Developer Chat

inkycatz Hey there! 12:32
inkycatz Looks like the gang is all here. :) 12:32
inkycatz Welcome to the last science chat of 2015! :D 12:32
inkycatz A few quick housekeeping items. First of all, chats are logged, so if you have to go don’t worry we post the log afterwards. 12:32
inkycatz http://fold.it/portal/chats 12:32
inkycatz Obligatory social media announcement - you can find us on Facebook 12:32
inkycatz https://www.facebook.com/folditgame/ 12:32
inkycatz And Twitter: @foldit 12:32
bkoep Hello everyone! 12:32
inkycatz FYI I am running pictures taken of the Baker Lab on Fridays until the end of the UW quarter via Twitter, but also repost them the next week to Facebook. bkoep took some sweet shots. 12:32
inkycatz (thanks for those, they’re so neat to look at( 12:33
inkycatz Always good to have our social media channels handy anyway in case of “issues”, if you know what I mean. 12:33
inkycatz You know what I mean. 12:33
inkycatz By the way, thanks jflat for the logging! 12:33
Wbertro :) 12:33
inkycatz Thank you to our team members (bkoep, vmulligan, free_radical) for changing their schedules a bit to accommodate this Friday chat! 12:33
inkycatz That being said, scientists may dive in or out of chat at any time! STAY ALERT. 12:33
inkycatz We have six questions on the list, and let’s try to keep follow ups to 3-4 per question if needed. 12:33
inkycatz After we get through the posted list - if there are any questions, we do have time scheduled for that. 12:33
inkycatz Please do keep them short and simple if you can so we can keep things moving - we want everyone who has questions to have a great opportunity to ask! 12:33
inkycatz I have gone on enough, and we’re glad you could make it today. Let’s do this! 12:34
inkycatz Let’s start with any updates from the team 12:34
inkycatz I see a free_radical - any updates for us on drug design? 12:34
free_radical hi! 12:34
free_radical We just got done previewing the first puzzle with the Foldit team 12:34
free_radical You should expect an update to devprev with teh first puzzle 12:35
free_radical The first puzzle is a very simple puzzle to test the waters with the new interface and get some feedback 12:35
free_radical Upcoming puzzles will include kinases (important to cancer), gpcrs(important to mood), and hiv proteases 12:36
inkycatz Excellent - any eta? 12:36
inkycatz Before Thanksgiving here in the US? 12:36
inkycatz After? 12:36
inkycatz Too soon to tell? 12:36
free_radical Monday for devprev 12:36
free_radical for the first puzzle 12:36
@TimovdL So just a test puzzle? 12:37
Wbertro any pre-puzzle info on the interface? 12:37
free_radical Yeah, it is a test puzzle. There is a blog post for the first puzzle that will be posted 12:37
free_radical it is a cool puzzle, though. We are testing if you can beat medicinal chemists 12:37
Wbertro good 12:38
free_radical in determining which small molecule binds better 12:38
free_radical so, you will design the small molecule by changing atoms and determine which one binds better 12:38
free_radical At the end of the puzzle, I will collect the results and have a blog post up showing how everyone faired against the medicinal chemists 12:38
free_radical sorry, I am an awful speller... 12:39
@TimovdL Also extra intro puzzle to get the hang of the tools? 12:39
inkycatz I’m holding you to these blog posts, but I’ll fix the spellings :D 12:39
Wbertro any LUA functions or just manual work? 12:39
free_radical of course! We have 6 new tutorial puzzles setup 12:39
free_radical but those will wait until the more complicated puzzles are released. The tutorial puzzles trace a design for VEGFR 12:40
inkycatz Cool stuff. :) 12:41
free_radical I am really excited to get everything out, but we are making sure that we dont relese broken things 12:41
free_radical there is nothing more frustrating than seeing something cool but not being able to use it 12:41
spvincent What is VEGFR? 12:41
free_radical vegfr2, its a kinase 12:42
free_radical so, for each puzzle, there is going to be an infographic explaining the protein 12:42
free_radical and what the disease is and who it effects 12:43
vmulligan The VEGF receptor.  It's involved in signalling for blood vessel growth, so it's an important target in cancer (where new blood vessels have to feed the tumour) and certain other diseases. 12:43
free_radical However, I havent created the media for that yet :D 12:43
inkycatz bkoep? vmulligan? any updates on stuff you’re working on or shall I dive into some questions? 12:43
free_radical thanks vikram! 12:43
vmulligan We can probably go straight into the questions.  I've got some fun stuff planned, but it'll be a little while -- there's under-the-hood infrastructure stuff that needs to be in place first. 12:44
inkycatz Heh, okay. 12:45
inkycatz speaking of VEGF 12:45
inkycatz Is there a current timeline for when the results might be published for the VEGF receptor 1 results mentioned back in July's chat? 12:45
bkoep That's all in the hands of our collaborators at this point 12:46
bkoep betahelix might know more (maybe we can ask him if he joins later) 12:46
inkycatz All righty. :) We’ll see if we remember to ask later! 12:46
inkycatz (Susume will, since it was her question. :D) 12:47
inkycatz Next up! 12:47
inkycatz In symmetry puzzles, should we consider also consider the monomer in isolation? 12:47
vmulligan Good question! 12:48
vmulligan We do that, when we're checking designs and thinking about what to produce -- but it's an after-the-fact check, not something that we typically do during design. 12:48
vmulligan The problem is that considering both the monomer and the oligomer simultaneously during design is hard.  It's called the "multi-state design problem", and it's something that we're not great at yet. 12:48
vmulligan But there has been some methods development in that direction.  Someday... 12:49
bkoep Right, the simple answer is: "No, don't worry too much about monomer stability" 12:49
vmulligan What we're doing instead is encouraging more polar interfaces between subunits, using things like the hydrogen bond network bonus. 12:49
vmulligan This helps to ensure that the monomer has a well-defined hydrophobic core and a somewhat less hydrophobic surface (even at the binding interface). 12:49
bkoep That said, we DO like to see that the monomer itself is a compact, globular structure 12:51
inkycatz Anyone have any followup questions on that? 12:52
spvincent Is there perhaps an overemphasis on getting structures globular? 12:53
spvincent Many naturally occurring proteins are very far from spherical 12:53
bkoep That's true 12:53
vmulligan Very true.  It's the drunkard and the lamppost, though: we look for the keys under the lammpost because that's where there's the most light.  We focus most on globular proteins, because that's what we can score most reliably. 12:54
bkoep I guess what's most important is that the monomer unit has a hydrophobic core 12:54
Skippysk8sirc would tentalcles that attach be useful? 12:54
vmulligan Indeed.  And we are starting to get away from globular proteins a bit with the repeat proteins and whatnot.  Maybe we should give players a repeat protein design problem at some point... 12:55
bkoep With the small proteins we're designing in Foldit, an extended protein is generally one without much of a core 12:55
inkycatz Skippy’s question makes me think of a lego octopus. :) 12:56
vmulligan >would tentacles that attach be useful? 12:56
inkycatz So thanks for that image. 12:56
Skippysk8sirc was thinking more of a tail that wraps. but welcome:) 12:58
bkoep Yes, if you're interested in more extended protein designs, you should check out this recent paper from the Baker Lab: 12:58
bkoep http://www.bakerlab.org/system/files/ageneralcomputational_Baker2015.pdf 12:58
spvincent tx: will take a look later 12:58
vmulligan A tail that wraps around the protein might be okay, as long as you don't get into knotted topologies or really extensive wrapping.  There's an entropic penalty to forming topologies like that (and we're really, really bad at scoring entropic effects like that, so the score function would not reflect that). 13:01
inkycatz next question, team? 13:01
vmulligan Sure. 13:01
inkycatz On the recent V3 hairpin puzzle, what is the potential benefit of mutating the hairpin to make it more rigid? 13:02
vmulligan Ah -- I can answer that one. 13:02
frood2IRC as a 13:03
frood2IRC a 13:03
frood2IRC oops 13:03
frood2IRC sorry 13:03
inkycatz it’s cool. 13:04
vmulligan If we can make an artificial HIV protein that presents the V3 hairpin locked in its pre-fusion conformation, then we can inject this protein into animals and try to raise antibodies that bind to that region of the protein, locking the NATURAL protein in the pre-fusion conformation (preventing it from switching conformation as it needs to when HIV invades cells). 13:04
vmulligan If we just inject the natural protein, though, it's much harder to get an antibody that binds the way we want to a floppy loop. 13:04
vmulligan Alternatively, a locked version would be useful as a laboraory tool for screening designed proteins intended to do the same thing. 13:05
vmulligan (Same thing as the antibody, I mean.) 13:05
@Susume2 cool! 13:06
kenzo92 Are there other techniques, aside from computing, to predict the ternary structure of a protein? How accurate are our predictions in relation to "real" proteins? 13:06
kenzo92 Errata corriger: other techniques to know the ternary structure ... 13:07
vmulligan There are experimental techniques (x-ray crystallography, nuclear magnetic resonance spectroscopy, and, more recently, cryo-electron microscopy).  There are unrelated computational techniques, like long-trajectory molecular dynamics simulatoins (but this is limited to very tiny proteins that fold very, very fast). 13:07
vmulligan The only non-computational technique that I can think of is giving amino acid sequences to gamers and asking human beings to predict sequences :) . 13:08
inkycatz ha! I see what you did there 13:08
bkoep Also, a lot of times we can use homology to predict structures 13:08
vmulligan *simulations (typo) 13:08
inkycatz ready for our next question? :) 13:09
bkoep I would still consider this "computing," but it still involves the use of empirical data 13:10
vmulligan Ah, yes.  There are many variants of the techniques that we use in Rosetta: homology modelling, modelling guided by limited experimental data (low-resolution electron density from x-ray or electron experiments, solution-phase x-ray scattering. incomplete NMR data, etc.) 13:10
vmulligan Modelling guided by evolutionary data, etc. 13:10
vmulligan But yeah, it's all still computing. 13:10
kenzo92 Ok, thank you :) 13:11
vmulligan The ways in which people solved x-ray structures before computers were pretty neat... They're a bit long to get into here, though. 13:11
inkycatz that’d make a great blog post sometime 13:11
inkycatz you know, in all that spare time you guys (don’t) have :D 13:11
vmulligan Yes, maybe I'll write about it. 13:11
kenzo92 I'm going to google that :) 13:11
vmulligan :P 13:11
inkycatz Here’s a good question from our thread! 13:12
inkycatz In opposition to proteases, are there also "glue" proteins or are they even thinkable? By that I mean proteins that could join the C- and N-termini of different proteins or even the very same one, forming a ring. 13:12
vmulligan Look up "Richards' box", though. 13:12
vmulligan (Sorry -- that was for the previous question.) 13:12
vmulligan Re: "glue" proteases. 13:13
vmulligan Yes, off the top of my head, I can think of examples. 13:13
vmulligan Not necessarily rings, but thinsgs that conjugate separate polypeptides together. 13:13
vmulligan There are the ubiquitin ligases, for example, that attach ubiquitin to many different proteins (albeit by a sidechain-backbone connection). 13:13
vmulligan There's also protein splicing, analogous to RNA splicing, in which a self-removing segment in the middle of a protein (an "intein") will splice itself out, ligating the ends of the flanking "exteins". 13:15
vmulligan And there are ways that we can take advantage of this in the lab to make N-to-C cyclic proteins. 13:15
vmulligan (We can also synthesize N-to-C cyclic peptides chemically.  I've got one going right now, in fact.) 13:15
spvincent What's the purpose: to make the peptide more stable? 13:17
vmulligan Finally, enzymes can only reduce the barrier to a reaction taking place; they don't set the directionality of a reaction.  So you can drive the protease reaction backwards under artifiical conditions. 13:17
vmulligan In our case, yes, we want to make N-to-C cyclic peptides to make the peptide more conformationally stable, and to protect from degradation (especially from proteases that "chew" from the ends). 13:17
inkycatz (what a great question) 13:19
inkycatz Everyone ready for some electron density chat? :D 13:19
hoglahoo I am ready 13:20
inkycatz Oh look at that, I said ED and betahelix arrived 13:20
frood2IRC grrr   lol 13:20
inkycatz hey there! 13:20
inkycatz great timing on that 13:20
betahelix I couldn't miss the ED fun! (sorry I'm late :-) 13:20
inkycatz ha! 13:20
inkycatz How much work is required in a typical research lab to solve a structure like in Puzzle 1152? 13:20
inkycatz There you go. Dive on in! 13:20
vmulligan Not to be confused with electron-dense chat, which is just a really boring conversation about electrons.  :P 13:20
@MikeCassidytoo    On the tutorials follow the instructions closely, don't overthink and do too much. Don't be afraid to reset and start again. On the later ones don't be in a rush to finish the level and move on this is your opportunity to play with the tools and find out how they work. There is lots of guidance on the wiki as to how to do the tutorial, 13:20
@MikeCassidytoo http://foldit.wikia.com/wiki/Tutorial_Puzzles 13:21
@MikeCassidytoo but you will learn more if you have a go at doing them for yourself first. 13:21
@MikeCassidytoo Sorry wonrg window 13:21
bkoep Also, we should introduce ScottH, who is joining us from the Bardwell Lab at the University of Michigan 13:21
inkycatz (Also hi to ScottH, joining us from UMich, yes!) 13:21
ScottH Hi! 13:21
betahelix Scott... this seems like the perfect question for you! 13:21
ScottH Hmm, for this particular puzzle 13:22
ScottH I'd say that if I were doing it by myself, I would guesstimate about 8 hours 13:22
ScottH To do it right 13:23
ScottH I know people that I bet could do it in a lot less, but then again, I know other people that would take a fair bit longer 13:23
ScottH Depends on experience as well as strategy 13:23
frood2IRC and machine 13:23
ScottH yes, that too 13:23
ScottH And also to clarify, that's not including any of the getting crystals and collecting data 13:24
@Susume2 is that after you have the ED data?  or including the crystallography part? 13:24
@Susume2 jinx 13:24
ScottH yeah... that first part is often the long part of a project 13:24
ScottH it depends though. 13:25
ScottH I have another protein where we got crystals in two weeks from starting, but it took another year to solve the structure even though those crystals were great 13:25
Wbertro why was that? 13:26
ScottH well, it turned out that one of the proteins in the crystal was moving around a lot. Took a lot of work to figure out what conformations it was in and where it was. 13:26
ScottH as opposed to if it had been in one static position, which is much easier 13:27
Wbertro like taking photos of a subject and not finding it in the same place each time 13:28
Wbertro compared to the surroundings 13:28
ScottH yeah, that's basically what we encountered 13:28
Wbertro ok 13:28
ScottH maybe someday I'll put a puzzle like that on foldit... it would definitely be a challenge 13:29
inkycatz :D we’d be glad to work on that with you! 13:29
@TimovdL we like challenges 13:29
frood2IRC u r refering to areas that are called ;disordered’? 13:30
Wbertro but wouldn't it be in only one place in a one shot of a crystal? 13:30
Skippysk8sirc slow lens LOL 13:31
Wbertro could be? 13:31
ScottH Proteins can actually move a lot in crystals. It's one of the reasons why people often freeze them before collecting data 13:32
ScottH But even if you freeze them, if there was motion before freezing- now it's frozen in a slightly different position in each copy of the protein in the crystal 13:32
ScottH so the signal you see from the protein starts to blur and average out, eventually descending into the noise if it's too much motion 13:33
ScottH at which point, the "disordered" label gets applied 13:33
frood2IRC that makes perfect sense - thanyou Scott 13:33
Skippysk8sirc are there types of protein parts most likely to be slippery? 13:34
@Susume2 are the disordered parts usually on the surface? 13:34
ScottH I think those last two questions are asking the same thing- the answer is typically yes, surface residues are more likely to be disordered 13:35
ScottH the more buried, the less motion it will typically have. Not always true, but a decent rule of thumb. 13:35
frood2IRC I really don’t want to see one where the center is disordered!   LOL 13:37
inkycatz Now we do have just one more question for the team then we can revisit or open things up for more questions. :) I know everyone probably has come up with some. 13:37
inkycatz here it is: How do other protein copies affect contact or ED data? How do you decide where to cut an electron density cloud for Foldit puzzles? 13:37
bkoep Good question 13:38
ScottH You want to handle that one, bkoep? 13:38
bkoep Sure 13:38
bkoep We'll start with predicted contacts data 13:38
bkoep Sometimes, when we have evidence that two residues are in contact, we're not sure if those residues are in contact on the same copy of a protein 13:39
bkoep If we see that residue A is predicted to be in contact with residue B, it's possible that residues A and B are on opposite sides of a protein, and only contact one another when multiple copies of the protein bind one another 13:41
@Madde Is there a rule of thumb, e.g. if both are hydrophilic then they are probably on different copies? 13:42
bkoep Without more data, we can't know for certain if this is the case. 13:42
bkoep @Madde, that's an interesting idea, but I don't think it would hold out 13:43
bkoep There are plenty of examples of intra-molecular salt bridges, for example 13:43
@Madde but more probably than two phobics? 13:44
bkoep @Madde, maybe, but there are also plenty of examples of exposed hydrophobics that make inter-molecular contacts 13:44
bkoep I would hesitate to make any generalizations 13:45
bkoep The other half of the question addresses electron density 13:45
bkoep When we calculate electron density maps, we know the size and dimensions of the "unit cell", but we don't necessarily know the boundaries between individual protein copies within that unit cell 13:46
frood2IRC which causes us players probs with areas we need to ignore 13:47
bkoep Usually, we have a "molecular replacement" model 13:47
bkoep But in the case of Puzzle 1152, that density map was calculated without a model, and we normally wouldn't know where the protein boundaries are until we developed a model for the density 13:49
bkoep If we know where the protein boundaries are in a density map, we will often cut out the density of a single protein copy for Foldit puzzles (mostly to reduce the size of the puzzle) 13:49
frood2IRC is this why it is refered to as ‘unrefined’ ? 13:49
bkoep But other times, we have provided an entire unit cell, and left it to Foldit players to figure out the boundaries between protein copies 13:50
betahelix Those puzzles had huge ED clouds! 13:52
@TimovdL Do you usually know parts of the boundaries, so you dont need to give us too much density? 13:52
bkoep For Puzzle 1152, we thought it was pretty clear where the protein boundaries were (even without a model), so we cut out the density of a single protein copy 13:52
frood2IRC so all EDcloud i n1152 should be filled? is that what u r saying? 13:54
bkoep @Timo, yes, usually by the time we have an electron density map that is clear enough to build into, we have a molecular replacement model that tells us something about protein boundaries 13:54
ScottH so to clarify an frood2IRC's point, there may be some density that remains unfilled 13:56
ScottH but the vast majority of it will be filled if completed properly 13:56
frood2IRC kk - thx 13:56
inkycatz BTW we have reached the end of our scheduled chat time, and I’d like to thank all our scientists (who may or may not be able to stick around now but are welcome to) and all of you for coming. :) This has been super fun. 13:58
Wbertro in reference to pre-computer solving of x-ray structures: 13:59
Wbertro https://en.wikipedia.org/wiki/Frederic_M._Richards 13:59
inkycatz Oh hey thanks berto for the link. 13:59
inkycatz Good callback from earlier. 13:59
ScottH thanks all- I look forward to seeing the structures on 1152! 13:59
betahelix And an especially big thanks to ScottH for stopping by! 13:59
inkycatz yes thank you ScottH! :D 13:59
bkoep Yes, thanks ScottH! 13:59
betahelix Good luck to your students ;-) 14:00
Wbertro and thanks for the chat, very interesting 14:00
frood2IRC I have to go - many thanks to the usual suspects (u know who u r) and scott 14:00
@TimovdL tx, it has been informative 14:00
@MikeCassidytoo TY 14:00
@Susume2 beta, any word on when the VEGF receptor 1 results might be published? 14:00
inkycatz Ah there that is, thanks Susume :D 14:00
inkycatz I knew I’d forget to re-ask but you wouldn’t! 14:00
inkycatz (teamwork) 14:01
betahelix I keep asking them, but they are still working on it. I will point them to this chat log when I ask again! :-D 14:01
@Susume2 :-D ok 14:02
vmulligan Cheers, everyone.  Got to run! 14:02
betahelix ScottH, maybe we should ask you: "How much work is required in a typical research lab to PUBLISH a structure like in Puzzle 1152?" ;-) 14:03
inkycatz thanks vmulligan! :D 14:03
inkycatz aww but he left. :) 14:03
inkycatz any questions for betahelix and bkoep before they dash off? :D 14:06
inkycatz or just beta! 14:07
inkycatz haha 14:07
betahelix Thanks again to all of you and keep up the great folding (especially in the final week of 1152!)... 14:07
betahelix We hope that has been a fun challenge! 14:08
Wbertro yes, bring some more 14:08
inkycatz Appreciate the hard work you all put in to help making that happen. :) 14:08
betahelix That's the plan, Wbertro! 14:08
Wbertro great 14:09
betahelix Take care, everyone... bye bye! 14:09
Wbertro bye 14:09
inkycatz bye! :) 14:09
@MikeCassidytoo bye betahelix 14:09
betahelix Thanks again for organizing a great chat to end the year, inkycatz! 14:09

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