Supercomputer Sets Protein-Folding Record 63
Nicros writes with this snippet from Nature News:
"A specially designed supercomputer named Anton has simulated changes in a protein's three-dimensional structure over a period of a millisecond — a time-scale more than a hundred-fold greater than the previous record. ... The simulations revealed how the proteins changed as they folded, unfolded and folded again. 'The agreement with experimental data is amazing,' says Chandra Verma, a computational structural biologist at the Bioinformatics Institute of the Agency for Science, Technology and Research in Singapore. Simulating the basic pancreatic trypsin inhibitor over the course of a millisecond took Anton about 100 days — roughly as long as computers spent toiling over previous simulations that only spanned 10 microseconds."
Even though it was published in Nature News... (Score:5, Interesting)
..it's a rather poor article. It talks in very basic terms about proteins and their folding, talks a bit more about the scientist who founded the institute behind the computer, and says fuck-all about the construction of the computer itself.
Bah. For a publishing house of Nature Publishing Group's (intellectual and economic) muscle, one should expect more.
Re:Even though it was published in Nature News... (Score:5, Informative)
Here this should give you more information.
http://en.wikipedia.org/wiki/Anton_(computer) [wikipedia.org]
I think the article was alright though. It told what was going on, and why it could be important. It wasn't written for a computer nerd demographic so the exact specs weren't really relevant.
Re: (Score:2, Funny)
It wasn't written for a computer nerd demographic so the exact specs weren't really relevant.
You are writing for a nerd demographic though so you should have said "It wasn't written for a computer nerd demographic so it missed out all the most important information."
Re: (Score:1)
It wasn't written for a computer nerd demographic so the exact specs weren't really relevant.
You are writing for a nerd demographic though so you should have said "It wasn't written for a computer nerd demographic so it missed out all the most important information."
A true computer nerd demographic wouldn't need to have it written out for them. They would just figure it out for themselves.
Re: (Score:2)
Re:Even though it was published in Nature News... (Score:5, Informative)
Re: (Score:3, Insightful)
The best way is to just compare them to the actual structure which is known from x-ray crystallography and NMR studies.
And so far, this is the only way that most researchers are willing to trust. There is a very good reason why these folding studies tend to focus on a small group of well-defined model systems, because the folded native structure is already very well understood, and it provides an essential constraint on the interpretation of results. Using ab initio physics calculations like this for trul
Re: (Score:3, Informative)
I believe the article is published in Science not because of its computer utilization (i.e. using a bad-ass super computer), but because of its biological relevance. They managed to characterize not only the peptides conformations, but also their mutant's behavior in silico.
Re:Even though it was published in Nature News... (Score:4, Insightful)
This was not an ab initio, calculation. It's all atom MD, which itself is an approximation
Sorry, I meant "ab initio MD", although I realize that to a chemist or physicist this is a total oxymoron. (My background is molecular biology and bioinformatics, where we try not to think about quantum chemistry.) I should have written "physically-based", if you prefer, as opposed to the knowledge-based approaches that have been most successful for de novo structure prediction. (I think most MD "force fields" are ultimately based on genuinely ab initio QM calculations.)
There's other stuff too (Score:1, Interesting)
From experimental evidence we know the folding rates of certain proteins at various temperatures, we know the flow rates for ion channels, and so on. A lot of these macro-properties can't be tested in the short simulations that current computers can do, but they can easily be reached by the DE Shaw machine.
Re: (Score:2)
if the computer does a simulation on a known structure, then you can compare its accuracy to that of the known experimental results.
Re: (Score:3, Insightful)
..it's a rather poor article. It talks in very basic terms
That's because it's in nature news, which does rather high-level, short coverage of a wide range of topics for a very broad scientific audience. It's meant to be a "hey look this is cool" article, that you can read up more about if you are interested and have the right kind of background. Perhaps you were thinking it was a ahort or regular article?
Re: (Score:3, Interesting)
Nature and Science are not for hard science.
If you just get articles from citation search its not obvious why, but if you ever see a print issue it becomes obvious:
They cover a _huge_ range of fields. You can have articles about egyptian mummies, rainforrest status in south america, neutron scattering and virus chrystallography within 20 pages or so.
So people have to write the arcticles in a way that at least readers from most of the fields involved can understand it and see why it is important. Otherwise,
Re:Even though it was published in Nature News... (Score:5, Informative)
Nature and Science are not for hard science.
The actual research articles are hard science - this was just a news story for a general audience. The official publication of the results [sciencemag.org] in Science magazine appears to be a pretty serious piece of work, and it's significant enough that the editors allowed them to make it reasonably long instead of a (severely compressed) three- or four-page summary article like most of what they publish. There are lots of valid criticisms of those two journals, starting with their length requirements, but they're not Scientific American, and publishing in one of these is practically a prerequisite for getting a faculty position in biosciences at a major research university.
Re: (Score:2)
Re: (Score:1, Informative)
Here ya go, whinger:
David E. Shaw, et al "Anton, A Special-Purpose Machine for Molecular Dynamics Simulation," Communications of the ACM, vol. 51, no. 7, 2008, pp. 91–97. http://mags.acm.org/communications/200807/?folio=91
Jeffrey S. Kuskin, et al "Incorporating Flexibility in Anton, a Specialized Machine for Molecular Dynamics Simulation," Proceedings of the 14th Annual International Symposium on High-Performance Computer Architecture (HPCA '08), Salt Lake City, Utah, February 16–20, 2008. htt
Here (Score:1, Informative)
I didn't RTFA since I've already heavily researched these guys. D.E Shaw is the kind of billionaire I would be.
Summary: The actual atomic interaction equations are simulated very fast. Distributing the results of a local interaction to the rest of the simulation quickly, is hard.
http://www.deshawresearch.com/publications/Simulation%20and%20Embedded%20Software%20Development%20for%20Anton,%20a%20Parallel%20Machine%20with%20Heterogeneous%20Multicore%20ASICs.pdf [deshawresearch.com]
http://cacs.usc.edu/education/cs653/Shaw-msMD-SC09 [usc.edu]
Re: (Score:1, Redundant)
It's Nature News, not Nature. It's not supposed to be a research article. If you want that, there's a citation at the bottom. That's more than you can say for 99.99% of the popular science reporting out there.
Re: (Score:2)
It's Nature News, not Nature.
I know, and I think I wrote so myself in the OP. I read Nature News rather regularly at work (at the coffee table, believe it or not) and this article was, in comparison to the ones I read, sub par. It wasn't quite the kind of shite BBC News would publishes (online) regarding science and technology, but it deviated into that sad direction.
Future Computing Speeds (Score:2)
The fact that it takes 100 days to simulate a few milliseconds of molecular activity hints at the potential speed of future computers. I know the actual process isn't precisely analogous to the computation, but I suspect there are more elegant ways to compute than the methods we use today. Our brains "outperform" the best supercomputers, with energy requirements supplied by a bowl of oatmeal for a few hours of activity. The mind boggles at the possibilities.
Re: (Score:1, Insightful)
It's about someone (a rich someone) building a really big computer to tackle a really, really, really, really, really, really, really complex physical/chemical problem that we currently know dick all about.
If protein folding was equivalent to fluency in English, we'd be at "bwawubda?"
Re: (Score:1, Insightful)
Hundred-fold greater? (Score:1, Insightful)
over a period of a millisecond — a time-scale more than a hundred-fold greater than the previous record
This phrasing always confuses me where they say "It's this much faster so it's x times greater!"
So they're a hundred fold greater and they're a millisecond...? Does that mean the other guy took 1/100ths of a millisecond?
Re:Hundred-fold greater? (Score:5, Informative)
Processing power (Score:3, Funny)
The performance of a 512-node Anton machine is over 17,000 nanoseconds of simulated time per day for a protein-water system consisting of 23,558 atoms
So... how many libraries of congress per second??
Re: (Score:2)
since each of the 512 chips has six two-way links, one-way being 50 GBit/sec, we have roughly 6 * 50 * 512 Gbit/sec through each chip, or 154 Tbit/sec or about 19 Tbytes/sec.
If the printed LOC is 10TB, then almost two Library of Congress' worth of data being processed per second.
applause! (Score:3, Insightful)
This research is extremely important for finding new drugs, and therefore I applaud the originators of the project, especially D.E. Shaw who apparently put also a lot of funding into it. I wish more (rich) people put their money into such immensely useful projects. It is not just a noble thing to do, it is also smart, since we all could one day benefit from this kind of research.
not really (Score:3, Insightful)
This has been the promise of computer simulation - "in silico" drug design - for decades. It hasn't panned out. And I say this as someone who makes a living doing exactly what these folks have done. High throughput bench work is far more efficient, time and money wise, than computer simulation. Hard to say when or if that will change.
Re: (Score:3, Interesting)
According to the article, it now takes 100 days to do one simulation. If we had 100 times the processing power (maybe a little more accounting for overhead), then we could do it in one day. I'd say that would be possible today with sufficient financial support, or at least it could be a reality within a decade. In short, it still sounds promising to me..
Re: (Score:2, Interesting)
100 days is for a 'hero run', the bread and butter runs last 1-4 days apiece and account for more like 20-100 microseconds of simulated time. One of the big innovations of this machine is that those runs would otherwise take months on other machines.
Re: (Score:1)
I suppose that depends on what you are doing. The kinds of potential energy surfaces these simulations use are extremely crude - ball and spring (sometimes one ball for many atoms), electrostatics, short range repulsion, and no chemical reactions. Only good for rough trends in docking and stuff. If you need reactivity, and you will for it to really compete with bench work, there are some severely limited classical force fields that work - but they are at least an order of magnitude more expensive. More
Re: (Score:2)
Aren't these interactions only interesting at the small scale? If the bulk of the computation is in the long-range interactions, then I'm not sure if these more expensive computations will really increase the complexity that much... but I am not an expert.
Also, I've seen once a documentary (forgot the name), in which a couple of students were trying to figure out the folding by crystallization, and they were seriously concerned that the folding of the proteins would not happen correctly in the solid state.
Re: (Score:3, Insightful)
Re: (Score:1)
Re: (Score:1)
I do molecular dynamics. These are MD simulations. They are talking 2 orders of magnitude speedup. That's good, but not good enough to compete with bench work. Incidentally, people are claiming this sort of speedup by offloading some parts of the calculation to a GPU. NCSA has a cluster of GPUs for just this use - Lincoln if I recall correctly. MD is currently useful for a lot of stuff. Not drug design yet.
Re: (Score:1)
an 33 order of magnitude increase in a single typical MD timestep is of the order of the lifetime of the universe.
Re: (Score:2)
So where do I send my cheque?
Re:The folly of folding@home (Score:5, Informative)
That's a little unfair to Folding@Home. Shaw has a lot of resources to pour into this project - he's lured faculty members away from universities to work for him instead and has the equivalent of several large labs worth of advanced researchers. He also has an immensely larger budget than most non-profit labs, and he's self-employed so he doesn't have to answer to granting agencies or tenure committees. I think what he's doing is great but he's really one of the only people who could have pulled this off. It's difficult to know what approach will work best in advance, and both Shaw and Vijay Pande have been very innovative in approaching the problem from completely different angles.
By the way, this approach has been tried before with less stellar results - I'm thinking of the MD-GRAPE project in Japan. You're also assuming that every problem is equally well suited towards custom ASICs, but actually, molecular dynamics is far easier to do this with than many other methods. For instance, Rosetta (Rosetta@Home and Fold.It) is doing structure prediction, not folding, using a mostly statistics-based energy function and Monte Carlo sampling, and this isn't something you can trivially offload to a specialized chip. In that case, distributed computing is by far the most efficient solution.
Re: (Score:3, Insightful)
Actually, Folding@Home can also simulate these time scales by means of Markov state models. The trajectory is pieced together out of data collected from many short simulations, whereas the Anton trajectory is generated from a single MD run, but in practice that distinction is usually irrelevant. Protein dynamics are stochastic, so for any time scale longer than about 1 ns, both approaches given equally "realistic" or "valid" trajectories.
That's not to criticize Anton. It's an amazing piece of hardware an
Re: (Score:1)
For instance, Rosetta (Rosetta@Home and Fold.It) is doing structure prediction, not folding, using a mostly statistics-based energy function and Monte Carlo sampling, and this isn't something you can trivially offload to a specialized chip. In that case, distributed computing is by far the most efficient solution.
Right on the money. Because most of its applications use Monte Carlo as you mention, Rosetta requires lots of independent trajectories anyway. It's trivially parallelizeable (embarassingly parallel if you prefer) so distributed computing is the solution we use for pretty much everything. The Baker lab has the BOINC Rosetta@home and the rest of us use university-size clusters.
Re: (Score:3, Insightful)
Good thing F@H runs on the GPU, which is many times faster than the CPU at these operations.
Also, don't forget what it takes to build supercomputer capable of doing this, and that resources put into building supercomputers are then not available for the consumer market. Distributing this stuff allows for a compromise between absolute best performance and letting people have powerful computers at home.
Ah, the human body (Score:1, Insightful)
I love it how simple-minded tech geeks, usually IT guys, programmers and even people who should know better like electrical engineers, think that the internet is more complex than the human body... Here we have ONE molecule, simulated for a lousy millisecond, and it took more than THREE MONTHS. How many molecules in the human body? Our body is performing a truly staggering amount of computation. Actually, every bit of matter is, everything including "inanimate" matter, it's really all the same. We just happ
Re:Ah, the human body (Score:5, Interesting)
It is complex, but you are ignoring the relative isolation between levels that exists in the human, and rat, body.
Protein folding may be complex, but most of it is irrelevant detail. What's usually important is the final shape that one ends up with, e.g. But when wants to modify that process, then the details of that process become important. This is roughly equivalent to...at the level that I work, I pay no attention to how the compiler is going to optimize my code. If I wanted to modify that I'd need to pay attention to things at a much finer level of detail.
It *is* true that people tend to oversimplify things they aren't dealing with directly. But to make it a fair statement it needs to be made fully *that* general. (This doesn't make you original assertion false, but observationally it *is* false. I've never known a knowledgeable geek that oversimplified the biochemistry of life in the way that you painted. I'm sure they exist, but they aren't, as you implied, common. If they are common among your friends, well, then you have some uncommon friends.)
Re: (Score:1)
After all, can a mayfly explore a city? It'll be dead in three days. That's us, in space.
That's a horrible analogy IMHO. I don't see any need for humans to live longer than we do. We can pass on our knowledge to anyone that wishes to have it, and now have entire decades worth of knowledge, all at our finger tips. The only thing that matters is a persons thirst for such exploration. I'd much rather delve in something that stops the aging process (yes, i know, very far fetched) not to keep us alive for longer, but to keep us young till we die. Then you wouldn't have to worry about 80 year ol
Re: (Score:2)
I love it how simple-minded tech geeks, usually IT guys, programmers and even people who should know better like electrical engineers, think that the internet is more complex than the human body..
I love it how you think that the moon is made of cheese. Oh you don't? Ah that must be because I just made up some bullshit strawman to make a point.