Concerns Regarding Newly Obtained Raw Data From Cassava-Linked Paper
Recently, I obtained via Freedom of Information Law request a series of emails between Dr. Hoau-Yan Wang and the Editor-in-Chief of Behavioral Pharmacology (among many other emails). Prior, I had reached out to the journal on multiple occasions asking if they planned to publish the raw data that was analyzed in connection with their decision not to take any action regarding concerns raised on PubPeer about image irregularities in a 2007 paper published by the research lead at Cassava Sciences, which appeared to potentially contain recycled images from a 2005 Neuroscience paper published by Dr. Wang and Cassava’s Dr. Lindsay Burns. They declined to respond, and the data has remained locked away for more than a year.
The initial inquiry regarding these concerns was communicated to Dr. Wang on November 30, 2021, referencing issues raised in the aforementioned PubPeer post:
Part of the reason that these data may have been kept secret was a small note that Dr. Wang made at the start of his response, “this email and attachments are confidential and not subject to FOIL or FOIA.” I am not a lawyer, but I do not believe that this statement carries any legal weight whatsoever (and, indeed, these emails and attachments were obtained very simply via FOIL request; no appeal required). It is indicative, however, of the effort that those affiliated with Cassava Sciences have undergone in order to ensure that none of the raw data is scrutinized by the public.
The attachments, which can be found in the FOIL request response at the link provided at the start of this article, contain the raw data provided to Behavioral Pharmacology. A cursory look at these attachments reveals myriad issues overlooked by the journal, which, after additional issues spurred them to send a follow-up request, responded with the following:
It is certainly possible that the concerns which follow were identified by the journal and that additional clarification was provided, however there is no indication that that was the case. I am documenting the following in hopes that the journal and its editors will more openly communicate about their oversight process and address how and why the data was cleared.
Raw data from two separate experiments shows significant background noise overlap. Raw data provided in email:
Adjusted contrast, with regions of easily identifiable patterns matched using colored boxes. Many more such overlaps exist, but were not noted for brevity (a third panel also contains the same background noise):
Similar issues were identified by Dr. Bik in raw data sent to Neuroscience:
Raw data appears to contain significant differences from published data. The blot provided for the published data was inexplicably flipped horizontally (as noted in the email). However, this flip appears to also invert the direction of the angle between the two blot marks, which no longer appear to match the marks in the published data:
Possible signatures of cut/paste in critical blot sections. Though the effect is subtle, there appears to be rectangular patches underneath the two blots circled in red which are slightly darker than the rest of the background. Using the unexpectedly similar noise pattern as a guide, one can find the approximate location of one of these patches in the other blot, which does not appear to contain the same dark pattern. The two lower light brown boxes show matching patterns, whereas the third box on the upper left does not have obvious signs of matching background pattern.
Unexpected similarities in background noise between Neuroscience 2005 article and Behavioral Pharmacology 2007 data. Despite apparently being published two years apart, there are again striking similarities between the background noise signatures in the data provided for the two papers which were flagged for potentially containing reused images.
I hope that these irregularities are promptly addressed by the editors of both Behavioral Pharmacology and Neuroscience.