In December, another researcher contacted me with questions about the 9DS paper (he wanted to do some derivative research on the topic) and while carefully parsing the paper we noticed that in the methods section the ratio “DNA:PBD = 1:20” is incorrect, it should have been “DNA:PBD = 20:1” or alternatively “PBD:DNA = 1:20”.
To give a little bit of context, in this experiment, we add the compound of interest to DNA and see what effect it has on the ability of DNA to unzip. A chemotherapeutic agent should make it more difficult for the DNA to unzip, and in theory, the stronger the agent, more difficult the DNA unzipping process is. In our experiment, we measured that sibiromycin is stronger than a reference compound (anthramycin) and in turn 9DS is stronger than sibiromycin. However, in the experiment you’re supposed to add less of the molecule (PBD) than DNA, because that is a more reasonable simulation of what happens when you treat a patient than dousing the cell with drug. According to the paper as written, I added more of the molecule (PBD) than DNA, which is not the right thing to do.
So the question is, is it a transcribing error, either when I wrote it down in my notebook, or when I transferred notebook to manuscript? or did we actually set up the experiment incorrectly?
Unfortunately, since none of the researchers are any longer affiliated with the Univerisity of Maryland, it doesn’t seem like a good idea for either of us to go there and engage in a possibly futile archaeological session to figure out where exactly the error occured. We entered into an informal agreement with the aforementioned researcher to redo the experiment (via indysci, as soon as it gets started), and were planning on reporting the error* then (and also the researcher is planning to publish a paper on the matter with our cooperation) but ACS Chemical Biology has contacted us and asked us to issue a correction sooner than later, so the correction has been put into motion, and I’m disclosing this now ahead of the correction.
So what does this mean? Although we don’t have any confidence in these data, they are more of a ‘supporting experiment’. The main gist of the paper (reporting the production of 9DS, confirming its structure via NMR, and the NCI-60 data) are untouched, and the data that are being stricken are a retrospective attempt to explain why 9DS is more powerful than sibiromycin in the still-valid NCI-60 data. Personally, I am not certain that changing the ratio will have a dramatic effect on the qualitative differences between the three compounds we tested. However, to be sure, we’ll redo the experiment.
In terms of indysci.org, I think it’s still reasonable to move forward with the experiments. I’ll be producing an extra batch of 9DS and sibiromycin to support the revision of these data and will pay for any unshared expenses out of my own pocket.
*in December, I attempted to report this via the wonderful Pubpeer, but my comment on my own paper appears to have been moderated into oblivion.