 2 lacks the extra bands below the main bands in lane 3 and 5. Padmanaban himself notes this in his response to me: "One can see the artifactual bands below the main bands in figure 7c, but not in figure 6a."
But the extra band is visible in lane 5 (numbered 6.9) of both published figures!
What sort of post-processing in PhotoShop could introduce such bands where they didn't exist in the original autoradiogram? One may expect genuine features to be washed out (by tuning the brightness/contrast, for example). In fact this seems to have happened with a smear below lane 6 (both "original autoradiograms") that is missing in the published images. But one does not expect new features to be introduced.
There are other curiosities, minor by themselves. In lane 6 (numbered 1.5 and 1.6), the published images have a clearly asymmetrical band, with a prominent "bump" on the right hand side. This is not visible in the autoradiograms. Lane 1 is barely visible in both published images, except for the right edge; but in the original autoradiogram, the left edge is darker than the right. The extra band below the main one in lane 3, in the autoradiogram for Fig 7c, is darker than the band in lane 1; but in the published figure, the band in lane 1 is visible and the extra band in lane 3 is not.
So I reiterate my original question: even assuming the authenticity of these autoradiograms (and ignoring the poor image quality and signs of possible manipulation), how did the Padmanaban Committee verify that these correspond to the published images?
Padmanaban in fact acknowledges the possibility that they don't, and raises another possibility -- that the same blot was given different exposures -- but does not comment on it. He argues that if one admits such possibilities, "the sky is the limit and beyond anyone's imagination". I have to agree here.
Nevertheless, I am pleased that Padmanaban has chosen, overleaf from my reproduction of the figures, to defend his analysis that proclaims the authenticity of those figures. Even if he does so without any evidence, it takes a certain bravery to do it at all. And it takes a certain effrontery to argue with an international journal, using such "evidence", that their decision to withdraw the paper was unjustified.
One final note: Padmanaban comments that "the target for these people [me and Sohan Modak] is the committee (or me) rather than Kundu's data or the explanations offered by the authors." He is right. I can't speak for Modak (I have never corresponded with him), but my reason is simple: fraud happens everywhere, and isolated occurrences of such things, in a country of a billion people, is expected. A whitewash, by a committee consisting of the country's top scientists is, however, not expected.
August 8, 2007
(Note: I believe I have said all I have to say and would really prefer not to get any further correspondence unless it contains something new and interesting that hasn't been said before. I'm at best an "in silico" biologist, and would like to leave further discussion of the issue to "real" biologists, in appropriate media.)