Thursday, August 12, 2010

Where I disagree with Drugmonkey ...

... and I'm sure he's devastated. However, he's just plain wrong in celebrating the demise of Supplementary Material from the Journal of Neuroscience. I say as much in his comments section, but I'm going to start a discussion (for the two or three people who read this blog) here anyways. It's my perogative!

At any rate, Bobby Brown aside, here is what I had to say to his celebrating.
I seriously hope that not all journals go this route, and I’ll tell you why . As part of my postdoc, we did an MLST project on a common pathogen in the food industry. We managed to get strains from all over the world, from state and federal agencies, and from international organizations. In the end, the project had several hundred isolates. Where were we supposed to list them and all the supporting data (date collected, place collected, food it was isolated from, serotype, etc etc)? In the article itself? It would have taken a dozen or so pages just for that one table! And that’s not including the several phylogenetic trees we generated all of which took up an entire page. It would have been half the issue if we had incorporated it all into the text. It seemed that that was a PERFECT use of the supplementary material section. Instead of the editors taking proper control of their reviewers and authors, they’ve banished the practice altogether … bye bye baby, but at least we don’t have to worry about that bathwater any longer.
I'm not the only one who thought that there were good uses for Supplementary Material, and I do think that my anecdote is one good example of why Supplementary Material needs to remain as an option. In an age where we can generate millions of base pairs of data sequence in a few short hours, but where journal issues are not increasing accordingly in size, something has to give. I know there are repositories for most data of this sort (such as GenBank for sequence and GEO for microarray data), but I think we still need the Supplemental Material option for certain descriptive elements of a manuscript.

I agree with J. Neurosci when they said that additional experiments added as Supplemental Material were allowing things to get out of hand, but as I said in my comment ... instead of cracking down on such practices, they just threw the baby out with the bathwater. If all journals take this route, things are going to get a bit hairy. I know a couple of my manuscripts, where I provide background information that would dominate a manuscript's page total if I could not include it as Supplemental Material, is descriptive in nature, and is provided as a courtesy to the reader* would have a hard time finding a place to published. So, why is it too much to ask that journals, and their editors, spend a bit more time sorting the issue out, rather than torpedoing it wholesale? I bet part of the issue is money ... it's probably no small fee to host that data in perpetuity.

*It also saves me countless emails being asked for that last bit of information which some people might hope to glean some ideas from for their own data sets, but which for me didn't factor at all in my own study.

10 comments:

Anonymous said...

Same for mathematical models... Do you think it is possible to provide most of the equations of a mathematical model in a Journal of Neuroscience paper given the very strict size limit of the articles. Maybe they don't want mathematical models anymore...

Anonymous said...

Why don't you just put all that stuff on your web page or your lab's web page? That's what we do in computer science, and believe me, we are able to generate far more data than what you're talking about. None of our journals have "supplementary material" though (at least not that I've ever heard of).

Thomas Joseph said...

Anonymous number 2: that is a bad BAD idea. I have seen that happen in bioinformatic journals. A paper that detailed a new program I wanted to test which had it supposedly on their uni website had a permanent "Page Not Found" once the 1st author moved on ... Three months after the paper was printed. Took me forever to hunt the guy down. That should never happen and if it were on the journals website it wouldn't have.

Thomas Joseph said...

Anonymous number 1: I honestly don't know what some journals want anymore. I suspect they want to save money and that is their driving factor.

Nat Blair said...

Did you even read Maunsell's explanation? Doesn't seem like it. You haven't addressed any of the reasons they rejected other approaches, and he did specifically state that it wasn't expensive to host the info (0.4 MB of supplement avg per paper).

Anony 1: The only stated space limits in J Neurosci are word limits of the abstract, intro, and discussion. So yes, you could put all the required eqns in the Methods, or Results.

Thomas Joseph said...

Nat: Yes, I did read Maunsell's explanation. It's horrid. Essentially what J. Neuro is going to do is force the authors to host the information on their own website (which Anon 2 and I discussed). Which means, when they leave that institution ... that data will be lost FOREVER. That is simply brilliant! Ok, not exactly brilliant as it is completely asinine.

The best fix to the issue is to allow descriptive supplementary material*, and ask the editors to police the reviewers demands. I discuss that in the blog entry, and it seems pretty reasonable to me.

I think my comments are reasonable and appropriate, whether I read his lame justification or not (and I did).

*Why the journal can't limit the TYPE of material submitted went unaddressed.

Nat Blair said...

Nat: Essentially what J. Neuro is going to do is force the authors to host the information on their own website

No, what they will force authors to do is to include all important data in the paper. Anything that's not important enough to include in the paper, isn't worth worrying about whether the authors keep it available. The paper will stand on its own.

The best fix to the issue is to allow descriptive supplementary material*, and ask the editors to police the reviewers demands.

Why not put the descriptive material in the actual paper? In places like "Materials and Methods" or maybe, the "Results" section? You haven't addressed why this isn't a solution.

Keeping a supplement that is peer reviewed maintains the work of determining what can be put into that onto the already busy editors. And when the editors can't make this determination, who do you think they ask? The reviewers. Which again gets to his point that the reviewers already have too much to review, and don't always adequately look over the supplementary material.

Secondly, this doesn't address the potential for reviewers to continue to ask that new experiments be added to the paper. You cannot expect the editors to police that, because to do so requires that the editors actually reviews the paper herself. Again, it doesn't change the workload.

*Why the journal can't limit the TYPE of material submitted went unaddressed.

As I said, to determine this requires that the material actually be reviewed by somebody. Thus negating one aspect of the benefit of doing away with Supplementary Material.

Thomas Joseph said...

Why not put the descriptive material in the actual paper? In places like "Materials and Methods" or maybe, the "Results" section? You haven't addressed why this isn't a solution.

I already explained why this isn't a solution in a number of cases. In microbiology, if you have a strain table which runs a couple of hundred lines, it's going to take up several pages of text. Many journals have page limits, or force the authors to page exorbitant fees. Including it as Supp Material is the best place to put these descriptive data sets. If my strain table runs a dozen lines, I put it in the M&M. When it jacks my printed manuscript to 30+ pages (that's journal pages, not my Word document) I'd be a fool to think it'd be included in the main text.

If I host it on my own website, the minute I leave the institution and the website is removed, all the phylogenetic trees in the body of the manuscript are worthless because now no one knows what the strain is, where it came from, what it was isolated from, its serotype, etc etc.

Their "all or nothing" approach was the wrong move IMO. And if the editor can't tell a strain table (descriptive) from a gel (experiment), he or she shouldn't be an editor.

Nat Blair said...

Well, that kind of tabulated material is pretty rare to see in a neuroscience related paper, and isn't related to the various issues surrounding supplementary material that are being addressed. I'm not suggesting that this approach is appropriate for all disciplines and all journals.

Maybe in certain fields it would be trivial to have the editor review the proposed supplementary material/appendix to see if it were a type that is permissible.

But that's quite a different suggestion that what I took from your initial post, which was keep supplementary material as it is, and expect the editors to do a better job weeding inappropriate things out.

Thomas Joseph said...

I imagine it'd be pretty rare in the neuroscience field, and admittedly depending on the study, it can be rare in microbiology as well ... though if the researchers really care about their MLST (multi-locus sequence typing) results, they'll include as many non-redundant strains as they can get their hands on.

My original comments on DM's site cautioned against other journals from taking the approach that J. Neuro took. I honestly think that a good set of guidelines for Supp. Material would work and would take minimal effort on the part of Editors and Reviewers ... especially if J. Neuro is already peer-reviewing Supp. Material, the workload (if you had this guideline in place) would drop.

I agree with J. Neuro that Supp. Material could be abused, especially when the reviewers ask for more experiments, and sympathize with every author who has had to undergo such treatment (probably because I have been there myself ... but successfully argued against needing to do so).

If I was not clear that I take the Descriptive Data Not Experiments approach to Supp. Material, I apologize ... though I thought I did so in the blog entry and DM's comments. I don't expect journals (even J. Neuro) to keep things as is ... I just think they went too far in their "solution" which in the long run IMO, will result in making a mess of things a few years down the line.