Developer Chat

Scientist Chat

@inkycatz Welcome everyone to our Scientist Chat. I know it's been a while, and we're excited to bring today's topic your way. I'd like to introduce cymbal_king, who is our guest today along with jaclyn. 13:59
@mimi2 greetings and salutations 14:00
@inkycatz Why don't you tell us a little bit about yourself and what you're up to before we dive into the couple of posted questions. 14:00
karstenw hi king, jac 14:00
vmulligan Hello. 14:00
cymbal_king thanks, I'm a first year grad student at Dartmouth college 14:00
cymbal_king in the program in experimental and molecular medicine 14:01
JacAttack and I'm a 4th year graduate student in the same program at Dartmouth 14:01
cymbal_king I have some background from my undergrad in protein modelling and folding and thought I could help jaclyn uncover some aspects of regulation of the protein she is interested in 14:01
cymbal_king (Id2) 14:02
@inkycatz Bruno Kestermont couldn't join us today but posted a series of questions in our forum, so I'd like to tackle those first then open up the floor. 14:03
@inkycatz Sound good? :) 14:03
JacAttack sure 14:03
cymbal_king sounds good 14:03
@betahelix Good call :-) 14:03
@inkycatz First up (there are multiple parts so just tell me done or something so I know when to move to the next part!) 14:03
@inkycatz Do we work on the complete protein? or on a subset of it? 14:03
cymbal_king The Fold it puzzle is the entire 134 amino acids of Id2, in a homodimer 14:04
cymbal_king the starting HLH region is from the crystal structure found on the PDB website 14:04
@inkycatz <if anyone wants to ask a question, send me a message so I can queue you up, btw> 14:05
cymbal_king (done) 14:05
@inkycatz Do you believe that the terminal could also include sheets? 14:05
cymbal_king i'm not too sure about sheets, these types of molecules usually include a lot of helices and loops/turns/coils 14:06
cymbal_king Some partial Circular dichroism data suggests that the terminal regions have a lot of the coil/loop/turn structure 14:06
@TimovdL Can you explain partial Circular dichroism? 14:07
vmulligan Quick interjection: circular dichroism (CD) spectroscopy is a lab technique for estimating the helix/sheet/coil content of a protein, without knowing its full structure.  The helix signal, though, tends to swamp out the sheet or loop signal, so proteins with a mixture tend to look helical. 14:07
cymbal_king the top hits we have gotten so far seem to have nearly all of the terminal regions in helices, the structural assessment of those hits suggest that there could be less helical structure 14:07
spvincent The SS predictions aren't very illuminating in this respect 14:07
@betahelix indeed the secondary structure predictions from the SAM server are consistent with the CD data: 14:08
@betahelix http://fold.it/portal/node/997867 14:08
cymbal_king (Next question) 14:08
@inkycatz I think this one is more of a comment/wonder, but if you or any of our usual scientists in the house have a comment here it is: Concerning dimers: each branch should correctly fold separately. However, we experienced that the link between the dimers gain a lot of (foldit) points in symmetric puzzles (the branches of some of our top symmetric puzzles could certainly not be separated without 14:09
@inkycatz losing a lot of points). I wonder how our results would be different separating the elements or working on the all dimer. 14:09
@inkycatz (Any thoughts?) 14:10
cymbal_king yeah, it is a little more complicated of a question 14:10
cymbal_king functionally, Id2 acts in a heterodimer against various other proteins 14:11
cymbal_king however, the published crystal structure was in a homodimer. Hence why we choose to use a homodimer here for the puzzle 14:11
@betahelix I think it comes up because in the past with symmetry design puzzles we got some very interesting topologies that would not fold up well as monomers. 14:12
cymbal_king Having the symetry function made it so we would have an identical structure on either side 14:12
vmulligan It's not unusual to have a protein with disordered regions that only take on structure in the context of larger assemblies. 14:12
bkoep I 14:12
vmulligan It's possible that the tail regions do that, here. 14:12
bkoep Thanks vmulligan, I was going to mention that as well 14:13
cymbal_king good points 14:13
@TimovdL So suspecion is that it is partial symmetric? 14:13
spvincent I take it efforts to crystallize the dimer have failed, apart from the central bit 14:13
cymbal_king yes, having multiple intertwinned structures would be very difficult to analyze, especially if we were to take a step back and try to rationalize interactions with other proteins 14:14
cymbal_king @spvincent, yes crystallizing the terminal regions has not been fruitful 14:15
@betahelix Here is a link to the solved central bit, for anyone who didn't see it yet: 14:15
@betahelix http://tinyurl.com/pdb4aya 14:15
cymbal_king however a lot of work on these proteins has been with truncated versions as well 14:15
cymbal_king so they lack the c-terminal region altogether 14:15
@Madde do you have any contact predictions or ED clouds? 14:16
@TimovdL We are good with ED clouds 14:16
karstenw ;) 14:16
cymbal_king our lab isn't heavily involved in the actual crystallization of protein 14:16
cymbal_king we are a very translational lab and stumbled upon some regulatory modifications that affect stability 14:17
cymbal_king that's why the puzzle has a phosphorylated amino acid 14:17
vmulligan The electron density data would be from the crystal structure that gave rise to the partial structure, and would probably be obtainable.  Cymbal_king, do you know whether the solved structure was based on a full-length protein in which the tails just weren't visible in the electron density, or whether it was a truncated protein used for crystallography? 14:18
cymbal_king the crystal structure had a n-terminal tail, but not a c-terminal tail 14:18
@TimovdL So let us solve that first 14:18
karstenw i think the first 8 segs coming out of the structure are tricky to determine. i wonder if the cloud would give us a clue. (not during casp though). 14:19
cymbal_king @TimovdL, clarification, you would like to solve the truncated protein first? 14:19
@Madde I suppose N and C-terminal won't interact with each other anyway? 14:19
@betahelix Agreed that there will be no ED puzzles until CASP11 is over :-) 14:20
@TimovdL If given as an ED we might be able o solve that 14:20
@Susume2 we'd like to work from the ED on the truncated protein 14:20
cymbal_king ah that's teh kicker, the N-terminal is phosphorylated, while the C-terminal has a binding site for a ubiquitinase 14:20
vmulligan Electron density is unlikely to help if the region of interest -- the tails -- aren't visible in the electron density. 14:20
cymbal_king so we think there is some form of conformational shift/crosstalk between the two tails that prime the protein for degredation 14:21
cymbal_king upon phosphorylation 14:21
@TimovdL The lack of given sec structure lets us really shoot in the blind 14:21
* rav3n_pl it is way over my english skills, waiting for something that I can comment :) 14:22
alcor29 You said the presence of  helices make it seem helical, So the fact that server predictions show mostly coil would indicate that there are no more than the 3 helices shown? 14:23
karstenw my two tails kept trying to "shake hands" in one way or another. i separated them by hand. any idea if that is the right move? 14:23
cymbal_king the lack of 2ndary structure has been a problem even before fold it. The computational algorithms were having difficulties with the model as well 14:23
@TimovdL On that phosphor thing, should that be on the inside, the outside, any clues 14:23
spvincent Does the lack of any confident SS predictions suggest that, as mentioned earlier, that the momomer would have no fixed structure? Or is there no correlation 14:24
cymbal_king @timvodl, the phosphorylated protein should be accesible by the solvent, so it shouldn't be too buried 14:24
vmulligan @alcor29  I don't think that we can guarantee that there are no more helices. 14:25
cymbal_king @alcor, there should be a significant amount of coil/loop nature to the tails, but I wouldn't rule out any extra helices 14:25
vmulligan @spvincent  A negative result in prediction tends not to be definitive.  It COULD be the case that the termini are intrinsically disordered, but we don't know that from the evidence at hand. 14:25
cymbal_king the computational modelling programs all generated some extra helices in the tails 14:25
cymbal_king (in the best models) 14:26
cymbal_king even as a monomer 14:26
@mimi2 but are you completely eliminating the possibility of sheets? 14:26
@TimovdL Foldit rewards helixes a lot, that might be a problem 14:26
cymbal_king I don't want to eliminate anything so as to not skew things the wrong way, but I doubt there would be a lot of sheets 14:26
rav3n_pl If we not want helices scoring can be "fixed" in puzzle, we see this few times 14:27
vmulligan @mimi2:  No, sheets are possible.  Not seeing a sheet signal in a CD experiment is not definite either, particularly when your protein has a lot of helices.  There could be a bit of sheet there. 14:27
Skippysk8s i do agree that scoring rewards helices. its too bad, as it discourages trying sheets. for example, gargleblasters did well with weak helices by sitcking to helices.... 14:28
@TimovdL I ran a seperate prediction on the tail and the head, it gave 2 sheets 85-88 and 105 110 14:28
cymbal_king The CD spectra of a truncated Id2 showed no sheets at all, but inkycatz talked about those subtlties earlier 14:28
cymbal_king (the truncation was lacking a c-terminal tail) 14:28
@TimovdL But i didnt score at all 14:28
cymbal_king it would be interesting if there was a sheet at 105-110 14:29
cymbal_king 100-107 is a "dead-box" 14:29
cymbal_king the binding site of a ubitquitin ligase 14:29
frood2IRC Is it fair to assume that the helices in the starting model start and end as shown? 14:29
cymbal_king @frood2, I think that is a reasonable assumption 14:29
karstenw iits trickt for us because of scoring. i have a wandering loop that fell into place very nicely, but i converted it to a pair of sheets for points.  the protein has a meandering quality that is hard to work with imo. 14:30
@Susume2 do ubiquitin binding sites have a typical shape, or pattern of hydrophobes? 14:30
spvincent The whole thin is hard to with: very slow 14:30
spvincent *thing 14:30
cymbal_king @susume, jaclyn is looking into it 14:31
cymbal_king I know i've probably missed some questions, any in particular? 14:31
@betahelix @karstenw if you ever find any interesting topology that scores poorly (or can easily get outscored) please use the Share With Scientists button! 14:32
karstenw kk 14:33
alcor29 Do you have any info on whether the dimer consists of 2 discreet units or is it interlaced? 14:33
@betahelix Did this question get answered? <karstenw>: my two tails kept trying to "shake hands" in one way or another. i separated them by hand. any idea if that is the right move? 14:33
vmulligan @alcor29 We don't know, unfortunately. 14:34
vmulligan Though that might be related to karstenw's question. 14:34
cymbal_king yes those questions are related 14:34
vmulligan Domain-swapping what that's called.  We see it sometimes in crystal structures.  Sometimes it's an artifact of crystallization, though we think it sometimes happens in real, functional proteins, too. 14:35
cymbal_king it is possible that teh tails wrap around their interacting partner, but it would be much simpler on my end if they didn't 14:35
karstenw lol, ok. 14:35
vmulligan But things in science are rarely as simple as we'd like them to be :P . 14:36
cymbal_king definately 14:36
cymbal_king (it was a joke if nobody realized) 14:36
@inkycatz I sort of want vmulligan's quote on a coffee mug, actually. :D 14:36
vmulligan :) 14:36
@inkycatz Don't be shy with those questions, everyone! 14:37
karstenw we'll get another crack at this one i hope? 14:37
cymbal_king as a small summary of our main interest, phosphorylation on the n-terminus causes degradation of the protein. There is a known binding site on the c-terminus for a ubiquitin ligase. 14:38
@MikeCassidytoo Jenny McCarthy told me science is simple 14:38
@betahelix That's the plan, if CASP will give us some breathing room :-P 14:38
vmulligan (Background: ubiquitin ligase is an enzyme that attaches ubiquitin, a small protein which, when attached to other proteins, serves as a signal that it's time to degrade the other protein.) 14:38
cymbal_king So the phosphorylation site should definitely be accessible to a phosphorylating enzyme 14:38
cymbal_king thanks vmulligan 14:39
@Susume2 so the dead box attaches to the enzyme, not to ubiquitin directly? 14:39
@TimovdL So there should be a score penalty if we bury it 14:39
Skippysk8s so how do I know that is what I am looking at? 14:39
@betahelix Are there any more questions for our guests? :-) 14:39
cymbal_king another option for phosphorylation to enhance degredation is to weaken the affinity to protective proteins (we know of one of them) 14:40
cymbal_king the dead-box binds to the enzyme that attaches a ubiquitin 14:40
Skippysk8s Timo's suggestion is great if possible 14:40
cymbal_king yes, an interesting point 14:40
JacAttack the dead box is where id2 binds to the ubiquitin ligase (APC) but the ubiquitin is attached to the n-terminus 14:41
cymbal_king @skippy, please clarify your question 14:41
vmulligan @TimoVL: not necessarily, unfortunately.  There are proteins that function dynamically by burying a region that interacts with something else, and only exposing it when it's time to interact with that thing. 14:41
Skippysk8s just trying to be sure we know how to identify that structure. I am not a chemist, so I am a little slow. 14:42
@Madde it just gets better and better (sigh) 14:42
cymbal_king A phosphate probably doesn't want to be burried anyways, it is highly charged 14:42
@TimovdL I am always to quick with conclusions -:( 14:42
vmulligan It was a good suggestion.  Things are rarely as simple as we'd like them to be, though :( . 14:42
cymbal_king @skippy, the structure of the dead-box? or the phosphorylated amino acid? 14:42
Skippysk8s the latter 14:43
@Susume2 so the question remains whether a ubiquitin ligase binding site has a typical shape 14:43
Skippysk8s I can live with Timo's suggestion that the score always drops if you try to bury it. 14:44
cymbal_king A cursory search doesn't show a particular shape 14:44
Skippysk8s kk 14:44
@Susume2 thx 14:44
@MikeCassidytoo Are there specific side chains that need to bind 14:44
cymbal_king often when talking about binding between ~7 residues only the primary sequence matters 14:44
cymbal_king 100-107 is the dead-box 14:44
cymbal_king but there could also be other ways the protein is primed for degradation 14:45
@TimovdL But those binding ones should also be neear the surface or not? 14:45
cymbal_king @skippy, the phosphate is in the puzzle near the n-terminal end 14:45
vmulligan Ubiquitin gets added by a covalent bond onto a lysine residue.  So yes, a lysine that gets ubiqutinated needs to be exposed, but then, lysine residues usually are surface-exposed. 14:46
cymbal_king It is more likely than not, that the dead box is near the surface 14:46
@mimi2 I think that after this I need another week to try out different structures 14:46
porkythepundit hi 14:46
@inkycatz Hey porky, any questions for our scientists today? 14:47
@betahelix welcome to the Scientist Chat, Porkythepundit. 14:47
Skippysk8s i would have liked more time on all the structure ones. I felt rushed to narrow things down. I wouldn't mind seeing these puzzles again 14:47
cymbal_king i would be ok with another week if the fold it team is ok with that. I would prefer the right data to quick data 14:47
@betahelix @mimi that was something we were considering... 14:47
porkythepundit thank you, I didn't realise...bit out of the loop atm 14:47
@betahelix but we weren't sure if we should extend this puzzle, or post a new one with some modifications based on the analysis of this puzzle. 14:47
KarenCH I like the second idea 14:48
@TimovdL Maybe a repost where 14 and 100-107 should be on the outside to score well? 14:48
rav3n_pl put best (most promising) structures as templates on another puzzle 14:48
porkythepundit may I ask which puzzle? 14:48
cymbal_king The brain cancer-related puzzle 14:49
KarenCH brain cancer id2 14:49
porkythepundit ty 14:49
@Susume2 sounds like the N terminus should also be at or near the suface 14:49
jeff101 please post a transcript of this meeting on the Foldit site 14:49
karstenw or at least repost with the best scoring segments coming out of the helices at either end would help. it would at least point the tails in a direction. 14:49
cymbal_king THe top hits of the current puzzle had great values for one of the structural assesment programs I ran (based on bond angles/lengths). but the other program based on secondary structure was very unfavorable 14:50
@inkycatz Of course, we are totally going to post a transcript of this soonish when we're wrapped up :) 14:50
rav3n_pl 103-119 can be sheet, my old script shows many structures there 14:50
cymbal_king There were other unfavorable factors in the program as well 14:50
alcor29 Currently I have the dead box outside and most is coil but I have 2 distal segs as helix. Would that be possible? 14:51
porkythepundit I would like to know about another puzzle, is that ok? 14:51
cymbal_king yes that's possible 14:51
alcor29 Tx 14:51
@betahelix (sorry, looks like internet has issues today) 14:52
cymbal_king (that was @alcor) 14:52
@inkycatz Sure, porky, after we wrap up here which should probably soon. 14:52
porkythepundit ty 14:52
cymbal_king does anybody have any burning questions I missed? 14:53
@inkycatz (now's a good time to ask) 14:53
rav3n_pl I`m never sure, in this puzzle we should try to bind both parts together? ie sheet bond between ghost protein and main? 14:53
cymbal_king could you clarify that rav3n? 14:54
rav3n_pl in 903 we have mirror protein 14:54
cymbal_king oh the ghost is the mirror 14:54
rav3n_pl we can do bonds between it and protein we re working on 14:54
cymbal_king Yes bonds between them are possible 14:55
cymbal_king i don't know if they are competely wrapped around eachother or completely separated 14:55
rav3n_pl yes, but desired? 14:55
rav3n_pl ah, ok 14:55
cymbal_king they shouldn't be too intertwinned, because they should be able to come apart easily 14:56
rav3n_pl i remember some pussle, wher s1 get best score when both parts was far away ; 14:56
cymbal_king but they are likely to bind eachother more than just the core 14:56
karstenw lol, we haven't ruled out much. we most likely don't have a bunch of long sheets. otherwise we have free reign. 14:56
cymbal_king lol yeah 14:56
cymbal_king I saw one top hit that was just like a rope of helices just like a myosin chain, probably not that either 14:57
cymbal_king (one outstreched line of helices) 14:57
@betahelix yeah, they should probably be pretty compact 14:57
porkythepundit my guess is when hydrophobes are near the surface interface then close is better and vise versa 14:58
KarenCH you could do crazy horrible mess like my solution - middling hit there :/ 14:58
@TimovdL I made everthing helix except where prolines are 14:58
cymbal_king that's probably a little excesive on the helices 14:59
cymbal_king especially with the sec structure predictions 14:59
rav3n_pl proline... nemesis.... ;] 14:59
@betahelix any last second questions before we end today's Scientist Chat? 14:59
cymbal_king thanks everybody, this has been a blast! 15:00
karstenw thanks for the help. 15:00
@inkycatz Well, it looks like we're about at the end of our chat here - I'd like to thank cymbal_king, JacAttack, our lurking scientists here, and of course all of you for coming today! 15:00
cymbal_king you guys are the ones helping! 15:00
JacAttack Thanks! We appreciate the help on the structure of Id2 :) 15:00
@mimi2 thank you for talking to us 15:00
porkythepundit thank you, looking forward to reading the transcript 15:00
alcor29 Thanks. 15:00
KarenCH thanks! 15:00
jeff101 thanks 15:01
vmulligan Take care! 15:01
rav3n_pl now lets beat this cancer :D 15:01
porkythepundit may I ask that quwstion now? 15:01
JacAttack :) 15:01
vmulligan @porkythepundit: Sure, I'll stick around. 15:01
cymbal_king i should go make dinner, bye 15:01
@inkycatz Thanks, all! 15:01
porkythepundit kk, 897 continued to rise in score for days after close, is that common? 15:02
@inkycatz Look for more great chats coming in the months to come! As always, if there's any burning topics you need to hear about or have a great idea, drop me a private message and I'll see what we can do. 15:02
@inkycatz (No guarantees implied, other than I'll do my best.) 15:03
porkythepundit in other words,"is the puzzle complete?" 15:03
alcor29 how about a dev chat re the darned default prob, inky? 15:03
@betahelix that was a particularly large and nasty CASP target, porky! 15:03
porkythepundit were you satisfied that it was enough? 15:04
@betahelix I bet we could have left it open for another week and the scores would go up even further! 15:04
vmulligan @porkythepundit: You mean the scoreboard on the website showed new, higher scores appearing after the puzzle closed?  I'll ask the other members of the Foldit team -- it sounds odd, but I don't know about that particular puzzle. 15:04
porkythepundit yes, I would like to know if enough scientific outcome was derived 15:05
@betahelix I don't know if it was enough, but it was more of a test to see if Foldit could handle such large CASP targets: 15:05
@betahelix http://fold.it/portal/node/997715#comment-27760 15:05
porkythepundit no, I kept mine going after expiry to test how far it would go 15:05
YoyoParis YES MUCH PROBLEM 15:05
alcor29 beta: do you folks learn much from all the refined achieved by better scores. or is the fold the thing? 15:05
rav3n_pl 257 omg 15:05
jeff101 what do you mean rav? 15:06
YoyoParis hy Rav you come back 15:06
vmulligan Oh, gotcha.  Yes, it's not unusual, given infinite time, to keep refining a structure.  Qualitatively, did you see major shifts in the conformation of the protein? 15:06
@betahelix Rav, it was 201 residues 15:06
alcor29 *refinement 15:06
@betahelix From the feedback we got from that puzzle, we will not post CASP targets that large 15:07
@betahelix @rav because the sequence was trimmed: 15:07
@betahelix http://fold.it/portal/node/997715#comment-27769 15:07
@Susume2 it helped that the other targets that week were small 15:07
@betahelix @alcor29 we'll find out the answer to your question after CASP :-P 15:07
alcor29 Tx, beta 15:08
porkythepundit good, so the answer is still undefined atm 15:08
@betahelix It is important to learn our current size limit, and hopefully in the future we'll come up with ways to solve that issue... but during CASP, we have enough targets to deal with already! 15:08
@betahelix exactly:-) 15:08
rav3n_pl jeff101: 257 residuse will be slooooow, hat this one, always ;] 15:09
rav3n_pl hey yoyo 15:09
rav3n_pl 200 is still big :G 15:09
porkythepundit thanks for staying and attending to this question :) 15:09
@betahelix it sure was! 15:09
alcor29 903 brain is also very very slow but I think it may be the ghost/dimer thing doing it? 15:09
@TimovdL That is also the problem with the science puzzle, even on my big machine it is slow 15:09
rav3n_pl dimer not slowing much 15:09
@Madde at least it was a refinement puzzle, so only one start 15:10
@betahelix Of course! Thank you all for the hard work you are putting in with CASP11... we are trying to spread out the puzzles as much as possible (and limit the number to 5 at the same time, and not closing all on the same day) but the CASP organizers aren't making it easy for us! 15:10
alcor29 rav. There are large gui delays. 15:10
porkythepundit bedtime for me, oink evryone, nice chatting :) 15:10
alcor29 nite 15:11
@betahelix @timovdL that might be because of the additional constraints, right? 15:11
rav3n_pl alcor29: im on it now, not notied. it is not listening to me anyway, cant make staight sheet :P 15:11
@betahelix I have to go as well... gotta send your 902 solutions to the WeFold team since the deadline is coming up! 15:11
@TimovdL That might be the cause, also for the slow reactions to any hand pulls 15:12
alcor29 that's it rav. The constraints. Thanks. 15:12
rav3n_pl 2 days left, or im missing something? 15:12
@inkycatz back to work, beta! 15:12
@inkycatz :D 15:12
@betahelix thanks to inkycatz for organizing such a great chat, and thank you all for coming! 15:12
@betahelix Bye bye everyone... Keep up the great folding! 15:12

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, Amazon, Microsoft, Adobe, RosettaCommons