Mike the Mad Biologist

detroitindustry
Detroit Industry by Diego Rivera

To put this post in larger context, Paul Krugman stirred up quite a ruckus with a column that argued that a lot of jobs for college graduates are being rendered obsolete by technological change. For scientists, this is not a new phenomenon. At a recent celebration type-of-thing, a colleague explained how a Prominent Genomic Researcher realized that the next leap forward in biology was going to happen when biologists would view their science as an information science. The future was not going to involve benches filled with dozens of Ph.D.s furiously pipetting (his phrase). Which is why this section from a column by Adam Ruben bothers me:

I realized recently that if I examine it in a day-to-day sense, I have one job in science. It’s not curing malaria, which is what my grant says it should be. My job, in essence, is to move small amounts of liquid from one place to another. That’s it.

That’s it? This is what smart people do? This is our reward for withstanding years in the trash can?

Currently, we have many informatics problems to solve–the sheer scale of data has overwhelmed current analytical tools and techniques. When we will look back on the last decade or two, the transformation of much of biology into an information science (or, perhaps the incorporation of informatics into biology?) will be the dominant theme. But automation, I think, will be the next ‘great leap forward’ (to use a phrase), although it probably won’t begin to seriously happen for five to ten years.

Robots and other high-throughput systems are faster, more reliable, and don’t get bored. These tools would allow data production–and thus, biological discovery–to occur at a qualitatively greater scale (e.g., genomics). This would also, I think, decrease the need for graduate students and post-docs as cheap labor.

Obviously, some areas of biology and certain kinds of experiments aren’t amenable to robotics or or high-throughput methods. But the more we can incorporate these methods, the more time we would have to spend on these ‘unmechanizable’ activities. It would also increase the future job prospects of Ph.Ds. How? Training would focus not on technique (although one should have good technique), but on analysis. Those skills can be applied widely (i.e., DNA is DNA, whether it’s produced by Sanger sequencing or Illumina).

Anyway, that’s enough futurism.

Discuss.

Comments

  1. #1 Adam
    March 10, 2011

    I think there are already prototype robots that even do some experimental design reasoning, for example in protein crystallography to choose environments that will produce suitable crystals for X-raying. To my shame, I have forgotten the name of such projects though…

  2. #2 leyo
    March 10, 2011

    When I started my PhD, i was once struck by the amount of useless time wasting work I did, pipetting tubes after tubes, plates after plates, endlessly. I felt like an intellectual McDonald’s employee. I became highly depressed from my work when I realized that around me where folks who would rather do this manual work and were so absorbed by it that they were antisocial, had incredibly difficult time even saying hi at the coffee table. That’s when I realized that research was somewhat doomed until everything turned into kits, robots and automation. There is a reason Celera sequenced the human genome 10x faster that the Broad, the broad has opted for your usual academic approach, heavy and nonsensical. To this day we still do that but in a different way, now automation enables scientist to do all sorts of expensive large scale project with almost no return what so ever. Just look at genomics, almost zero return after this 3 revolution, cancer genomics is still a nightmare, complex, multivariant disease that it was, prognsostic is mostly a one marker one mutation affair and scientist are still promising things like your genome on an Iphone without proving any sort of advantages to human kind, while people still die from bad water quality in third world countries. Shame on us ! Every single large screening facilities I have seen in my life has been a stalled, money pit machine. I have never seen a robot actually work and I have seen a lot of large installation. Why, people fantasize about automation and forget the large requirements upstream and downstream of that process. Same with genomics, folks don’t know how to store all of this data almost all of the time… Scientist are just not meant to manage process.

  3. #3 S. Pelech - Kinexus
    March 10, 2011

    Biology has been primarily a descriptive science and the sequencing of genomes and determination of the 3D structures of the proteins encoded by their genes has really just carried on this tradition. However, we are at the brink of a major paradigm shift in which at least the study of molecular and cellular biology are on the verge of become much more constructive and predictive.

    With a growing avalanche of data and improving computing capability, complex problems that only a few decades ago were calcitrant and untractable are yielding to solutions. The development of synthetic or artificial intelligence will further accelerate scientific research and development. This new found power can be channeled to spur on innovation and creativity that can truly transform health care and many diverse industries including those for food, clothing, shelter and energy production. Such a biorevolution will be achieved by a work force of scientists that will require a lot more training than what we actually offer today.

    Those contemplating a career in this direction should appreciate that dedication, a very broad and deep base of knowledge, and the ability to rapidly learn and master a new subject are key factors to ensure continuing employment success. The graduate and post-graduate experiences remain the best opportunities to acquire such capabilities, but those beginning to embark on this career path should expect to work very hard and long as the bar is raised. Too often, the relatively relaxed and unhurried lifestyles of many graduate students and post-graduate fellows does not adequately prepared them for what is coming and needed.

  4. #4 S. Pelech - Kinexus
    March 10, 2011

    Biology has been primarily a descriptive science and the sequencing of genomes and determination of the 3D structures of the proteins encoded by their genes has really just carried on this tradition. However, we are at the brink of a major paradigm shift in which at least the study of molecular and cellular biology are on the verge of becoming much more constructive and predictive.

    With a growing avalanche of data and improving computing capability, complex problems that only a few decades ago were calcitrant and untractable are yielding to solutions. The development of synthetic or artificial intelligence will further accelerate scientific research and development. This new found power can be channeled to spur on innovation and creativity that can truly transform health care and many diverse industries including those for food, clothing, shelter and energy production. Such a biorevolution will be achieved by a work force of scientists that will require a lot more training than what we actually typically offer today.

    Those contemplating a career in this direction should appreciate that dedication, a very broad and deep base of knowledge, and the ability to rapidly learn and master a new subject are key factors to ensure continuing employment success. The graduate and post-graduate experiences remain the best opportunities to acquire such capabilities, but those beginning to embark on this career path should expect to work very hard and long as the bar is raised. Too often, the relatively relaxed and unhurried lifestyles of many graduate students and post-graduate fellows does not adequately prepared them for what is coming and needed.

  5. #5 Janne
    March 10, 2011

    “This would also, I think, decrease the need for graduate students and post-docs as cheap labor.”

    No need to guess. Just look at computer science, statistics and mathematics department for comparison. How much does their ratio of grad students and post-docs to faculty differ from the biological sciences?

    My completely anecdotal guess is, the difference is not very large, and the impact of automation on this ratio will be limited.

  6. #6 kevin R
    March 10, 2011

    lab automatons Adam and Eve have gotten a lot of press

    http://www.scientificamerican.com/article.cfm?id=robots-adam-and-eve-ai

    It seems inevitable that robotics will assume most routine lab duties. The capabilities of robotic systems are advancing rapidly and costs are declining.
    While it is true that the sheer scale of data has overwhelmed current analytical tools and techniques, it is also true that huge amounts of R&D efforts are focused on those problems. Huge payoffs await the leaders in the field.

    IBM’s Watson is famous for winning on Jeopardy, but its primary role was to showcase IBM’s efforts, and success, at pulling specific information from huge amounts of disparate data sources and formats. The data overload problem will be solved and the benefits for science will be immense.

  7. #7 Paul Orwin
    March 10, 2011

    When I read comments threads like this I realize that many people just don’t take a very long view of their lives. Instead of bemoaning the essentially “descriptive” nature of Biology (BS, by the way; of course their are descriptive studies in Biology, but there are also hypothesis driven experiments and efforts – they are complementary), or how your labmates were asocial a-holes, think bigger. Do you really want to run sequencing reactions, or make plates, or clone and subclone genes? Is that what you set out to do, or is it a tool for the discovery you want to make? We don’t purify our own restriction enzymes anymore, or measure plasmid sizes by electron microscopy, because that wasn’t the bloody point! Now, instead of spending a year making a knockout, you can call or e-mail the Kelo collection or Bob Hancock at UBC and get one (in E.coli or P. aeru, respectively). And that’s good, because now you can ask serious, complex questions about those bugs! But, you can also do traditional molecular biology in a non-model, and work out whether it works the same way. And you can get it sequenced pretty quickly and cheaply, which is nice too. That quote about pipetting is just the grad student blues – it gets better, eventually! The key is to use your damned brain once in a while. Given that the easy stuff can be done by a machine, why don’t you try some hard stuff? What do you have to lose? That’s what it’s about. And anyone who thinks people won’t be pipetting in 10 years is smoking some serious crack. There will always be a place for people who are technically adept and meticulous in laboratory science.
    Automation won’t take people’s jobs away, it will make them much, much better!
    IMHO, of course :)

  8. #8 Paul Orwin
    March 10, 2011

    When I read comments threads like this I realize that many people just don’t take a very long view of their lives. Instead of bemoaning the essentially “descriptive” nature of Biology (BS, by the way; of course their are descriptive studies in Biology, but there are also hypothesis driven experiments and efforts – they are complementary), or how your labmates were asocial a-holes, think bigger. Do you really want to run sequencing reactions, or make plates, or clone and subclone genes? Is that what you set out to do, or is it a tool for the discovery you want to make? We don’t purify our own restriction enzymes anymore, or measure plasmid sizes by electron microscopy, because that wasn’t the bloody point! Now, instead of spending a year making a knockout, you can call or e-mail the Kelo collection or Bob Hancock at UBC and get one (in E.coli or P. aeru, respectively). And that’s good, because now you can ask serious, complex questions about those bugs! But, you can also do traditional molecular biology in a non-model, and work out whether it works the same way. And you can get it sequenced pretty quickly and cheaply, which is nice too. That quote about pipetting is just the grad student blues – it gets better, eventually! The key is to use your damned brain once in a while. Given that the easy stuff can be done by a machine, why don’t you try some hard stuff? What do you have to lose? That’s what it’s about. And anyone who thinks people won’t be pipetting in 10 years is smoking some serious crack. There will always be a place for people who are technically adept and meticulous in laboratory science.
    Automation won’t take people’s jobs away, it will make them much, much better!
    IMHO, of course :)

  9. #9 Kaleberg
    March 10, 2011

    When Herschel was doing astronomy in the late 18th century, it was all about grinding telescope mirrors. That’s amazingly tedious work, but it’s now all done by machine. I expect biology to take a similar path. People will still grind mirrors or pipette fluids, but the heavy lifting and raw tedium will be automated. Of course, that just means that scientists will find new heavy things to lift and tedious tasks to perform until they can be properly automated.

  10. #10 Paul Orwin
    March 11, 2011

    Sorry for the double post, Mike! Please remove it if you can

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.