Soul searching in bioinformatics

Photo by mychatham on Flickr

I empathize with bioinformatics. We were born around the same time, and our parents and grandparents – computer science, statistics, molecular biology, and mathematics, in the case of bioinformatics – had been around for a while. We enjoyed fairly idyllic childhoods, well-supported and full of promise. The world was our oyster; there was nothing we couldn’t do. And yet here we both are in a quarter-life crisis.

In my case, the problem is colored by what is known as “impostor syndrome,” a way of thinking which is apparently quite common in people who by most standards should not think that way. Half the time I believe that anything is possible – I am smart enough, skilled enough, strong enough, creative enough. The other half I am plagued by feelings of inadequacy and believe it is only a matter of time before I am “found out.”

Recently I’ve come to realize that bioinformatics suffers from its own case of impostor syndrome. It knows it has potential, it knows it has unique talents, and most of the time it pushes forward confidently, but deep down there is a fundamental insecurity about its place in the world. The insecurity isn’t completely unfounded; bioinformatics often has to defend itself to those who would judge it. “Tell us why bioinformatics should be its own field,” they say. Its critics are unconvinced of its ability to stand on its own, scornful of its promises, and quick to demonize its missteps.

Photo by strangeones on Flickr

Small wonder, then, that bioinformatics as a whole has developed a sort of complex. It’s not quite sure what it is yet or where it’s going, at least not to the satisfaction of others. My awareness of bioinformatics’ troubled psyche came to a head during a panel discussion at the Pacific Symposium on Bioncomputing, where the conversation quickly turned to the place of bioinformatics in the existing academic establishment. How do we get biologists to collaborate with us? What is the value of informatics outside of the biological findings? What can we do to increase people’s respect for bioinformatics?

The responses from the panel were interesting and varied. “Biology is about getting new data, and bioinformatics is about getting new science out of that data,” quoted one. The informatics is just as important as the biology, insisted another, because the methods can have broad applicability. We should see a promising method through to utility just as we would see promising new biology through, another agreed. Sure, informatics is often relegated to a supporting role in most published studies, but there are plenty of cases where the informatics made the paper possible. Panelists brought up Chip Lawrence’s seminal paper on Gibbs Sampling, published in Science, as an example of pure bioinformatics getting prominent notice.

But dost thou protest too much? An audience member pointedly summarized the mental zeitgeist thus:

At biology conferences, discussions and panels revolve around experimental protocols and the results, but at bioinformatics conferences, they tend to revolve around questions of how we can [define] our science better – why?

There was a heavy pause as the words reverberated through the room and the grains of truth settled. It was as if we were looking through a mirror and really seeing ourselves for the first time. Are we so unsure of ourselves as a field that we obsess over how to get our work done and how to get it appreciated by others? Do we still not know who we are?

Photo by ildanish on Flickr

Photo by ildanish on Flickr

The panel offered up some palliative comments. The problem is the lack of true bioinformatics departments, opined one, for we are not being judged by our peers but by others’ standards which may not be appropriate. Yes, continued another, we have to defend what we’re doing because not everyone values it. Yet another admonished: the boundaries are constantly being redefined, and if you want something “safe”, you’re in the wrong place. So be strong, O conflicted warrior, for bioinformatics is not for the faint of heart.

Perhaps the room was too large or the audience too drained; either way, there was too little time left to come to a satisfying conclusion to the Big Questions. So what’s in store for bioinformatics? Will it gain respect as a field, or slowly fade? Lincoln Stein, who 5 years ago predicted the demise of bioinformatics, now says it is alive and kicking, but it wasn’t a complete about face. What he claims is that the practice of bioinformatics will be alive and well, but bioinformatics as an area of study will be obsolete.

In some ways, I agree with Stein – it’s difficult to classify bioinformatics as an area of study if what you value is declarative knowledge, knowledge about facts. But bioinformatics is as much about the process as it is about the end discoveries, and the inclusion of method makes it a square peg in a world of round holes. As long as exclusively round holes persist, and its emphasis on method persists, then bioinformatics (and necessarily all emerging sister disciplines of informatics) will never find the right fit. How might the metaphysical journey end? Either the holes will change, or bioinformatics will change, or both. Which way is better is probably another one of those Big Questions.

Photo by stuckincustoms on Flickr

Photo by stuckincustoms on Flickr

13 Responses to Soul searching in bioinformatics

  1. nsaunders says:

    Very nice discussion of a topic that troubles many of us. For now I’ll just say what I said at FriendFeed – it’s not you, it’s them :-)

  2. Deepak says:

    What Neil said.

  3. Iddo says:

    Here’s my take on this: as long as there are masses of data to analyze, there need people dedicated to specifically analyze them. But these people may find themselves in a shifting organizational framework. Stein does not say that Bionformatics as an area of study will disappear; quite the opposite: bioinformatics is becoming more embedded in life science academic programs. What he says is that there will be no more “dry labs” because all labs will have wet and dry components; at least those that concern themselves with asking biological questions. He is not referring to the “CS” brand of bioinformatics, where one’s peers in the department are CS people that emphasize algorithm development rather than asking biological questions.

    It *is* true that more young faculty are creating a wetlab component to their initial drylab, or setting up a two component lab. It is easier to manage a molecular biology lab now than it was ten years ago. Also, the “critical mass” of people, equipment and reagents needed is much lower for a “two component” lab, making a faculty candidate attractive in terms of start-up funds the department would shell out. Sequencing is getting cheaper. Various wetlab protocols are more easily performed. So it is relatively painless to set up a small wetlab where you can ask biological questions of your own and supplement database sequence data with data you generate. And the bottom line: funding agencies love this kind of setup.

    That being said there is also the path of collaboration: many high impact bioinformatic papers today do not come out of one lab: the work described is usually a collaboration, with one lab giving more of the wet component, and the other more — or completely — computational.

  4. nsaunders says:

    I’m more troubled by wetlabs with no interest in a drylab component than the converse.

  5. AP says:

    I think my analysis is a little darker then the ones described here. I spent 10 years in bioinformatics, authoring and co-authoring Recomb and Science papers and I pulled out last year. My role was ancillary to that of the biologists, and I saw no chance of having my own lab. My boss did not care for good bioinformatics, did not trust it and did not believe it. We collaborated with top biology labs all over the country and I found the situation exactly the same. When an analysis did not confirm the PI intuition, someone else was put in charge until the results converged to the desired ones. Bioinformatics journals have lost traction and biology journals have a bias against dry research. For one, they are incompetent to review it. I put the question to a genome research editor. She said verification in the wet lab was almost a necessity for publication. Forget p-values, theorems, anything: a biologist has to see a stain on some gel. A father of the field told me: “They [biologists] just want to push a button”. It is a systemic issue: where are the great independent bioinformatics departments? When I interviewed I was being dragged to biology, biophysics, medicine depts where no one had a clue what I was talking about. Where are the bioinformatics companies? I made a call it was time to call it quits, despite the problems being so incredibly interesting. I wish the best to whoever hangs in there.

  6. rbaltman says:

    I have recently been introduced to the idea of “Bio envy” among bioinformaticians. As I understood it, it is the tendency of bioinformaticians to seek validation almost entirely based on their biological credentials/contributions. I find this odd, because many of us chose the field because we are good at slightly (very?) different things. I would recommend seeking validation in a more diverse set of ways. For example, after 20 years in the field, I realize that I am more of an engineer than I realized. That was a very liberating discovery for several reasons.

  7. AP says:

    The issue is not validation or envy, it is publication, tenure and funding. When you are fresh out of grad school those are the issues, not validation. I took a quick look at your top 20 medline entries and 18 are in biology journals, just on a first pass, I apologize if this is not accurate. My guess is that most of your papers have been reviewed by biologists, not peers. If you want a career you’d better publish in Nature, not JCB, but that means that the algorithms are just a means to an end, not a subject of research of their own.

  8. shwu says:

    AP, as long as Bio-centric “discovery-based” journals are held aloft by those reviewing the publication record of young bioinformaticians, we’ll continue to have a bit of this desire for validation. It’s simply a consequence of how we’re being measured. But what I tried to get at was that what informatics is about is sometimes orthogonal to the discovery – and these means can be just as exciting as the ends. So until academia recognizes this (and by academia I mean those on the hiring and evaluation committees), bioinformatics folk will continue to have this metaphysical struggle.

    Granted, we’re at the point now where there is a generation of bioinformaticians now in senior roles, so things could potentially start changing.

  9. Pingback: The curse of BLAST : business|bytes|genes|molecules

  10. MedInformaticsMD says:

    “Demonize its missteps?” No. Rather, take it to task for not achieving what it might if its practitioners were more collaborative.

    Critique in biomedicine is not demonization. It is a necessary tool to ensure reflective practice.

    From the psychological viewpoint, yes, that “demonization” interpretation does seem to show a complex. Get over it!

  11. After doing some sole searching in bioinformatics, I decided to rounding out my perception about bioinformatics.

  12. shwu says:

    Scot (MedInformaticsMD), thanks for your comment. I do have a post in the works that will address more specifically the problems you wrote about. “Demonize” is probably too strong a word; I more take issue with what appears to be a blanket opposition to an emerging field that is a somewhat unfair.

    Much of bioinformatics does have the higher goal of advancing research that may benefit human health (and translational bioinformatics has been developed to aid in this goal). I disagree with requiring all biological research to have the express goal of advancing medicine. Yet this is how the vast majority of life scientists, including bioinformaticians, must frame their grants in order to get funding, leading sometimes to overhyped research.

    I wouldn’t limit the hype to bioinformatics. I’m sure that clinical research is sometimes overhyped and results in unexpected, negative outcomes – which perhaps could have been avoided with a little more insight into the biology underlying the disease or the treatment. That, to me, seems irresponsible – to develop drugs that appear to work but the mechanisms behind their action are not understood. I agree with you that better collaborations between medical and biological scientists will help.

  13. MedInformaticsMD says:

    “I more take issue with what appears to be a blanket opposition to an emerging field that is a somewhat unfair.”

    I agree 100%.

    “I agree with you that better collaborations between medical and biological scientists will help.”

    That was the major point of my post!

    Finally, much of healthcare requires a more reflective approach. For example, see my academic site on Health IT difficulties at http://www.ischool.drexel.edu/faculty/ssilverstein/medinfo.htm and the many posts on other healthcare topics by my colleagues at Healthcare Renewal blog.

    — SS

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s