4 ms·
I think there is this (from my limited experience rather naive) belief that if a bunch of smart computer people get together, they can solve problems that have
by eshvk 14y ago
I think there is this (from my limited experience rather naive) belief that if a bunch of smart computer people get together, they can solve problems that have flummoxed biologists for long. This is something pervasive which afflicts even the smartest ones (See Steve Yegge's talk http://www.youtube.com/watch?v=vKmQW_Nkfk8 http://www.youtube.com/watch?v=vKmQW_Nkfk8 on how The Data Mining needs to be used more to solve problems in biology).
Having spent some time in grad school foolishly believing that all biology needed was a sprinkle of math and a generous dollop of python skills, I think the problem is that in Computer Science we have problems that are sometimes insanely hard but due to the several layers of abstraction, are inherently rather deterministic and much less noisy than real-world systems. I think the best comprehensible example comes from the physical layer of most electronics where several of the modeling equations are naive approximations designed to work under specific conditions. The complexity becomes even more overwhelming when you look at biological systems: The nonlinearity especially in say genetic hacking becomes apparent when you realize that each tiny modification has so many unintended effects (making modeling in a computer environment especially difficult) which is why biologist go through the grind of running all those "boring" real world experiments in labs. I think there are quite a few research labs that are using data mining and machine learning techniques to try to solve problems but we are at most quality places well beyond needing "smart people with computer skills". After a while, fundamental research requires, like you said, money on a massive scale which VC's really can't afford.
- Dn_Ab 14y agoYou make good points but swing the pendulum too far the other way. No, a bunch of smart computer people won't solve problems that have flummoxed biologists for long. But a bunch of smart computer, biology, math, physics and chemistry people will. You are right in that ignoring bench work + domain knowledge and thinking one can come in write a bunch of programs and solve cancer is not only an insult but foolish. Not all the problems have flummoxed biologists for long. The problem of sequencing improving faster than computers is a new one. And one where computer people can help. In fact modern sequencing techniques rely on computers to be viable. Modelling phenotype from gene sequences and predicting possible drug interaction will become real soon. Biology is complex yes but some of it is complexity introduced in an attempt to avoid difficult but simplifying concepts like non linear PDEs. Funny that you mention electronics. See what Yuri Lazebnik has to say about this. http://protein.bio.msu.ru/biokhimiya/contents/v69/pdf/bcm_1403.pdf http://protein.bio.msu.ru/biokhimiya/contents/v69/pdf/bcm_14... Computer people can help with tools, modelling, analysis and prediction.
- eshvk 14y agoI am sorry if I came off as disagreeing with your main points because I actually don't. Just to say a couple of things: At least from looking at such cross-functional bio/c.s. teams work here in Grad. school, I have seen two types: 1. Where the bio (in this case wildlife) people have extremely limited ideas about what happens during a simulation or modeling experiment, this leads to ridiculous attempts to say develop multi-agent models for hyena cooperation which if you read the paper make so many ridiculous assumptions that it doesn't really make sure what if any utility can be obtained as inferences from those models. 2. The other much more rarer situation where the bio people have a keen understanding of the basics of both C.S. and Math and are well aware of the limitations of what they are getting into. Such collaborations usually turn into using a C.S./Math (usually grad. student) person in order to delve deep into the technical specifics which the bio guy could understand with some help. Yuri's article is interesting and while I do agree that some cross pollination of ideas (such as is happening in systems biology) is useful. I am not so sure I am that enthusiastic as he is about drawing so much hope from the engineering analogy (because there is an element of silver bulletism to it:).