Bioinformatics and experimentation

I have been involved in a number of discussions recently about the proper role of bioinformatics in biomedical research.  A few themes emerge that bother me, and upon which I feel compelled to comment.  I will first take on this proposition:

“Bioinformaticians should do wet lab experiments.”

I disagree.  Experiments are hard and require as much training and expertise as algorithm design and implementation.  To suggest that they are getting “easy enough” for even bioinformaticians to perform misses the point.   It is a very rare person who will perform at the highest levels of technical virtuosity and innovation in both algorithm development and experiment development.   Why ask or expect a person who is great at computation, to add this credential?  My fear is that they will become a mediocre experimentalist and (as a side effect) their algorithmic work will also suffer, thus creating a “jack of all trades, master of none” disappointment.    Of course, bioinformatics algorithms and databases should often be tested by experiments, and this is why we have collaborations (or even subcontracts), but expecting young bioinformaticians to do it all is risking drowning out their real expertise with distractions.  It is also very critical that bioinformaticians understand the experiments that they analyze in great detail–otherwise risking irrelevant methods that don’t make valid assumptions about the data.   I could go on forever, but my advice to a talented young bioinformatician is  to stay away from pipettes until your career is really secure.  After that, you can do what you want, but recognize whether you are doing something because it is an ego trip or because it is the most effective way to get something done.   Or perhaps you don’t want to share credit, and so want to bring everything in-house under your own control.  That’s a whole ‘nother discussion…

As a footnote, I have recently been convinced (by some students) that attending biological group meetings is probably the best way to fully understand the data and how biologists think about it without having to actually do the experiments.  That will require earning the trust of the biologists to let them sit with you and speak openly about their data and its shortcomings.   But it is probably the single best way to understand a biological experiment.


  1. Thank you Russ for being a voice of reason 🙂 My argument against doing wet-lab experiments myself has always been that it is better that I spend my time on what I am good at (bioinformatics) and leave the wet-lab work to those who are good at that.

  2. Hmm, I have to disagree a bit, from personal experience.
    I’m a computational biologist with my first degree being computer science and engineering, my second a Ph.D. in molecular biology.

    I didn’t have any wet-lab experience when I started my doctorate, but did spend 2 painful years at the bench during it.

    You’re right, I’m a mediocre wet-lab scientist, and I don’t do lab work anymore, I leave it to those with better hands.

    But it was not a waste of my time. I understand the basic limitations and caveats of the wet-bench techniques in a way that would have been very hard without that experience.

    As a result, I can better evaluate to true biases of the data presented to me for computational and algorithmic analysis. This is critical, because bad assumptions about accuracy can invalidate such analyses.

    Some comp. biologists may be able to learn these pitfalls without the work at the bench, but for most, I would advocate some time and pain, because it can pay off.

  3. Russ,

    Great blog! I agree that Bioinformaticists don’t need to be expert biologists but I do think they need to have some minimal wet lab experience so they know where data come from and how easy it is to generate bad data. This experience could come from a 2-3 month wet lab rotation during graduate school, for example. The ability to doubt data is an acquired skill that only comes from wet lab experience. Jason

  4. Russ,
    I agree with you to a certain extent. However, a significant number of bioinformatics papers today have not been validated in the wet lab (I feel a good number will fail). Bioinformatics looks like a mere publishing industry and good work is drowned in too. A level of wet lab experience might help.

  5. Jason, I strongly disagree with you that the ability to doubt data is an acquired skill that only comes from wet lab experience. I do have some wet lab experience from my M.Sc. degree (I studied chemistry), though, and fully agree that it is important to have an understanding of where the data comes from.

    I would claim that my ability to doubt data came from the hours spent in front of the computer, reanalyzing other people’s published data (in particular data from microarrays and protein interaction screens). That is where I learned to always question the quality of the data (and the wet-lab biologists’ interpretations of them), irrespective of whether it has been through peer review or not.

  6. Very nice post, but considering the fact that most of academic position prefer people with wet lab training with experience in dry lab don’t you think it will be not good for our careers. what we need is change in mindset that we don’t need highly trained bioinformatics people

  7. Experiments are hard and require as much training and expertise as algorithm design and implementation. To suggest that they are getting “easy enough” for even bioinformaticians to perform misses the point.

    Rather than missing the point, isn’t this the point exactly? As Tara above says, you may remain a mediocre wet-lab scientist, but you will almost certainly, in my experience, be a better ‘dry’-lab one.

  8. I think it is overkill to do the experiments. Sure, it will give you this expertise, but you can find out what you really need to know much more efficiently (e.g. group meetings).

  9. Generally I agree that bioinformaticians don’t need to work in the lab to be successful… However, I do think there is a world of benefit from not only attending group meetings with the folks who you collaborate with- but also to spend a day or so following them in the lab. Take notes on the process, ask questions, get a sense of what your data producers are up to. At the very least you have a better foundation to communicate with the lab folks, and you’ll probably understand the data you end up receiving a little better.

  10. Re validation, some would say that this sword cuts both ways. That is, if results from bioinformatics need wet lab confirmations, results from the wet lab also need bioinformatics confirmation…

  11. I don’t know if I agree with this clear distinction between wet/dry lab. I am a biochemist and did a Phd in bioinformatics working in a wet lab. I am doing a postdoc again in a lab were most people are doing experimental work.
    I did some (not a lot) of bench work during the PhD and from my experience learning a method in the lab is not that different from learning a computational approach. The biggest difference is that there is much more online support for the computational stuff while for the wet lab the best/only way to learn is seeing someone else doing it.

    It seams to me that you are mostly arguing for specialization and wet/dry is just an easy and clear line of distinction for this argument. You could as easily replace wet/dry with some subfields within bioinformatics and the arguments would still appear logic. I think that learning a wet lab method (ex how to use a plate reader to measure fluorescence) does not take more effort than learning a computational approach (ex. how to use a machine learning method). So the question is not so much in wet vs dry lab but more the time it takes to pick up different orthogonal skills.

  12. I agree on Pedro Beltraos comments. This is more about spezialisation.
    I think we need both pure bioinformaticians and “half-breeds” 🙂 and it’s not always it’s half wet-lab, it can as well be half matematics, physist, statistics, genetics, or whatever!
    One of the major contribution of a half-breed is communication. It is not trivial to collaborate across departments and many collaborations fail due to different “languages”.
    Don’t learn things for its own sake, learn because you are interested and do what you are good at.

  13. Maybe because my basic degree is in molecular biology and did a wet-bench research project before, I agree with Pedro that it is not hard to pick up biological experimental techniques and understand the principle/concept behind them. However, I do not think it is easy to pick up computational knowledge. It is one thing to use a tool, it is totally another matter to understand how they actually work, their assumption and limitations. The limitation and caveats of many experimental techniques can be obvious but many computational methods can be based rigorous theorems and subtle assumptions that are not immediately apprehension. I guess as a bioinformatician, the wet-lab experience is most rewarding as holistic understanding of how data are derived. Beyond some time, the law of diminishing return kicks in and time thereafter, I think, is better spent on understanding your computational techniques further or learn a new one.

  14. Given that you’ve made another blog post, stating that bioinformatics is NOT computational biology, I agree that “wet bench” work is not needed or desired for a bioinformatics programmer. Good knowledge of biology (what is a gene? how do genes and proteins evolve? what is a population? how do we sample from populations? etc), and more importantly the ability and willingness to go get a textbook and learn more if needed for the project you are working on, is important.

    This is equally true of the “wet bench” biologist who wants to use the computational tools. They should know something about the tool, and not just put data into a black box.

    All too often, I see teams of authors where some have done the computational work without really understanding the biology, and others generated the data to give to the computational people without understanding the analytical methods. The end result might be something like using software for studying the evolution of a population. The computational tool is designed for diploid, sexually reproducing, well mixed populations; but it get used to study an organism such as a virus, which is not diploid, does not reproduce sexually, and defining the “population” and it’s mixing can be problematic.

  15. Hi there, nice discussion.
    I am biologist by training but have been working as bioinformatitian for many years now.
    The biggest problem I’ve found is that dry-scientists trust basically everything that it is peer-reviewed (i.e.: early protein-protein interaction data based on Y2H) published. Then, when dealing with the data, they are not capable to decide which experiment is worthing to use or not “it is good because it is a nature paper”, which brings another discussion about Impact factor et al (not going there…).
    This is risky, because then wet scientists will use whatever we generate based on wrong data.
    So, in the end-> Catch 22.
    My point is if somebody is generating hypothesis about biological questions, she or he should know about the Biology, and the experiments available to address the question.
    It doesn’t imply one should do wet science.
    But I strongly recommend dry-scientists to do very simple experiments in the lab in order to get a notion how hard wet science can be. 😉
    You can ask Andreu Alibes, a pure comp. scientist at CRG doing some experiments for the sake of it. You’ll be surprised about his views.

  16. I see your point. In fact, I’m in that very situation. Honestly, I’m not sure if I agree with you or not.

    Here is my take on Bioinformatics: it is a tool of the trade, rather than a separate discipline. It acts as the bridge across a huge body of water, connecting two separate yet dependent continents. While I am certain this statement angers more than a few, this is true for nearly all cross-disciplinary areas of research. Most are not separate areas of study, though they may seem like it due to the level of interrelatedness of the fields, but rather taking one data set and looking at it through a different perspective. After all, perception is all in the eyes of the beholder.

    However you look at it, Bioinformatics remains a new and incredibly diverse field — most programs, books, papers, etc. on the subject vary so greatly in the level on content there is no way to say “Bioinformaticians should do “.

    Currently, I am a PhD student in Bioinformatics with a heavy interest in Immunology and Virology. Unfortunately, this field currently sees Bioinformatics as number crunchers. Only one lab would seriously look at me without considerable wet lab experience. In fact, although most of the Bioinformatics labs have collaborations with wet labs for their experiments, the only-dry labs are rare.

    In other words, I certainly agree with the need for a specialty, yet in a cross-disciplinary field there is no substitute for learning the ins-and-outs than to actually do the work. Each situation is different and depends entirely on where and what work your do. These type of imperative statements are the ones always eaten in the end.

  17. To Stephen, I too was interested virology/immunology lab and I too had exactly the same sentiment, except that I have gone one step further down the process. After finishing my graduate degree in Bioinformatics from a pure ‘dry’ lab, I am doing my post-doc in a pure ‘wet’ virology lab with the intention to better equip myself for future career. The learning process was fruitful, and I have deeper appreciation of the difficulty that experimental biologists face. However, I have come to exactly the same conclusion as Russ:

    “Experiments are hard and require as much training and expertise as algorithm design and implementation. ”

    Granted that following established molecular biology standard protocols to generate data may not be complicated, yet proper protocol establishment and trouble-shot require keen intuition that can only come with prolong immersion into experimental work. This process would easily translate into several additional years of post-doctoral training for young bioinformaticians before they accrue enough credential to be considered for faculty position.

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 )

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