Posted by: Kay at Suicyte | May 11, 2007

Collaborations in bioinformatics

Recently, PLoS Computational Biology published an editorial entitled “Ten Simple Rules for a Successful Collaboration”. Chris at Fourth Floor Studio wrote a comment on that piece, adding two more rules to the list. The topic of scientific collaborations is very dear to my heart, as >90% of my work involve joint projects between bioinformaticians (us) and wet lab people (them).

I have to admit that I was somewhat disappointed by the PloS editorial, as it mostly states the obvious. It is hard to imagine that anybody would seriously disagree with their 10 rules, except maybe for difficulties with implementing some of them in a real-world setting. I would have preferred if the authors had taken position on some of the more controversial issues of scientific collaborations, e.g. assignment of authorship.

I completely agree with Chris’ first additional rule “don’t wait; find a suitable problem and offer it to potential collaborators“, this is something I try to follow all of the time. At least in my area of research, it works surprisingly well and I can only recommend to consider this rule as soon as you have started your own group.

I am not so sure about the second new rule “beware of farming-out“. If I understand this advice correctly, Chris suggests that a collaboration should be on equal terms. During last years collaborations, I have made the experience that the more equal collaborations were also the more difficult ones. I have seen more than one fruitful collaboration end in a catfight about who’s going to be the first and last author. More on authorship in a moment.

I think that most collaborations are unequal by design. Typical collaborations between wet lab groups are started because the group initiating the project needs a certain expertise or technique that is not available in their lab. Typically, the initiating group stays in the lead – this is not necessary connected to seniority or even hierarchy, but rather to who had the original idea and who is doing most of the work. These collaborations are the least problematic ones, unless one of the involved parties starts acting up. More equal, and maybe more problematic are collaborations that have been planned from the onset, e.g. when two groups team up to study a certain phenomenon in two different model organisms.

Collaborations between pure bioinformatics groups are rare. I never tried one myself, so I cannot really comment. What I am doing most of the time are collaborations with wet lab groups. Our group’s expertise is mainly in sequence analysis, and as you might imagine this is a welcome addition to many experimental projects. Theses collaborations come in various shapes and sizes, but there are two recurring scenarios:

  1. The wet lab people identify one or more new proteins/genes and ask you to do some analysis, e.g. find some common properties, make a functional prediction, something along these lines. In some cases, your results just add some icing to the final publication, let’s call this scenario 1a. In other, more interesting cases, your analysis returns a testable prediction, which then leads to another series of experiments. This is scenario 1b.
  2. By doing some bioinformatical work in your own area of interest, you make an interesting discovery – maybey some kind of prediction that can be tested experimentally. In scenario 2a, you could publish this prediction as such, without any collaboration. You could also go for option 2b, which means contacting experimental groups, convincing them that your predictions have a good chance of being correct, and suggesting to them a number of experiments that should be done.

Now, let us have a look at the likely outcome of these scenarios – assuming everything works according to schedule. In scenario 1, there is little doubt on who is leading the collaboration – the lab guys. Irrespective of having scenario 1a or 1b, the outcome will be a joint publication with the author order “lab rat 1, lab rat 2, lab rat 3, bioinformatics guy, one or more wet lab PIs“. In my opinion, this is acceptable but one should take care that the wet lab people do not demote the bioinformaticians to the acknowledgment section.

In case 2b, where the initial idea comes from the bioinformaticians, the situation is less clear. In an ideal world, you would expect that the inventive bioinformatics group takes the lead and eventually claims senior authorship on the resulting publication. Unfortunately, I found out that it doesn’t work like this. What you typically get is exactly the same as in scenario 1: a paper with the bioinformatician sandwiched between the wet lab students/postdocs and the wet lab PIs. Obviously, you could try to argue. Most likely, you will get the reply that the wet lab part took much more time, manpower, resources and money than the bioinformatics part. Often, this is true. Another answer you might get is a reminder that only their (highly sophisticated) wet lab work has given you the chance to publish your work in a high-impact journal, instead of dumping it to the Journal of Inferior Bioinformatics Research. This might be true as well.

I am not convinced that this outcome is really fair – I think there should be some visible difference between a 1a and a 2b scenario. Unfortunately, there is not much that can be done. If you don’t agree with their terms, most experimentalists are not even interested in collaborating with you – why should they invest a significant amount of resources in a project that doesn’t even get them a senior-author paper? In the end, the wet lab people always tend to have the upper hand – unless you fancy going for JIBR.

If any of you have similar or different experiences, or maybe a piece of good advice, I would love to hear about it!



  1. For the farming out rule, I was thinking along the lines of extremely unequal collaboration, when you are virtually reduced to technician status along the lines of your scenario 2b.
    Many of the more dominant/established/grasping people you may come into contact with don’t seem to be able to resist getting something for nothing, and resisting the gravity of a larger, more established group is hard. But that’s what you have to do if you want to build a reputation as a researcher in your own right.
    That’s not to say that you should avoid collaboration – far from it. But picking collaborators who are also partners and mentors (due to either seniority or expertise) will have very different outcomes to getting involved with someone who sees you as cheap labour.

  2. The ability to make the contribution of each collaborator explicit is an important benefit of carrying out Open Science using tools like wikis. Even if and when a paper does come out from the collaboration, links back to the experiment pages will show who knew and did what when.
    For example, our own synthetic organic chemistry group has been looking for bioinformatics collaborators to help us with docking and we have made some very helpful connections:

  3. First of all, I am a computer scientist. I had a rather “interesting experience” working with medical professionals. Let just say that at the end of the day, I decide to continue my future bioinformatics work by myself. The only way that any bioinformaticians to be really visible to the academic world is to test your ideas on simulated data and public-domain data. I manage to do that in my recent article in BMC Bioinformatics 2009, 10: 294. And yes, this time the lab rat’s name is behind a computer scientist’s name. Moreover, the head of wet lab is not the last author either.
    P.S. I really enjoy writing the “authors’ contributions” part in the article.

  4. Why not train yourself to do experiments? If you know how to do wet lab, you can do it all yourself and be the first author after if you have made some computational predictions.

    • Yes, this is clearly a possibility. However, many people in bioinformatics don’t have a wet lab at their disposition.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: