Developer Chat

jflat06 Ok, let's get started! We have a number of questions today, and we're also here to talk about the Aflatoxin puzzles, as well as the new experimental build. 12:01
jflat06 To start us off, Susume asked: Monomer design update: How many of the ~75 foldit monomers you ordered genes for last summer ended up passing the wet lab tests? Did the successful ones have a variety of different shapes? Have you ordered genes for more foldit monomers? Has our rate of passing the Rosetta tests gone up further since last spring's blog posts? 12:02
bkoep Yes! Last summer we ordered DNA genes encoding 76 Foldit player designs (but two of them were never delivered). 12:02
bkoep Of those, 31 successful designs have passed all our tests showing that they are well-structured and stable (this is an excellent success rate for tested protein designs). But we don’t have crystal structures with definitive evidence that these are folding accurately. 12:02
bkoep Optimistically, we have no reason to doubt these 31 successes, but we would like to have few representative crystal structures showing whether these proteins indeed fold accurately (i.e. that the proteins look exactly as designed). 12:03
bkoep These 31 designs represent about 15 different protein folds, which is REALLY impressive diversity. 12:04
bkoep Looking forward, we have a growing list of good looking designs that we’d like to test, but we have not ordered any more yet. 12:05
bkoep We might be able to set up a high-throughput experiment to test lots of these in parallel, but at the very least we’ll probably test the most interesting ones (those with unique or unusual folds, for example). 12:05
bkoep Computational analysis of Foldit designs is still showing about the same level of success in recent puzzles as previously. 12:06
bkoep Players might have noticed we’ve been trying slight variations on the design puzzle setup, so we can tune the parameters that give the best designs. (For example: Is the Ideal Loops filter necessary? How much protein “core” is enough core?) 12:06
jflat06 Thanks bkoep - does anyone have any further questions on that specific topic? 12:08
@LociOilingIRC Susume is not here today 12:08
frood2IRC bkoep: so U T saying that nomonomer designs since last summer have been thro the wet lab - is that what we should undersstand? 12:08
frood2IRC *U R 12:08
bkoep @frood, that's correct, we have not started wet lab testing for any additional proteins since the batch of 76 designs we started testing last summer 12:11
frood2IRC thx 12:11
bkoep There are plenty of more recent designs in early (computational) stages of analysis 12:12
bkoep But for experimental wet lab testing, it usually makes sense for us to test designs in batches 12:12
@LociOilingIRC I see a couple of new names in chat, welomce everyone 12:14
jflat06 Ok - next question from S0ckrates: Any plans to refine the ranked leaderboards? As it stands, with the leaderboards being more or less static, I'm finding it a bit boring to grind. Any ideas to 'sweeten the deal' or 'shake things up'? 12:14
jflat06 I'll go ahead an take that one. 12:14
jflat06 We have been talking about this recently, and have some ideas. 12:14
jflat06 The idea of seasons has definitely come up, and was originally intended to be a part of the leaderboard rework when we created the category leaderboards. There are some technical hurdles that need to be overcome for this to work, as well as making it feel relevant by showing your season rank and progress in the right place. 12:15
jflat06 Another idea we’ve been throwing around is a non-competitive player ‘level’ that is a measure of your total experience playing the game. Right now the game is pretty brutal for newer players who hit a wall when they have to compete with our vets. 12:15
jflat06 Allowing them to still see a form of progression may help them stick around long enough to become a vet themselves. 12:15
frood2IRC sorry if I’m speaking out of line here….surely this is a game/tech question….not suitable for a science chat. 12:15
jflat06 I understand that argument, but we don't really have separate Science and Developer chats like we used to 12:16
frood2IRC Ahhhhh 12:16
jflat06 So it is reasonable to ask and talk about here. 12:16
frood2IRC so cev chats have now been cancelled? 12:17
frood2IRC *dev 12:17
@S0ckrates I would also argue frood, that nothing is in isolation. When the data relies on an active playerbase to be generated, the game portion cannot be neglected as it be a science in and of itself to perfect 12:17
jflat06 Not cancelled, but we're using them together like we did in the old days. 12:17
@S0ckrates That's why I've been asking these sorts of questions. 12:18
jflat06 Coming back to the topic: 12:18
jflat06 Rewards are something we’ve also talked about, both within the context of levels and seasons. We’ve been brainstorming some ideas, and we’d love to hear yours! 12:18
frood2IRC I’m not being awkward….just that this is the first I have heard of no future dev chats….I’m sure I’m not alone in that 12:18
jflat06 To address Froods question and concerns about player number counts: 12:18
frood2IRC so perhaps the chat description should be changed in future 12:19
jflat06 Retention is an important part of the game. It’s been a long time since our last huge player influx, and our player count has been slowly declining since then. The game needs to be fun for both existing players and for new players the next time we get a big influx. 12:19
jflat06 @frood2IRC, yeah that is probably reasonable. We aren't ruling out devchats for the future, but for the time being, it makes sense to combine them. 12:19
frood2IRC ok 12:19
jflat06 We appreciate that some players see the game as a purely scientific endeavor, but many players and perspective players are motivated by the game side of things, and it is a core premise of the project. 12:19
frood2IRC then - to be fair- this should have been made clear already 12:20
jflat06 I'll make a note of that, thanks. 12:20
jflat06 Next question is by jeff101 - Aflatoxin Puzzles: I'm curious if any of our designs looked interesting 12:20
jflat06 and if any have been synthesized or even crystallized. 12:20
jflat06 What features did the most interesting designs possess? 12:20
jflat06 If there will be another round of aflatoxin puzzles, 12:20
jflat06 how would you change the way our designs are scored 12:20
jflat06 to better reflect which ones you found most interesting? 12:20
jflat06 Finally, which aflatoxin orientation gave the best results? If any of our designs have been synthesized, 12:20
jflat06 do they seem to fold properly? Do they bind 12:20
jflat06 aflatoxin better than previous designs? Do 12:20
jflat06 they bind aflatoxin as well as you wanted? 12:20
jflat06 Do any degrade aflatoxin? 12:20
jflat06 (sorry for the formatting!) 12:21
bkoep These are great questions! And... 12:22
bkoep You should stay tuned until next week, when we'll have a blog post from the Siegel lab and a new round of Aflatoxin puzzles! 12:22
jflat06 Next question is from Bruno: 12:24
jflat06 How is the numbering of the residues determined? 12:24
jflat06 Would we find the same solution if the numbering was reversed? 12:24
bkoep Great question! Actually, LociOiling posted a great response to this question here: 12:25
bkoep http://fold.it/portal/node/2004837#comment-36355 12:25
bkoep (Thanks LociOiling!) 12:25
@LociOilingIRC the power of wikipedia 12:26
bkoep The short answer is, the two ends of a protein chain are indeed different 12:26
@S0ckrates Actually, I wanna clarify, is it numbered from N-Terminus to C-Terminus, or vice versa? 12:26
bkoep @S0ck, by convention, the ordering is from N-term to C-term (so the N-terminal residue is residue 1) 12:27
@S0ckrates Got it, makes sense. Thanks! 12:27
bkoep So if you were to make two designs, but with reversed numbering, those would be meaningfully different 12:27
bkoep *two designs that are otherwise "identical" 12:28
frood2IRC but the chirality could be bad perhaps 12:29
frood2IRC ? 12:29
bkoep @frood, yes! 12:29
bkoep In addition to "polarity" of the protein chain (in which the N- and C-termini are distinct), there is also the issue of "chirality" 12:30
bkoep You could also make two "identical" proteins that have reversed chirality; they would be mirror images of one another 12:31
frood2IRC perhaps I may ask a question….sorry not posted B4….I treat chirality as inportant….am I wrong to do this? 12:31
frood2IRC kk - I can see that - symmetry etc 12:31
rmoretti Chirality is very important in biological systems. The standard analogy is one of putting your left foot into a right shoe (or right hand in a left glove) -- you might be able to make it work, but not as well as if you used the correct chirality. 12:32
bkoep All natural proteins use L-amino acids; to make the mirror image proteins you would have to use "artificial" D-amino acids, which are uncommon in biology but no problem for a chemist 12:32
frood2IRC these are not catered for in foldit I guess 12:33
frood2IRC but enough from me….thx for the answers 12:33
bkoep @frood, actually, the new update includes some support for D-amino acids 12:34
frood2IRC not sure whether to cheer or not   (if u see what I mean)    LOL 12:34
bkoep We're looking forward to trying some mixed-chirality protein design in Foldit, but that's still a little ways down the road (not a high priority) 12:35
@LociOilingIRC D-glycine should be easy 12:35
frood2IRC just make sure there is a really good explanation….that will help   :) 12:35
jflat06 Great - another question from Bruno - When starting de novo, I try to make the hydrophobicity on the same size for successive sheets. 12:36
jflat06 However, the starting flat protein often reverse the hydrophobicity between successive sheets. 12:36
jflat06 Is there a reason for this default alignment? 12:36
@jeff101 if we're designing man-made proteins, we might as well try D-amino acids 12:36
bkoep If I understand correctly, Bruno is asking about the extended starting structure in De-novo Freestyle puzzles 12:37
@jeff101 not sure. maybe it is about sheets tend to be alternating h-philic and h-phobic residues 12:39
bkoep When we make these starting structures, we simply set the the backbone torsions of all residues to a constant value (very similar to a beta sheet, with sidechains alternating between opposite "faces" of the chain) 12:39
@LociOilingIRC so all segments except the first and last are at the same point on the Rama map 12:40
bkoep So, I wouldn't necessarily derive any meaning from the fact that patches of hydrophobics tend to be on one face or the other 12:40
bkoep @Loci, exactly (and the only reason the first and last are different is because, without neighbors, some torsions are undefined) 12:40
@LociOilingIRC I think Bruno may have been asking about rotation 12:41
frood2IRC i think he is actually asking about chirality 12:41
@LociOilingIRC yes, related 12:41
@LociOilingIRC you don't simply turn one sheet 180 degrees in the plane to align with the next sheet 12:42
@LociOilingIRC you usually have to turn the sheet over as well 12:42
frood2IRC exactly - but not always   >:) 12:42
@LociOilingIRC generally 12:42
@jeff101 has someone made a video about how best to combine a group of b-strands into a sheet? 12:43
bkoep Possibly to this point, if there are an odd number of residues between two sheets, then in the starting structure their hydrophobic patches will appear on opposite faces of the extended chain 12:44
bkoep (because in the starting structure the residues simply alternate faces) 12:44
bkoep Does this clarify anything? If not, could you elaborate on the question? 12:46
@jeff101 if bruno isn't here, maybe you should e-mail him to ask what he meant and then make a blog post with a response 12:47
@LociOilingIRC Bruno is not here unfortunately 12:47
jflat06 Ok, let's move on. 12:48
jflat06 The last question is by S0ckrates: 12:48
jflat06 Is it possible to simulate a hydrophobic region in Foldit for the sake of puzzles involving proteins with external hydrophobic domains for cell membranes? I've always wondered about designing little cellular molecular spaceports for receptor molecules and having a little fun with that. 12:48
rmoretti This is something that we're actively interested in pursuing. 12:49
rmoretti The Y1 puzzles from several months ago were actually of a protein that's normally found in a hydrophobic membrane. 12:49
rmoretti Scientifically, the results of those puzzles weren't all that great, mainly because we lacked proper modeling of the membrane environment. 12:50
rmoretti Up to now, we really couldn't treat membranes appropriately, due to technical limitations. 12:50
rmoretti But the Rosetta updates that are currently being testes in the experimental update group have a new way of modeling membranes which (hopefully) should be more compatible with Foldit. 12:51
rmoretti We're hoping to be able to properly model membranes with future puzzles, but it may take a bit of testing and tweaking of the GUI/game interface to make things work properly. 12:51
@S0ckrates Awesome, that sounds promising! Hope it doesn't lag too much for a lot of systems; this could be a great way to learn about  membrane protein principles for new players 12:53
@jeff101 if we had membrane protein design puzzles, how woud you test the designs in the lab? 12:54
bkoep @jeff101, with much difficulty... 12:54
@jeff101 :) 12:54
frood2IRC LOL 12:54
rmoretti Testing would be in a similar fashion to regular design puzzles, though the expression systems for membrane proteins are much harder to do than soluble proteins. 12:55
rmoretti But at least for the moment, the thought would be to use Foldit for membrane structure prediction, rather than design. 12:55
rmoretti (But structure determination of membrane proteins is also much harder than soluble protein determination.) 12:55
@LociOilingIRC do we have an idea when the experimental release will move to devprev? 12:56
frood2IRC or, indeed, main 12:56
bkoep For those interested in designed membrane proteins, there is an exciting new paper published just last week by some of my colleagues in the Baker lab: 12:57
bkoep http://science.sciencemag.org/content/359/6379/1042 12:57
bkoep If you don't have access to Science, you might be able to access a PDF of the article on the Baker Lab website: 12:58
bkoep https://www.bakerlab.org/index.php/publications/ 12:58
jflat06 Ok - with that, we want to make a quick announcement - we're planning to release the Experimental client on Tuesday of next week. Like other large changes that concern the scoring function, this change will need to happen all-at-once to devprev and main in order to avoid scoring issues between people using the old client and the new client. 12:58
jflat06 As such, all puzzles will end before the transition, and new puzzles will be posted when the new client is released. 12:59
@LociOilingIRC ok, good to know 12:59
@S0ckrates I'm assuming the new scoring rebalance will affect the new Aflatoxin Puzzle round 12:59
rmoretti We're excited about this update. Not only does it fix some long-standing bugs, but it also includes a new-and-improved scoring method, which should hopefully improve designs and predictions. 12:59
rmoretti See this blog post for more details, if you haven't already: http://fold.it/portal/node/2004787 13:00
@LociOilingIRC latest experimental update seems good, running nicely on old two-core laptop 13:00
jflat06 That's great to hear! 13:00
rmoretti @S0ckrates - That's correct, we're planning to wait for the updates to release the new Aflatoxin puzzles 13:00
jflat06 We'll make a news post about this as well, for any that didn't catch the news in the chat here. 13:01
rmoretti While we don't know for sure, we hope that the new scoring will make the new aflotoxin designs that much better. 13:01
frood2IRC Hmmm - I guess it will all depend how much it sloes down many players. I cross my fingers….I have seen this sort of sea change B4….I hope it works out better than B4 13:01
@LociOilingIRC the membrane protein article is available on bakerlab.org 13:02
@LociOilingIRC https://www.bakerlab.org/wp-content/uploads/2018/03/Lu_Science_2018.pdf 13:02
@jeff101 It all sounds promising, but fingers crossed. 13:02
frood2IRC well I hope the lessons of NC introduction have been well learned 13:03
@S0ckrates The rebalance doesn't sound like it's going to look like things changed playerside, but I hope it does yield better results scientifically 13:03
@jeff101 I'm hoping less local minima to get stuck in. 13:03
frood2IRC always a good thing:) 13:03
jflat06 Great - we're over the hour now, so we'll end the chat here. 13:04
@LociOilingIRC ok, thanks 13:04
jflat06 Thanks to everyone for the great questions, and thanks to our scientists for their responses. 13:04
@jeff101 Thanks for having this chat. 13:04
jflat06 A log should be posted in a bit. 13:04
frood2IRC many thx jflat, bkoep, rmoretti…..well handled and informative    :) 13:05
bkoep Yes, thanks for the great questions! And thanks to all of you for playing Foldit! 13:05
@S0ckrates Thanks for the chat folks! 13:05

Generated by irclog2html.py 2.13.1 by Marius Gedminas
- find it at mg.pov.lt!

Get Started: Download
  Windows    OSX    Linux  
Windows
(7/8/10)
OSX
(10.7 or later)
Linux
(64-bit)

Are you new to Foldit? Click here.

Are you a student? Click here.

Are you an educator? Click here.
Search
Only search fold.it
Recommend Foldit
User login
Soloists
Evolvers
Groups
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, Microsoft, Adobe, RosettaCommons