01 November 2013

A scary specter of scientific research: When your paper has an error.

This week at work, I encountered something far more terrifying than the costumed ghouls wandering my neighborhood last night for Halloween.

My paper on transposable elements in Asparagales just came out in Genome as a part of a special issue on Genome Size Evolution. Apart from being the first bit of research from what is the main part of my research career these days, it's also the first manuscript on which I'm first author. Even scarier…I'm the only author. Suffice it to say that there is no one to blame but myself for any and all errors which occur in this piece of research and writing. Of course, just after the paper came out in print, I found some problems.

Let me preface this attempt at a self-effacing apology by saying that the NESCent journal club just read and discussed a few real humdingers of articles last week: first, the now-classic paper "Why Most Published Research Findings are False," and second, a recent article from The Economist describing the alarming inefficiency of science (more accurately, the scientific community) at self-correcting errors or misinterpretations. The two articles cited above strike fear into my scientific heart because they indicate our implemented standards for scientific research don't match our expectations as professionals.

This admonition brings me to my own sad confession: there are two problems with my paper that, while not deal breakers for any of my results or conclusions, are bothersome and irritating issues that may cause headaches for other researchers down the line. A good portion of my graduate training was spent reconciling with myself the knowledge that I would eventually be publishing research that was imperfect. It rankles me that those imperfections are going to persist beyond "lack of data" or "insufficient sampling," and perhaps affect priorities near and dear to my heart, like reproducibility and transferability to other systems.

The first issue involved me overlooking a small change while revising the manuscript. The genome size data in Table 1 is actually Mb/1C, instead of pg/1C. Oh, the humanity! Granted, the correction was made in Figure 2, and the error is obvious if you have any frame of reference about genome sizes in plants, but still…I'm quite irked at myself. The second issue is something for which I couldn't have anticipated an problem. The software I used for genome assembly, MSR-CA, has been modified and re-released since I sent back the proofs for the article. My paper includes the website from which I obtained the program, but that website is no longer functional. Moreover, the program name has changed (MaSuRCA), so trying to reproduce my methods would be a bit difficult. However, the paper describing the software is now available, so this will hopefully be less of an issue as I continue to publish.

I understand that these issues like these happen far too commonly in scientific publishing. I've read papers with legends for figures mixed up, obvious typos, and figures where you can't tell if the dot was a data point or printing error. I suppose I feel these errors in my own work more keenly because I am so early in my career. I have more at stake regarding their success, and I'm more invested in their survival and propagation through scientific literature. Are these worth contacting the journal to issue a correction? I'm not sure. It makes me feel a bit better, though, that in the meantime, this post might get picked up by someone trying to figure out what the hell I was talking about with that "MSR-CA" stuff...

No comments: