CASP11 starts next month!
The CASP organizers have shaken things up a bit for the upcoming CASP11 experiment!
One of the major changes is that for each prediction that we submit, we will be required to provide how confident we are with each residue in the chain.
For example, we might report for our model1 submission of a particular target (with length 181) that residues 1-80 are 100% accurate, but residues 81-85 are 50% accurate and 86-181 are 80% accurate. Then for model2 we might have completely different confidence values: residues 1-20 are 50% accurate, 21-180 are 70% accurate, and we have a confidence of 25% for the last residue, 181.
The organizers have told us that predictions with fewer than 2 different confidence values will be rejected, so this is going to add another layer of complexity to CASP!
The more exciting changes are the new initiatives in CASP11:
Modeling based on predicted contacts,
Modeling based on simulated sparse experimental data,
Structure prediction based on only the correct predicted contacts
The Assisted Structure Prediction Category was first introduced at the last CASP, but unfortunately Foldit was unable to participate in this new category because the puzzle load of the regular CASP10 targets was already too much.
Since this year they are expanding this category to 3 different rounds (up to 4 rounds if they able to include real cross-linking data for certain targets) we want to make sure Foldit is able to participate in these.
Our plan for CASP11 is to post Foldit puzzles for as many of these Assisted Structure Prediction targets as we can, in addition to the targets in the Refinement Category.
For the Template-Based Modeling (TBM) targets, Foldit has never been able to participate in all the TBM targets, but the organizers have added a new initiative for CASP11:
Evaluation of models based on a perfect alignment to the template.
Instead of giving you 3-5 different alignments (that may or may not be correct) in the Alignment Tool like in previous CASPs, the organizers will provide us with the perfect alignment!
Our plan for the TBM targets in CASP11 is to participate in these cases where we can give you the correct alignment.
This just leaves the Free Modeling (FM) category (where there are no templates), which is of course the most difficult category at CASP.
We are happy to announce that the best-performing FM group at CASP10, keasar, will be teaming up with Foldit at CASP11 via WeFold. You can see how well keasar performed at CASP10 in the pdf of the official rankings
The table below shows all the CASP10 groups that submitted at least one model deemed good enough by the CASP10 assessors of the FM category. At the top, with 8 different targets, is keasar followed by everyone else below (including FOLDIT and 2 of the Foldit teams):
What the keasar team is able to do very well is pick out the best server models (denoted by the corresponding green squares: the server models used by keasar for each target that they won) but the keasar team isn’t as good at improving these server models. This is why we are excited about joining forces with keasar by creating a WeFold team where keasar will provide us with their best server picks, and we’ll post those as Foldit puzzles for you to improve even further!
This is our plan for the FM targets.
We also hope to collaborate with other WeFold groups for all the non-FM puzzles that we post, to see if they are able to pick out your predictions better than we can.
In summary, our plan for CASP11 is for Foldit to participate in the Refinement Category, all the new CASP11 categories, and the FM category via WeFold.
Speaking of categories, there will be a new Foldit category for CASP11 (and perhaps a sub-category for puzzles using contacts) http://fold.it/portal/categories( Posted by beta_helix 76 1220 | Fri, 04/04/2014 - 18:20 | 4 comments )
Foldit Science Café in San Diego
Hey everyone! I will be leading a Science Café about Foldit at the Southpaw Social Club, in San Diego, on April 28. The purpose of the event is to engage with the community in a casual, comfortable setting to talk about the exciting science of protein folding and the Foldit computer game. There will be a short presentation on the science behind Foldit, but the event is mostly dedicated to Q & A and open discussion with the audience.
We're very excited about this opportunity to share Foldit in such a personable format! This is also a great chance for any Foldit players in the San Diego area to come together and talk to a member of the Foldit team. Or come share your experience with others who are looking for a way to contribute to cutting-edge scientific research!
The event is free and registration is not required. You can find more information here.bkoep 76 864 | Wed, 04/02/2014 - 07:58 | 2 comments )
Improvements in Foldit designs
Hi all, I wanted to share some exciting results we've gotten from folding predictions of Foldit designs!
As many of you know, after a design puzzle closes we submit a selection of Foldit player designs to the Rosetta@home distributed computing project. Rosetta@home distributes your design sequence to 100,000s of home computers all over the world, so that each computer can calculate a prediction about how that amino acid sequence might fold up. This huge dataset of predicted structures tells us a lot about the weaknesses of a design, making this the most rigorous test available to validate designs before we construct the actual proteins in the lab.
The plots below show Rosetta@home datasets from two Foldit monomer design puzzles. Each red dot represents a different predicted structure, and is positioned according to its RMSD (root-mean-square deviation in Cα position; closer to 0 means closer to the designed structure) and its score (a calculated potential energy; more negative score means a more stable structure). What we like to see is a "funnel" running from the upper-right to the lower-left of each plot. This indicates that predictions very different from the design structure are unstable, and that more similar predictions are more stable.
The top-most plot represents the top-scoring solution from Puzzle 798, which we ran in October of 2013. Note that the closest prediction has an RMSD of >2 Å, meaning that no prediction even got close to the designed structure. Furthermore, the closest predictions were not even the best-scoring; the lowest-energy prediction for this structure has an RMSD of 7 Å, representing an entirely different fold.
The lower three plots represent the three top-scoring* solutions to Puzzle 854, which closed a couple weeks ago. Each of these plots shows that the lowest energy prediction is <1 Å RMSD from the designed structure—an incredible result (the first funnel is stronger than many of the designs we come up with in the Baker Lab). Perhaps even more exciting than the quality of these folding funnels is the fact that they were derived from the best-ranked Foldit solutions, whereas in previous puzzles, scientists in the lab have been able to identify poor-ranking designs that fold better than the top-ranked solutions. These are all very exciting results, and a batch of designs from Puzzle 854 is being fast-tracked to lab production presently.
We appreciate all the effort that our Foldit players have invested in adapting to the recent changes in gameplay, and a big thank you especially goes to all those players who have been helping us troubleshoot and fine-tune the latest design tools. Note that we are still working to slim down the client and optimize these tools to be efficient as possible. Likewise, there is still plenty of room for improvement on the side of the Foldit players (we'd love to see some more beta-sheet designs** :P). Stay tuned for results from the lab!
*These are the three top-scoring designs that did not come from the same group, since players from within a group often have very similar top-scoring solutions. These designs are all significantly different from one another.
**We're working on this as well. Foldit is inherently biased towards helices, so this will be a bit of an uphill battle.
Auto-Wiggle Power and Idealize Secondary Structure
We're introducing several new features soon, and we're here to give a brief description of what they do.
Auto Wiggle Power
We've heard requests by many players to make Low Power Wiggle the default. Our reasons for not making this change might not be obvious. There are tools in the game that allow you to introduce unideal peptide bond lengths and angles, and we want the default behavior to be capable of resolving these issues. Low Power isn't capable of addressing these issues, but Medium Power can.
However, as many of you have noticed, there are a couple of problems with Medium Power Wiggle. The first is that it is slow. This is because Medium Power Wiggle modifies bond lengths and angles for every peptide connection in the protein. Thus, there are more variables for Wiggle to try changing, which takes more time. The second is that since Medium Power Wiggle has more freedom of which variables to change, it can find better shapes and get 'stuck'.
We'd love to have the unideality-resolving power of Medium Power with the speed and looseness of Low Power - that's the purpose of Auto-Wiggle. When you start a Wiggle, Auto Wiggle Power figures out the bare minimum of extra computation to resolve the unidealities, instead of just including all bond lengths and angles like Medium does.
The result is that instead of adding ~400 extra degrees of freedom like you would with Medium Power, you're now only adding on average ~4 with Auto Wiggle Power. And if you have NO unidealities, it will function identically to Low Power Wiggle.
We hope you like it!
Idealize Secondary Structure
Another request we've seen is to make forming Helices and Strands easier.
Since Secondary Structures are so critical as building blocks for proteins, we agree! Our current tools for forming a perfect helix or strand are Rebuild and Tweak. Neither of these tools is particular good at that task, so we've decided to give you a dedicated tool: Idealize SS.
Idealize SS turns a helix or strand into a perfect helix or strand. Like Idealize, it simply sets these values, so there will be global changes even when run on a local region. If you want to prevent this, you'll have to use cuts.
Idealize SS is available in the right-click menu of the Original Interface or by any non-empty selection in the Selection Interface.( Posted by jflat06 76 638 | Thu, 03/06/2014 - 20:51 | 6 comments )
Scientist Feedback on the Hotspot Finding Puzzles
Hi, folks. I just wanted to write a quick note thanking everyone who has been playing the hotspot finding puzzles. Players produced some very neat designs in both the SOD1 and Ebola puzzles. The Ebola one particularly has yielded some great starting points for design, and we currently are extending players' designs using Rosetta to try to make peptides and proteins that neutralize the Ebola virus. I was also very interested to see common features emerging in the diverse designs. Many players seemed to find, independently, that there was a "shelf" in the Ebola glycoprotein that could accept an amino acid with a flat side-chain, and that this was was adjacent to a "chasm" that could accommodate something with a long side chain. These players created designs accordingly with many different "shelf/chasm"-filling pairs. This is the sort of insight that only human intuition can provide. Automated algorithms have a very hard time discovering patterns like this.
Thanks also to those who have been using the "share with scientist" button. In both puzzles, there were some very interesting designs that were shared. I noticed that in the SOD1 puzzle, a few clever players noticed the two cysteines (cysteine 111) located on opposite sides of the dimer interface, and created peptides with two cysteines that could form disulfide bonds to both of these amino acids, thus covalently linking the dimer. These designs weren't the highest-scoring (the scoring function isn't smart enough to recognize this as a good design), but they're definitely useful designs, and they're exactly the sort of thing that the scientists want to see. Good idea, and good work!
We've got a lot more targets for which we need hotspots, so if players are enjoying these puzzles, we can post more. Feedback on this hotspot-finding puzzle format is welcome, too. It's our hope that these puzzles will give us the starting point that we need in order to allow us to design drugs to treat some very nasty diseases, so your participation is greatly appreciated.
--Vikram K. Mulligan( Posted by v_mulligan 76 2145 | Thu, 02/27/2014 - 08:04 | 7 comments )