Developer Chat

@betahelix Hi everyone! And welcome to our Winter Science Chat, thank you all for coming! 12:30
jflat06 Hello! 12:30
bkoep Hi all! 12:31
alcor29 HI 12:31
horowsah Hi all! 12:31
bkoep Thanks to everyone who posted questions in advance! 12:31
bkoep We'll get started with those, and see where the chat takes us! 12:32
@betahelix "It's probably been asked before, but will we ever see another exploration mode puzzle, or are they extinct? How about sketchbook puzzles?" 12:32
@LociOilingIRC sounds familiar 12:32
bkoep Exploration puzzles are probably extinct 12:32
bkoep Directly rewarding exploration turned out to be tricky, often leading towards an exploration-at-any-cost approach that made unrealistic structures. 12:33
bkoep Funnily enough, sketchbook puzzles were one of the directions that we took from the lessons learned in exploration puzzles. 12:33
bkoep The sketchbook puzzles may make a comeback! 12:33
bkoep The Sketchbook puzzles were intended to shift gameplay, with an emphasis on early- and mid-game handfolding, and much less late-game optimization. 12:33
bkoep When we ran all of the Sketchbook puzzles in summer 2018, we found that they didn’t have much effect on puzzle results (looking at prediction accuracy and sampling diversity, for example). 12:34
Susume2 I thought the partition function puzzles might be improved by a big exploration bonus 12:34
@betahelix Did you enjoy the sketchbook puzzles? 12:34
bkoep However, we still think that late-game optimization has little scientific value, and might unfairly favor players with more computing power—so we might revisit this approach! 12:34
alcor29 If they didn't have much effect why not keep them instead of having to work to optimize? 12:34
jflat06 @alcor29, we've had that same thought - and that's part of the reason we'd like to revisit them and maybe see if we can't improve them 12:35
alcor29 yup 12:35
bkoep @Susume, that's a good point! We might have to consider that if we do more partition puzzles. 12:36
bkoep I would also like to try "blind" partition puzzles, which don't explicitly reward exploration 12:36
BletchleyPark Is there news on Aflatoxin ? 12:37
bkoep But I wonder if players could design sequences that try to mislead opponents in Phase 2 (building decoy solutions) 12:37
bkoep @BletchleyPark, not at this time! The Siegel Lab is still analyzing Foldit solutions from the last puzzle! 12:38
BletchleyPark ok, thank you 12:38
@betahelix "The category scores on fold.it have been a mess for a while now. Group scores are a perennial problem. Several puzzles have failed to award points without poking and prodding. Is there hope for restoring data integrity?" 12:39
jflat06 Good news on that front - we're working on a new website! 12:39
robgee @betahelix Yes sketchbooks are fun 12:40
Skippysk8sIRC jflat, sounds good. chat isn't working correctly now either... 12:40
@betahelix thanks robgee! 12:40
jflat06 This will be a completely new site, with a more modern design and much more robust backend, so it shouldn't break easily. 12:40
jflat06 @Skippy, hopefully this will help alleviate some of those chat issues as well. 12:41
disturbedxxx666 omg I made it to the chat 12:41
alcor29 Are the broken GUI by stride recipes ever going to be fixed? Can the problem be fixed? 12:41
@betahelix welcome, disturbedxxx666! 12:41
jflat06 The new site has a bunch of cool new features (like an in-site recipe editor and a personal 'recipe' library) 12:42
disturbedxxx666 unable to use IRC so im stuck with this username. sorry folks 12:42
BletchleyPark I noticed solutions can be saved in LUA, but not deleted, is that in the making ? 12:43
Susume2 editing recipes outside the game client, yay! 12:43
jflat06 @alcor29 - I'm not sure what's going on with that. We'll be taking a look at it soon 12:43
alcor29 Personal library, yes! 12:43
alcor29 Thanks jflat. It's causing a lot of extra work. 12:44
jflat06 You'll be able to create new, edit existing, clone existing recipes all from the site 12:44
BletchleyPark @Susume, you could always do that by copying it to an editor and back 12:44
jflat06 And then you can manage your recipe library on the site and download it to any client with the click of a button in the client 12:44
BletchleyPark What if I want to keep using the in-client editor ? 12:45
disturbedxxx666 is the topic on scrips? because I have a question 12:45
bkoep @disturbedxxx666, ask away! 12:46
jflat06 We don't have any plans to eliminate the in-client editor, but that may be something we re-evaluate as the client ages. 12:46
disturbedxxx666 why cant we select text? I think it would be much easier and faster if we were able to copy and paste. 12:47
alcor29 Search function in editor would also help. Or have Control F work on it. 12:48
BletchleyPark You can copy the text to wordpad and edit there on windows, then cut & paste back 12:48
bkoep @disturbedxxx666, that's a big limitation of the current in-client editor. But this should be enabled with the editor on the new website! 12:48
jflat06 Part of the issue is the existing in-game editor is very poorly written and isn't very maintainable or update-able 12:49
alcor29 Just realized could probably search if on web sie. 12:49
jflat06 But the new recipe editor is just one small part of the new site. We'll be revealing some more details soon, and we'll have a beta where you can check it out before it goes live to help track down any issues. 12:50
Susume2 (psst, line numbers in the editor would be fab, since the game gives line numbers with errors) 12:51
jflat06 Do note that, since this is a brand new site, there will be some changes from what you're used to, but most functionality from the old site should be present in the new one 12:51
jflat06 (and even some new stuff, too!) 12:51
jflat06 (The new site lua editor does have line numbers :) 12:51
Susume2 :) 12:52
BletchleyPark Will the content of the old site be preserved ? 12:52
jflat06 Yes 12:52
BletchleyPark very good 12:52
jflat06 There may be some odd bits that dont get carried over, but it is our intent to preserve existing content 12:52
Skippysk8sIRC I don't code, but want to know about the science of symmetry puzzles. with the new bands between the monomers function is it good or bad to have dimers/trimers tightly bonded 12:53
Skippysk8sIRC 10,000 foot answer is fine :) 12:53
bkoep @Skippysk8sIRC, 10,000 ft is probably a little too much... 12:54
Skippysk8sIRC kk big picture. what is our goal 12:54
bkoep The monomer units should make close contacts! Ideally, orange residues should pack right up against orange residues. Blue residues should make H-bonds with other blue residues (so they should also be pretty close to one another) 12:54
Skippysk8sIRC so we want to kill the voids, not make limited contact like that one series of puzzles 12:55
jflat06 I think the key is to think about it like is interacting with any other bit of protein (always with the disclaimer that I'm not a biochemist :) 12:55
spvincent Is there any progress on tracking down add/delete residue problem on symmetry puzzles? 12:56
bkoep @Skippysk8sIRC, yes! Limited contacts are still nice, but those points of contact should be close together! 12:56
disturbedxxx666 symmetry puzzles are my fave 12:56
alcor29 In general what does the coming year hold for foldit? 12:57
jflat06 @spvincent - I think there's support for that in Rosetta now, which means that we have the potential to look into adding it into foldit 12:57
bkoep The problem comes when there are many, many orange contacts between units. 12:57
@betahelix @alcor29 hopefully more fun puzzles and Nature papers :-P 12:58
alcor29 Or, will AI have a growing role? 12:58
BletchleyPark Is it correct for us to claim we are co-authors of a Nature paper ? 12:58
BletchleyPark I got a nasty response to that claim a while ago 12:59
@betahelix yes indeed, BP! 12:59
BletchleyPark thank you betahelix 12:59
@betahelix Especially for everyone who's real name is in the supplemental material 12:59
disturbedxxx666 I personally would love to work on pdb 12:59
BletchleyPark there you go 12:59
jmbrownlee333 I would like to know more about the poly proline stuff. Sell me on its importance and will there be new tools? 13:01
horowsah Actually, I do see job postings from the PDB with some frequency, and I would think Foldit would be an awesome thing on a resume for then 13:01
Susume2 cool, where do you see them? 13:01
horowsah The job postings are usually in a mailing list called ccp4bb that is mostly for crystallographers... not sure if they post elsewhere too 13:02
@betahelix @BP this Physics paper famously has over 5000 co-authors: https://journals.aps.org/prl/pdf/10.1103/PhysRevLett.114.191803 13:02
bkoep @alcor29, we're still learning about how AI is useful for protein structure prediction and design, but it could very well show up to help out Foldit players! (Probably not in the next year, though) 13:02
Susume2 thx horowsah 13:03
alcor29 Thanks. 13:03
BletchleyPark @betahelix, I'll check them all for duplicates ;-) 13:03
Susume2 I'm fascinated with the polyprolines too - I was reading that they often dock with SH3 barrels, which are my favorite little barrels 13:04
@betahelix lol BP... here's an article about it: https://www.nature.com/news/physics-paper-sets-record-with-more-than-5-000-authors-1.17567 13:04
@betahelix but not as cool as beta helices, right Sus? ;-) 13:04
bkoep @jmbrownlee333, the poly-proline puzzles are an open field! We have a few ideas about how we might use them, but we're mostly curious to see how they can be incorporated in designed proteins! 13:04
BletchleyPark thank you ! 13:04
jmbrownlee333 Love to hear your ideas. 13:05
BletchleyPark Please do look into the memory leak issue I reported in feedback, it hampers work on 1766 13:05
bkoep @Susume, yes! Poly-proline helices are common binding partners for SH3 barrel domains in biology. Some recent work has shown that these binding interactions are important for T-cell activation. 13:05
disturbedxxx666 Any of yall know why my mibbit IRC isnt working anymore? I cant get into the foldit channels. due to the changes? 13:06
@LociOilingIRC (p.s., I like "Fruit-fly paper has 1,000 authors") 13:06
@betahelix I'm using mibbit right now 13:06
jflat06 @BP - we just talked about it in today's meeting, and we'll try to replicate it on our machines 13:06
Skippysk8sIRC all the irc chat not working right now... on jflat's list 13:06
BletchleyPark @jflat06, great, thank you ! 13:06
bkoep Designed proteins with poly-proline helices could be useful for replicating some of those interactions in synthetic systems. Or one day, we could potentially use them to modulate T-cell activation itself. 13:07
BletchleyPark Any more papers in the pipeline ? 13:07
jmbrownlee333 Might we see a ideal SS tool for proline helices? 13:07
Susume2 do the SH3 barrels bond to the same backbone atoms we are supposed to try to satisfy? I'd like to see how they do it 13:08
bkoep @jmbrownlee333, that's a good idea! That sounds like a project for neilpg628! 13:08
Skippysk8sIRC by designed protein is our goal to get a good link to part of the polyproline helix with a well folded "attachment"? or to make as many bonds as possible 13:09
bkoep @BletchleyPark, nothing currently in writing, but the symmetric design project is moving along nicely! 13:10
jmbrownlee333 Yay for SH3 barrels, Yay for TIM barrels! 13:10
Skippysk8sIRC tim must shrink to be workable.... 13:11
BletchleyPark very good, I'll keep an eye out 13:11
BletchleyPark I have to leave, very early rise tomorrow. Thank you for organizing and joining ! 13:11
bkoep @Susume2, yes (more-or-less). There's usually some small deviations in natural poly-proline helices (they are not perfectly ideal), but the binding interactions should be pretty similar 13:12
Susume2 an ideal SS tool for polyproline would be nice, that part of the rama map is very common and we don't have an easy way to get to it besides dragging individual dots from the ideal beta sheet position 13:12
BletchleyPark I'll read up the rest in the chat logs tomorrow, bye for now. 13:12
@betahelix good night BP! 13:13
@betahelix Speaking of papers... here is a question "Is there any plan to give us unsolved Cryo EDs?" 13:13
bkoep We would love to challenge Foldit players with unsolved cryo-EM maps! 13:14
bkoep The main hurdle is soliciting datasets from EM microscopists. 13:14
bkoep We hope that the most recent paper Foldit paper (about cryo-EM density puzzles) will reach other scientists working on EM, and lead to more collaborations!   13:14
@betahelix So if you know any experimentalists with unsolved structures... please send them our way! :-) 13:15
@betahelix "What is the limiting factor for Cryo ED? The density measurement or the structure solving? (I would understand that players might be useful for prediction only if a mass cryo measurement becomes available to feed a limited number of crystallographers)." 13:15
Susume2 I got the impression somewhere that cryo EM data is piling up because people are generating it faster and there are not enough experts to interpret it...? 13:15
horowsah @Susume2, I don't think that it's so much a pile-up as it is that it's slow 13:16
horowsah most people want to interpret their own data, but don't have either the tools or experience to do it quickly and well 13:16
disturbedxxx666 not too sure where that tim barrel came from. maybe a redesign from foldit? 13:17
jmbrownlee333 TIM barrels ancient and common in nature. 13:17
horowsah the EM data collection is getting faster and easier all the time, and I'd say the interpreting part is getting a lot of attention right now, but still needs more 13:17
jmbrownlee333 The cryo-EM data we got was idel and gorgeous 13:20
@betahelix The 4 cases from the paper, or this current puzzle? 13:20
Susume2 is EM free of the phase problem that plagues x-ray? 13:20
jmbrownlee333 the 4 cases in the paper. current is ok  too. 13:21
Susume2 the EM data from the paper was beautiful 13:21
horowsah @Susume2, yes the EM is pretty much phase-problem free, which happens to be helpful for Foldit puzzles 13:21
jmbrownlee333 That one gnarley huge thing with multiple proteins sucked though. 13:21
@betahelix There is one limitation in terms of giving you unsolved cryo-EM maps to work with, and that is their size. The 4 subunits from the paper were part of a huge complex with 300 and 500 residue proteins. 13:21
horowsah as in crystallography, we often update the phase calculations and therefore the map as the model improves, but we don't have to do that for EM 13:22
@betahelix Yes, jmbrownlee333... let us never speak of that one again ;-) 13:22
jmbrownlee333 I am good w that. 13:22
@betahelix That doesn't mean we don't plan to tackle these larger cases eventually, but before getting to that we'd need to address many of the important ED suggestions that you have brought up. 13:23
Skippysk8sIRC any idea when we will hear about IL-7R results? be interesting to see how folds worked versus points 13:23
bkoep @Skippysk8sIRC, it will still be some time before we have data from the IL-7R experiments. It's probably too early to estimate a date, but rest assured, bcov is working on it as we speak! 13:25
Skippysk8sIRC let bcov know that points versus best results would be interesting 13:26
@betahelix "Re. bcovs' high-throughput method, is this ability to test lots of proteins at once only applicable to a certain type/class of protein?" 13:26
bkoep This is a great question! There are some limitations for bcov’s high-throughput experiment (yeast-display cell sorting), although it should work for any kind of protein fold. 13:26
bkoep The main limitation of this kind of experiment is that we need a visual indicator for “successful” designs. 13:26
bkoep In bcov’s experiment, a “successful” design is one that binds to the target IL-7R protein. In this case we’ll put a fluorescent tag on the target IL-7R protein, and we’ll “display” all the designed binders in a population of yeast cells. 13:27
bkoep When we mix the two, then the fluorescent target will stick to successful binders, and we can sort out the yeast cells that are surrounded by fluorescence. 13:27
alcor29 I believe that Bcov had said that he couldn't handle more segs 13:27
bkoep @alcor29, yes! Another limitation for this experiment comes from the DNA-synthesis technology required to make a large library (10,000s) of designed genes. 13:28
bkoep So far, this has been limited to short genes (only big enough to encode a 60-residue protein). But the technology is improving, and soon we should be able to test libraries of larger proteins (100+ residues)! 13:28
@betahelix That was the last pre-chat question... anyone have any additional questions/comments? 13:30
bkoep @Skippysk8sIRC, yes! I'll pass that on to bcov! I'm sure he will also be interested if Foldit score correlates with real binding. 13:30
alcor29 Seems odd tome that attendance todays seems low. Need an earlier time? 13:30
@betahelix That is a great suggestion... 13:31
@betahelix we want to have chats at different times, so that everyone can at least make it to one of them! 13:31
@betahelix We'll make sure to have the next one earlier in the day :-) 13:31
alcor29 And more frequency? 13:32
@betahelix That's what the Lab Reports are for! ;-) 13:32
Skippysk8sIRC It may be chat and client issues.... TIM barrel is crashing often for many. My IRC chat keeps shutting down. But do keep up the Lab report you tube videos 13:32
alcor29 love the lab reports. 13:32
jmbrownlee333 The lab reports are great addition. 13:32
robgee Lab reports rock! 13:32
bkoep @alcor29, I think this was a good chat, and we got lots of good questions! We'll try to do another one in a couple months! 13:33
Skippysk8sIRC Thanks all 13:34
alcor29 agree 13:34
jflat06 Thanks! 13:34
pletsch good info, thanx 13:34
@betahelix and look forward to bkoep's next lab report in the new year! :-) 13:34
jmbrownlee333 And maybe something from the Siegel lab in 2 months? 13:34
alcor29 Happy Holidays to all! 13:35
bkoep @jmbrownlee333, fingers crossed, but we don't want to rush the Siegel lab. Good science takes time! 13:35
@betahelix bye everyone... keep up the great folding! 13:36

Generated by irclog2html.py 2.17.0 by Marius Gedminas
- find it at https://mg.pov.lt/irclog2html/!

User login
Download links:
  Windows    OSX    Linux  
Windows
(7/8/10)
OSX
(10.12 or later)
Linux
(64-bit)

Are you new to Foldit? Click here.

Are you a student? Click here.

Are you an educator? Click here.
Social Media


Search
Only search fold.it
Other Games: Mozak
Recommend Foldit
Topics
Top New Users
Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, Boehringer Ingelheim, RosettaCommons