PC Gamers Save The World, A Bit

Well, this is all perfectly clear.

Perhaps you’re not familiar with the search for the molecular structure of a protein-cutting enzyme from an AIDS-like virus found in rhesus monkeys. But groups of scientists have been trying to fathom this mystery for years. And it’s just been cracked by PC gamers.

Why is this exciting? Because it’s believed that this information could be crucial to better figuring out how the AIDS virus works, and how it can be tackled. The findings, discovered via the game Foldit, are published online ahead of the next issue of Nature: Structural & Molecular Biology, with the authorship listing both scientists and gamers.

The results being discovered by this game are useful in many significant areas of research, the big three being AIDS, cancer and Alzheimer’s. All three diseases are related to proteins, the subject of Foldit’s puzzles, and the more that is known about how they behave, the more can be done to tackle the diseases they’re related to. And their behaviour is very much determined by their structure.

It’s the second paper Foldit has seen published in Nature, which is a remarkable achievement. The game involves folding proteins, attempting to manipulate three-dimensional models of proteins into the most energy-efficient structure possible. As such, it’s competitive, with players attempting to out-score others, and in the process getting closer to the unknown structure of the protein sought.

According to the paper, for years scientists have failed to uncover the crystal structure of M-PMV retroviral protease. However, Foldit players struck upon the solution. As the paper explains,

“Foldit players leverage human three-dimensional problem-solving skills to interact with protein structures using direct manipulation tools and algorithms from the Rosetta structure prediction methodology. Players collaborate with teammates while competing with other players to obtain the highest-scoring (lowest-energy) models. In proof-of-concept tests, Foldit players—most of whom have little or no background in biochemistry—were able to solve protein structure refinement problems in which backbone rearrangement was necessary to correctly bury hydrophobic residues. Here we report Foldit player successes in real-world modeling problems with more complex deviations from native structures, leading to the solution of a long-standing protein crystal structure problem.”

Got that?

Explained in the Sydney Morning Herald’s article, when looking down a microscope all you’ll see is a 2D splodge of the molecule you’re trying to understand, but to do anything useful you need to know about it in 3D. Which is where the puzzling comes in. It seems that computer models are not as capable as humans when working within spacial reasoning, hence the advantage of gamers approaching these challenges. Because, well according to the paper,

“Foldit players — most of whom have little or no background in biochemistry — were able to solve protein structure refinement problems in which backbone rearrangement was necessary to
correctly bury hydrophobic residue.”

Thankfully MSNBC explains it all rather well. According to the site, there are millions of ways for an enzyme’s atoms’ bonds to twist, and the secret to getting it right is finding the lowest-energy configuration for such a structure. Because that’s the one that Horace picked when he designed the universe. So 236,000 players working together and against each other for the last three years have proven rather useful. And it took them only 10 days to unravel (or indeed re-ravel) a mystery that had stumped the experts for over a decade.

I just cured cancer!

And this isn’t a one-off. Foldit can continuously be used in this way to solve the structures of mysterious enzymes, and being equipped with such knowledge is extremely powerful. Knowing the correct arrangement of a molecule’s spaghetti is crucial to understanding its behaviour, how it bonds, and then tackling any shenanigans it may get up to. This time it could be a significant step toward finding better means of treating AIDS. Next time, something else.

As the paper concludes, this is a unique approach, this recent result demonstrating great potential for crowd-sourced gaming as a means for research,

“The critical role of Foldit players in the solution of the M-PMV PR structure shows the power of online games to channel human intuition and three-dimensional pattern-matching skills to solve challenging scientific problems. Although much attention has recently been given to the potential of crowdsourcing and game playing, this is the first instance that we are aware of in which online gamers solved a longstanding scientific problem. These results indicate the potential for integrating video games into the real-world scientific process: the ingenuity of game players is a formidable force that, if properly directed, can be used to solve a wide range of scientific problems.”

If you want to join in, you can get Foldit from here. It begins with a tutorial that explains the concept in great detail, and then you’re away into attempting to find the most energy efficient structures, either alone or in teams, competing against others around the world. And saving lives.

Thank you to everyone who tipped us about this.


  1. Bodminzer says:

    This is the greatest thing of all.

  2. hello_mr.Trout says:

    +1 gamers
    +1 science!

  3. HerrKohlrabi says:

    A million monkeys, with a million typewriters… :)

    Actually, I’d rather say brains vs computers 1-0.

    • sinister agent says:

      As soon as we can convert all the world’s nuclear weapons to some sort of protein code based triggering system, the future war is won!

    • MrMud says:

      Its more along the lines of the second explanation.

      Protein folding is generally regarded as an NP-complete problem (nodeterministic polynomial) which means that there must be a way to verify a solution in polynomial time but that there is no known algorithm for solving the problem in polynomial time on a non quantum computer.

      Polynomial time means that the time it takes to solve a problem increases with a polynomial factor. Examples of this is that sorting a list of values takes up to n^2 operations where n is the number of items in the list. A list of 10 items could then take ~100 (10^2) operations to sort. Because n^2 is a polynom this counts as solving the problem in polynomial time. As n grows so does the number of operations, but not excessively so.

      A non-polynomial growth (for example exponential) is 2^n. This grows very fast, n= 10 gives1024 operations and n= 20 is over a million. For a complex protein the number of n can be so big that it becomes impractical or impossible to calculate.

    • CMaster says:

      Indeed, this is basically the opposite. Computers can do the “infinite monkeys” pretty well. The problem is that there are so many “possible” conformations, it would take a very long time (possibly age-of-universe time, if the protein is big enough) to generate and test them all. Humans are a lot better than computers (especially with spatial problems, for example) at throwing out the “obviously” rubbish possibilities and focusing on the few of interest. I’d be intrigued to know how well using people to do this sort of thing compared to using say, genetic algorithms. Individual teams on foldit would also be quite prone to getting stuck in a secondary minimum one would imagine.

    • CMaster says:

      And I’ve just read the paper. Apparently local minima were in fact a very big problem, especially if they gave the players a “best we’ve got” solution. The way around this of course, was to give the players an “eh, it looks a bit like this” structure and let them take it from there.

    • Lambchops says:

      @ CMaster

      Plus giving them a vaguer starting point makes for a more potentially rewarding and entertaining “game” for whoever is playing it as there’s a lot more possible variation and room to experiment. Definitely seems like the way to go to improve the utility of this sort of thing. Obviously there has to be some restraints or most of the results would be garbage. Looks like half the battle with this sort of project is finding the sweet spot of where to pitch the starting point.

  4. HexagonalBolts says:

    You solved the AIDs virus, ‘Congratulations!’ (multi-coloured star firework).

  5. Jharakn says:

    Interesting article and a clever use of crowd-sourcing. It would be interesting to hear from someone who works as a researcher in molecular biology to find out if this is as big a deal as the article reads or if its just a small discovery hyped up by the gaming community to give us some nice press.

    • Thermal Ions says:

      Well given I came a cross the story in mainstream news media initially, then jumped over to RPS to see if they’d seen it, I’m thinking it’s somewhat noteworthy.

    • CMaster says:

      The paper is also published in Nature, which means this isn’t some really obscure piece of work, it’s anticipated to be of some broader scientific interest.

      Edit: Contrary to what the above article says, it’s published in Nature Structural and Molecular Biology, not Nature itself. So I guess it is still fairly niche.

    • Sinomatic says:

      I also heard of this elsewhere first. Not to say that actually indicates the scientific significance of the story at all, I think ‘gamers cure AIDS’ is a whimsical enough line for any outlet to use regardless of it’s veracity.

    • John Walker says:

      It just so happens that my fiancee is a researcher in molecular biology. She says:

      “Retroviral proteases (PR) are the enzymes that are responsible for cutting up viral precursor proteins into their active (and therefore virulent) forms. If scientists could stop these enzyems from working, they could help combat retroviruses, including HIV.

      One of the main ways to stop these enzymes working is to stop them being made by the cells. Enzymes are proteins themselves, and act on other cellular proteins to alter them.

      PR enzymes are “homodimeric” which means they are made up of 2 protein subunits of the same “type” (hence “homo”). Scientists would like to prevent this dimer from forming. In order to do this, they need to know the structure of the monomer (single subunit) that dimerises to make the complete enzyme.

      Once they have this, they can create a molecule that mimics this monomer. It will compete to bind to the monomers that the virus creates in the body, and if it binds at a higher rate than the native monomer does, it will act to sequester the monomer and prevent the active enzyme from forming. Thus defeating the viral spread.

      Therefore, in answer to your reader’s comment: Yes, this is an important discovery, and not just hype.”

      She is brainy.

    • Sinomatic says:


      And yes, she is brainy.

      Also, congratulations on the imminent binding ;)

    • CMaster says:

      Simplifying that a little:

      Retroviral proteases are enzymes (“tools”) that are needed by the virus for them to multiply and spread.

      These enzymes are made of two parts that need to come together to work.

      If you know how the joining part works, you could make a drug molecule which the two parts of the enzyme would prefer to join with than each other, so the enzyme never works.

      So hopefully the virus won’t be able to work properly.

      (Correct me if I’m wrong with that one. Trained scientist, not a great biochemist however)

    • Sinomatic says:

      So would that be some sort of allosteric regulation/inhibition? (sorry if my terminology is way off, has been a while).

    • 12kill4 says:

      So essentially, Enzyme-1 and Enzyme-2 want to get together. However, if you know what turns Enzyme-1 on you can hire a super sexy, pro-choice enzyme hooker to come in and cock-block Enzyme-2. Enzyme-1 will forget all about Enzyme-2, thus preventing the birth of their awkwardly named children and AIDS.

      (Correct me if I’m wrong with that one. Trained sociologist, not a great biochemist however :P)

    • YohnTheViking says:

      12kill4; In an extremely simplified way, yes.

      In slightly more brainy terms you’d use a competitive substrate for the one the enzyme would normally bind to. One with a higher affinity, but one with a high kinetic requirement thus it remains locked in an enzyme-substrate complex. More commonly known as an enzyme inhibitor.

      With the information that has now been uncovered it’s possible to tell what this substrate would be and how it would be changed by the enzyme, because the active site of the enzyme is known all the way up to quaternary structure (the arrangement of the subunits).

      (Someone who spent his workday running PCR’s on chlamydia samples.)

    • Sassenach says:

      Simplifying that, they’re finding the carbon monoxide to the enzyme’s haemoglobin.

      (I think)

    • Lambchops says:

      Imagine that as a chemist i wanted to make a mimic to bind to a certain protein. Often it’s actually just some very small hotspots on the protein that are involved in interactions. So imagine the protein as a random shape with little bits of velcro on it where the interactions occur. As a chemist I could just make a shitload of molecules with the requisite velcro, chuck them at the ball and hope one of them sticks. This is a legitamite tactic. Sometimes it works. Often it doesn’t. On the otherhand if I knew the shape of the ball and where the velcro was I could design types of molecule to stick to it, and if it only sticks slightly I can try and rationalise why and continue to improve my molecule until it damn well sticks and becomes an inhibitior and suitable drug candidate.

      So the research is useful to a point.

      Why a point? Well because it’s a predictive tool that has been used largely on already solved structures. So if I’m doing some drug design I’m just going to look at the solved structures. Granted it could be used to predict the structures of other proteins and it might do it well. But I’m not going to expend effort designing and making molecules when there’s other structures out there that are already solved and perhaps more worth my time. PS iI don’t actually do this type of chemistry so I’m tallking in hypotheticals!

      Which brings me to another point. I don’t care! Why does everything have to solve AIDS or cure cancer? This (appears to be to me until someone more knowledgable in the field tells me otherwise) is an impressive piece of work on its own merits and therefore worthy of praise. As an example of impressive programming and being able to harness human intuition to elucidate complex structures it’s definitely fascinating. I think I’ll stop before I enter rant territory as I am changing the subject somewhat, but to cut it short, research from curiosity is (unless it veers into unethical territory) always worthwhile.

    • YohnTheViking says:

      Yeah, this never would have gotten the coverage it did if the words; “GAMERS CURE AIDS!!” couldn’t be attached to it.

      As for the research itself it’s a novel method, and important. But do remember that the result is exclusionary. It’s not about saying what will work, but what we can guarantee won’t.

    • tnankie says:

      blergh, thanks media another bash at science. “Baffles scientists for years” NO. IT. DIDN’T. It eluded a computational solution. But that isn’t a story, and don’t we like to feel superior to those scientist know-alls who don’t really understand how the world works in spite of all their years of study of an area, stupid ivory tower academics. Yes it is a story about science (so any publicity is good publicity) but it also a story about superiority and how science can’t get the answers but ordinary people can, therefore ignore science.

      Let us not mention the fact that there are drugs that work on HIV-1 Protease already, lets not mention that protease inhibitors are already a key part of current aids therapy.

      Sorry everyone, as a chemist who spends most of their day modeling HIV-1 reverse transcriptase (a different HIV enzyme) this article has left a fairly bad impression on me and now I am venting on you all.

      As for the structure, while it useful it doesn’t say anything of the dynamics of the system. There are 4 currently used drugs to treat my enzyme, several more in development and the crystal structure has been solved for 20+ years and yet HIV still kills. More to the point the HIV protease crystal structure has also been solved for quite some time, (1989, link to rcsb.org ) and HIV still kills, having a structure is not the solution, having a drug is not the solution (or 20 drugs in the case of HIV therapy).

    • Superbest says:

      I’m only a grad student, but I guess I might as well chime in: It’s important to note that while the protein itself is very interesting, the main “novelty” of this paper comes from the method which they used. John has simplified a bit when talking about 2D images in a microscope- you can’t see molecular structures at all. The best you can do is something called X-ray crystallography (which was famously used to deduce the structure of DNA), which is basically looking at a very distorted “shadow” of a whole bunch of the molecules and trying to guess the structure from that. Unfortunately, for that you need an extremely pure sample of protein (which can be very difficult to obtain), and interpreting the X-ray diffraction pattern is an art in its own right, and even then there’s the possibility that maybe the protein folds differently when crystallized (it must be a crystal ie dry to cast an interpretable shadow) as opposed to floating around inside cells. And besides, purifying a protein and then getting an X-ray crystallograph is kind of expensive, especially compared to things like this, which are as I understand it, well, free.

      Proteins are like beads on a string, except each bead likes to be near some beads and hates being near others. There are 20 basic kinds of beads, and they are sometimes further modified, and each protein has thousands of beads, so it is a bit of a problem to figure out which folding pattern will make the most beads happiest (because happy beads are low energy beads and we live in a universe where everything tries to have as little energy as possible). No-one has come up with a very good algorithm yet (or maybe we just don’t have powerful enough computers)- Folding@Home solves that problem by letting people donate their computers temporarily. FoldIt solves it by making human brains do some of the work- and apparently (I’m not very good at FoldIt myself) with good starting conditions the brain is quite capable of figuring out the solution.

      This is certainly a “big deal”- the structure of a protein is everything, it is literally like a lock and key in the sense that function is extremely dependent on what the protein will do. And one of the major bottlenecks in just about any kind of biological research nowadays is the question of, “So this gene makes this protein… But what is this protein for? What does it for?”. I’ve been excited about FoldIt for years, but it’s very nice to see them actually demonstrate that it can be used, in the present state, to effectively solve useful problems. This isn’t the first time a biologically significant protein has been solved by FoldIt, by the way. As I recall it a year or so ago the players figured out a cancer-related protein as well.

      I don’t really do protein modelling, so please excuse any inaccuracies.

  6. Dana says:

    What a troll argument to use in pc vs console “debate” :D

    • MiniMatt says:

      Oh indeedy, I’m sure it can be twisted into the most wretched non-sequitor. PC games cure AIDS, therefore XBox causes cancer.

    • Radiant says:

      Console players /are/ the lowest-energy configuration.
      etc. etc.

    • Joshua says:

      However, this is a ‘casual’ pc game.

  7. Luis_Magalhaes says:

    While I understand that RPS is a blog about PC gaming, I think it would be remiss not to point out that a significant part of the FoldIt gamer community does it with their Playstation 3.

    (for example, I leave my PS3 folding while I play on my PC :p )

    • henben says:

      I think you might be confusing Folding@Home with Foldit. Folding@Home is a program you can run as a “screensaver” which tries to solve folding problems automatically using “spare” computing resources, and it has a PS3 version, but there’s no PS3 version of Foldit.

    • Man Raised by Puffins says:

      This isn’t Folding@Home, it seems to draw on human input to manipulate proteins rather than the distributed computing approach of @Home.

    • Luis_Magalhaes says:

      Ah, true enough. I stand corrected. :)

    • YohnTheViking says:

      Folding@Home is pure tertiary structure as far as I can tell. It’s all about the 3D arrangement of the peptide chain, and doesn’t involve the arrangement of subunits which FoldIt seems to do.

  8. Anthile says:

    They should use crowd-sourcing stuff instead of hacking minigames in video games. Yes.

    • sonofsanta says:

      Yes. Yes they should.

      “Mass Effect 3: released March 6 2012. Thyroid cancer: cured March 11 2012”

    • Belua says:

      THIS must be the best idea I have ever seen. Not just related to gaming, but in general.

      You, my dear Sir, win 9001 internets and a free hat.

    • henben says:

      The problem is that this is way harder and more complicated than anything you’d get in a hacking mini-game. No-one would be able to solve actually useful structures in a reasonable amount of time. Maybe the very first tutorials, or a simplified version, would make a good hacking game, though – then you could solve more difficult problems for optional bonus items in-game.

    • Pijama says:

      Bloody fucking hell Anthile, this is seriously one of the best damn ideas I have ever read on the internet.

    • aerozol says:

      This could actually work… Devs, pay attention!

  9. Post-Internet Syndrome says:


  10. Ankheg says:

    Still, no cure for Pratchett?

  11. Ergates_Antius says:

    I’d be interested in seeing the microscope you can “look down” to see individual protein molecules (even as a 2D “splodge”).

    Back to school Mr Sydney M. Herald…

    • CMaster says:

      Yeah, I was thinking that one. It would have to be one hell of a big protein to see it in a microscope – although with say a confocal microscope, somewhat more conceivable. You’re not getting any structural data, 2d or 3d out of any optical technique though.

    • Ergates_Antius says:

      Oh, and “AIDS virus”?

      That’s Mr HIV to you and me.

      In all seriousness: It’s a bit hypocritical for us all to laugh and point at the bad reporting surrounding video games, then to do exactly the same thing with science reporting. OK, we’re far far from FOX News’ standards here, more like a local newspaper story about people who play RPG games like “Quake”.

    • Askeladd says:

      He was not accurate in that then, but he explained the point why we can’t just look at it and say “Hey, thats it!”.
      /pedant >

    • John Walker says:

      Oh phew, I thought you were saying that *I’d* got it wrong.

    • CMaster says:

      @Askeladd Not really. The reason you can’t just look at it, is that there’s simply no way to look at individual atoms – which is what protein structures are – the arrangements and bonds of individual atoms within a large, complex molecules.

    • Peptidix says:

      Atomic force microscopy (AFM) and Scanning tunneling microscopy (STM) get a high enough resolution to see atoms, but your protein will be stuck to a surface and as a result its shape will not be the one you are interested in. Electron microscopy doesn’t quite get to atomic resolution, but you can zoom in enough to study structures (eg keyhole limpet hemocyanin protein assemblies).

    • CMaster says:

      AFM, STM and the whole range of electron techniques aren’t really something you look down, though. Never mind that they’re all pretty unsuitable for looking at proteins. STM’s out right away, as proteins don’t conduct. AFM is only really atomic resolution height wise, you need something pretty flat to get that resolution out of it, and most proteins are probably too soft anyway. And electron microscopy tends to cook proteins, never mind the vacuu requirements. You can get somewhere with techniques like cryo-TEM and E-SEM, but you’re not going to be able to determine the exact location and orientation of say, the 7 arginine that way.

    • Lambchops says:

      @ Peptidix

      “Electron microscopy doesn’t quite get to atomic resolution”

      In some cases it can image small organic molecules. Saw a talk at a conference recently on it, if you’re interested and have access to journals and such you should check out this groups work (link to chem.s.u-tokyo.ac.jp) at the bottom of that page, it’s seriously impressive stuff. Slightly out of my realm of knowledge but it was definitely one of the most inspiring talks I’ve seen recently.

    • Lambchops says:

      Actually never mind imaging molecules, they’ve actually imaged reactions! There’s a good publication on it from this year, even without subscribing you can still watch movies of it (link to pubs.acs.org.) which I guess is pretty much only of interest to chemistry geeks!

      Kind of research that looks like it’s gunning for a Nobel price really.

    • Garg says:

      It is definately possible to image “atoms” with a HR-TEM, at least as long as they’re part of a lattice. Low voltage aberration corrected systems now have the resolution for the imaging of individual carbon atoms in a graphene sheet for instance.

    • Peptidix says:

      Thanks Lambchops, I know some people in Europe that have tried to measure single molecule reactions with STM and other techniques, but I cannot find any nice results by them.

  12. psyk says:

    How do they know it’s correct?

    • Askeladd says:

      That wasn’t explained but I guess you know it “when you see it”.

    • Demiath says:

      Oh, that’s easy; when you solve it correctly God intervenes with the appropriate “Achievement Unlocked” notification…

    • psyk says:

      Another thing if a “win screen” comes up then they must already know the “win conditions”.

    • CMaster says:

      According to the paper – in some cases they’ve been able to get the actual structure experimentally at some later point – here, the gamers haven’t been perfect, but they’ve been much closer than computer simulations. In the case of the Monkey virus, apparently the gamer structure was the only one to allow solution via “molecular replacement”, but I don’t really know what that means, and they seem to have no experimental data to compare with.

    • Burning Man says:

      It’s about evolving a model that could fit. If you develop a structure that would work under the conditions of a normal pH environment, that has an appropriately low total energy level, which means the binding energy is really strong…. If something could work, they test it via simulations. If it fails at some point, they iterate upon it. The better their model gets, the more complicated are the tests they can do to them. It’s a combination of direct (“I saw it under the microscope and it looked like this”) versus indirect (“But that protein would never react with that DNA sequence”) evidence.

    • foop says:

      The traditional way to work out a protein structure is to crystallise the protein (not a trivial exercise, can take years), then fire X-rays through the crystal to get a set of diffraction patterns. The atoms in the crystal diffract the X-rays in the same way that the slits in the Young’s slit experiment that you may remember from Physics lessons diffract light, and you end up with a pattern caused by constructive and destructive interference between the X-rays.

      There’s a problem with this, however. This only gives us half the information we need to solve the structure – we get the intensities of the X-rays in the diffraction pattern, but not the phases. To get the missing information, we usually use one of two techniques. We can stick heavy metal atoms in the protein, which gives such a stonking big signal that we can work out the missing phase information. This can be quite tricky to achieve. Alternatively, we can use the known structure of a protein that we think is pretty damn close to our protein and use molecular replacement in an iterative process to work out the phase information.

      The crowd-sourced structure was close enough to the crystal structure for it to be used as a starting point for molecular replacement. After a quick glance through the paper, it seems that the crowd-sourced model was pretty good.

      Of course, the structure of a crystallised protein may not exactly mirror that of a protein in solution inside the cells in your body. Nonetheless, crystal structures are excellent starting points for the design of drugs and the understanding of biological processes.

    • Lambchops says:

      As far as I got from the paper the crystal structures on the majority of the proteins they were working on had been (or were nearly about to be) solved and were being offered up to scientists wanted to test their prediction models against the real deal.

      Seems the results for FoldIt have been variable but in some cases very impressive, plus it makes a nice story for a paper and as anyone in science knows that’s half the battle when it comes to getting a good publication!

      Of course crystal structures are all well and good (I’m a chemist, I like ’em, they’re nice and definitive for small molecules) but I always suspect the biologists like them because they’re suckers for pretty pictures! They can give you a reasonable idea how a protein might interact but trying to predict how molecules interact in the body is a tricky business.

    • YohnTheViking says:

      Yeah, adding my voice to the already understood crystal structure.

      Problem is that a protein is much more than just the overall structure, and the devil truly is in the details with an enzyme. The most minute of changes in the binding site changes the activity of the enzyme (going from a “binder” to a “cutter” can take just a single amino acid jutting out).

      In addition to this it adheres to certain laws of thermodynamics (low energy cost), and ability to float around (hydrophobic elements being buried inside the enzyme).

  13. MiniMatt says:

    You know, those piccies look staggeringly similar to the tangled mess I call cabling inside my PC – using my patented “feck it, wodge it all in there, force the case shut and forget about it” method.

  14. sonofsanta says:

    But the important point is, how many of the players later went out and deliberately infected other people with viruses after completing the game? Computer games can only do bad, never achieve good!

    • pipman3000 says:

      They’ve actually invented new AIDS just so they an crowd-source them away, addicted to curing aids, tetris, etc, etc

  15. Jimbo says:

    Winning the platform war by curing AIDs, cancer and Alzheimer’s. I didn’t see that one coming.

  16. Joe Duck says:

    This is awesome news, specially because it looks as if the research results are worthy by themselves and not because of how they were obtained. However, we need someone to hide this news from Jane McGonigal and other telepreachers, as it is just exactly what they need to keep on selling bullshit and “predicting the future”.

  17. Skg says:

    I have to say, I’m a little confused here. While I haven’t as of yet read the papers on this, it seems to me that they’re only solving one part of the problem. The energy minimisation approach to protein alignment isn’t something that is that difficult to solve (ok, it’s a pain, but it’s not horrible, and has been solvable for years), usually using Bayesian techniques, which get pretty close to the local/global minimum bond energy, but it doesn’t always give the full structure of the protein at hand, because there are other factors in play – and it’s these factors that more recent techniques have been trying to account for.

    The energy minimisation approach works, and often does give, if not the best, but a pretty damn good approximation to the protein structure, but I’m slightly confused that the information as presented (and I’m not blaming RPS for this, but more the way it’s being presented by the scientists and the other media outlets that RPS are sourcing) seems to be portraying these results as really significant….when I’m not quite sure they are, in the face of all the current research in bioinformatics. Because in the end, it’s just solving the energy minimisation problem, and can never really account for the functional variations we sometimes see that stray away from the minimum bond energy case.

    • CMaster says:

      Go read the paper. It’s not behind a paywall on the above link and is pretty easily readable.
      But in general – the foldit players have been found to get closer to real structures than existing energy-minimization simulations. In the case of the monkey virus protease, no experimental crystal structure has yet been reported, despite several attempts. Supposedly the foldit guys produced the first “possible” structure to have been suggested although I don’t fully understand some of the terms used when they talk about verifying the result.

    • Skg says:

      CMaster – crystal structures aren’t really that great an indicator of functional protein structure though, as the process of crystalisation can distort your results. Well, they’re good, because it’s a nice way of seeing what’s happening, but they aren’t necessarily a representation of the functional structure.

      I’ll grab the paper tomorrow, as I’ve others to read tonight, but the point is that it’s still treating the whole problem under an energy-minimisation framework. As you say, the restults apparently come in to be slightly better than that of standard energy-minimisation search techniques, but from the way RPS has presented it, it’s still, in the end, an EM technique, just using distributed geek brainpower. And if it’s still just an EM technique, it still shares the same problems that other techniques have been trying to resolve since the early 2000’s.

    • CMaster says:

      Well yes, but biochemists do seem to love their crystal structures so. It’s not like there are many practical ways of determining actual physiological solution structure.

    • Skg says:

      Yeah, fair point. I just pointed that out as an aside, because while I think there are other ways of determing the structure, I cant remember enough to be sure, and I had the structure under crystalisation thing pointed out to me last week. But the point still is – if it’s treating it as an EM problem, no matter if it gets a lower energy than the current crop of EM techniques, it’s still bound by the problems of EM techniques. It’s an interesting exercise, sure, but, I’m curious to see the level of difference.

      That said, I keep on prattling on about the problems of the EM algorithm, but in the end, minimum energy does, in most cases, give a pretty damn good representation of the structure as a whole. So while I don’t think it’s particularly efficient or really practically useful bioinformatics technique, I’m still curious about it. Moral of the story – I need some sleeeeep

    • YohnTheViking says:

      They do seem to be going for the lowest energy requirement model, but there’s a lot more to the actual problem solving here than the first two paragraphs of the paper (which has been quoted). And that part of the paper really only deals with how the FoldIt program works in general.

    • Ergates_Antius says:

      I think that the way these results were generated is actually more important that the results themselves.

      It’s kind of a proof of concept for crowd-sourcing complex problem solving.

      The distributed computing projects like SETI@home and folding@home have been running for quite a while now. This is the next step – applying the same principle to problems that are better solved by humans than computers. If you can work out how to gamify* a problem (plus add a competative slant) you can get people to volenteer to work on it for free…


  18. Joe Duck says:

    Foldit sounds like a early prototype of GLADos, just saying…

  19. KillahMate says:

    Fun how the comments on that MSNBC article devolved into a Stupid Conservatives vs Whiny Liberals debate, of all things – and within one page or so.

  20. Xocrates says:

    I wish people stopped saying they cured AIDS through this. That’s the equivalent of saying you caught a criminal while what you did was find a picture of his brother.

  21. Lars Westergren says:

    A similar game-for-science of folding developed by Carnegie Mellon University and Stanford University is

    I played it for one evening, and then I drifted away from it. Feeling terrible.

    • roy7 says:

      I thought EteRNA was pretty cool. Very nice interface, and the leading designs get created in real life now and then for actual testing. Never played FoldIt. Will have to check it out. :)

  22. Spooner says:

    I’m curious how this human system compares with brute-force systems like Folding@Home. As far as I’m aware they are working towards the same thing, aren’t they? The main difference is that even dumb people can “gift” a CPU core for the latter system :D (I’ve had one core of my quad working for 3 years on folding and I have no idea if it has really made a difference).

    • YohnTheViking says:

      Lower time requirement.

      To make the whole thing more understandable for a PC gamer. Which is faster; brute forcing a mod, or reverse engineering the code with a bit more manual labour?

      You wouldn’t even dream of trying to brute force it because the possibilities are too damned many, and that’s kinda what Folding@Home does. But a lot of those proteins don’t have very good existing models and games like FoldIt (which require a base structure to be in place) simply will not work.

  23. Hmm-Hmm. says:


  24. Peptidix says:


  25. ArcaneSaint says:

    So, does this mean that the Nobel prize for video gaming is finally going to be introduced?

  26. AlexB2015 says:

    John Walker-
    Your article is very amazing and reached out to the gaming and science community. This is a very touching article that reached out to me in so many ways also. Not only was I proud of all the gamers out there but I was also amazed that they had created a game for assisting curing fatal diseases such as AIDS. Maybe someday we will be able to teach in schools about how this helped a ton to curing diseases. There will be so much glory! The world will be so happy we can finally find a vaccine to eliminate certain diseases. Wouldn’t that be so fantastic if the modern day world was AIDS/Alzheimer’s free? This piece of writing was over all, one of the most awesome pieces of writing, it spreads its wings and gives credit to two great communities. Gaming and Science Communities. Keep up with your writing i enjoy reading it!

  27. Bongs says:

    seo service
    Awesome ty so much. I think that is pretty much exactly what I was looking for. Couple quick questions and I swear I will leave you in peace

    As far as RAM goes is there a particular style that I need to make sure to always purchase? For example is there only a certain type that my Motherboard accepts or is it pretty general? There are so many different types but sometimes it only seems as if you’re paying for a prettier color.

    Lastly will my case require any additional fans for cooling? I am sure if I OC’d my CPU I would but if I leave it stock for the time being and run the 560 graphics card will the case provide adequate enough cooling?