Playing Games For The Good Of Humanity, With Phylo

By John Walker on March 7th, 2012 at 4:48 pm.

Science!
Hopefully you remember FoldIt, the protein manipulating game that instantly solved years-old biological puzzles. It’s probably the most well known example of a crowd-sourced citizen science project that had immediately remarkable success. It’s in this same field that Phylo exists – a game designed to sequence genomes, while the player is enjoying some fairly familiar match-3-style colour matching. I spoke to designer Jerome Waldispuhl, Assistant Professor at the School of Computer Science McGill University, Montreal, to find out how it is that fun can be recycled into progress. This is gaming literally changing the world, perhaps even saving lives.

Multiple sequence alignment. That’s the key phrase to stare at in some confusion here. It means, to the best of my limited understanding, to compare the DNA, RNA or proteins (anything made up of nucleotides – the building blocks of life – basically) from different species, to seek commonality, and from that, infer a common ancestor. This is phylogenetic analysis, and it’s from this that the game, Phylo, gets its name. And it’s all pretty bemusing stuff, depicted as sequences of what look like the scribblings of a lunatic, madly complex collections of capital letters in row after row, with highlighted colours representing the most complicated washing up rota anyone has ever constructed.

Crowd-sourcing/citizen science is certainly not a new phenomenon. In fact, there are examples going back over a hundred years, with projects like Audubo’s annual Christmas bird count. Getting lots of people to do a small amount is a great way to get lot of stuff done in a small amount of time. And observation is one field where computers haven’t so neatly stepped in. It seems that working out the best patterns for aligning genomes is best done if human imagination is involved. But for a research group, that’s an improbable volume of man-hours to get done.

So turning that outward requires incentive. Getting the general public to do your research for you requires a couple of elements – interest and entertainment. And if that can be combined, you may well find an enormous audience willing to help with your multiple sequence alignment genome project. In Phylo you’re asked to move coloured blocks in horizontal rows, to create groups of matching colours while avoiding leaving gaps.

“If the game is not fun nobody will play it and it’ll be useless,” explains Waldispuhl, who has already seen remarkable success with Phylo. “The success of these approaches reside in the entertaining value of the game. It is nicely quoted in a paper of Luis von Ahn, arguably one if not the most important pioneer of this field, on human computation.”

Of course, that’s true of any game. If you want people to play your game, it needs to be good, right? So why especially here? “It is even more important for a game like Phylo which aims to fully abstract the research problem to a tetris-like game,” the scientist told me, stressing that they were focusing on constantly improving this.

That’s the key here. I’ve attended a talk by Waldispuhl, asked him questions after, and exchanged emails, and I still don’t understand how it is that my playing a puzzle game directly correlates to my helping to sequence the nucleotides of various species in a way that could dramatically advance scientific knowledge. And while I have a peculiarly good memory of DNA structure and synthesis from exams I failed sixteen years ago, and indeed a wife who works as a research biologist, I’m fairly sure that actually knowing what it’s all about would require a few years of study, rather than a scan of a Wikipedia article and some well-worded bedtime queries. If I’m to usefully play this game, it’s because I’m going to be enjoying myself, not because of an informed altruism.

But can that get in the way of pragmatic progress? Doesn’t needing to focus on being fun mean you’re distracted from the core purpose, or at least taking the long way around? “I don’t think so,” says the professor. Instead he argues that it requires someone to be more creative, and stresses the need to work with game design experts. “It actually opens a new avenue of research,” Waldispuhl suggests. “What can be solved with fun?”

It strikes me that there could be another issue here. At the start of the article I asked whether you remember FoldIt, and not whether you were still playing it. While I’m sure a few readers are, it seems realistic that what received a lot of play when the press covered it and word spread, will see a significant drop off. Keeping that interest long-term is a lot harder. And while that’s normal for the average puzzle game, it seems to have slightly farther reaching consequences for a project that has such significant consequences.

“That’s a very good point, replies Waldispuhl. “That’s also why I wanted to design a casual puzzle game that could be a ‘classic’.” Which is a big ask. “We’re still working actively on improving the game design. My goal is to have a Tetris-like puzzle that everybody can play (from children to seniors), that is easy to understand and that will never look outdated.” Ambitious, certainly, but then it would seem foolish to go into such a project without those ambitions, especially with the intended goals.

And those goals are being met. The game has already had 600,000 visits, with 450,000 puzzles completed. And perhaps more importantly, 70% of regions analysed have been improved. That’s 521 genes better understood. And the resources were free – the consequences of that are enormous. If you’re looking for a grant, and you can tell your committee that you can get half a million people to work on your project, without having to pay them, you may well get some attention. Waldispuhl describes this as recycling. A casual game that recycles the energy spent by gamers to solve a scientific problem.

There are gaming problems to solve too, and that’s partly found in competitiveness. With high score tables, measuring both immediate points scored playing, and usefulness of the time spent playing, there’s an incentive to have another go, to do better, and therefore to increase the findings of the project. But I wonder whether the largest problem here is the sell. Might it not be tough to ensure I care about genome sequencing in the first place, motivating me to opt for this game, rather than any number of others?

“That’s a risk,” says Waldispuhl. “But I think that very soon genome sequencing will be everywhere.” But it’s harder to convince people that the multiple sequence alignment is important, surely? “Having its genome will be essential for public health and understanding the risks for the individual, thus to prevent them. I think it’s easier for the public to understand the importance of genomes which has a broad impact on our health, than the specific interest of a group a researcher on a particular molecule. That’s a personal point-of-view of course.”

But really, the success or failure of the project really falls on whether people enjoy playing the game. Which strikes me as, well, a remarkable situation. Progress in scientific fields (and when you look at FoldIt’s astonishing results, and the success Phylo has already shown, we’re talking genuine progress) depending upon the appeal of a puzzler. Or as Waldispuhl puts it, “Playing games for the good of humanity.”

__________________

« | »

, , , .

48 Comments »

  1. AmateurScience says:

    I did a double take there when I alt-tabbed away from my multiple sequence alignment only to be confronted with multiple sequence alignment…the game.

    I love it when work and play overlap!

    Edit: It has just dawned on me how much fixing an MSA after the computer’s had a go is like a sliding tile/match 3 puzzle. I shall now complain less loudly when the workstation spits out an alignment that is off-key.

    • Mirqy says:

      I used to do this for a living! Ah, the As, the Gs, the Cs – even the Ts. How I miss them. And the acrylamide. If only lining up sequences had a peggle-like ode to joy moment when you get it right and I might still be doing it now.

    • Bhazor says:

      Not the U’s though. Never work with those backwards fuckers.

    • lnbxzsg says:

      With the constant beeping, limited time to work on a puzzle, and lack of feedback, it’s like they want to penalize you for thinking or learning. http://no.nl/hEYS

    • YohnTheViking says:

      Acrylamide? You really have been out of the game for a while. Now it’s all about pyrosequencing, or failing that; capillary electrophoresis.

    • G-Lord says:

      I’m pleased to see that I’m not the only RPS user who is in life sciences. Have to try out the game soon.

  2. Cryptoshrimp says:

    This is progress. It’s also awesome. Awesome progress.

  3. pixelprime says:

    I often wonder (and hope) that many of the world’s problems will soon be solved using this method. Although my initial impression judging from the screens is that it might be quite complex, like the Bacillus game on Kickstarter (which is all about Microbiology), it might only appeal to certain audiences.

    That said, I applaud their efforts in bringing a consumer-friendly product like this to the masses, as it only helps further the cause that gaming isn’t just about warfare and violence – it also has tangible, noble goals.

  4. krisanto says:

    It would be interesting once someone figures out a way to make this work with an MMORPG. A collaboration with Blizzard perhaps?…or Zynga.

    • Torgen says:

      Oh, could you *imagine* the crapstorm from the creationists and fundamentalist masses if EVOLUTION was HIDDEN IN THEIR GAMES??? (Yes, mapping genome sequencing is different, but beyond the average Facebook user’s willingness to comprehend.)

    • pixelprime says:

      Help world farming by playing FarmVille:Nations!

    • mickygor says:

      Torgen, it’s funny you should say that, because it probably is. Genetic algorithms have their uses in video games.

    • Kdansky says:

      Actually, there’s a SC2 Zerg roach-rush build that was figured out by a genetic algorithm. For a while it was really common to see it, because it’s incredibly hard to defend against if you don’t know it.

  5. jezcentral says:

    This needs to be an App. I would happily play this on the tube for XX hours per week (and upload the results later) instead of Angry Birds. The problem with it being on PC is why would I play this when I have the option of playing something like Mass Effect 3 instead? (Yeah, yeah, helping my fellow man, for science, medical breakthroughs that would extend my life for decades, yadda yadda yadda).

  6. Post-Internet Syndrome says:

    Nice concept and a fine game, but I got really frustrated by the timed element. It’s a flash game, and so has terrible performance because flash sucks, so there is a slight mouse lag. Forcing me to speed puzzle under such conditions is just sadistic.

    In addition, checking the ancestor patterns and par score require you to hover the mouse over objects that are far away from each other. Proceeding to the next stage requires you to click a button that is in yet another place. In addition, mass-selected boxes are deselected after moving them, requiring you to reselect them if you want to test another configuration.

    Also, the intervals between updating the score just baffles me. Why delay information that the player needs in a timed game? Calculating the score yourself can be done, but for most people (including me) probably takes much longer than the handful of seconds of the delay. This leads to me sitting and waiting for the delay so I can see wether a given change was an improvement or not.

    These are all small but very annoying design flaws that will turn off people like me right away, and people who doesn’t notice it right away will soon enough.

    • pixelprime says:

      Saying ‘Flash sucks’ because the game has terrible performance is an odd bit of reasoning. It’s more than likely the game’s implementation of mouse handling that results in less than agreeable mouse performance, as there are multiple ways of getting fantastic performance with all input methods through Flash.

      You make good enough observations about the game’s playability later on in your post to avoid having to make such sweeping generalisations as that, but everyone is entitled to their own opinion.

    • Torgen says:

      I think PIS is spot-on with the UI observations. It seems that these things were deliberately added to eat up time and increase stress to the user. There’s no reason at all not to include the info that you’re forced to hold the mouse button down on the polygon to get.

      I do not feel inclined to play any more, as the time limit is completely arbitrary to the supposed goal of getting free help with genome sequencing. Perhaps I *want* to take time to ponder the best sequence and help cure cancer? Punishing me for that is the worst possible thing to do. (I am also apparently rubbish at saving humanity while racing against the clock.)

    • Chris D says:

      I agree with most of what Mr Syndrome says. The game seems intent on adding an artificial competitive element with the time pressure and hidden information at the expense of allowing me to actually learn the game. It would be really useful to be able to see all the information in real time. If I’m supposed to be comparing common ancestors it would be nice to actually be able to see both of them together.

      Maybe also let me see a better solution so I can figure out what I was doing wrong. If I had enough time I could probably work it out on my own but it’s not letting me do that either.

      I really wanted to like this and was prepared to cut it a lot of slack but as it stands it’s a massively frustrating experience.

    • LCinn says:

      yeah, what the hell is up with the time limit? There have been sequences that really intrigued me, and that I would have gladly tried to figure out the `solution’ to. Is it because rushing us through puzzles, hoping we stumble upon solutions to some of them is more efficient than letting us ponder the ones we do not easily get? Because if so, I think that is a rather destructive way to deal with gamers-as-resources.

    • Harkkum says:

      I have to join the bandwagon; already the first level on the easiest setting was an unsurmountable task for someone like me and entirely in spite of my best intents, I cannot force myself to play this. I have no clue whatsoever as of why the combination I figured out provided me with +2 points and an another -2 although they were for me quite identical. I tried to play through the tutorial but it was n-times easier than the actual puzzle element of the game, which even accentuates the entirely arbitrary nature of the timer.

      Remove the timer and scoreboards from us mere mortals and just provide us with a brain puzzle of trying to solve those tricky sequences. If someone has to compete on DNA-sequencing just provide an option for that with achievements linked, but if the goal is lofty, better make it accessible to us normal people as we’re ready to chant: “The internet is for porn, not for DNA-sequencing!”

    • Lacero says:

      UI wise it needs little expanding +2 or -10 numbers coming from the place where you get the points from (coloured green + red (with the green a blue green for the colour blind)).

      And yes the time limit seems really stupid.

      Finally flash sucks because aftr playing it once it no longer loads. This doesn’t happen with unity.

    • Torgen says:

      Lacero, one of the buttons overlays symbols on top of the colors. I found it totally by accident.

    • Caiman says:

      Playing this game is like sitting through a lecture given by a taxonomy student talking about their thesis – ie. I got lost within 2 minutes because it seemed to assume a certain level of prior knowledge that wasn’t provided. The tutorial is fine, but then come level 1 on the easiest setting and I have little idea why something works or doesn’t work. This is probably why I never continued with taxonomy! Neat concept, but needs far better implementation.

  7. Strange_guy says:

    I wasn’t getting any high scores for any of them, so I guess I wasn’t contributing anything, and the games not that fun. So I gave up.

    • Skabooga says:

      Welcome to the world of science.

    • Xocrates says:

      One of the main reasons I gave up on biotechnology (the other being unemployment).

    • Harkkum says:

      @Skabooga

      …but the very purpose of this is to make it not-science so that the contribution of the masses will ameliorate the sequencing in an unforseeable fashion. It is much like that age-old Condorcet’s formulation where even slightly better than arbitrary choice (us commoners) will lead to absolute certainty if iterated enough of times. As the graphical cues ought to provide us with guidance to make our choice something else than a mere guess, it should be possible to achieve great certainty by just using our numerous repetitions. For this, highest scores ought to be inconsequential but rather gradual improvement (for which there is none on this game).

  8. MOKKA says:

    I really like the idea, but the developer should get rid of this picture in the bottom right corner. This is not an accurate illustration of humand evolution, or of how evolution works in general.

  9. Urthman says:

    It’s really hard for me to believe that a computer can’t be taught to play this game much faster and better than a human.

    • AmateurScience says:

      The tools we use at the moment have some pretty sophisticated maths behind them, but without fail, every time I do this with a handful of shortish reads (5-10 sequences of ~1000 letters) the workstation will get the jist of it, but I’ll have to spend the rest of the day sifting the sequence and shifting bits and bobs around to get a really good match to use as consensus.

      People are doing this kind of thing with 20-30 species and big chunks of chromosomes (couple of million letters) and it’s these little variations that cause ‘noise’ in the results.

  10. Xocrates says:

    Ok, what the hell? A perfect match between the top and bottom sequence isn’t good, but adding a massive gap was the answer?

    Either the game is broken or it is very poorly designed.

    • Mirqy says:

      Depending on what exactly they are trying to solve, a massive gap might be exactly the answer. Two genes from two individuals might be different only in that one of them has a big insertion in it. Obviously looks very different on paper, but this reflects the ways genes can change across generations, ie by having bits of other genes randomly stuck in them.

      This is making me misty-eyed for my student days.

    • Xocrates says:

      Trust me, I know that.

      The problem is that the game treats a perfect match as being wrong without further elaborating. So either I, as a player, lack the correct context since the game fails to provide it, or the game is outright broken.

    • InternetBatman says:

      I had that problem too. The game doesn’t telegraph its rules very well, they just say that gaps will cost you when they don’t actually cost you, not all of the time. I got a high score that’s probably gone now, but hell if I know why doing something which should penalize me increased the score.

      Also, it’s annoying to see absolutely beautiful matches where you can tell exactly where a mutation was fail in the scoring.

    • Chris D says:

      I suspect the issue is that scoring is based on similarities between each pair of ancestors, not just what you see in front of you but unless you watched the video tutorial and not just the playable one there’s no way you’d know that. Even when you do know that the game doesn’t exactly go out of its way to make it obvious what’s being compared to what.

      It’s a nice idea but this one definitely needs some work.

  11. Roxton says:

    Having spent a fair amount of time on it now, I’ve put a review together (click my name). Shameless self-promotion aside, I find myself agreeing the comments above. The core mechanics are ok – nothing special, but it’s for a good cause – but there are just so many irritating design decisions from the delayed score updating to the UI in general that playing it for any amount of time just becomes frustrating.

    And I agree with Urthman: I find it really hard to believe that a computer couldn’t do it better than me: hell, half the time I couldn’t make the computer-generated par (but maybe I’m just really stupid).

    It didn’t help that it’s timed, and that the scoring was badly explained and couldn’t be viewed while playing.

    Oh, also: [insert mandatory 'designer genes' pun here] .

  12. Xaromir says:

    I’m always excited to hear about such things and participate – thank you RPS!

  13. Baines says:

    I tried it, and it just doesn’t make sense.

    Something is missing in the explanation of what they want, or what we are supposed to do. So far, the most successful strategy simply seems to be to smash everything together, and then move some stuff at random to make matches. Then I find myself ten points off of par by the end. I retry, smash everything together, move stuff around again, get an equally arbitrary solution that visually doesn’t seem to make any more sense than my previous failure, and make par.

    The game makes a point of not wanting gaps, except when it wants them, and it is hard to judge when it really wants them. Sometimes aligning colors will kill your score due to introducing a single gap. Other times, introducing multiple gaps is essential to getting a score, and from multiple plays, it all just feels arbitrarily chosen.

    The timer and scoring kills my ability to learn any nuances of the system, as you only have so many tries before you lose your work, and you only see the results of any change after a length of time. If score updated in real-time, then maybe I’d better understand how each piece of movement affected things.

    The crowd-sourcing doesn’t seem effective, either. In the time it takes for me to complete a “solved” match, a computer could have calculated the score of every possible alignment. Sure, this likely becomes less true when you get to larger situations, but I cannot help but feel a computer could still handle it better than a person.

    For the heck of it, I tried an eight sequence puzzle, which has seven stages. I easily broke par for the first four stages, sometimes by fairly large values. The sequences were rather easy to match, and fairly obvious. Then I hit the fifth stage, where I fell 6 short of par. The thing is, there seemed to have been no point to the first four stages I played, as to make par on the 5th stage, it looks like you have to be willing to sacrifice any of the prior work that you’ve done aligning stuff. I could have started at the 5th stage and been no better or worse off. The first four stages were nothing but a waste of time once you got to the larger set.

    All in all, it just feels like a very poorly executed attempt to follow in the footsteps of Fold.it.

  14. explodeydendron says:

    With the constant beeping, limited time to work on a puzzle, and lack of feedback, it’s like they want to penalize you for thinking or learning.

    • Baines says:

      Playing FoldIt immediately after Phylo is like night and day. No time limit, real-time score updates for all actions, a well-structured tutorial that explains what they are looking for…

  15. mrcalhou says:

    This is what I’m teaching my Biology II class about right now! Hopefully I can get this on the computers in the computer lab.

  16. Sahib says:

    Great article, I think everything is good here.
    tubal ligation reversal

  17. T4ffer says:

    Time limit: two thumbs down
    Hiding information that could as easily be displayed at all times: one thumb down
    Science: three thumbs up
    Sum total: meh
    tubal litigation reversal: what?

  18. Moonracer says:

    I like the idea but I’m not good at it and I assume you don’t actually help science unless you beat the top score for any given puzzle/set.

    I do like these sort of research/games. It would be interesting if any edutainment MMOs like NASA’s could include these as interactive research terminals in the game world.

Comment on this story

XHTML: Allowed code: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>