2013-09-20

Interview with a PeerJ author (me)

This summer I submitted a paper to the innovative new open access journal PeerJ, where it was published this week (Cock et al. 2013). I decided to write up the experience in the style of the PeerJ's Interview with an Author blog posts. I've copied the questions they normally ask, and written up my own replies - other than some rough edges in their current submission system it was all good.


Update: This has been reposted on the official PeerJ blog, with responses which I have inserted below.
PeerJ: Tell us a bit about the research you published, and why it is interesting?

Peter: This work isn't directly about biological research, it is some software engineering work designed to make some of the plant pathogen genetic analysis work we do at the James Hutton Institute easier, and more reproducible.

The falling costs of generating sequence data mean we have more and more genetic information available, but over the same time period the man-power available for its analysis has remained fairly static. Each new generation of biologists gets more computational and programming training, but for now bioinformaticians are still in short supply.

In our paper we describe how we've used the web-based Galaxy platform ("Data intensive biology for everyone") to make tools of particular interest for analysing plant pathogen sequences available via a web-browser, and used this system to construct reusable workflows or sample data analysis pipelines which can be shared between researchers. This makes it easier to repeat existing protocols on new datasets, and it is possible to create new workflows from the existing tools collection without hands on programming. Our primary goal is to empower more biologists to perform their own sequence analysis, in particular plant pathologists - but actually many of the tools are extremely general.

Another part of what I wanted to do in this paper was encourage researchers in this area to include data analysis workflows with future papers in a reusable form - not just as an outline text description in the Materials & Methods.

PeerJ: With this research, why did you choose to publish in PeerJ rather than some other venue?

Peter: This work is about scientific software, which we're releasing as Open Source Software (OSS), and as a matter of principle I wanted the paper to be published as Open Access. We don't yet have a formal policy on this where I work, but the principle that any tax-payer funded research ought to be freely available to the public seems a good one. (The James Hutton Institute is largely funded by the Scottish Government.)

Also, I've often run into journal pay-walls at home or where our Institute does not have a journal subscription preventing me from reading other people's work - and I want people to be able to read my work.

One option was to write an "Application Note" for OUP Bioinformatics, but these two page articles don't give much room for any details, so I ruled out that journal. OUP Briefings in Bioinformatics could have been a better choice, even though we don't have a subscription to it, but for both these journals OUP charge a hefty open access fee of £1750/ $2800 / €2275.

Instead, the first drafts of the manuscript were written with PLOS ONE in mind, which would have charged us a more reasonable open access publication fee of $1350, less than half OUP.

While the manuscript progressed, PeerJ was launched by some of the same team behind PLOS ONE. It had a similar ethos, more open peer review, and an interesting and far more affordable business model. It would cost us just $99 per author, making just $396 in total - and that wouldn't just cover this paper but a lifetime PeerJ basic membership.

In addition to being cheaper, PeerJ was also promising a quick turn around, open peer review, and seemed to be getting a lot of good press and doing a great job promoting papers published there. It seemed worth investigating, so I emailed PeerJ to see if they'd take software articles. The reply was yes, formal guidelines were pending, but basically would follow those at PLOS ONE.

PeerJ: How did you first hear about PeerJ?

Peter: I'm pretty sure it was on Twitter, see @thePeerJ (I'm @pjacock).

PeerJ: And what were your initial impressions (i.e. before you submitted)?

Peter: PeerJ sounded like PLOS ONE with the good things like Open Access and not caring about "impact" for reviewing, but with an innovative funding model making it far cheaper for authors. And PeerJ had an intriguing novel far more open approach to peer review as well.

PeerJ: What persuaded you to submit to us?

Peter: It came down to a choice between PLOS ONE and PeerJ, and both journals are Open Access with similar scope and review policies. PeerJ clinched it by being more cost effective, and I was curious to try their non-anonymous peer review system.

PeerJ: Were the author instructions and policies clear to you?

Peter: Not entirely, no. In particular the "Aims & Scope" didn't talk about software focused papers but I was reassured after I enquired by email.

PeerJ: What was your experience of the submission process?

Peter: I was actually disappointed, in part from reading glowing praise from other authors. Mostly this was down to how the PeerJ system currently supports LaTeX and figures, which falls well short of the usability standard set by arXiv.org (the preprint server popular in Mathematics and Physics).

I prefer to write scientific papers using LaTex, a typesetting language very widely used in Mathematics and Physics (which I learnt as an undergraduate), and moderately popular with bioinformaticians. I don't usually have that many equations in my work (which is where LaTeX really shines), but I like how it handles cross-referencing and the bibliography. LaTex is also great for use under source code control, and we used a (private) Bitbucket.org git repository to manage group editing of this manuscript.

I appreciate that LaTeX will not suit most biologists, but nevertheless one of the criteria I use when selecting a journal is if it accepts LaTeX for submissions. Most of the OUP journals supports LaTeX and provide templates (including OUP Bioinformatics) and likewise PLOS ONE have clear guidelines and a LaTeX template. While PeerJ does accept LaTeX documents, I found their current system a bit clunky (Summer 2013).

First of all, I'd like them to provide a recommended LaTeX template, including a BibTex style to get the references in the preferred PeerJ style (this will save work for the production team too), and something like \usepackage[mathlines]{lineno} to include line numbers for the review process.

Separately, I found it frustrating that the PeerJ online submission system requires the paper's title, abstract, and the figures & table and their captions to all be entered separately from (and not duplicated in) the main document. For my first attempt at uploading I had to copy and paste these bits from which was error prone (e.g. fixing quotation marks and italics for species names), and comment them out in the LaTeX document - which breaks automated figure and table cross referencing.

I hope that one day PeerJ will have arXiv-like LaTeX support, but with a modern web interface.
My compromise solution was to comment out the title and abstract in my LaTeX document, but include the figures and table (and upload the image files as supplementary figures).

PeerJ's system insists on EPS (Encapsulated Post Script) rather than the more mainstream PDF format for vector images used as primary figures. I use Mac OS X which makes producing PDF images really easy - but converting them into EPS gave me bloated files and produced a degraded image in the composite PDF prepared by the PeerJ system. I'm pretty sure most submitting authors would agree, providing PDF images is far easier than as EPS (people have complained about this for PLOS ONE too, e.g. here). Happily the production team could work with my PDF figure instead (which I could upload as a supplementary figure).

The other image surprise was the online submission system enforced minimum image sizes on PNG files for primary figures, where-as the guidelines (at the time) implied these were just recommendations. Taking web-browser screenshots big enough to pass these limitations was quite fiddly (requiring a zoomed browser window spanning two high resolution monitors).

Some parts of the meta-data entry were also unnecessarily difficult, for example when entering suggested reviewers (something which I regard as a questionable practise) it was not possible edit a suggestion - you had to delete them and re-add them instead. Also selecting possible Academic Editors was hindered by the fact you can only filter on one topic area at a time - it would have been useful to be able to search for editors with expertise in (for example) both Bioinformatics and Plant Science. Also there was no way to move an uploaded figure from the primary files to secondary files section.

Overall I didn't find the PeerJ submission system significantly better or worse than last time I submitted a paper elsewhere. On the plus side, the PeerJ team were prompt and helpful via email (and on Twitter), and did a great job with the final layout.

PeerJ Note: Thank you for this detailed feedback, as we already commented on your original posting, we are taking this feedback into account as we further develop our functionality! Also, it is worth noting that only a small minority of our authors currently use LaTeX.

PeerJ: And what was your experience of the review process?

Peter: Excellent. We had constructive comments from our Academic Editor and two named reviewers back in just two and a half weeks, and once we'd returned the manuscript the Editor's acceptance was also very prompt. You can see the reviews and our replies online.

PeerJ: What did you think to the overall speed of the process? And what about the appearance of the published article?

Peter: The overall speed was exemplary - I think the longest stretch of time on the PeerJ system was from initially starting the submission to actually clicking submit (trying to submit a multi-author paper during summer conference season is a bad plan). In fact this took us so long PeerJ co-founder Pete Binfield even emailed me to see if I needed any help :)

PeerJ: Was there anything that surprised you with your overall experience?

Peter: Yes, I was hoping that after submission I could opt into releasing the paper immediately as a PeerJ pre-print. It seems only the reverse route is possible, publish a pre-print, and then submit that for formal publication with PeerJ.

PeerJ Note: We do encourage authors to put their journal submissions into the preprint server (as has happened, for example with https://peerj.com/preprints/54/). At the moment, staff facilitate the transfer on the behalf of authors, but ‘one click’ functionality is planned for the future

PeerJ: And of course, would you submit again?

Peter: Yes - and at my current membership level I can publish one manuscript per year with PeerJ at no extra cost, which is a nice bonus.

2 comments:

  1. Thanks for the "author interview," Peter!

    Sorry for the troubles of using LaTeX with the submission system, it is definitely one of things we have been slower to refine compared to the more dominant Word doc/docx use cases. You make some good suggestions that I hadn't seen before, so will add those to our roadmap.

    Congratulations on your publication by the way.

    Jason - PeerJ

    ReplyDelete
    Replies
    1. Thank you - and I do understand the more dominant Word doc/docx use cases deserve priority :)

      Delete