Developer Chat

inkycatz Welcome everyone to our dev chat today. :) A few quick housekeeping items. 12:59
@jflat06 Hey everyone 12:59
betahelix Hi everybody! 12:59
inkycatz First of all, chats are logged, so if you have to go don’t worry we post the log afterwards 12:59
@jflat06 logging should be on 12:59
inkycatz jflat is awake so that’s also good! 13:00
inkycatz Topic list is pretty short today 13:00
inkycatz Our next chat is science, and in June. 13:00
inkycatz I think that’s all I’ve got on that front. 13:00
inkycatz Let’s start today with an update on scoring anomalies Hbond bridge filter. 13:01
inkycatz Since I know jflat’s been working on some fun stuff 13:01
inkycatz :D 13:01
@jflat06 I've been investigating the various issues that people have reported 13:01
@jflat06 happy to say, I've found the problem(s)! 13:02
inkycatz yay! 13:02
BletchleyPark very good, thank you 13:02
IgorS nice 13:02
@jflat06 One fix, which should handle the scoring being different on loads, is very easy 13:02
@jflat06 There may be other issues that are also contributing to the problem, in which case, this change alone wouldn't fix it. 13:03
@jflat06 the other issue, which BP reported, was the NAN showing up when you manually run the filter with the 'run' button 13:03
@jflat06 that problem is harder to fix, and is something I've been working on 13:03
betahelix Thank you all for finding and reporting these bugs! 13:04
@jflat06 As far as I'm aware, those are the main issues with the hbond network scoring at the moment 13:04
inkycatz How’s that blog post coming? :) 13:04
BletchleyPark thanks for picking that up quickly 13:04
@jflat06 I've been preparing a blog post in coordination with the mastermind behind Hydrogen Bond Networks in the Baker Lab 13:05
@jflat06 It's pretty close to being released - we will try to release the blog post at the same time that we make the new puzzle available in main 13:05
BletchleyPark is that a hbond bridge puzzle ? 13:06
@jflat06 And hopefully we'll be able to get the scientist that has been doing this work in the Baker lab in for a future chat 13:06
inkycatz awesome, I’m looking forward to that 13:07
IgorS sounds great! 13:07
@jflat06 It will be identical to the one in devprev, with possibly some minor changes. Our intent with the first puzzle is to give you a very good backbone where you can focus on just making the networks. The protein in the devprev puzzle was given to us by the Baker Lab as an exemplary protein backbone for constructing a variety of networks. 13:07
betahelix and we can't wait to see the new designs all of you will come up with! 13:08
@jflat06 From there, we plan on taking player designs from past puzzles, and having you guys improve the interface by integrating hydrogen bond networks. 13:08
BletchleyPark We will first have to upgrade to a fixed client 13:08
@jflat06 @BP, a new client will be posted to devprev for testing before we introduce hbond networks to main 13:08
BletchleyPark excellent 13:09
@jflat06 Susume asked about small molecule design - 13:10
free_radical hi guys 13:10
betahelix Hi free_radical! 13:10
free_radical I have some updates I can give 13:10
free_radical hi betahelix! 13:10
inkycatz cool, we’ll hit those updates after the small molecule design question :) 13:11
free_radical We released a client to our medicinal chemists at Vanderbilt. The good news is that they love it! 13:11
free_radical this was to test a new scoring feature called QSAR scores. They tell you whether the small molecule you design matches known features of other small molecules 13:12
free_radical So, we are hoping to release the new features to you soon; however, there are a couple of topics that we were thinking about first 13:13
free_radical 1) almost 60% of known drug targets are membrane proteins 13:13
free_radical so, we would like to give you a puzzle with membrane proteins. We have two lined up that will be tested by our experimentalists 13:14
free_radical the first one is a design puzzle for symmetric membrane proteins 13:14
free_radical we are having trouble designing symmetric membrane proteins and we would like to see if you guys could help 13:14
BletchleyPark how big are they ? 13:14
free_radical the second puzzle, which will be a segway into the drug design proteins, will be modelling of a gpcr protein 13:15
free_radical @bletchyPark, I am not sure. 13:15
free_radical one of the first drug design puzzles will be to design a drug for the model you created for the gpcr protein 13:15
free_radical so, you will act just like a scientist. Create a model, then design a drug into it 13:16
jeff101 have we ever done membrane proteins w/Foldit? do they have hydrophobic or hydrophilic cores? 13:16
free_radical hydrophobic scores, but it is just a modified solvation potential 13:16
free_radical almost like an inverse solvation potential 13:16
betahelix We have not done Membrane proteins, other than CASP targets we didn't know were not globular proteins 13:17
free_radical I am really excited to see how it goes 13:18
jeff101 will Foldit naturally compensate and tend to put blue residues inside and orange ones outside? 13:18
free_radical Foldit sampling techniques rely on the score function to guide its placement of residues 13:19
free_radical So, yes. When you run folding in Rosetta, you just swap the score function out 13:19
free_radical the same for design 13:19
free_radical So some bad news. I am moving across country right now 13:20
free_radical I will not be able to get much done until I settle into my new position, but as soon as that happens, I promise we will get the puzzles out 13:20
betahelix We wish you quick and painless transition, free_radical! (Ok gang, I'm sorry but I have to go) 13:21
jeff101 bye beta 13:21
BletchleyPark bye Beta 13:21
betahelix Keep up the great folding everybody. Take care! 13:21
betahelix Bye bye... 13:21
ComputerMage god luck! 13:21
ComputerMage good ;) 13:22
jeff101 is gpcr a membrane protein? 13:22
inkycatz bye beta! :) 13:22
free_radical yes, they are 13:22
ComputerMage always. 13:22
inkycatz (thanks for coming) 13:22
free_radical gpcr are the most important drug targets of our time 13:22
free_radical they control everything in your body 13:22
ComputerMage their 3d structure was revealed only 2-3 years ago 13:22
free_radical from depression, anxiety, apetitie, 13:22
free_radical you name it, they are responsible for it 13:22
free_radical Our lab helped crystalize the human one :D 13:23
ComputerMage main drug tagets 13:23
ComputerMage targets 13:23
ComputerMage I remember it ;) 13:23
ComputerMage I have a good mamory. And I know qhat technique was used as well ;) 13:23
ComputerMage qhat = what ;) 13:23
ComputerMage I am curios how we suppose to calculate hidrophobic belt that is burried in membrane ?) 13:24
ComputerMage If it's de-novo ;) 13:25
jeff101 being membrane proteins, do they have extra things like fatty acids attached? 13:25
ComputerMage that what I am exactly asking! ;_ 13:25
ComputerMage ;) 13:25
free_radical for membrane proteins, there are secondary structure predictions that can tell you what residues are in the membrane 13:25
free_radical one such service/product/algorithm is called octopus, anther is jufo, which is developed by our lab 13:26
free_radical when you do de novo folding, those preditions are used to determine how to place the protein into a "membrane" 13:26
free_radical just like soluble proteins for foldit, we do not use explicit solvation (water molecules) 13:27
free_radical we implicitly "show" them with the solvation score 13:27
free_radical so, with membranes, we implicitly show the membrane fatty acids in the membrane potential 13:27
free_radical the amino acids shown in the prediction are mapped as membrane residues 13:28
ComputerMage Can you show them slightly different visually so we will know that they are membrane burying sites? 13:28
ComputerMage It will be easier to fold them for us. 13:29
jeff101 or have them fixed in place where they are supposed to bind 13:29
free_radical because membrane proteins are aliphatic on the "outside" and "hydrophilic" on the inside 13:29
free_radical you have an inversion where they hydrophilic residues will be in the core and hydrophobic will be on the outside 13:30
free_radical they are really interesting proteins 13:30
free_radical for visulizations, we will not have to do anything major 13:30
free_radical we can just invert the idea that hydrophobics will be on the outside 13:31
@Susume2 and an inverted scoring function will be available in foldit for these puzzles? 13:31
free_radical correct, membrane proteins have been in Rosetta for a long time 13:31
ComputerMage but this membrane hydrophobic layer look like a belt on a protein so when I see it - then it will be easier for me to fold it. 13:31
free_radical we might be able to do a belt 13:33
free_radical I havent really thought about it, but thats why we have these chats :D 13:33
jeff101 do both sides of the membrane have the same properties? 13:33
free_radical correct @jeff101 13:33
free_radical in a general sense 13:33
free_radical So, I think we need a blog post! 13:33
inkycatz I do love a good blog post 13:34
free_radical to introduce membrane proteins and a first puzzle on it 13:34
BletchleyPark Will the membrane designs take up more computing resources 13:34
jeff101 like neurons exploit different concentrations of ions on each side of the memrane 13:34
free_radical @BletchleyPark, the membrane designs will be symmetric design problems, so they will be comprable to soluble proteins 13:35
BletchleyPark how many residues ? 13:35
jeff101 one side of membrane is intracellular while the other is extracellular 13:35
free_radical I think the symmetric ones are 100 amino acids, but I am not entirely sure. I would have to ask our designer 13:36
BletchleyPark ok 13:36
@Susume2 100 times how many symmetric copies? 13:36
free_radical I believe 2 13:36
free_radical so 200 residues in total 13:36
@Susume2 thx 13:36
free_radical 100 symmetric. It might be slightly more, but I cant remember 13:37
free_radical I helped design them a long time ago 13:37
free_radical and we are just now getting some cool experimental data from them 13:37
free_radical the whole goal of the membrane proteins is to introduce everyone into the concept of these proteins 13:38
free_radical for when we start doing drug design 13:38
jeff101 is the distribution of AA's different in soluble vs membrane proteins? like do membrane ones have more alanines? 13:38
free_radical Honestly, I think the distribution is similar, there is just an inversion of hydrophobic residues on the surface and hydrophilic in the core 13:39
free_radical there might be a preference, but we have optimized the score function for that 13:39
BletchleyPark 20 minutes 13:40
jeff101 just in design puzzles, we pick which AA's to use. Would be good to pick realistic distributions of them. 13:40
eusair FR, are you involved in game design/programming? 13:41
free_radical @eusair, I am not sure how to answer that question. I am a programmer and programmed during my phd career 13:42
free_radical I wrote a score function for Rosetta called the partial covalent interactions score function 13:42
free_radical I am now working on drug design (programming) and designing the interface for Foldit 13:43
free_radical does that help? 13:43
eusair so are you the guy to talk to if i've ideas for new game features? 13:43
inkycatz The best place for those would be our feedback forum :) 13:44
BletchleyPark Please see my questions / suggestions on the bandwisth usage by the interface. 13:44
eusair ok 13:44
BletchleyPark can we please touch on the deaf client / dead chat issue update ? 13:44
inkycatz so that we can keep track of them and discuss them better - we love reading everyone’s suggestions 13:44
inkycatz (even if we can’t always implement them for reasons) 13:44
IgorS are you guys all in Seattle? 13:45
@jflat06 @BP, we may get to that at some point, but a low-animation build isn't soemthing that's a high priority. Bandwidth concerns for remote access are a special case, and time spent on that front will affect a minimum of players. 13:45
BletchleyPark ok, deaf client / dead chat issue ? 13:46
inkycatz I think this is a topic that’s probably better diagnosed outside of chat, from my understanding of it. :) 13:47
inkycatz jflat, did you have any topics you needed more feedback on from our community? 13:47
BletchleyPark This is a logn standing very annoying issue 13:47
IgorS :-) 13:47
@jflat06 for the client connection issues, we have tried extensively to reproduce problems with the connection light. Foldit doesn't actually maintain a persistent connection. It connects through http requests any time communication is needed. 13:48
@jflat06 the light is just an indicator about the status of recent attempts 13:48
BletchleyPark The light is often green while no connection exists 13:48
BletchleyPark trying to open a puzzle or shared solution will then crash the client 13:48
@jflat06 There may be issues with how the light is representing the failure/success of recent connections, but we have not been able to reproduce those. 13:48
Skippysk8sirc seems random... 13:48
BletchleyPark just unplug the network cable for a few minutes ? 13:50
Skippysk8sirc I never know if I have to manually upload to get credit for a puzzle 13:50
@jflat06 For chat reconnecting, this is another case where we are having trouble reproducing the issue. If you can provide very precise reproduction steps including the state of all your connected accounts (whether they are identified, whicih channels they are on, etc), that may help. 13:50
@jflat06 @BP we have tried that. Every indication with our tests is that the light is functioning as intended. 13:50
BletchleyPark it always fails here 13:51
BletchleyPark we can take the conversation offline and try to resolve it, we will make friends with that 13:51
inkycatz Always appreciate the suggestions for repro steps, jflat. Good to have when reporting any issue, especially connection related ones. :) 13:51
inkycatz We’re all friends here, just wanted to make sure we hit any other burning questions before we wrap up for today! 13:52
@Susume2 are there any new ED features in the works? 13:53
Skippysk8sirc can we agree we will keep at least one smallish puzzle open at a time. our group has lots of new players. 200 AAs is too big for them 13:53
@jflat06 @Susume, I think that beta_helix has some students that are working on some minor improvments for ED. 13:53
@silverberg Centroid? 13:53
@Susume2 nice 13:53
@jflat06 We also agree that editable density would be a worthy goal - this is somewhat hard to implement in practice. 13:54
@jflat06 I tried it once, and had something working, but it wasn't in a state that was up to the quality needed for release. 13:54
inkycatz Thanks for the suggestion, Skippy - I don’t know about making a promise here without our scientists, but we’ll make sure they hear the suggestion. 13:55
@jflat06 For actual scientific puzzles, there's constraints on what we're able to post 13:55
BletchleyPark why ? 13:55
BletchleyPark legal ? 13:55
@jflat06 we don't always have a small puzzle to post, which is why we tend to keep revisited puzzles open 13:56
BletchleyPark okk 13:56
@jflat06 no, it's a matter of finding problems that fit well into foldit and can be framed correctly 13:56
BletchleyPark clear, thanks 13:56
inkycatz And that looks like a good note to wrap things up on. Thank you all for coming and thank you to jflat and free_radical today for sharing a bunch of interesting stuff with us. :) 13:57
BletchleyPark how about previous science puzzles to see if we can come up with better results this time 13:58
BletchleyPark with the new tools ? 13:58
inkycatz We look forward to seeing everyone at future chats! Stay tuned to the news for our next chat date. :D 13:58
@jflat06 @BP that might be a good direction to look into 13:58
jeff101 thanks for having these chats 13:58
@jflat06 alright, thanks for attending, everyone! 13:58
free_radical thank you! 13:58
BletchleyPark thank you, goodnight 13:58
Skippysk8sirc thanks 13:58
IgorS thanks 13:58

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

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, RosettaCommons