Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Biotech Medicine

Crowdsourcing HIV Research 52

biolgeek writes "In recent years, HIV has been managed with a collection of therapies. However, the virus will likely evolve around these drugs, making it crucially important to get a better understanding of the virus itself. An important step in understanding the virus is to get a handle on its genetic blueprint. William Dampier of Drexler University is taking a novel approach to this research by crowdsourcing his problem. He is hosting a bioinformatics competition, which requires contestants to find markers in the HIV sequence that predict a change in the severity of the infection (as measured by viral load). So far the best entry comes from Fontanelles, an HIV research group, which has been able to predict a change in viral load with 66% accuracy."
This discussion has been archived. No new comments can be posted.

Crowdsourcing HIV Research

Comments Filter:
  • Wow (Score:3, Insightful)

    by phantomfive ( 622387 ) on Wednesday May 05, 2010 @04:39PM (#32104102) Journal
    Wow, I would love to get involved with this and help find the cure for AIDS. Unfortunately I don't really have the expertise to do ANYTHING related to it, and I'm not sure many do.

    I'm not sure you can call it crowdsourcing when the number of people who can get involved are so small. Maybe a contest or an open research project or something. Either way, I wish them luck.
    • Re:Wow (Score:4, Informative)

      by Capt.DrumkenBum ( 1173011 ) on Wednesday May 05, 2010 @04:49PM (#32104276)
      Well you could get involved with:
      http://www.worldcommunitygrid.org/ [worldcommunitygrid.org]
      and donate some processor time to FightAIDS@Home

      Just a thought.
      • Re: (Score:3, Funny)

        by Chris Burke ( 6130 )

        But I already fight AIDS at home by engaging in an alternative to one of the most common transmission methods.

        Why, I've probably prevented over 100 cases of HIV in just this year to date!

      • Re: (Score:2, Interesting)

        by drinkypoo ( 153816 )

        I find this kind of stuff amazingly ironic. We're going to burn more fuel and produce more cancer-causing emissions so that we can produce more CPU cycles to fight illness. I don't know about you but my operating systems send HLT instructions to my processors when they're idle.

        • Re: (Score:2, Insightful)

          by tignet ( 1303483 )

          Now that you mention it, all research requires fuel that, at some level, produces cancer-causing emissions. All research should be stopped! We've known about this [wikipedia.org] for a long time, why produce all those cancer-causing emissions looking for 'better' treatments?

          Although, I suggest that instead of sending HLT instructions to the processors as part of the idle loop, you should turn your computer off when it's idle. Think about all the energy you're using; the cancer-causing emissions are too much to bear. Wait!

    • Re: (Score:1, Insightful)

      by Anonymous Coward

      I'm not convinced you NEED any knowneldge of AIDS. At the end of the day, it's just a data mining problem.

    • Unfortunately I don't really have the expertise to do ANYTHING related to it, and I'm not sure many do.

      Judging from TFA, if you're posting here you probably have at least most of the expertise to do it, quoting:

      Models can be trained using the records of 1,000 patients. To predict an improvement in a patient's viral load, competitors will be provided with data on the nucleotide sequences of their Reverse Transcriptase (RT) their Protease (PR) and their viral load and CD4 count at the beginning of therapy. There is a brief discussion of the science of these variables in the Background section, but no knowledge of biology is necessary to succeed in this competition. Competitors' predictions will be tested on a dataset containing 692 patients.

      Sounds like they give you the data and tell you what you're looking for, and then you tell them when you've found them.

      I'm not sure you can call it crowdsourcing when the number of people who can get involved are so small. Maybe a contest or an open research project or something.

      That may be why they don't, but "biolgeek" does. The page specifically calls it a "competition" and "contest." No mention of crowdsourcing, which doesn't mean a lot to most people.

      • People often refer to 99designs logo competitions as crowdsourcing designs (http://webworkerdaily.com/2008/04/28/99designs-crowdsourcing-works/). On that basis, I think this classifies as crowdsourcing.
    • by Yvanhoe ( 564877 )
      I have the feeling that you are only 3 hours of wikipedia-surfing away from being able to translate it into a pattern-matching and algorithm-finding problem. If you know how to code and are good at algorithms, you could help.
  • by jameskojiro ( 705701 ) on Wednesday May 05, 2010 @04:39PM (#32104124) Journal

    Isn't this what caused so many people to come down with it in the first place?

  • by jameskojiro ( 705701 ) on Wednesday May 05, 2010 @04:41PM (#32104156) Journal

    Infect Huge Numbers of people with HIV and after 10 years breed the survivors and infect them again, after 100 years the crowds are dwindled down and whomever remains is immune, HIV epidemic, SOLVED!

    • Re: (Score:3, Insightful)

      I was under the impression that some African countries are more or less doing this with little to no success in gaining immunity.

      • Actually, we've already bred people with HIV resistance [wikipedia.org]. Why nobody is intensively studying these people to document WHY they remain asymptomatic for so long is beyond me.
        • Re: (Score:3, Informative)

          by Anonymous Coward

          We are being intensively studied. See www.zephyrfoundation.org .

          • I stand corrected, and thank you for doing so. The study of Elite Controllers sounds very promising; I'm surprised I've never heard of the Zephyr Foundation before. Also, the fact that the Elite Controllers they were studying were all infected by transfusions destroys the myth that HIV is spread through "irresponsible behavior", as well as my personal theory that long-term non-progressors developed immunity though exposure to low levels of the virus over an extended period of time. So it looks genetic -- so
      • No, it's going fairly well from a long term perspective.

        Just seems that some people object to the pain and suffering necessary for this method to work.

    • Re: (Score:3, Insightful)

      Or just produce people with the CCR5-32 gene variant.

      • Or just produce people with the CCR5-32 gene variant.

        Didn't we find out that would work through a screening process in humans pretty similar to what jameskojiro suggested? In other words, do we know that people with the CCR5-32 gene variant are immune because out of the millions of people with HIV, there are a few people we've found are immune and those people have that gene variant?

    • so I see. A giant 'catch a bug' party for everyone? Do you think majority would do this to themselves maybe to solve the problem for the future generations? Why would anyone sacrifice themselves for the future generations ever, that's a stupid thing to do.

  • Lowest bidder ... (Score:4, Insightful)

    by DeadDecoy ( 877617 ) on Wednesday May 05, 2010 @04:48PM (#32104262)
    They're really going for the lowest bidder if they want to crowd source this problem:

    There is $USD500 up for grabs, and the winner(s) will also have the opportunity to co-author a paper with the competition host. The winner must supply their methodology before any prize money is awarded.

    $500 amounts to around a week or so worth of work, not counting resources used like hardware and computing time. And also, the prize is you get to be a coauthor? If you develop a novel algorithm that has a statistically significant improvement over prior methods, you should damn well be the first author with the host being the coauthor. A more interesting crowd-sourced competition should involve a >$100k prize with a publication in some significant journal like nature, bioinformatics, or new england journal. That would at least attract the hardcore statisticians to your cause.

    • $500 amounts to around a week or so worth of work

      Yeah, they can't afford much. The budget is kind of small - most of the money was allocated to fighting baldness and impotence.

  • If you crowdsource AIDS Research/Treatment, you'll find that upwards of 50% of treatments will be complete crackpottery/homoeothermy/witch doctors, 45% of finds will be from religious groups/AIDS denialists, and the remaining 5% will be paid astroturfing from pharmaceutical companies. A vanishingly small percentage of suggestions will be from actual medical researchers.

    The internet is a cesspool of misinformation, junk and lies. Sorry but that's just the way it is. If you give everyone on earth their own pe

  • Will the solution be mine, his or some company who will charge money for it and say they invested a LOT in it and make profit from it?

    • The algorithm will be published as an open-source paper ... likely BMC Bioinformatics but I guess that depends on any number of factors. btw. This is Will Dampier, I'm hosting the competition with Kaggle.com and Drexel University
  • Gross (Score:3, Funny)

    by davidbrit2 ( 775091 ) on Wednesday May 05, 2010 @04:54PM (#32104358) Homepage
    This strikes me as the sort of disease where you'd want to stay away from phrases like "viral load".
    • by treeves ( 963993 )
      Is there a sort of disease where you *wouldn't* want to stay away from phrases like "viral load"?
  • A small typo, its actually "Drexel University" in the Philadelphia Area
  • From the forum page [kaggle.com] over at Kaggle:

    The Fontanelles is a group which does HIV research professionally and so has some specialized information in this area. We're disqualifying our entry, but have put it in just for fun as a target. We may be back if someone beats it.

    So despite the appearance of the professional entry, this looks like an interesting contest that anyone might enter.

  • In an effort to actually decrease the barrier for entry I'm working on a quick example function for predicting the dataset. I'm hoping to have it posted in a day or so. I'm actually defending my PhD thesis within the next few weeks so I'm kinda pressed for time. btw: This is Will Dampier, I'm hosting this competition at Kaggle.com
  • by jd ( 1658 ) <imipak@yaTOKYOhoo.com minus city> on Wednesday May 05, 2010 @08:34PM (#32106598) Homepage Journal

    Can't help but wonder, though, if they're trying to solve the wrong problem. There's research out that suggests that virus-related cancers are exploiting what are effectively a small set of security holes in the way DNA handles the "junk" portions. HIV is a retrovirus, IIRC, which also means it installs itself into the DNA. The first line of attack I'd have thought of, based on those two pieces of information alone, would be to see if the SAME security holes are responsible for both the virus-caused cancers and HIV breaking into the DNA. If there is a common attack vector, across multiple viruses, then that attack vector becomes far more interesting than the specifics of each virus.

    Assuming the attack vector cannot actually be patched in mainstream cells, to fix the flaw, then perhaps it can be fixed in T-Cells, which are essentially disposable and it doesn't matter a whole lot if they're non-standard. HIV crashes the immune system through a massive DDoS attack via the immune system itself, by using the T-Cells. If the T-Cells are closed to that specific attack, then the virus can mutate all it likes but it can't crash the immune system. IF it is invariant across multiple viruses, then it's likely invariant across all of HIV strains. Merely preventing a DDoS on the immune system should massively slow the virus down and improve the chances of additional treatments actually ridding the body of the virus.

    The ideal would be to fix the security hole in total, for all cells. I'm not sure that's possible, as evolution has required the mechanism to inject new code into the DNA strands. Indeed, a lot of evolution would be impossible without such a mechanism, and you can't exactly install X.509 certificates into all harmless or potentially beneficial RNA and DNA sources on the off-chance they need to integrate. Besides, cell defenses don't usually include SSL. The best I think you can probably do is bio-engineer a new DNA strand, which you can install in an organelle (organelles are just places where cells used to have DNA before all the useful bits were pushed over into the nucleic DNA), which provides some sort of Intrusion Detection System. As I see it, you've two options - a honey-pot (an extra-vulnerable DNA strand that causes the whole cell to self-destruct if infected by a retrovirus), or a Tripwire-like IDS that looks for mutations in any given strand of nucleic DNA =and= monitors for virus-production. If both conditions are satisfied (ie: it's not a benign insert, but a malign one), then the strand is broken up.

    Again, not sure if this is remotely possible. Sure, there are enzymes which break up DNA - they're used all the time for sequencing, as you can't sequence long strands. But to identify a malign region in the DNA =and= have the enzyme only break the DNA at that point =and= have this done in a way that won't cause the end result to do strange and undesirable things -- that's going to be tough.

    So if this approach is so tough, why go for it? Because researchers have tried targeting the virus directly and have failed utterly. Deactivating it only results in it reactivating itself, making vaccines extremely hard to produce. When they are produced, the virus has mutated and the vaccine is useless. In other cases, the virus has even used the immune response to hijack more immune system cells, so as to spread faster. It also mutates so fast that what worked one week won't work the next. Direct attacks have no serious shelf-life and just won't work.

    That leaves indirect attacks. To beat the mutation problem, you need some aspect of the virus that will never change. If one such aspect is the mechanism for breaking into nucleic DNA and inserting rogue sequences, controlling that entry-point will not only beat AIDS, but it will beat some cancers too. Since uncontrolled entry into DNA is why some gene therapies cause cancers, controlling the entry point will also be critical to gene therapy being successful for a wide range of conditions.

    Thus, this is the obvious place to focus on. Ign

    • by comm2k ( 961394 )

      Sure, there are enzymes which break up DNA - they're used all the time for sequencing, as you can't sequence long strands. But to identify a malign region in the DNA =and= have the enzyme only break the DNA at that point =and= have this done in a way that won't cause the end result to do strange and undesirable things -- that's going to be tough.

      There are some approaches with modified/fine-tuned Zinc-fingers cleaving (your own) DNA - one example see this article: http://www.health.am/aids/more/proteins-put-personalized-hiv-therapy-within-reach/ [health.am]

      • by jd ( 1658 )

        Again, this is exceptionally interesting and because it doesn't directly attack the virus itself but attacks a vector being used, there is a greatly reduced risk of the virus mutating to work around the fix. It's a relatively novel approach, which in my opinion is critical to any actual solution. And because it's effectively for building a bullet-proof T-Cell, it shows that my thinking wasn't totally off in left field, that closing attack vectors is an area researchers consider interesting.

        I don't know how

  • The research is actually being done at Drexel University in Philadelphia PA. If anybody was wondering...

If you suspect a man, don't employ him.