5 ms·

I came to biology from physics over a decade ago thinking that theory would matter. But modern biology is not waiting for a Grothendieck. We are not wanting f

I came to biology from physics over a decade ago thinking that theory would matter. But modern biology is not waiting for a Grothendieck. We are not wanting for a deep theoretical view to reinterpret our data with. We are wanting for cheap, high quality data about cells and organisms in the first place. Those of you who have not done biology can't realize how primitive things still are. A CS analogy is that we're trying to reverse engineer a non-deterministic alien architecture from the future via a remote debugger over a noisy line that has no error correction and runs at a fraction of a millibaud, where every peek or poke costs hundreds to thousands of dollars. With the advent of high-throughput sequencing we just figured out how to get noisy ROM/RAM dumps (genotype), but still have precious little ability to quickly interact with the biological systems under study (phenotype). Biology is much more like hacking and reverse engineering than it is like physics or mathematics. A working exploit is worth much more than general pontification. The transformative events in biology these days are the clever hacks that enable whole new classes of fast experiments to be done: quick homologous recombination induced by CRISPR, optical interrogation of neural systems using light-sensitive ion channels, the development of adeno-associated viruses for rapid introduction of genetic material, etc. In the coming decades the combination of such clever hacks with cheaper automation and sensors will hopefully bring a flood of new data that will make a predictive, quantitative science of biology possible. With cheap and plentiful data, perhaps clever minds might discover general quantitative principles that we can't see yet. But the data is what's lacking, not cleverness.

TisButMe 12y ago
While I absolutely agree with the need for cheaper way to obtain data, I don't think this is the major problem we're going to face. We now have ways of obtaining mountains of data (RNA-chIP comes to mind, but high-speed sequencing is a big one as well), but we don't know how to process it. We end up with thousands of candidate proteins, but then someone has to go through them one by one. Things are incredibly primitive. The huge pain point coming up is our ability to process information, which is very limited. How do you deal with machine generating TBs of data/day ? I think what's lacking is a culture of automation. Physicists do it, chemists do it, but biologists still do everything "by hand", and not only is that imprecise, is also slow and costly. We also need better ways of obtaining information about other's work. When there are 1000 paper a year published on your subject, how do you filter ? How do you even find the time to read papers in other subjects, which could be interesting in their methods, or just for general intuition and knowledge ? What we need is good algorithms, able to replace curate our sources of information well, and machines able to do the bulk-processing now needed in biology. Some things are coming up (I'm co-founding a start-up to do exactly this), but it should have happened 10 years ago.
- daemonk 12y ago
  We have ways of obtaining mountains of sequencing data that represents only a single dimension of an unknown system. While, it is a step in the right direction, this data by itself is still lacking. It's like trying to model the trajectory of a bouncing ball with extremely high resolution X coordinates, but horrible Y and Z coordinate data. I've been working with sequencing data for the past 5 years. The problem really isn't with how to interpret the data, it is that the data itself sucks because it's extremely myopic. And yes, there are people who are trying to analyze the various *-seq methods holistically. But I haven't really seen anything substantial coming out of it.
busyant 12y ago
This is the best analogy of modern biology that I've ever heard. Just to add one bit of color commentary... > but still have precious little ability to quickly interact with the biological systems under study (phenotype) In case it's not obvious, part of the inability to interact 'quickly' is that organisms take their damned time to grow. Even E. coli will take about 20 minutes to double (under the best circumstances). Fruit flies take a few weeks. Mice? even longer. I suppose the field is ripe for some breakthrough to speed that up, but I don't know what it is.
akulesa 12y ago
I agree with you that its essential to lower the amount of time and cost that experiments take, and find needles to replace our sledgehammers. I work in technology/methods development, so I'm happy that I should a job for the next decade! However, I have to disagree with the claim at the end of your response: once we get the methods right, we can start building biology into a predictive and quantitative science. I think the debate needs to start at, what does it mean to make biology into a predictive and quantitative science? Sure, once that flood of data comes in (and I'd argue its already here), we can build huge dictionaries of perturbations-responses in every system we study. And sure, this will be extremely useful for drug development, synthetic biology, et cetera. But is it "understanding"? Is it intellectually satisfying? Right now it's unclear to me how to even evaluate where are our current understanding of biological systems are incomplete <b> in ways that matter </b>. Sure, in any system there are remaining things to be discovered, but maybe they are 2nd order, 3rd order, 10th order effects on the behavior of the system. The way people are incentivized though, they will do all they can to convince you that the 3rd order effect they study is really just as important as the 1st order, and it often takes more effort than is worth it to see through the smoke screens. Developing more methods and technology further fractionates the field more than it serves to unify it (with the exception numbering what I can count on my hands). Until we have any idea where we're going, it's futile to just wait for more data to come in.

TisButMe 12y ago

While I absolutely agree with the need for cheaper way to obtain data, I don't think this is the major problem we're going to face. We now have ways of obtaining mountains of data (RNA-chIP comes to mind, but high-speed sequencing is a big one as well), but we don't know how to process it. We end up with thousands of candidate proteins, but then someone has to go through them one by one. Things are incredibly primitive. The huge pain point coming up is our ability to process information, which is very limited. How do you deal with machine generating TBs of data/day ? I think what's lacking is a culture of automation. Physicists do it, chemists do it, but biologists still do everything "by hand", and not only is that imprecise, is also slow and costly. We also need better ways of obtaining information about other's work. When there are 1000 paper a year published on your subject, how do you filter ? How do you even find the time to read papers in other subjects, which could be interesting in their methods, or just for general intuition and knowledge ? What we need is good algorithms, able to replace curate our sources of information well, and machines able to do the bulk-processing now needed in biology. Some things are coming up (I'm co-founding a start-up to do exactly this), but it should have happened 10 years ago.

daemonk 12y ago
We have ways of obtaining mountains of sequencing data that represents only a single dimension of an unknown system. While, it is a step in the right direction, this data by itself is still lacking. It's like trying to model the trajectory of a bouncing ball with extremely high resolution X coordinates, but horrible Y and Z coordinate data. I've been working with sequencing data for the past 5 years. The problem really isn't with how to interpret the data, it is that the data itself sucks because it's extremely myopic. And yes, there are people who are trying to analyze the various *-seq methods holistically. But I haven't really seen anything substantial coming out of it.

busyant 12y ago

This is the best analogy of modern biology that I've ever heard. Just to add one bit of color commentary... > but still have precious little ability to quickly interact with the biological systems under study (phenotype) In case it's not obvious, part of the inability to interact 'quickly' is that organisms take their damned time to grow. Even E. coli will take about 20 minutes to double (under the best circumstances). Fruit flies take a few weeks. Mice? even longer. I suppose the field is ripe for some breakthrough to speed that up, but I don't know what it is.

akulesa 12y ago

I agree with you that its essential to lower the amount of time and cost that experiments take, and find needles to replace our sledgehammers. I work in technology/methods development, so I'm happy that I should a job for the next decade! However, I have to disagree with the claim at the end of your response: once we get the methods right, we can start building biology into a predictive and quantitative science. I think the debate needs to start at, what does it mean to make biology into a predictive and quantitative science? Sure, once that flood of data comes in (and I'd argue its already here), we can build huge dictionaries of perturbations-responses in every system we study. And sure, this will be extremely useful for drug development, synthetic biology, et cetera. But is it "understanding"? Is it intellectually satisfying? Right now it's unclear to me how to even evaluate where are our current understanding of biological systems are incomplete <b> in ways that matter </b>. Sure, in any system there are remaining things to be discovered, but maybe they are 2nd order, 3rd order, 10th order effects on the behavior of the system. The way people are incentivized though, they will do all they can to convince you that the 3rd order effect they study is really just as important as the 1st order, and it often takes more effort than is worth it to see through the smoke screens. Developing more methods and technology further fractionates the field more than it serves to unify it (with the exception numbering what I can count on my hands). Until we have any idea where we're going, it's futile to just wait for more data to come in.