Developer Chat

betahelix So everyone, I'd like to welcome Zeckendorff, who is the Scientist in the Baker Lab working on Sepsis 14:01
Zeckendorff Hello everyone 14:01
cassandraberry81897 hi :) 14:01
CFC hey Zeck, welcome :) 14:01
BletchleyParkirc hello thank you for being with us 14:01
Zeckendorff Thanks for having me. 14:01
Wbertro Zeckendorff: Hi 14:02
betahelix So we've been looking at your solutions from the previous Sepsis puzzle and we'll start by talking about those... 14:02
CrazyShades hello 14:02
betahelix Here is the top-scoring solution from the Hand-folding round: 14:02
betahelix 14:02
Zeckendorff So, if you take a look at this solution... 14:03
Zeckendorff overall, we are happy with this, but it could be improved 14:03
Zeckendorff As you'll notice, this solution does not incorporate any new H-bonds 14:03
Zeckendorff And the additional residues added form a non-structured loop 14:04
eromana m 14:04
Zeckendorff We think this solution has potential, but what we'd like to see is something more like this...... 14:04
betahelix 14:04
Zeckendorff And just so we're all on the same page, let me briefly describe the goal of this project. 14:05
Zeckendorff These sugar binding puzzles are part of a DARPA funded project, the aim of which is to make blood filtration devices 14:05
DM996 DARPA? 14:05
Zeckendorff We in the Baker lab are attempting to design proteins that are able to bind to molecules found specifically on the surface of pathogens. 14:05
betahelix Defense Advanced Research Projects Agency (DARPA) 14:05
BletchleyParkirc @DM, wikipedia that 14:05
DM996 thank you 14:05
Zeckendorff That will be our contribution to the DARPA project. The other teams involved in the project are focused on the hardware, the actual filtration devices. 14:07
Zeckendorff So, we have taken a look at several proteins that can bind sugars. 14:07
Zeckendorff Sugars that are commonly found on the surface of many pathogens, including bacteria and viruses. 14:07
Zeckendorff The first puzzle we set up a few months ago did not turn out so well, and that was primarily our fault because the model didn't give the players much to work with 14:07
Zeckendorff So, we have now identified several other candidate proteins that are known to bind sugars, and we are hoping to improve upon them. 14:09
betahelix This is why you have more slack on these recent puzzles :-) 14:09
Zeckendorff The current puzzle is a sugar binding protein from the snowdrop plant Galvanis nivalis (spelling?) 14:09
Zeckendorff And we feel this one offers a much greater chance of success than our previous puzzle 14:10
betahelix Are there any questions about the general purpose of these Sepsis puzzles? 14:10
BletchleyParkirc its Galanthus nivalis 14:10
Zeckendorff Right, thanks :-) 14:11
betahelix thanks BP ;-) 14:11
cassandraberry81897 that's a mouthful 14:11
BletchleyParkirc yw 14:11
BletchleyParkirc this is to become a generic 'cleaner'for biological agents ? 14:12
Zeckendorff Yes, that's mostly correct. We would like our designs to be as broad-spectrum as possible, though in practice we won't be able to design a protein that will recognize all pathogens. 14:13
BletchleyParkirc would fewer residues in the capture rea be better (more robust) ? 14:14
BletchleyParkirc area 14:14
Zeckendorff We believe that if have a strong binder for mannose and related sugars, we should be able to capture many of the pathogens responsible for blood infections. 14:14
Zeckendorff Thanks for that question BP 14:14
BletchleyParkirc yw 14:14
Zeckendorff If you take a look at the second solution we just uploaded 14:14
Zeckendorff ... 14:14
betahelix 14:15
Zeckendorff What we like about that one is that the solution incorporates a stable secondary structural element.... 14:15
Zeckendorff and, perhaps more important, two new hydrogen bonds to the sugar are introduced by the solution 14:15
Zeckendorff These are the kinds of things we're looking for, stable secondary structural elements (alpha helices are probably most appropriate here), and hydrogen bonding with the sugar. 14:15
Zeckendorff So, whether than means more or less residues is not entirely set, but the lowest number of residues that can incorporate those two ideas is probably an optimal solution 14:17
BletchleyParkirc The sugar is portrayed in that specific position, should we anticipate the sugar to be 'floating'and aim at capturing it ot keeping it in its current place ? 14:17
Zeckendorff Good question... 14:17
Zeckendorff The location of the sugar in the structure we used for the puzzle is where the protein would bind that sugar in the natural context. 14:19
Zeckendorff So yes, the sugar should be kept in the location depicted in the puzzle. Does that answer your question? 14:20
BletchleyParkirc yes it does, thank you. Are there any particular sidechains that are more suitable to bind to sugars than others ? 14:20
Wbertro so the sugar on these proteins are sticking out? 14:20
Wbertro to be able to go around that #23 solution helix 14:21
Zeckendorff For making interactions with the sugar, the polar and charged side chains (Arginine, Lysine, Glutamine, Glutamate, Aspartate, Asparagine, Serine, Threonine, Histidine) are most appropriate. 14:21
Zeckendorff Those are the residues capable of making hydrogen bonds. 14:22
Zeckendorff Hello Wbertro. The sugar will be on the surface of a pathogen, most likely a bacterium. The protein will bind to that sugar on the pathogen. 14:23
Wbertro so bonds with the backbone are not preferred 14:23
Zeckendorff Bonds with the backbone are ok, but we would prefer bonds with side chains because that forces some rigidity on the protein. 14:25
Zeckendorff You can see the types of preferred interactions highlighted in the second solution we posted. 14:26
BletchleyParkirc what is the intended end effect this binding will have on the bacterium ? 14:26
Wbertro so a puzzle with scoring favoring sidechains instead of backbone maybe 14:26
BletchleyParkirc (kill it, make it heavy, 'mark' it) 14:26
Zeckendorff that solution you'll also notice the purple residue, a methonine. That methionine could most likely be changed to a polar residue. 14:27
Zeckendorff BP... the end goal is to tightly bind to the pathogen, via its sugars, with the end goal of removing that pathogen from the blood stream. 14:27
TheGUmmer or delivering an agant into the pathogen? 14:28
betahelix @bertro are you asking if we should set it up in the puzzle to upweight sidechain h-bonds? 14:28
TheGUmmer agent 14:28
mimi2 the problem as ever is that the solutions that you want are not the ones that score most 14:28
BletchleyParkirc filtration, you mentioned that earlier, thank you 14:28
Zeckendorff TheGUmmer. Thanks for that question. That is something we have thought about, but for this particular project we simply focusing on designing strong binders to sugar molecules. 14:29
Wbertro a throwaway fly trap for sugar 14:29
betahelix @mimi exactly... we will bring this up at the end of the Sepsis discussion :-) 14:29
Wbertro betahelix: yes 14:29
Zeckendorff Wberto, that 14:29
Zeckendorff that's a great way to put it, a flytrap 14:29
betahelix @bertro what we did for these recent Sepsis puzzles was upweight the sugar by a factor of 10 14:30
Zeckendorff The idea for this sugar binding protein is that it will be specific to sugars on the surfaces of bacteria (and viruses), but will not recognize sugars on the surfaces of human cells. 14:30
betahelix but maybe downweighting the backbone h-bonds might help (or upweighting the sidechains) 14:30
Wbertro betahelix: that helps a lot 14:30
Susume2 assuming there are some H bonds to the sugar, are additional H bonds to the frozen portion of the protein useful for rigidity? 14:32
Zeckendorff Hmmm, that's interesting, and not really something we have considered here. 14:32
Zeckendorff I would say... 14:32
TheGUmmer and can you weight rigidity? 14:33
Zeckendorff it's possible to have some stabilizing additional H-bonds, but it's probably unlikely for this particular protein 14:33
Zeckendorff For future puzzles it may be beneficial. 14:33
betahelix please do note that Valine 32 can be mutated to other amino acids... we noticed in the last round of puzzles that not many solutions had changed this residue 14:35
Zeckendorff Yes, you can also take a look at V32. But, perhaps a valine at that position is optimal. 14:36
BletchleyParkirc It usually does not change when using mass mutate 14:36
BletchleyParkirc regardless of the other things around 14:36
CFC flytrap with a lid maybe? 14:36
Zeckendorff OK, good. That's not surprising. We just thought we would make that residue flexible, in case something is better there. 14:37
betahelix but it is hard to beat mother nature ;-) 14:37
BletchleyParkirc glutamate at 27 might be a nice candidate for that 14:37
Zeckendorff We looked at that residue, and thought it could be interesting. But because of the way the loop is pointing away from the sugar at that location in the protein, we ruled that out. We felt whatever was placed at position 27 would be facing away from the sugar. 14:41
betahelix are there any last questions for Zeckendorff? 14:41
BletchleyParkirc as it is a combined project, is there a fixed deadline to when the last modification is due ? 14:42
BletchleyParkirc (how many rounds can we expect ?) 14:42
Zeckendorff We had a meeting with the DARPA people last week, and they would like to see designs tested soon :-) 14:43
Zeckendorff But.... 14:43
Zeckendorff We will likely be posting a few rounds of sugar binding puzzles 14:43
BletchleyParkirc could we use more tools than just the gui scripts ? 14:44
betahelix We will of course post Round2 after the Hand-Folding round is over :-) 14:45
BletchleyParkirc fortunately, thank you. It takes a lot of time to do the bulk work 14:45
BletchleyParkirc and jobs require attention too ;-) 14:45
betahelix We'll always post a round2 after any hand-folding puzzle! 14:45
BletchleyParkirc keep up the good habit ! 14:46
betahelix Ok, thank you Zeckendorff for joining us today 14:46
TheGUmmer maybe make the handfolding shorter 14:46
eromana is my relatively high scoring 697 all glycene solution valid 14:46
marie_s the round 2 of the first handing give result? 14:46
Zeckendorff I would very much like to thank everyone for participating in this puzzle. Sepsis is an important unmet medical issue. If we are able to make contributions in this area, it could have a profound positive impact on patients suffering from blood-borne infections. 14:46
Wbertro Zeckendorff: thanks you 14:47
TheGUmmer thank you 14:47
marie_s thanks 14:47
BletchleyParkirc thank you, appreciated 14:47
Zeckendorff Eromana. We will certainly be investigating your solution further. Thanks everyone. 14:47
betahelix @TheGUmmer the only reason we make the hand-folding long is because it takes some time to do everything by hand! 14:47
betahelix I want to address mimi's comment from a few minutes ago: 14:48
betahelix "the problem as ever is that the solutions that you want are not the ones that score most" 14:48
Wbertro yes, some days cannot do hand work, some days yes 14:48
Wbertro exactly 14:48
Hanto no doubt DARPA involved for military/D.O.D. reasons. interesting to know that. 14:49
betahelix As David Baker mentioned in his blog post about experimentally testing 3 designs from Puzzle 691, it's true that we will not pick the three top-scoring solutions! 14:49
betahelix We are trying to resolve this with tamirh's latest blog entry: 14:50
betahelix 14:50
auntdeen2 then… to keep us all interested if not scoring well… please post what you are using as soon as you decide :-) 14:50
auntdeen2 otherwise it feels lie folding in a black hole ;-) 14:50
auntdeen2 like even 14:51
betahelix Of course! We'll post the actual pdb files that we pick so that you can see/manipulate them 14:51
auntdeen2 good 14:52
auntdeen2 when will mutate / rebuild be added to the gui recipe? 14:52
betahelix If this works well, we can use this system for other design puzzles (like the Sepsis ones) as well! 14:52
betahelix tamirh, can you answer that one? 14:53
tamirh auntdeen2, i'm still working on some features for design puzzles, we will probably push out a new devprev once those are done 14:53
auntdeen2 thanks, tamirh 14:53
Wbertro :d 14:53
Wbertro :D 14:53
tamirh the GUI script changes (along with all the other bug fixes + features) should be in that devprev push 14:53
betahelix On a similar note (and we'll end our Scientist Chat on this topic) I want to introduce another Baker Lab Scientist who is going to ask you for some of your solutions :-) 14:53
auntdeen2 will the gui scripts eventually be able to run while minimized? 14:53
Wbertro auntdeen2: good one 14:53
BletchleyParkirc @tamir, can you please also fix the popup text bug I reported ? It gets in the way with design puzzles 14:53
betahelix I'd like to welcome, bk, who is analyzing the CASP10 recipes that you all used the most 14:57
tamirh auntdeen2, if you're on osx or linux they should work without any changes. on windows there's an options.txt flag to set to have them run while the client is minimized 14:57
bk Hi everyone 14:57
BletchleyParkirc hello bk 14:57
betahelix 14:57
tamirh BletchleyParkirc, PM me the link to the feedback for that. i'll look into it 14:57
Wbertro tamirh: what is it? 14:57
BletchleyParkirc will do, thanks 14:57
tamirh Wbertro, one second. i'll look it up 14:57
bk so, one of the things that flounders us programming people about the recipes you have all made 14:58
bk is that they are context dependent 14:58
bk basically, that you have made different recipes to use in different situations 14:58
tamirh Wbertro, "update_while_minimized 14:58
Wbertro tamirh: thanks 14:58
bk what we'd like to understand is which situations are best for which recipes 14:59
auntdeen2 thanks, tamirh 14:59
auntdeen2 and hi bk :-) 14:59
Wbertro hi bk 14:59
auntdeen2 bk - some of that is intuition 14:59
auntdeen2 and some depends on how stiff a protein is when presented 14:59
auntdeen2 some of us use loose "routines" - but they change depending on what the puzzle responds to 15:00
BletchleyParkirc The more green, the fewer rebuilds 15:01
bk @auntdeen exactly 15:01
bk often times, you guys see things that the computer isn't so easily able to pick up on 15:01
auntdeen2 in the spring - beta posted some "old" puzzles that didn't respond to any techniques I use 15:02
bk so we've looked at a few of the most popular recipes you've created 15:02
auntdeen2 so I started doing a rav "Push" first to open them up, with sucess 15:02
auntdeen2 not something I would do with any other puzzle I've worked on 15:02
betahelix speaking of Rav... 15:02
bk and we'd like you to submit "saves" of structures that are much improved by a particular recipe 15:03
bk 15:03
betahelix we thought we'd start with just one recipe for starters (since it was highly used during CASP10) 15:03
auntdeen2 beta - that shows no parent which is not true 15:04
betahelix Rav3n_pl GAB v0.6 loss 15:04
auntdeen2 there were many parents to that script 15:04
Hanto and many descendants 15:04
auntdeen2 crashguard, I believe, did the original gab 15:04
auntdeen2 and rav must have had at least 6 prior versions 15:04
betahelix tamirh, I believe it only finds a parent if the previous recipe was modified... not if someone started a new one in the lua editor, right? 15:04
marie_s w have to edited in external editor recipelike this 15:04
BletchleyParkirc correct 15:04
bk you can follow tamirh's instructions here ( to submit your saves, just for the name enter "recipe_GAB_1" 15:04
bk or "recipe_GAB_2", and so-on 15:04
auntdeen2 :-) but… I still use many of the prior gabs 15:04
BletchleyParkirc GAB-II V1.8 Crashguard303 132 378 08/18/10 None 15:04
betahelix we just picked that GAB because it was the one used the most by so many different players during CASP10... this is not claiming that Rav invented GAB or anything! 15:04
auntdeen2 depends on the puzzle :-) 15:04
auntdeen2 we can try, bk 15:07
Susume2 do you just want the save after the script, or also the one before it? 15:07
bk that would be very helpful, thank you 15:07
betahelix In fact, this was Rav's reply to me about that particular recipe: "GAB - Genetic Algorithm on Bands. Idea got from Cartoon Villan GA script and CG303 one" 15:07
bk we would like the save just BEFORE running the script 15:08
betahelix that way we can compare how the recipe does compared to running Rosetta on those exact same starting Foldit saves 15:09
Susume2 you know if you run the script against that save you will not get the same results, because of the random selection of residues and bands? 15:09
auntdeen2 :-) true Sus 15:09
betahelix that's what supercomputers are for ;-) 15:09
Susume2 :-D 15:09
Hanto and that is why I use so many clients, variation 15:09
auntdeen2 I want one of those :-P 15:09
betahelix and we plan to do this for many different recipes (this is just an initial test, and we didn't want to pick Blue Fuse again :-)) 15:09
betahelix exactly ;-) 15:09
bk @suseme yes, but it's still extremely useful for us to see what cases are best served by these recipes 15:09
markm457IRC if the GAB script was modified to print the "random seed" and that was submitted too, it could be repeated with same seed 15:11
betahelix great point! 15:11
betahelix We'll post a news item about this tomorrow as well, but we wanted to give you a heads about it today. 15:11
Susume2 is there a way we can indicate to you what options we selected in the script (Rav always provides many editable options) 15:11
Wbertro maybe attach the script in a segment note 15:13
bk these are great points 15:13
bk we're still working out how to best get saves and options from you guys 15:14
Hanto sounds to me like you're trying to improve rosetta with foldit ideas/tools 15:14
TimovdL Just getting home, and scanning the conversation, do you want more info how we use which scripts? 15:14
betahelix @Hanto that has always been one of the goals :-) 15:15
bk we'll post more details in the blog soon about how to submit saves 15:15
betahelix well as any more info we'd need (we might have to resort to PMs for that, or the segment note idea) 15:15
ManVsYard options selected could be sent to output at the beginning to the script along with the name of the script 15:15
betahelix Ok everyone, if there aren't any last minute questions... we'll wrap up today's Scientist Chat 15:15
Wbertro make a short module we can insert into recipe for info you would like 15:15
betahelix Please make sure to check out today's Blogpost, and we'll post the recipe one tomorrow. Until then, send us your coolest 691 designs and we'll test at least 3 of them! 15:18
betahelix 15:18
katfish Thanks everyone for coming out today! 15:18
Wbertro great chat! 15:18
bk Thanks, everyone! 15:18

Generated by 2.12.1 by Marius Gedminas
- find it at!

Download links:
  Windows    OSX    Linux  
(10.12 or later)

Are you new to Foldit? Click here.

Are you a student? Click here.

Are you an educator? Click here.
Social Media

Other Games: Mozak
Only search
Recommend Foldit
User login
Top New Users

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