On Twitter

The Australian Bioinformatics Network was initiated by

ABACBS (pron.ˈabəkəs) runs the Australian Bioinformatics Network

Handy links...
Open Jobs

By type:

By location:

« Terry Speed: a pattern is emerging | Main | A new website that may help to you find software about NGS »

I've got issues! Take 1: AMATA 2013

As part of this year's excellent AMATA conference, we held a pre-conference workshop aimed largely at early career researchers.

As part of that workshop, I convened a discussion forum where (among other things) we asked participants and presenters to share their "burning issues".

Taking into account Elizabeth Kane's comments on how brainstorming can easily reduce the quantity and creativity of ideas, we asked everyone to write their issue on a PostIt note. That meant the responses would be (a) less prone to "group think" (b) somewhat anonymous, thereby avoiding "evaluation apprehension" a bit and (c) easy to re-organise and compile.

Here (with a bit of rearranging) is what we wound up with:

I posted this on the ABN Yammer to see if the issues resonated, and indeed they did! Here is what two ABN members had to say:

Comments from Nathan Watson-Haigh

Thanks for putting this together, it makes for interesting reading and reflects many of my own thoughts and experiences too. I wanted to comment on a few points in particular:

1) Cultural Issues. Unless you work in a bioinformatics group (of say >4 people) it can be hard to get across the idea that you are not IT, but "scientific computing". This extends to the expectation that there is some magic "black box" software already out there for analysing the data set you are given and all that is required is the press of a few buttons. People don't appreciate that writing robust software/code requires design and experimentation with different parameters; just like a wet lab experiment.

2) Multidisciplinary Breadth. As with all multi-disciplinary fields, you have to make a choice about where in the bioinformatics spectrum you would like to position yourself in terms of skills. Do you want to be focused more on the biology, infrastructure, stats or computer science side of bioinformatics. You may prefer to take up a "jack-of-all-trades, master of none" position and sit squarely in the middle. Each has its own advantages and disadvantages but understanding that it's not possible to be an expert in everything is probably a good place to start. After all, if you think you can be an expert in programming, an expert in statistics as well as an expert in biology probably means you underestimate what it takes for someone to be truly an expert in any one of those disciplines. I think collaborations here are key. Whether this is between members of the same group, between groups or institutions. It's important to have a solid group with sufficient overlap in skills/knowledge that the lines of communication remain open but are different enough to maximise impact. i.e. a solid team. Try to align yourself with local people that you can approach for help. Contact them by phone or arrange a 15-30min meeting rather than email - the latter takes more time to explain complicated ideas and is prone to misinterpretation/misunderstanding by novices. Simply talking expedites the knowledge transfer.

3) Irreproducible Results. People interested in reproducible research should try to follow the activities/comments of some of science's biggest open access/research/review advocates such as Titus Brown. On the programming side of things, I'd also encourage people to take a look at Software Carpentry, learn how to use version control (e.g. git) and think more about writing tests for their code. If you write R code for your analyses in R, take a look at Sweave or knitr for creating reproducible documents containing your code and figures. Warning, you'll also have to learn some LaTeX!

4) Publication Issues. Have a look at the Directory of Open Access Journals (DOAJ; and also our trusty friend Wikipedia ( for journals with hybrid and delayed open access journals too. If you're a budding OA advocate, you may also want to publish pre-prints of your papers and would also like to know what journals are OK with you doing this when it comes time to publish your papers in a traditional journal. For this I'd recommend…

I hope my musings are useful to someone :)

Comments from Terry Bertozzi

Hi Nathan,
I can certainly relate to the comments and scenarios you posted. I’d like to add to a couple from my own experiences:

Cultural issues. I think this also includes the expectation that you can provide the finished magic “black box” software solution so that others can simply pick it up and analyse their own data, without any concern for the heterogeneity of the data itself. In my work environment I have yet to see two experiments that have the same aims, let alone come from the same sequencing run, platform and sequencing provider.

Irreproducible Results. Something that I have found particularly frustrating is journal articles that state “ procedure x was carried out using an in-house script…” which of course is never included with the article. The authors may as well say “… which we multiplied by a random number and subtracted the number we first thought of…..” There is no way to repeat, assess or learn from the analysis. While we can influence this culture by implementing some of the suggestions you’ve made, I think that some of this falls squarely on the shoulders of journal reviewers and editors who should insist that there is sufficient information to make the study reproducible e.g. ALL code is published.

In closing

I want to thank the AMATA Worskshop participants, and Nathan and Terry for their candour.

Given the nature of bioinformatics, bioscience, and maybe just modern life in general, I think it is quite common to wonder "is it just me who feels this way?".

While there is often no easy solution to the issues and challenges we face, I find great comfort in knowing that I'm facing challenges alongside others.


Reader Comments

There are no comments for this journal entry. To create a new comment, use the form below.
Member Account Required
You must have a member account on this website in order to post comments. Log in to your account to enable posting.