As I mentioned yesterday we are officially announcing a new version of Foldit for use with Leap Motion! One of our devs working on the project, Dun-Yu, answered some questions about the release.
How close are we to releasing Foldit for Leap?
We plan to release it in the next month or so.
What's different about Foldit for Leap?
It provides an opportunity to get closer to the true hand folding experience. You can manipulate the 3d geometry of the protein with your fingers and drag it to a true 3d position. You can essentially twist and move the protein as if you were manipulating it with your own hands. Like a real world object.
How long has the team been working on this?
Since last Fall.
What types of players might be interested in this?
People who like hand folding for example. It gives you better control when playing. For example, you don’t need to do so many camera rotations to position the protein. It's all pretty intuitive.
What’s the difference between Foldit for Leap and Foldit for Kinect?
Kinect allows multiple people to play the same protein but Leap caters to a one player experience. It's more precise and to top if off, you don’t need a big workspace. You are working very close to the screen (and your protein structure physically). You're using your hands instead of your whole body. With the Kinect you have to play at a distance of 2-3 meters from the protein itself, but with Leap you can get right up close to the protein. I have a video of some Leap gameplay from last year when we first started working on it that people might be interested in seeing.
What's also cool is that you can point your finger at the screen to make mouse clicks. Just another cool gesture for controlling the game.
What's required for the Leap version?
You'd just need to purchase the Leap camera, install their drivers, and run the Leap version of Foldit which will be coming out soon.
What has been the most difficult part of designing this new version?
It’s all about working through human interaction. For example, if I'm thinking about working in a pinch gesture, everyone pinches in slightly different ways. We are using a vision based method to track human gestures, and it’s a big task to accommodate all kinds of human gestures. Optimizing detection for everyone is never easy, but getting it to work is really rewarding.
Any last features that you think players will be excited to hear about?
We have some cool extra tools for Leap like the pin tool. It lets you pin part of the protein structure to the workspace and move the rest of the structure around it. Kind of like freezing sections but more handy when using your fingers to manipulate the structures.
Thanks, everyone! If you have any questions feel free to leave them in the comments.
It’s time to work on some larger proteins.
We’ve just posted a new puzzle with 398 residues. This protein is too large for the normal Foldit client to handle. Because of this, we've introduced “centroid mode”. Where normal “full atom mode” puzzles have to calculate the score based on individual atom positions of all the sidechains, “centroid mode” uses approximations to speed up the process.
This new method of scoring is a radical change compared to the normal “full atom mode” score function. Tools such shake, mutate, and wiggle sidechains have been disabled because the data required to run them is not longer being generated. Other tools such as wiggle and clashing importance are ostensibly the same, but under the hood they are working in a very different way. You will find these differences require your Foldit strategies to change. We encourage you to try out new methods of working with the protein that take advantage of these differences.
One of the most apparent changes will be in how clashing importance works. In “full atom mode” when you turn the clashing importance down, there are many other elements of the score that normally take over and generally compress the protein. In “centroid mode” these scoring elements do not exist to the same extent. Because of this, you will see that the clashing importance slider does not perform exactly as it did before.
We would appreciate your input on these changes, and want to work with you to improve this new method of folding proteins. This is the first step towards Foldit players being able to tackle entirely new sets of problems that were untenable before.
Back in 2011 we congratulated Bletchley Park for being one of the first players to surpass the one million moves required for the Perpetual Motion Machine achievement. Two years later, BP has reached a whopping fifty million moves -- and then some! We've created a brand new achievement to mark the occasion...
Congratulations, Bletchley Park!
So, the CASP10 results have been up for a while on the CASP webpage as most of the natives have been solved and released.
If you looked at the results you saw that Foldit did quite well in the Refinement category!
The Template-Free category is always a very tough one, and unlike in CASP9, there was no amazing de-novo prediction in CASP10. In CASP9 there was one amazing prediction that was highlighted by the assessors: T0581. It had been generated by the RosettaServer and the best prediction came from the Void Crushers (this was in the NSMB paper). But this year it was a very tough category and no group really "nailed" any template-free target.
The Template-Based category was different, where lots of CASP10 teams were able to do well. These are the targets where there is already a close structure that you can start from, or many different templates. This category was a lot harder for Foldit, because unlike the other CASP10 teams (who get to use many bioinformatic tools) all we gave you was an extended chain and the Alignment Tool. Even with just that, though, many Foldit players were able to do very well. The main issue is that we have trouble selecting these models.
This was also an issue in the Refinement Category, as you can see in the table below:
Each row represents each of the a different CASP10 refinement targets that Foldit was able to participate in (some of the proteins were too big for Foldit).
The second column is the GDT of the starting model given to us by the CASP organizers.
The third column shows you the GDT of the best Foldit prediction in the set of models that was filtered by the WeFold team.
The last column is the GDT of the model that the CASP organizers deemed to be the best prediction from any CASP10 team (not just us).
You clearly generated some amazing predictions (most of them are a lot better than the starting refinement model!) and had we been able to pick them out, they would have beat the predictions selected by the other participants at CASP10. In terms of the refinement category, the last column highlights the winners of each of those targets, but clearly you generated better models than what they were able to pick out and submit! Unfortunately, we are very bad at selecting those solutions.
What is interesting is that you also seemed to get better as CASP went on, but that could be because the first refinement targets were smack in the middle of CASP10 (during the Template-Based and Template-Free puzzles) whereas eventually you were able to focus solely on the refinement targets.
Sadly, we've known for a while that we are very bad at selection, which is why three Foldit Groups asked us (before CASP10 started) to be able to pick out THEIR OWN Foldit group's submissions.
The next table shows you which CASP10 team submitted the best prediction for each of the CASP10 refinement targets that Foldit participated in:
Here, the second column is sorted by the "Improvement in GDT over the starting refinement model."
This table shows: how much did the very best prediction submitted to CASP10 (by any team) actually improve the starting refinement model that the organizers initially gave us.
It is obvious from this table that the FEIG group won the refinement category in CASP10, but you can see that a lot of their "winning predictions" didn't actually improve over the starting model that much.
[For anyone interested: Michael Feig's CASP10 team utilized many independent explicit solvent molecular dynamics simulations, which Foldit doesn't have access to, since Rosetta currently doesn’t include explicit water molecules]
But, if you look at which refinement predictions had the best improvements in CASP10: Foldit is at the top!
Anthropic Dreams, Void Crushers, and one of the WeFold branches were all able to find, select, and submit those amazing predictions! When Foldit wins, it wins big, but when we select poorly (because we were bold in our selections) then it really hurts us.
I think the take-away message is that selection is still the main issue... but that you are much better at it than we are! It is important to note that I'm sure the other CASP groups will argue that they too have this problem, and they generated better models as well that they weren't able to select.
This leads me to Hand-Folding:
The above figure is an RMSD plot for Puzzle 689b: Hand-Folding CASP10 T0711 Repost. It was a 33 residue freestyle puzzle, with templates, and had 3 disulfide bridges (which you totally got perfectly!). The rosetta energy (y-axis) is what you see in the game as it corresponds to the Foldit score (except that it is negative on this plot). On the x-axis is the RMSD representing how far from the native each Foldit solution is (RMSD = 0 is a perfect match, RMSD = 33 is completely wrong).
You can see above that the lowest Rosetta Energy—the top-scoring Foldit solution—the one that is easy to pick out by score, is actually one of the closest to the native.
Now compare that RMSD plot to the one for Puzzle 694: CASP T0711 Disulfide Repost Round 2:
This is the same plot as above, except for Round 2 of this puzzle (when Lua scripts and sharing was allowed) where you can see that if we now selected the best Foldit score (lowest Rosetta energy) it would not be the solution that is closest to the native. What we want is for the lowest points to be as far to the left as possible.
[The reason we reposted these puzzles after CASP10 is that when you formed the disulfide bridges during CASP10 it didn't score as high as if you didn't form them, so even though there were many Foldit predictions that were correct, they were not easy to pick since they were not the top-scoring ones. This has been fixed with the new disulfide bonus that was on these recent puzzles.]
These are obviously preliminary results (as this was only the fifth Hand-Folding puzzle we've posted) so we need to look into this some more, but this might explain why we were having so much trouble picking out your solutions that were closest to the native during CASP10.
We want to give you the opportunity to send us your favorite 691 designs. We realize that it might not be your top-scoring solution (there is a feedback about this), so you can send us your favorite manual saves and we will look at them. Details on how to send us your saves are in the comments of this news post. The deadline for submissions is Thursday, May 18th 17:00GMT.
If you want to know what we'll be specifically looking for in order to experimentally validate your designs, please read the Symmetric Protein Design Guidelines blogpost.
Looking forward to seeing all of your great designs.
NOTE: If you submitted your designs prior to 18:00GMT on April 10th, you will need to resubmit the solution.
Happy Friday, everyone. It's time for another edition of our podcast, Let's Foldit!
In this episode Foldit dev Tamir Husain shares tips for playing Foldit -- perfect for new players! Learn helpful keyboard shortcuts, how to customize the visual look of your proteins, and strategies for climbing the leaderboards.
(transcript coming soon)
I know that Puzzle 691: 60 Residue Tetramer has not been easy. As I mentioned in my recent blogpost we are working on fixing these issues for future design puzzles (hopefully the new Puzzle 695: 70 Residue Trimer is a step in the right direction), but I wanted to mention something about Puzzle 691:
Many of the top-scoring designs so far look amazing! I have been blown away with what you have been able to come up with. In fact, the rest of the Foldit Team had to ask me to be patient because I wanted to already start testing some of your designs before this puzzle even closes! That is how excited I am about them!
So I know that 691 has been frustrating, and we really hope that will be solved soon, but until then: please keep up the great folding you have done on this one, because WE WILL EXPERIMENTALLY TEST OUT AT LEAST 3 DESIGNS FROM PUZZLE 691 ONCE IT CLOSES!
Today we are proud to announce the latest groundbreaking addition to Foldit which will revolutionize gameplay for folders of all experience levels.
We give you the Foldy 9000, an intelligent in-game AI assistant who can help you in any step of the game. Foldy has been programmed to intuit your needs and will appear with helpful tips when you need them most.
And that's not the half of it! We have also expanded Foldy's expertise to include mastery of lip reading, art appreciation, and chess, making him a well-rounded in-game companion. But don't be fooled, Foldy didn't always look as polished and crisp as he does now. As a special treat, here is some early concept art for the Foldy 9000.
If for any reason you do not wish to use Foldy, rest assured a deactivation feature will be released as soon as we iron out the kinks. Thanks everyone and have a fabulous April Fool's Day from all of us here at Foldit headquarters!
Woohoo! Foldit received honorable mention in Technical Excellence at the 2013 IGF awards in San Francisco. We are thrilled to be included among so many great titles!
On the heels of our last devchat comes a highly requested scientist chat! Come chat with our scientists in #global on Thursday 4/4 from 1:00-2:00pm PDT / 20h00-21h00 GMT. We’re on Pacific Daylight Time here in Seattle (-7 GMT) so don't forget to double check your time zone conversion. A transcript will be available on the site after the chat for those who can't make it.
If you need help getting into #global chat, check out this helpful guide for assistance.
Thanks and we’ll see you there!
Edit: A transcript of the chat is now up for those of you who missed it.
*Thank you to everyone for correcting my GMT conversion. Sorry for the epic derp!
Your de novo protein designs are looking extremely good! Our goal since we introduced design puzzles was to have the best solutions go straight into experimental testing. This requires synthesizing a gene encoding the design, putting the gene into bacteria to direct them to make the protein, and then purifying the designed protein away from everything else in the bacteria and determining whether it folds up the way you designed it to. As you can imagine, this involves a fair amount of effort, and up until recently there have been issues with the top scoring designs that would have had to be fixed before ordering the genes.
What are the issues that we would like to resolve before ordering your designs? There are several “rules of thumb” about proteins that earlier designs were not consistent with. The main one is very simple and familiar to most of you-in proteins the buried residues should be hydrophobic and the exposed residues should be polar. Another perhaps less familiar one is that glycine residues, which are very flexible, should only be in turns and loops, not in helices or sheets. Finally, there should be few alanine residues in the core because they are too small to stabilize the protein much.
The filters which have been added over the past month or two simply enforce the above rules. You might wonder why the score function itself doesn’t take care of these problems. The answer is that naturally occurring proteins break all the rules some of the time-nature had 3 billion years to fine tune things and can get away with things that designers simply can’t. If we encoded the rules directly in the score function, we would fail to correctly predict the structures of many native proteins.
The exciting result is that the top designs now do not break the rules, and so are close to being ready to order. On the other hand, there are clearly problems with the filters that make the game less fun to play, and we are working hard to fix them. The main issue is that they are slow to compute, but I think this can be fixed pretty easily (the way we are determining whether a residue is buried is more time consuming than it needs to be, for example).
So please keep designing, and we will work hard to improve speed. I’m really excited about the beautiful symmetric structures you are creating and can’t wait to start testing them. If the sequences of your designs fold up to the structures you have created it will be a major scientific milestone! There will be immediate applications, for example to stabilizing the HIV surface protein, which is a trimer, in a form which can be used as a vaccine.
Happy Friday, folders!
We have a new opportunity for you to come work with the creators of Foldit. We are currently seeking a System Administrator to work on Foldit and other Center for Game Science titles. Know a good candidate? Send 'em our way.
Our new System Administrator will be helping us make sure our servers continue to run smoothly -- no more surprise crashes!
Foldit was recently featured on the University of Washington's Science and Engineering site. Take a look at the video for an inside look at Foldit, what inspires us, and how citizen scientists are solving hard scientific problems. Enjoy!
It's time to gussy up those computer screens with a new Foldit wallpaper! A big thank you goes out to our artists who took the time to create it for us. Enjoy!AttachmentSize 1024x768357.25 KB 1280x1024521.96 KB 1400x900524.8 KB 1600x1200692.49 KB 1680x1050681.1 KB 1920x1080716.9 KB 1920x1200766.44 KB
Happy Tuesday, everyone!
This morning the Center for Game Science's director Zoran Popović was interviewed on National Public Radio. Listen to the whole story which features a segment on Foldit here!
Sepsis is a potentially lethal condition affecting hundreds of thousands of people every year in which a severe infection spreads into the bloodstream, leading
to multi-organ systems failure and often death. Despite years of research, clinicians are in need of new treatments for sepsis.
Research teams at the University of Washington and at the Wyss Institute for Biologically Inspired Engineering at Harvard University are developing protein-based pathogen capture reagents to be used for the removal of circulating pathogens from the blood of patients. The approach combines online gaming as a platform to introduce and improve upon innovative protein designs with state-of-the art protein manufacturing capabilities critical in translating such designs for clinical applications.
Our original approach to this problem was to have FoldIt players focus on redesigning mammalian Mannose-Binding Lectin. This approach proved problematic, and we are now refocusing our efforts on a lectin from the snowdrop plant, Galanthus nivalis (GNA lectin). The goal here is to use Foldit to improve GNA's sugar binding capabilities, allowing for its potential clinical use. You will work towards this goal by redesigning specific regions of GNA in order to increase the number and strength of protein-sugar interactions.
At the conclusion of the challenge, the most promising Foldit designs will be manufactured and tested at the Wyss Institute in order to determine their pathogen capture capabilities, and whether they represent an improvement over natural GNA. If the current approach of using Foldit to target pathogens proves successful, we plan to continue our effort by designing additional capture reagents.
Join us in helping to solve this important public health issue!
Try out the new Sepsis Design Puzzle now: http://fold.it/portal/node/994591
Widespread use of clean-burning hydrogen for fuel would be an amazing step forward for society. Have we mentioned recently how amazing? Oh that’s right, we have.
To recap: molecular hydrogen (H2) produces water when it’s burned, and can be re-made from water plus a bit of energy. Well, a LOT of energy, but that’s where hydrogenase catalysts come in: if we can reduce the cost of producing hydrogen from water, we can power everything with clean, near-unlimited hydrogen that we can make cheaply at our power plants or in-home plug-in hydrogen makers. If we get a good enough hydrogenase catalyst working, the future becomes really exciting.
We’ve shown you a hydrogenase catalyst before (Puzzle 644: http://fold.it/portal/node/993779 ). Now we’ll be working with a different catalyst with proven activity; a nickel-phosphine dimer inspired by the active sites of some natural enzymes. We can watch hydrogen bubble off it when we apply a small current to feed it electrons. This catalyst was developed at PNNL (Pacific Northwest National Labs, in south-central Washington State) by a group led by Daniel DuBois. It’s a really neat molecule, and more information can be found at the PNNL website here: http://www.pnl.gov/news/release.aspx?id=883
The good news is that we’ve already proven we can attach peptides to this catalyst without making it worse. (This is a bigger hurdle than you might realize!) With peptide attachments –specifically, with YOUR help building good peptide attachments – we can make this a much better catalyst.
So, how do we make this catalyst a game-changer? For starters, we’re just looking for interesting, compact designs. Just putting this catalyst into a protein-like environment and determining its structure would be a huge step forward. But this isn’t just about making our catalyst a cozy little apartment out of protein. We need to feed protons (hydrogen atoms) to the active site so we can assemble hydrogen faster.
We also need to hold the catalyst in place so that it can’t wiggle. The catalyst is more flexible than it appears in our puzzle. It can bend into other conformations where it stops working as a catalyst. But with a strong, compact protein support, we’ll be able to hold it in the working conformation so that it stays active more of the time.
Many natural enzymes do exactly this sort of thing: they encase their active site in a protein machine that keeps it stable, and “feed” it protons or other molecules it needs. If we can get a Borg-like hybrid protein/small-molecule catalyst working, we won’t just make a big step towards a hydrogen energy economy, we’ll also have a unique hybrid bio-machine to brag about. :)
Future puzzles will provide bonuses for reaching either the central metal or nearby nitrogen with polar hydrogen atoms. But for this first puzzle (#675: https://fold.it/portal/node/994490) just have fun building up peptide scaffolding to wrap around the catalyst.
Technical advice (for puzzle #675 and those like it):
Feel free to look to previous two-chain symmetric design puzzles for inspiration, but don’t limit yourself to what’s worked before, since this big central molecule linking your two chains provides new opportunities and new challenges. Specifically, the two chains are now permanently stuck together, so you don’t have to worry about two separate chains finding each other; they count as one chain now, so they don’t have to be able to fold up on their own. On the other hand, the bulky molecule in the middle means that the available geometries for packing the chains are more limited, and to get a good score you’ll probably have to include at least part of that big molecule in your core.
Even more technical: be warned that it’s going to be trickier than normal to wiggle and rebuild segments near the N-termini of the two chains, since they’re locked together at the catalyst.
Happy Wednesday everyone! We're proud to present the first episode of our new podcast all about Foldit and the science behind it. In this episode we sit down with Brandon Kier, one of the scientists working on Foldit, to discuss his latest puzzle: 675 Hydrogen-producing Catalyst Dimer. A text transcript will also be uploaded soon.
Be sure to check out Brandon's companion blog post as well!