1 reply [Last post]
Joined: 11/05/2019
Groups: Foldit Staff

Hi Foldit players! Here are the Office Hours transcripts: Here is the most recent one. Please scroll down to see past Office Hours transcripts.

Office hours 6/26- led by Beta_helix and SethCooper

6:00 PM] beta_helix:
Hello everyone!
[6:00 PM] :
donuts554 hello how are you?
[6:00 PM] :
SethCooper hello!
[6:00 PM] beta_helix:
Seth Cooper and i will start off today's Office Hour
[6:00 PM] :
jeff101 hi
[6:00 PM] :
Formula350 Wait it's just Beta Helix? I thought it was Gamma Helix who'd be here. :( Welp, I'm leaving then......... lol
[6:00 PM] beta_helix:
Doing good, donuts554! Hi Jeff101!
[6:01 PM] beta_helix:
hahahhahahahhaa
[6:01 PM] :
donuts554 how are you seth?
[6:01 PM] :
Formula350 :P
[6:01 PM] :
SethCooper doing pretty well
[6:01 PM] :
Formula350 First thing's first.... Chocolate or Vanilla?
[6:02 PM] :
Formula350 And then: Who would win in a fight: Beta or Seth? :}
[6:02 PM] beta_helix:
Gamma Helix would beat h of us!
[6:02 PM] :
SethCooper probably a tie
[6:02 PM] :
Formula350 Alright I've gotten all my shenanigans out of the way ;)
[6:03 PM] :
donuts554 Chocolate and idk
[6:03 PM] alcor29:
Question: Currently Foldit Rosetta?) does Wiggle, Wiggle sidechains, and Shake sidechains as discrete operations. Has anyone ever looked at coding that would do all three as a single fluid motion, as one might imagine happens in a real environment? Would that be desirable? If so, even doable?
[6:03 PM] :
Formula350 Donuts we're asking THEM questions, not each other lol
[6:03 PM] :
donuts554 oh ok
[6:04 PM] :
Formula350 Wiggle does "Wiggle Backbone" and "Wiggle Sidechains" at the same time. Just to point that out.
[6:04 PM] alcor29:
k
[6:04 PM] beta_helix:
It's all good
[6:04 PM] :
SethCooper yeah, I think early on we had considered having an operation that does h shake and wiggle in one operation
[6:04 PM] beta_helix:
That is a very good question, alcor29
[6:05 PM] :
SethCooper but I think even at that level, it would have to alternate wiggle and then shake somehow
[6:05 PM] beta_helix:
One thing that the Rosetta community noticed, and I know you have as well, is that there is often a big difference between Wiggle-Shake or Shake-Wiggle
[6:06 PM] beta_helix:
That is why Blue Fuse (and Relax in Rosetta) do these alternations with different thresholds (of clash importance, for example)
[6:07 PM] beta_helix:
I agree that it would be a lot more realistic to have one fluid action... like in Nature. That is what molecular dynamics simulations try to do.
[6:08 PM] alcor29:
Thx.
[6:08 PM] :
donuts554 My Question is part of something else, but is it necessary to render aromatic compounds without the see-through holes?
[6:08 PM] :
donuts554 In Foldit
[6:08 PM] :
donuts554 I mean amino acids not compounds
[6:09 PM] beta_helix:

[6:09 PM] beta_helix:
So, here is a TRP in Foldit, right?
[6:10 PM] :
donuts554 Yes
[6:10 PM] :
donuts554 I can't see the image though
[6:10 PM] beta_helix:
https://foldit.fandom.com/wiki/Amino_Acids
Foldit Wiki
Amino Acids
Amino acids (also called segments or residues in Foldit) are the building blocks of proteins. Each amino acid has a unique sidechain, except for glycine. Each protein is identified by a unique...

[6:11 PM] :
Formula350 I looked at it, it's simply the Tryptophan as you know and love (or I guess in your case, hate lol)
[6:11 PM] beta_helix:
(sorry, I'm in Discord
[6:12 PM] :
donuts554 ok then yes it is a TRP in FOldit
[6:12 PM] beta_helix:
You can render it without the hole with the Sphere view, right?
[6:13 PM] :
donuts554 Yes I can
[6:13 PM] beta_helix:
Are you asking if we should do the same in stick mode?
[6:13 PM] :
donuts554 But it its more puffy and takes more space and makes the surface covered
[6:13 PM] :
donuts554 Yes I am asking that
[6:14 PM] beta_helix:
If you look at all these amino acids from a cartoon view, such as TRP:
[6:14 PM] beta_helix:
https://en.wikipedia.org/wiki/Tryptophan
Tryptophan
Tryptophan (symbol Trp or W) is an α-amino acid that is used in the biosynthesis of proteins. Tryptophan contains an α-amino group, an α-carboxylic acid group, and a side chain indole, making it a non-polar aromatic amino acid. It is essential in humans, meaning the body can...

[6:14 PM] beta_helix:
Then we are essentially just drawing lines for each bond... which in this case creates these 2 rings.
[6:15 PM] beta_helix:
As you mentioned, if we didn't draw it that way it would make the surface covered and takes more space.
[6:15 PM] beta_helix:
Our Cartoon and Cartoon Thin are designed to give you enough information about the sidechain, without so much detail that you can't see what is going on.
[6:17 PM] :
donuts554 Oh ok, thanks for the answer! I have another question but i'll let others ask their own
[6:17 PM] :
Formula350 If your followup is directly related, it might be good to ask it.
[6:17 PM] :
donuts554 Oh ok then I'll ask it
[6:17 PM] beta_helix:
There are 2 of us, so we can handle multiple questions
[6:19 PM] :
jeff101 one question folks were asking the other day is if Foldit accounts for pi stacking, and if so, which energy subscores reflect it best?
[6:22 PM] :
donuts554 I think thats kinda related to my question somewhat
[6:22 PM] beta_helix:
I know Foldit does for sure... but I don't know which subscores show that.
[6:23 PM] :
jeff101 sorry donuts
[6:24 PM] beta_helix:
I will find that out and post it in the Forum for you (unless that question was already posted somewhere)
[6:25 PM] :
donuts554 Its ok, I meant that I think it was related to my previous question about aromatic compounds without the holes
[6:26 PM] :
donuts554 The next question im going to ask is going to be a bit long because I have to describe things
[6:26 PM] :
Formula350 We need "Arcade Mode"... With big text popups and an announcer. Displaying like: GREAT PI STACKING!! or an annoucer proclaiming "You just flipped Covid the Zinc Finger!"
[6:26 PM] beta_helix:
Type away, we'll answer any other questions
[6:26 PM] beta_helix:
It's funny that you mention that, Formula350, because the early versions of Foldit had that!
[6:27 PM] :
Formula350 Donuts, just emember that Foldit can only send messages that are so-long before it gets cut off.
[6:27 PM] beta_helix:
Seth, am I right in remembering that most players turned that off?
[6:27 PM] :
SethCooper i think the majority of options are left on the default
[6:27 PM] beta_helix:
I mean, it didn't say "Finish him!" (this Protein!)
[6:28 PM] :
jeff101 what things are each of you working on these days?
[6:28 PM] :
Formula350 haha I mean, I have sounds disabled, but I wouldn't mind visual stuff, as I think it being pointed out when you create a.... "motif" (?? I guess that's what it'd be considered as), would actually be a learning experience for some of us.
[6:29 PM] beta_helix:
donuts, feel free to break your posts up so they don't get cutoff!
[6:29 PM] beta_helix:
Seth, you go first (your stuff is cooler ;-))
[6:29 PM] Susume:
I would love it if foldit could recognize a motif - it mostly crows about smaller things like H bonds
[6:30 PM] beta_helix:
ohhhhhh, so you mean USEFUL popups?
[6:31 PM] :
Formula350 :P Yea, like Rank Up. However, adding some comedic flair would probably make it more appealing to the younger (or more childish, in my case) crowd since it'd provide some "action".
[6:32 PM] alcor29:
A little Wagner leitmotif?
[6:32 PM] Susume:
dystopian foldit popups
[6:32 PM] beta_helix:
@jeff101 Scott Horowitz and I just submitted an NIH grant yesterday to make Foldit Electron Density state of the art (ie incorporate the tools used in the best ED programs out there) fingers crossed it gets funded

1
[6:32 PM] :
donuts554 Is it necessary to have a new "View Protein" option with aromatic compounds without the holes, the proline without its hole, the thick helices as cylinders, the sheets without the zigzag contours --
[6:33 PM] Susume:
drools for upgraded ED tools
[6:33 PM] beta_helix:
Right?!?!?!
[6:33 PM] :
jeff101 I've asked questions like this in previous Office Hours. Not sure if I asked you guys.
[6:34 PM] :
jeff101 the papers I've seen don't generally show where the h-bonds go
[6:34 PM] beta_helix:
I just want to say that I've been trying to get ED funding since I left Seattle in 2013... but it is hard!
[6:35 PM] :
jeff101 I hope you get the funding.
[6:35 PM] :
SethCooper one thing we have been working on is something like the tutorials but more dynamic
[6:35 PM] beta_helix:
Me too! I'm submitting another grant to the NSF next month as well. (edited)
[6:35 PM] Susume:
do you feel like there is a "market" for foldit to work on ED? do scientists maybe feel like they're doing just fine without us?
[6:36 PM] :
spvincent Get the word Covid in the funding application somewhere and hopefully people will throw money at you.
[6:36 PM] beta_helix:
@spvincent don't worry... it's in there!
[6:36 PM] :
SethCooper (in this case "we" means mostly Josh)
[6:36 PM] :
spvincent good!
[6:37 PM] beta_helix:
@Susume it's a very good question... but I honestly think that the issue has been how unorthodox Foldit is (especially when asking for grant money)
[6:37 PM] :
Formula350 Call us a Think Tank and it might go over a bit better?
[6:38 PM] :
donuts554 --to eliminate all convex angles in sidechains, so a lysine pointing straightout is not a zigzag, instead its a striaght line, a valine looking like a orange triangle, and a arginine looking like a blue triangle on top of a blue pole with blue ends --
[6:38 PM] beta_helix:
You can imagine that when a panel has 100 applications and only enough money to fund 10... it's not hard to believe that they will go with (what they feel is) the "safer" bet.
[6:39 PM] beta_helix:
@Formula350 You know, you are not far off... because a lot of grant reviewers reject the notion of "gamers"
[6:40 PM] beta_helix:
I talked to a Program Manager at the NSF, and he agreed that if I only refer to Foldit players as "citizen scientists", it would be received a lot better! sigh
[6:40 PM] :
donuts554 --and to have any H-bonds or disulfide formed between amino acids more solid and widened so that it looks like the amino acids are actually stuck together?
[6:41 PM] beta_helix:
(Anyway, I can ramble on about this for ever... so I'll stop and answer donuts)
[6:41 PM] beta_helix:
@donuts disulfides are a very good example of what you are talking about
[6:41 PM] beta_helix:
Because those are some very strong bonds!
[6:42 PM] beta_helix:
Seth, jump in here, but I think one of the reasons we didn't go with the h-bond glue (for example) is that we do want you to try different topologies.
[6:42 PM] :
donuts554 Not glue
[6:42 PM] Susume:
given how long some of us have been playing, we've almomst become a "curated collection" of citizen scientists (kinda like being pickled I guess)
[6:43 PM] :
donuts554 I mean like in this new view protein option, when you pull the two amino acids apart, the solid rod connecting the two charged atoms disappears
[6:43 PM] beta_helix:
If you have a protein with a lot of molecules stuck to one another, wouldn't you be less likely to break those?
[6:43 PM] :
donuts554 Like a normal H-Bond would in cartoon
[6:43 PM] :
donuts554 No It just looks like its stuck together
[6:43 PM] beta_helix:
but that rod would look like it was almost as strong as a carbon-carbon bond, right?
[6:43 PM] :
SethCooper well, I think bands can be used to hold h-bonds together if wanted, if that's what you mean
[6:44 PM] :
donuts554 No not hold H-bonds together, I mean, the same thing as now --
[6:44 PM] :
donuts554 But
[6:45 PM] :
Formula350 It's all about "spinning" it. Electrons spin, PR spin, as does Grants for funding. In reality, you're not lying by referring to us as Citizen Scientists, or Distributed Brainpower, or Community Think Tank. In the realm of science I can't ever see "Gamer" as being taken legitimately or seriously. Some of us (Susume) have literally been so inspired by Foldit, that they've shifted their life to pursue it. So I think calling Susume a "gamer" is arguably
[6:45 PM] beta_helix:
@Susume I like that term... I should use it in a grant be it would go way over their heads! Uh oh... do any grant reviewers hang out in veteran?
[6:45 PM] :
donuts554 The solidness used to render the bonds holding the amino acids together is used for the H-bond and disulfide bonds as well
[6:45 PM] :
donuts554 Basically so all bonds look the same
[6:46 PM] :
donuts554 only with different color of course
[6:46 PM] :
donuts554 and so that only h-bonds and disulfide bonds can be broken
[6:46 PM] :
donuts554 thats what I mean
[6:46 PM] beta_helix:
Donuts, why would you want them to look the same? The reason we color some orange vs blue (as you just pointed out) is to highlight their difference.
[6:47 PM] :
donuts554 Yes the color stays different, but the texture looks the same
[6:47 PM] :
SethCooper the hbonds and disulfide bonds are rendered a bit differently though
[6:47 PM] beta_helix:
The easier it is for players to differentiate between them, the easier it is to manipulate the structure, isn't it? If everything had the same texture (or say was all loop, instead of helix/sheet/loop) then it would be very hard to visualize
[6:48 PM] :
Formula350 Donuts if you would like the Disulfide Bond 'texture', I can make one for you to use instead.
[6:48 PM] :
donuts554 Yes I mean without the different rendering, same rendering for all bonds, just different colors
[6:48 PM] Susume:
I think donuts is saying he wants H bonds and disulfides to be drawn like covalent bonds are currently drawn, as solid cylinders - is that right?
[6:48 PM] :
donuts554 Like bright cyan solid bond holding as H-bond, bright yellow for disulfide
[6:48 PM] :
donuts554 Yes! That's what I mean formula
[6:49 PM] :
donuts554 They are bonds, and they should be treated fairly
[6:49 PM] :
donuts554 It follows the "bond" definition
[6:49 PM] beta_helix:
But they are different bonds...
[6:49 PM] :
donuts554 It's just their strengths are different
[6:49 PM] :
Formula350 I think you mean Susume? He wants everything to be "solid". Aromatics, Helices, etc. Not this "there's a gap, so something could arguably pass-through it"
[6:49 PM] :
donuts554 Like you have chemical reactions breaking bonds and making bonds, like in 1855
[6:50 PM] :
SethCooper we figured they were different enough they should look different
[6:50 PM] :
SethCooper plus, we are often encouraging the creation of hbonds so they are like a "goal" in the game
[6:51 PM] :
spvincent Can I ask a question about filter scoring?
[6:51 PM] beta_helix:
I completely agree that we are simplifying the reality of proteins... but that is done intentionally, for the exact same reason that you mentioned earlier: Sphere mode (while being more realistic) is almost impossible to fold in.
[6:51 PM] :
donuts554 This will make it look like there is actual chemistry, breaking and forming bonds
[6:52 PM] :
donuts554 Oh ok seth nvm about the bonds part
[6:52 PM] beta_helix:
Take electrostatics, for example, donuts... or even water, we don't even show those at all! So this really is a cartoon version of chemistry, as opposed to making it look 100% accurate. I hope that answers your question.
[6:52 PM] beta_helix:
@spvincent of course!
[6:54 PM] :
spvincent I was wondering why the bonus/penalty for filters such as Buns, Core Filter, etc follows a stepwise function as opposed to being continuous. It play havoc with scripts.
[6:55 PM] :
spvincent *plays
[6:55 PM] :
SethCooper I'm not sure about those specifically, but some bonuses are based on discrete things in the structure
[6:55 PM] :
donuts554 Well the intent is not for 100% accuracy, but also for less complicated things for the eye to remember and visualize in the brain memory, and to have less rendering so that it is less laggy and therefore less crashes
[6:57 PM] :
spvincent In the real world of proteins I'd expect smooth variation. A polar atom isn't either buried or unburied: it can't be that binary.
[6:57 PM] :
Formula350 Yea, like the uh... "IE" filter. I think it's called.
[6:57 PM] :
Formula350 Interaction Energy?
[6:58 PM] :
SethCooper for example, if a bonus is based on, say, a number of bonds or a number of residues, then it's more straightforward to score based on that
[6:58 PM] :
SethCooper there are some techniques use to smooth them out in some cases
[6:58 PM] beta_helix:
@donuts that is a very good point that you bring up, and something that we have struggled with for a long time... especially as we try to post puzzles with bigger and bigger proteins. We actually considered going the opposite direction of your suggestion: removing most sidechain atoms, which is when we tried out Centroid Mode (I know the mention of it just made a few players sick) needless to say, it did not go well.
[6:59 PM] :
Formula350 But I presume that it's based on actual Rosetta scoring, versus most Filters which are ad-hoc and side-calculations that would require more processing power (time) to provide "real-time" output. Is this a fair assessment, Seth?
[6:59 PM] :
Formula350 ("that it's" referring to something like the Interaction Energy filter)
[7:00 PM] :
SethCooper yeah, the filters I think can often be simplifications based on counts or thesholds
[7:00 PM] :
donuts554 Like the convex angles in valine and leucine are intuitvely harder to remember than as rendered as a short, filled orange triangle and a longer, filled orange triangle (cause the vertex point in that position in 3D is located in the inside of the atom according to sphere mode)
[7:01 PM] :
SethCooper the early filters/bonsuses were entirely binary, as I recall
[7:01 PM] :
SethCooper so adding in the steps in scoring does kind of smooth them out a bit from that
[7:02 PM] :
spvincent Its quite frustrating working with proteins when you're on a boundary and e.g. the core bonus keeps flipping from 2900 to 3000
[7:02 PM] :
spvincent and back again
[7:03 PM] :
Formula350 Yea. Not having a nice Subscore to gauge what we're doing on, makes thoses, and BUNS, etc, very hard to determine whether what we're doing is "good" or "bad".
[7:03 PM] beta_helix:
donuts, you might get a kick out of this (or be mortified) but some of the very early Foldit mockups didn't use actual sidechains at all... they were random shapes (or different fruit) or other cartoony representations!
[7:04 PM] :
jeff101 I would request the Ideal Loops Filter scores by residues in ideal loops rather than by the # of ideal loops
[7:04 PM] :
SethCooper it's quite possible some that are based on thresholds could be smoothed out a bit as well
[7:05 PM] :
spvincent that would be most helpful i think
[7:05 PM] :
Formula350 A "glow" factor, or "color transition gradiant" (for BUNS/Ideal Loops, and Core Existence, respectively) would help a lot, if that's doable as well.
[7:05 PM] :
Formula350 Acting much in the same way that the HBond line gets thinner as it grows weaker, or fatter as it gets really strong.
[7:05 PM] Susume:
(just try to get funding for something that represents sidechains as bananas!)

2
[7:05 PM] :
jeff101 being able to color by specific subscores rather than the total score would be helpful
[7:06 PM] :
donuts554 Also, showing the glutamine, arginine, asparagine as triangles on top of poles will make it easier to recognize the pattern of the shape of the amino acids, and to make it easier for espcially even children to recognize and categorize and theorize the properties of the amino acids and compare them, intuitively.
[7:06 PM] :
spvincent That reminds me: I do not like the way the Buns filter messes with the View Settings.
[7:06 PM] :
Formula350 Yea, they're working on that VIncent :)
[7:06 PM] alcor29:
Speaking of ideal loops. Is there some way of knowing whether mutating a single residue would make it ideal, instead of the sometime laborious process of trying to find a better shape?
[7:06 PM] :
spvincent oh good
[7:08 PM] :
donuts554 (ex. Histidine would look like a blue filled pentagon on top of a blue filled triangle on top of a pole., and Asparagine and glutimine would look like blue triangles on top of poles. I think, and even I think myself too, that they make think, --
[7:08 PM] beta_helix:
@donuts554 The team is working on a more specific Education version of Foldit that would be aimed more towards schoolchildren.
[7:09 PM] :
Formula350 Trying to smash a square protein through a round target hole with their feverous mouse clicking.
[7:09 PM] beta_helix:
That's what the big spiky ball is for!
[7:10 PM] :
Formula350 Who... ME?!
[7:10 PM] :
Formula350 IMAGE: http://fold.it/portal/files/chatimg/irc_817348_1593220214.png

[7:10 PM] beta_helix:
hahahahhahahhahahaha Nice!
[7:11 PM] beta_helix:
Well thank you all for stopping by Office Hours!
[7:11 PM] :
jeff101 will someone post a transcript on the Foldit web page?
[7:11 PM] :
jeff101 thanks for coming
[7:11 PM] beta_helix:
We've been trying to hold them at different times, to try to spread the timezones out as much as possible.
[7:11 PM] :
Formula350 Yes, we managed to bring donuts this time.
[7:11 PM] :
donuts554 "Oh! These three amino acids all have red and blue dots, and all similarly have a blue filled triangle on top of a pole! So, they may think, oh histidine is related to asparagine and glutimine, and histidine may have came from glutamine!'"
[7:12 PM] beta_helix:
Yes. We are trying to post the whole transcript, rather than just a discord link. We're still working on that, so thanks for your patience!
[7:12 PM] :
donuts554 And they think, "So that's why Histidine and glutamine have similar properties"
[7:12 PM] :
donuts554 You are welcome!
[7:12 PM] beta_helix:
Someone should really make an amino acid card game!
[7:13 PM] :
SethCooper that sounds like a good idea!
[7:13 PM] :
Formula350 I think Josh made an AA based Board Game (or maybe it was Proteins in general...)
[7:13 PM] beta_helix:
In that case, he should make it online so we can all play together!
[7:14 PM] :
Formula350 It's about to be released. It's on his website as Coming Soon
[7:14 PM] :
donuts554 I have more questions but Ill go to the next office hours
[7:14 PM] beta_helix:
Sweet... can I invest in that?
[7:14 PM] beta_helix:
Thanks again everyone... Thank you Seth!
[7:14 PM] :
Formula350 @Josh @Gamma_Helix wants to invest in your new boardgame ;)
[7:15 PM] :
Formula350 Yes, thanks Beta and Seth.
[7:15 PM] beta_helix:
Thanks donuts, I look forward to them
[7:15 PM] beta_helix:
Have a great weekend everyone... stay safe, and keep up the great folding!
[7:15 PM] :
SethCooper yes, thank you!
_________________________________________________________________________________________________________

Office Hours on 6/11/20 led by Bkoep:

5:02 PM] bkoep:
Hi everyone! I'm looking for office hours.
[5:02 PM] :
Serca i typed "11pm friday pst zimbabwe" according to formula350 message about friday and got no right time except google links to converters
[5:02 PM] bkoep:
I'm late and I can't find the office
[5:02 PM] :
Formula350 For example, when I'm at the cabin (in Alaska) on my hotspot, sometimes I'm located in Seattle.
[5:02 PM] :
wboler I'm guessing you forgot to bring donuts as well?
[5:02 PM] bkoep:
Is #veteran right? Can we set up an office here?
[5:02 PM] Susume:
this sounds like 90% of my bad dreams
[5:02 PM] :
Formula350 Yep the office is here.
[5:03 PM] :
Formula350 You can't see us becauase we're all in our cubicles.
[5:04 PM] bkoep:
Okay, great!
[5:04 PM] bkoep:
I'm bkoep
[5:04 PM] bkoep:
I'm a scientist on the Foldit team
[5:04 PM] :
Serca typed in google: "2:00pm GMT here" with no proxy and got 17 hours mistake. pretty sure my timezone in windows is +-1 hour from my ip timezone
[5:04 PM] bkoep:
I'm here to answer questions and chat about proteins
[5:04 PM] :
Serca hey bkoep
[5:05 PM] :
Serca bkoep why do we have so few prediction puzzles of the proteins with unknown and unpublished structures? can we have more of them? prediction is very interesting, what is the problem with getting that kind of puzzles?
[5:05 PM] wboler:
Do you have any plans on making a "continuous" version of the BUNS filter? The discrete stops seems to be overly harsh.
[5:06 PM] :
wboler steps* even
[5:06 PM] bkoep:
Good question @Serca
[5:07 PM] bkoep:
The short answer is, we think Foldit is most useful for protein design problems
[5:07 PM] :
alcor29 Does Buns count buns in the core also, not just the interface?
[5:07 PM] :
Serca bkoep so, de-novo only? no hope for the real prediction?
[5:07 PM] :
Formula350 It counts them wherever they are, alcor
[5:08 PM] :
alcor29 Tx bkoep
[5:08 PM] bkoep:
@Serca how do you mean de novo vs. real prediction?
[5:08 PM] bkoep:
@wboler that's an interesting question about "continuous BUNS"
[5:09 PM] wboler:
My line of thinking is attributing some sort of "distance" function to how close we are becoming to creating a BUN.
[5:09 PM] :
nspc IMAGE: http://fold.it/portal/files/chatimg/irc_959223_1591830575.png

[5:09 PM] bkoep:
I see
[5:10 PM] :
Formula350 lol alcor, I know but... here's an example (to the right of my Pin)

1
[5:10 PM] :
Formula350 IMAGE: http://fold.it/portal/files/chatimg/irc_817348_1591830613.png

[5:10 PM] bkoep:
Well, what it really comes down to, with BUNS, is whether or not the polar atom can make a hydrogen bond
[5:10 PM] :
Serca bkoep prediction puzzles is something with no pdb records. de-novo proteins is something very different that has no direct connection/links to its structure
[5:11 PM] bkoep:
And it's true that H-bonds come in all different strengths and energies
[5:11 PM] bkoep:
But the gradient between "can make an H-bond" and "cannot make an H-bond" is pretty steep
[5:12 PM] :
Serca my pc is not that good to new proteins design and i like to make something like prediction job on the puzzles with inlown secondary tertiary structure
[5:14 PM] bkoep:
What I mean is, very small changes in atom position can yield a huge difference in H-bond energy
[5:14 PM] :
Serca *with unknown secondary/tertiary strcuture
[5:14 PM] wboler:
True, I just wasn't sure if there was a better way. I did something similar for my master's thesis.
[5:14 PM] :
spvincent IMAGE: http://fold.it/portal/files/chatimg/irc_3010_1591830885.png

[5:14 PM] wboler:
(Not related to protein folding)
[5:15 PM] bkoep:
@wboler, well now I'm curious! Can you say more?
[5:16 PM] wboler:
I did work on optimizing the best placement of radio antenna with respect to "zones". If we gave the algorithm discrete zones, it would constantly get stuck in local optima. To guide it, I created a mix between continuous and discrete: small "energy" that grew the closer a zone edge was reached, until constant fitness for reaching within the zone.
[5:17 PM] wboler:
It made me wonder if something similar could be done for BUNS, but I wasn't sure how hard that would be.
[5:18 PM] bkoep:
@Serca I see. Unfortunately, I think the prediction puzzles are just not as fruitful, scientifically (compared to design puzzles).
[5:19 PM] bkoep:
So if we have to focus our efforts, we'd like to focus it where we can make the greatest contributions.
[5:20 PM] Formula350:
If BUNS aren't specifically your area, that's ok, but my own BUNS related question is: Would it be possible to pre-process the target by loading the ENTIRE protein with the BUNS filter, then remove them from the target in our puzzle? That way, the ones that show up by default (ie the 15 on 1846) would be factored out, thereby eliminating a bunch of visual clutter.
[5:22 PM] Susume:
some of those are reachable by the foldit design tho, would not want them factored out
[5:23 PM] :
jmbrownlee333 I am hearing that you no longer think that crowd sourcing is a way to solve the "protein folding problem". Is that true?
[5:23 PM] :
Formula350 Granted, but if they aren't BUNS when the entire target protein is present, then they shouldn't technically be anything WE have to worry about.
[5:23 PM] bkoep:
@wboler cool! In that sense, I guess a continuous BUNS score would be really helpful for gradient descent (wiggle)
[5:23 PM] bkoep:
The problem there is that BUNS calculations are really slow
[5:24 PM] alcor29:
I had asked because I thought the interface buns interfere with docking. The others might prevent proper folding but some of the others on the perimeter might bond with water.
[5:24 PM] wboler:
That's what I gathered
[5:25 PM] :
Formula350 Yea, my question was quite directly related to what you had asked Alcor :) I know that the Target's default BUNS cause a lot of confusion for players. So I figured if some of them didn't technically exist in the entire Target Protein, then having them omitted would help us understand.
[5:25 PM] bkoep:
@Formula350 Yes, that's a feature we'd like to add to the BUNS objective. It did not make it into the 1st generation BUNS prototype.
[5:26 PM] :
Formula350 Ah that's great news! Thank you :)
[5:26 PM] :
Serca t@Serca I see. Unfortunately, I think the prediction puzzles are just not as fruitful, scientifically (compared to design puzzles).
[5:26 PM] :
Serca Not that fruitfull to what? We now have 2 de novo puzzles and 1 revisting puzzle. Is predicting a protein with unknown scructure is worse than all that three puzzles? So predicting puzzle is even worse than pretty senseless work on a puzzle with known structure?
[5:27 PM] bkoep:
Like @Susume says, we'd want to be careful about BUNS that can be addressed in the target, so we wouldn't get too fancy—just a setting to let us exclude specific residues from the BUNS calculation
[5:27 PM] Susume:
is the coordinate system for the BUNS voxels anchored on the locked protein? or can it move when our protein moves? some players observe BUNS changes far from where they just made a minor change, I wondered if it is due to the voxels moving
[5:28 PM] :
Serca i am ok with the de-novo puzzles, just cannot undeerstand how revisting puzzles are better than precition ones
[5:29 PM] wboler:
Oh yeah, can we have an API to count bonds and buns programmatically in recipes?
[5:29 PM] :
Serca is that really the problem to find protein with the unkown structure? what is t you algorithm to find one?
[5:29 PM] wboler:
(Unless something like that already exists and I'm missing out)
[5:29 PM] :
Formula350 That may be a Josh and Neil question, I'm not sure.
[5:30 PM] :
jmbrownlee333 Here is a question from a player who can’t be here just now. Skippysk8 wants to know how to think about shape complementarity, vs overall square angstroms of the interface, vis specific electrostatic interactions at the interface.What’s more important to the design.? How close to atoms need to be at the molecules interface?
[5:31 PM] bkoep:
@jmbrownleee333, I'm not saying that crowd-sourcing is useless for protein structure prediction. But I'm convinced that Foldit players can make a greater impact with protein design.
[5:32 PM] :
jeff101 how do you plan to use our results from puzzle 1847? it seems like you already know what the structure should be.
[5:32 PM] :
jeff101 1847 is the ED puzzle for the player-designed protein
[5:33 PM] :
Serca bkoep that sound like that you won't have any prediction puzzles here anymore.
[5:34 PM] Susume:
I had asked bkoep for 1847 just for fun, since it's a foldit player design we haven't played before
[5:34 PM] :
Serca does it mean there is no need in any predictionion puzzle solution?
[5:35 PM] :
Formula350 Maybe it may be relevant and clear up some of Serca's confusion (which also touches on your answer to JMB): On the submitted predictions on say ORF8/6/3, what percentage of our results have proven useful?
[5:35 PM] :
Formula350 (as I think what we may need to hear is: we basically suck at prediction and our results show it, which is why we don't get many prediction puzzles)
[5:36 PM] bkoep:
@Serca I see your point. A big issue with the blind prediction puzzles is that we don't have a good way to evaluate the accuracy of Foldit predictions.
[5:36 PM] :
Serca Susume 1847 looks like a solution for you question in the last office hour about trying to use some unknown puzzles vs revisted ones
[5:36 PM] Susume:
Serca there are many unsolved proteins, but only a few that a researcher is about to solve, or has solved and not published - if we worked on predicting the ones no researcher is working on, we will never know how well we did
[5:37 PM] :
Migi-irc well Formula350 my suspicion is that AlphaFold is just better than the Foldit community at prediction puzzles
[5:37 PM] Susume:
the coronavirus ones are good because people are working to solve them right now, we might get results soon
[5:37 PM] bkoep:
In rare cases (like the MPMV protease from 2011), we have experimental data that we can use to confirm predictions.
[5:37 PM] :
jeff101 why not have a puzzle like 1847 before the Foldit Team solved the structure? Why not let us help you?
[5:37 PM] :
Serca the only question i don't understand why do we have no new puzzles with unknown sturcture, and have so much revisiting puzzles, that just seem to be scientificaly senseless
[5:38 PM] :
jmbrownlee333 To follow on Jeff's question.do you plan to look at the player models in terms of R and/or Free? Do player structures tend to move either of these more or less.
[5:38 PM] :
jmbrownlee333 ignore my question, if too wonky.
[5:39 PM] :
Serca Susume that sounds frnakly enough. that is why i am intersting why don't we have some puzzles that scientists are really working on
[5:39 PM] Susume:
I think it is hard to get non-foldit scientists to share their data
[5:40 PM] wboler:
My guess is the revisits have to do more with validating changes made to the program, which are useful.
[5:41 PM] :
Formula350 I can appreciate Serca's viewpoint. Assuming that there's no limitation on the server-end for having more than 3 puzzles running (as I know there are periodic automated submissions), it might be nice to exclude the Revisits as one of the "3", and include a Prediction puzzle, regardless.
[5:41 PM] :
Serca Susume i even tried to find a puzzle with the known environment an unknown 3d structure and length less than 200 to work on understood that it is not hat easy to find one
[5:42 PM] :
Serca that is the reason of my question and that is wju i try to clear that in me next questions: i get no the clear answer from bkoep (sorry dude, take my respect)
[5:43 PM] bkoep:
@Serca It's true the scientific value of the revisited puzzles is thin. They are definitely useful for long-term, continuous benchmarking (i.e. is Foldit performance consistent over time as the software and community evolves).
[5:44 PM] bkoep:
We also like the Revisited puzzles as a gateway for novice players, but that doesn't really have anything to do with scientific results
[5:46 PM] :
Serca and i also like the puzzles with unkown structure vs revisting ones to work as the only reason th present myself in foldit.
[5:46 PM] alcor29:
Any progress in the labs this week on Covid puzzles?
[5:47 PM] :
Serca but i am ok if you discriminate my 10y.o. cpu tand my hand folding skills to run just revisitng puizzles
[5:47 PM] :
Serca *tand=and
[5:47 PM] pc:
I like design puzzles
[5:49 PM] :
Serca guess i like them too if my cpu is ok with all the filters
[5:49 PM] :
jmbrownlee333 can I re-ask skippysk8's question. Skippysk8 wants to know how to think about shape complementarity, vs overall square angstroms of the interface, vs. specific electrostatic interactions at the interface.What’s more important to the design.? How close to atoms need to be at the molecules interface?
[5:50 PM] :
jmbrownlee333 to=do
[5:50 PM] bkoep:
@jmbrownlee333 @jeff101 Those are good questions about the ED puzzle! When x-ray diffraction is this good, it is less work to solve the structure automatically than to set up the Foldit puzzle.
[5:50 PM] :
Serca or they just switch their filters to gpu, to unload my cpu
[5:51 PM] Susume:
(thx jmb, you understood skippy's question better than I did)
[5:51 PM] bkoep:
The more interesting challenge comes from poorer quality or lower resolution data, where the automated methods fail. We hope to bring more of those puzzles to Foldit.
[5:53 PM] :
jeff101 if you took our 1847 solutions and removed the ED part of the scores, would the results be useful for understanding the energy landscape of the designed protein?
[5:54 PM] bkoep:
@alcor29 Yes, the lab experiments are progressing for h the CoV spike and IL-6R binders, but I still can't give a time estimate on when we'll have data
[5:54 PM] :
alcor29 Tx
[5:54 PM] bkoep:
Frankly, I probably can't offer a time estimate until we actually have the data
[5:54 PM] :
alcor29 k
[5:55 PM] :
jeff101 if you think of answers to some of our questions later, please make a blog post about them
[5:55 PM] :
Serca bkoep can you please advice any algorithm to find a puzzle with the unknown structure?
[5:55 PM] :
Serca I want to have some fun on the sandbox puzzle with that.
[5:56 PM] bkoep:
@jmbrownlee333 @Skippysk8s That's a great question about which binder metrics to focus on.
[5:56 PM] bkoep:
We don't really know
[5:56 PM] bkoep:
We will try to set their relative weights to something reasonable
[5:57 PM] bkoep:
But the fact is that shape complementarity, and SASA, and BUNS, and DDG, and electrostatics all show some correlation with binding success rates
[5:58 PM] :
jeff101 there was another binding design project about a year ago ... non-COVID ... what is the status of that project?
[5:58 PM] bkoep:
For some of these, we have some target thresholds where maybe the correlation drops off
[5:59 PM] bkoep:
But I'm pretty hopeful that Foldit players can optimize all of them
[6:01 PM] bkoep:
The traditional way we design binders is pretty inefficient: we spend a bunch a computational time generating 1M designs without optimizing these objectives, and then we check which of those 1M designs meet the objectives and throw out the 99% that don't.
[6:02 PM] :
Serca @bkoep so, no algorithm to find a puzzle with the unknown structure and known environment? no answer looks like a positive answer for that. thank you.
[6:02 PM] :
Formula350 Do you guys/gals in the lab ever look at our designs and try to "Evo" the best candidates yourselves, or do you typically just feed them to an algorithm?
[6:02 PM] bkoep:
It seems like Foldit players might be able to consciously tailor designs meet all of these criteria, instead of blindly shooting birdshot at a tiny target
[6:03 PM] bkoep:
@Serca Sorry, I'm not sure what you mean about "known environment"?
[6:03 PM] :
jeff101 cytosol vs membrane?
[6:04 PM] :
jmbrownlee333 or excreted.
[6:04 PM] :
jeff101 favoring disulfides or not?
[6:05 PM] :
Serca i am just talking about any structure that is not listed in the pdb is interesting. just toooooooo tired of the revisiting puzzles. but "known environment" means just hydrophobcity score.
[6:05 PM] :
Serca anyway i guess most of us will be happy with just any random AAs that are not listed in the pdb
[6:08 PM] :
Serca but the best AA sequences are made by nature. that is the reason of the feeling to talk with god while folding them.
[6:08 PM] bkoep:
Oh I see. There are lots of protein sequences out there without structures. Like, the NCBI BLAST database of non-redundant protein sequences.
[6:09 PM] bkoep:
And people are discovering tons of new ORFs now from shotgun sequencing experiments
[6:10 PM] bkoep:
(Basically, take a bucket of seawater and sequence all of the DNA in it)
[6:11 PM] :
Serca so what stops you do add that to our puzzles and have us to fold something really known then?
[6:11 PM] Susume:
Serca you could find an unsolved sequence from ncbi and challenge players to solve it in sandbox, we won't know if anyone is right but we could compare results
[6:11 PM] bkoep:
@jeff101, the IL-7R binder experiments got held up by the COVID-19 shutdown, but I think we may have some of that data soon
[6:12 PM] :
Serca *really unkown then
[6:13 PM] :
jmbrownlee333 On 1847, s there any area where the x-ray structure varied from the design more? Were loops or sheets or helices farther off from the design model?
[6:13 PM] :
Formula350 Wonder if you can make a "Contest" using the Sandbox puzzle as a base.
[6:13 PM]
Serca Susume i don't like your solutions, sorry, but what is your the algorithm to find the AA sequence with unknown structure?
[6:14 PM]
Serca i see foldit guys don't have any algorithm to find the AA sequence no listed in pdb and that is their problem (sorry bkoep)
[6:15 PM] bkoep:
@Serca You can use NCBI BLAST to check if a sequence is in the PDB
[6:15 PM] bkoep:
https://blast.ncbi.nlm.nih.gov/Blast.cgi
BLAST: Basic Local Alignment Search Tool
The Basic Local Alignment Search Tool (BLAST) finds regions of local similarity between sequences. The program compares nucleotide or protein sequences to sequence databases and calculates the statistical significance of matches. BLAST can be used to infer functional and evolu...
[6:16 PM] :
jmbrownlee333 No like bkoep said, no way to know if you are on the right track, for who knows how long.
[6:18 PM] :
Serca bkoep so I want to fold a nice 150 length reisdue protein at the sandbox: what should I do to find one that is not listed in rcsb pdb?
[6:18 PM] :
Serca could you please tell me the buttons i have to press?
[6:18 PM] Susume:
Serca this is not something someone can teach you in 3 lines of chat, go read about blast and educate yourself
[6:19 PM] :
Serca very draft how you understand that
[6:19 PM] :
jmbrownlee333 @Serca i have some ideas, can chat in group or offline.
[6:19 PM] :
jmbrownlee333 Ie PM
[6:19 PM] :
Serca jmb that is interesting in prespective
[6:20 PM] :
Serca Susume you know... i guess the reason we have no any proteins with the unknown structure is that they just don't know how to find that
[6:20 PM] bkoep:
@Serca I'm sorry, I cannot tell you which buttons to press. But the NCBI tools are pretty nice, I think you will find them easy to use
[6:20 PM] bkoep:
Okay all, I'm sorry, I have to run
____________________________________________________________________________________________________________

Office Hours 5/26- led by Horowsah

4:00 PM] horowsah:
Hey all- just wanted to say that my "office hour" today is officially starting. This means that I can try to answer questions for the next hour. A couple disclaimers- I didn't put together the puzzles that are currently up so I'm not super familiar with their details, and I'm not a COVID-19 expert. My expertise is more biochemistry in general.
[4:01 PM] :
Formula350 Hello :)
[4:01 PM] :
jeff101 hi
nspc hi !
[4:02 PM] :
ipatrol awwww
[4:02 PM] :
jeff101 I've got a question about b-strands. Are the ideal zigzag structures stable if there are no hydrogen bonds?
[4:02 PM] :
jeff101 like what do b-strands look like before the hydrogen bonds form?
[4:03 PM] horowsah:
let's see if this answers the question- b-strands are the most extended form possible of a chain of residues
[4:03 PM] horowsah:
so when it forms h-bonds to form a sheet, it's basically the same shape for each strand before and after
[4:03 PM] horowsah:
just now with h-bonding
[4:04 PM] horowsah:
in real life, that can happen, but it will be very dynamic and moveable
[4:04 PM] horowsah:
there's a class of proteins called Intrinsically Disordered Proteins that are kind of like that
[4:04 PM] horowsah:
is that answering the question?
[4:05 PM] :
jeff101 when the b-strands form h-bonds, do they start near the b-hairpin bends or farther from the bends?
[4:05 PM] :
ipatrol Is there any way to ensure that puzzle solutions will be specific to the target and not just bind to other common protein motifs?
[4:06 PM] horowsah:
@jeff101 hmm- that's a question that I don't know a firm answer to, but I'm guessing it's not known for most proteins. But my thought would be that it would in most cases likely start from the loop and move outward
[4:08 PM] :
jeff101 I've read some about denaturation studies. Seems there are many different ways to denature sheets (chemicals, temperature, pulling on the ends).
[4:09 PM] horowsah:
@ipatrol: this is the big challenge of foldit in general. To have lots of different puzzles with different facets, it becomes hard to make it as specific for each case as we want. This is definitely an active area of development for the types of COVID puzzles we've been running. My thought is that the work Foldit players are putting in is great, but that there's definitely room for improvement to make this aspect work better from the game side.
[4:09 PM] :
ipatrol A strong oxidizer will denature darn near anything
[4:09 PM] horowsah:
Yeah, denaturation is a big deal in protein folding studies
[4:09 PM] horowsah:
I've used at least four methods of doing it in my lab work, and there's tons more
[4:10 PM] :
jeff101 do they all show the same intermediates?
[4:10 PM] :
ipatrol I think urea or a gunaidinium salt is usually the least destructive way of denatuging proteins at room temperature
[4:10 PM] :
ipatrol *denaturing
[4:11 PM] wboler:
Would it be too difficult to add filters for protein motifs we know won't work out?
[4:11 PM] horowsah:
that's a good question- in general, this is a point of discussion as to whether all forms of denaturation are created equal, or even physiologically relevant
[4:11 PM] :
jeff101 which way agrees best with molecular dynamics calculations?
[4:11 PM] :
ipatrol Well I mean, if it'll kill you, it's probably not very physiologically relevant :-P
[4:12 PM] horowsah:
that depends who you ask For example, oxidation is a common protein unfolder in cells. So that one's important. Does it show the same intermediates as temperature denaturation? Probably yes in a lot of cases, but not all.
[4:13 PM] horowsah:
@wboler: i know there's been a bit of discussion on that idea, but it's practically really difficult because there's so many different ways to do something that isn't great
[4:14 PM] :
ipatrol Hmm, I wonder, would it be possible to modify the game to allow us to play with some non-standard amino acid as sidechains?
[4:15 PM] :
jeff101 like phosphorylated ones
[4:15 PM] horowsah:
to the urea or guanadinium question, these are the most common ways to denature proteins in a lab environment without changing temp. The reason is because it's the easiest to reverse and it usually won't chemically modify the protein outside of unfolding it (and maybe causing it to aggregate)
[4:15 PM] :
Formula350 --- Poster child for "How to Do Stuff in a Not Great Way" :P
[4:15 PM] wboler:
Would it be possible to train a neural network somehow for things that don't work? My guess is one of the limitations would be available data. Maybe some other AI mechanism.
[4:16 PM] :
ipatrol I'm afraid to ask, Formula350
[4:16 PM] :
jeff101 what experimental technique is best for monitoring hydrogen-bonding between b-strands?
[4:17 PM] horowsah:
non-canonical amino acids! In theory, they do work in Foldit, but we don't give puzzles on them hardly ever. I'm not sure if there's a specific reason why, just the science hasn't trended that way yet. The one thing is that if we're trying to design something, it's harder to put non-standard amino acids in proteins than the regular ones in the lab, so it usually leads to things that will be harder to implement outside a computer
[4:17 PM] :
nspc this is high level questions yes oO
[4:18 PM] :
Formula350 I don't follow what's going on either, don't worry :P
[4:18 PM] horowsah:
@wboler on the AI- there have been a few attempts at similar things, but they've not gotten too far. It seems to me like the way Foldit players do things is especially hard for a computer to learn
[4:19 PM] :
jmbrownlee333 If I could ask a completely unrelated question, are there plans for small molecule design and coronavirus? I am thinking about a viral protease target, most specifically.
[4:19 PM] horowsah:
@jeff101: best technique for monitoring h-bonds: if you want a static picture of it folded, then usually crystallography, but NMR and cryo-EM do a good job. If you want to see it wiggle and jiggle in solution, then NMR is usually the way to go.
[4:20 PM] horowsah:
small molecules! I'm sure that's on the minds of the people working on that aspect of Foldit long-term (because it's on everyone's minds right now), but I really don't know if it's something on the horizon there or not.
[4:20 PM] :
jmbrownlee333 it has been ages since we had a small molecule puzzle.
[4:20 PM] :
Serca How do I find a protein with unknown tertiary structure (not listed in pdb) that has known environment (so I wouldn't try to fold a transmembrane protein with Foldit)? Is it really hard to find one?
[4:21 PM] horowsah:
@Serca: maybe uniprot.org?
[4:22 PM] :
jeff101 @horowsah what projects do you work on? that might help us ask better questions
[4:22 PM] horowsah:
@jmbrownlee333: small molecules present a whole different scientific set of problems than regular foldit puzzles, so it's hard to get it just right h from a scientific game development side, and from the perspective of letting players know what is the best way to tackle those problems
[4:23 PM] horowsah:
so it's something that is still under heavy development
[4:23 PM] :
Serca never heard about that site, thank you
[4:24 PM] horowsah:
what projects do i work on? Lots of things...
[4:24 PM] horowsah:
within foldit, i tend to work mostly on educational tools and integrating crystallography and cryo-EM tools
[4:24 PM] horowsah:
but I have a wet lab that studies how nucleic acids impact protein folding and aggregation, which I'd love to do in Foldit someday
[4:26 PM] horowsah:
@sarca: uniprot is my go-to whenever i hear about a protein that i need to learn more about and know nothing about it
[4:26 PM] :
Formula350 How well do 310 (3-10?) Helices fold in the lab (say, compared to the traditional alpha helix we use in the game), and does the Alpha Helix do the best or is there another helix type that does similarly well?
[4:27 PM] :
Serca Do you know how the Foldit Remix database is generated? When I try Remix tool I often get glycine and proline forming a shape that is detected as helix by Foldit autostructures. I read in a paper that glycines and prolines are pretty common in transmembrane helices because the environment is not that agressive to break the secondary structure. Is that the reason why Remix tool tends to make helices from h these AAs?
[4:27 PM] :
ipatrol So yeah, I understand the difficulty with synthesizing some non-canonical AAs. I think it's pretty easy to pin arbitrary AAs to the end of an existing protein though, so maybe they can be allowed for just the ends?
[4:28 PM] horowsah:
@Sarca: sorry, I don't know much about the details behind remix. I can't be too much help on that one
[4:28 PM] Josh:
Dang y'all ask a LOT of good questions! For the record, this all goes over my head too =P
[4:29 PM] horowsah:
@Formula350: I don't know specifically how well 310 helices do in designed proteins, but they're pretty rare in natural proteins, presumably for good reason. One thing to know is that Rosetta (which is the scientific engine Foldit is built upon) tends to have a love of alpha helices beyond what nature does. They are the most stable regular secondary structure, but it's not always the best for every scenario
[4:30 PM] horowsah:
@ipatrol: yeah, ends of proteins are easier to modify than in the middle, but that really restricts what you can do with them. It's way more fun to put them all over the protein, just harder. Still, that does sound like a fun idea to explore in future puzzles.
[4:31 PM] :
Formula350 I accidentally made one a couple weeks back, which piqued my interest. Then my fold on NSP6 Prediction we're working on ended up creating a 310 on its own, which certainly gave me pause. Thanks :)
[4:33 PM] :
jeff101 when you get a raw set of cryo-EM data for a multi-protein complex, how do you pick out which proteins go where?
[4:33 PM] :
jeff101 how do you decide what to trim from the EM cloud?
[4:33 PM] horowsah:
So there are tools that are used called segmentation tools, and they're job is to help you find gaps in the density where it looks like two proteins come together
[4:34 PM] horowsah:
That's typically the first thing one does with a cryo-em map
[4:34 PM] :
jeff101 are you working on any tools to help us solve ED/EM puzzles in Foldit?
[4:34 PM] horowsah:
Then once it's been split up, it becomes a rather tough problem of figuring out which protein could even go where
[4:35 PM] :
Serca And returning to the jeff's question about beta-sheets. Is that correct that even without H-bonds formed, beta-sheets tend to form extended straight lines? How likely loop-biased AAs tend to form that kind of straight lines?
[4:35 PM] horowsah:
to my understanding, this is mostly trial and error for most people. Start building something and then figure out which amino acids it is later. it used to be that the maps weren't high enough res to do that, but now it's possible. Before, you'd have to do things like attach antibodies to different parts of your structure to see what went where.
[4:36 PM] ZombieRaccoon:
Is there any possibility of being able to color the cloud with a paint program or something ? I think it would be very helpful to deal with it in small chunks and try to note areas by coloring bits of the cloud
[4:36 PM] :
jeff101 we had a puzzle #1667 with too much density. I think its ED cloud was for perhaps 10 proteins, and we were supposed to find where a single protein fit into it. We didn't know the sequence for the 9 other proteins there.
[4:36 PM] :
jeff101 https://fold.it/portal/node/2007726
[4:36 PM] horowsah:
@Jeff101: have you guys gotten a puzzle with the new radius tool for ED/EM yet? I actually can't remember.
[4:37 PM] :
Formula350 I loaded a Beginner ED puzzle a week or two ago and there was a Radius slider as I recall. Not sure if that's been there or brand new, though.
[4:37 PM] horowsah:
@Serca: so prolines for example tend to not be in beta sheets because they will cause a kink. Glycines are also usually too flexible. The other amino acids that don't show up in sheets often actually have more to do with the side chains and how they fit, usually.
[4:38 PM] :
Serca Formula350 i guess you got 3-10 on nsp6 because hydrophobis deformed the helix pretty much, trying to get buried
[4:38 PM] horowsah:
Ah yes, 1667. We really weren't sure what was going to happen with that puzzle, but we thought players might like a taste of what the whole problem was like before it gets broken up.
[4:38 PM] :
jeff101 I think 1797 was the last ED puzzle we got. It was in February.
[4:39 PM] horowsah:
So the radius tool is one to definitely use if you haven't before in ED puzzles. We have some more tools for ED we're starting on, but there's no real timetable for them yet.
[4:39 PM] ZombieRaccoon:
I like the ED puzzles, just have a hard time keeping track of area
[4:39 PM] horowsah:
On ED puzzles, I tend to use Q a lot
[4:40 PM] horowsah:
That's the easiest way for me to keep things in the right parts of my screen with the density
[4:41 PM] :
Formula350 Might be, Serca, but there's not a whole lot of them on the one that turned 310 lol (mind you, it's less 310 looking currently, than it had been 2 days ago)
[4:41 PM] :
Formula350 IMAGE: http://fold.it/portal/files/chatimg/irc_817348_1590532907.png

[4:42 PM] :
jeff101 below is a feedback I wrote after puzzle 1667 with ideas for ED Puzzles.
[4:42 PM] :
jeff101 https://fold.it/portal/node/2007742
[4:44 PM] :
Serca Formula350 i got some 3-10 like helix too near the phenylalanine parts of the helix
[4:44 PM] horowsah:
I remember that feedback. it's interesting, the way foldit has in the past done ED puzzles is really different from how crystallographers/em people do it. For example, I never use labels, but that's because I always thread the backbone first, and figure out sidechains later. However, in ED puzzles so far, we don't let players do that
[4:44 PM] horowsah:
If you're curious as to the reason, it's hard to enforce the "right" sequence if we give you a design puzzle
[4:45 PM] :
Formula350 That one in the screenshot is #49 (I deviated after running ""FormPRED"" on it lol)
[4:47 PM] ZombieRaccoon:
Have you considered using a generic em cloud for design puzzles to define where the designed protein should be, and penalize parts outside the cloud
[4:48 PM] :
ipatrol Isn't that already a puzzle type?
[4:48 PM] horowsah:
I've actually done that with classes before. I've gotten good feedback and the fun level, but it tends not to produce the best science.
[4:49 PM] ZombieRaccoon:
Bummer
[4:49 PM] horowsah:
I'm still working on a way to make it better, though!
[4:50 PM] :
jeff101 why doesn't Foldit let players thread the backbone in ED puzzles?
[4:51 PM] horowsah:
So threading the backbone would be like having a design ED puzzle in which you first thread it, then change the amino acids. In theory, this should work, but it requires a ton of detailed mutation work after the fact not to what would be most stable, but whatever the protein has originally. That means things like auto-mutate don't work right, and the score function will penalize you for getting the correct sequence.
[4:52 PM] horowsah:
So we'd have to put in a sequence filter on top of it, but not that would outweigh the other important things in an ED puzzle... not easy
[4:52 PM] :
Serca horowsah is there any characteristic of the helices/sheets like having different degrees of freedom comparing to the loops? I am trying to make the SS prediction in Foldit with getting some statistics about how they Remix in Foldit. I've noticed that the are some AAs sequences that have much more solutions from Remix database than others. That is correct for helix-propensity AAs, so I wonder if that is correct for sheets. What do you think about ponte
[4:53 PM] horowsah:
When you say different degrees of freedom, could you expand on that?
[4:54 PM] :
ipatrol I would guess he means in parameter space terms
[4:55 PM] :
jeff101 maybe remix gives more possible solutions for loops than for sheets than for helices
[4:55 PM] horowsah:
so i think it was designed to give more for loops
[4:56 PM] horowsah:
as to sheets vs helices, i'm not sure. Building good loops is notoriously tough.
[4:56 PM] :
Formula350 (Serca, for the record, your message cut off at "What do you think about ponte...")
[4:57 PM] :
jeff101 for threading ED, could we start with a poly-alanine protein, fit its backbone in, and then mutate to match the desired sequence?
[4:57 PM] :
jeff101 or poly-gly?
[4:57 PM] horowsah:
yep, that's the way it would work
[4:57 PM] horowsah:
i'll work on seeing if we can give that a try
[4:58 PM] :
jeff101 the mutation wouldn't be that much computation, because the AA's would always follow a certain sequence
[4:58 PM] :
jeff101 if one sequence doesn't match, just shift it along the polypeptide until it matches
[4:59 PM] horowsah:
I have to head out in a minute y'all, so maybe just one more question if anyone has one?
[4:59 PM] Formula350:
Sorta followup to "good loops are notoriously tough" (partial ramble here, I apologize) If ideal loops play an important role in our proteins folding in a lab, would it perhaps work better if we could instead work on an "Asymm Trimer"? Being able to make three individual helices that all want to strongly bond together, and effectively work as a Monomer, but being Asymm would then still let us determine one side as the docking side for the target (same as how we design what we do already). Or would this be too easy get destabilized and fall apart?
[4:59 PM] :
jeff101 phosphates on sidechains
[4:59 PM] :
mib_hgy8xr @jeff101 there was a puzzle like that 5 years ago: https://fold.it/portal/node/2000180
[5:00 PM] :
jeff101 (thanks mib)
[5:00 PM] horowsah:
Asymm trimer, like as three separate polypeptide chains?
[5:01 PM] :
Formula350 Yea, basically. Not connected by anything in the puzzle since they would be effectively individuals, and created individually, but later introduced next to each other to bond.
[5:01 PM] :
marsfan Formula350 I just had a crash trying to run all filters on 1840. What files do I need to send and how should I send them?
[5:02 PM] :
Formula350 Mars: Did you copy out the log.txt and/or debug.txt before starting Foldit up again?
[5:02 PM] horowsah:
Practically, that would be pretty hard, simply because you'd have to co-express h proteins simultaneously in E. coli to make them, which increases the number of things that can go wrong by some large number. But there are applications for things like this out there. Is it easier than finding a good loop? I'd say probably not in most cases.
[5:02 PM] :
marsfan Copied it out
[5:02 PM] horowsah:
Ok, I have to head out- thanks all!
________________________________________________________________________________________________________

Office Hours 5/15- led by Beta_Helix

10:00 AM] beta_helix:
Hi everyone. I'll be here for the next hour, with the first (hopefully of many) Foldit Office Hours!
[10:01 AM] beta_helix:
For anyone who doesn't know me. I'm here: http://www.cis.umassd.edu/~fkhatib/cv.html(edited)
[10:01 AM] beta_helix:
and I'll be happy to chat about anything you like!
[10:02 AM]
Skippysk8s hey beta helix....I think the theme of the day is hot cross buns,,,maddening to try to cover them
[10:03 AM] beta_helix:
Oh yeah... those are a pain in the BUNS
[10:04 AM] beta_helix:
But unfortunately, they are extremely important in design
[10:05 AM]
Skippysk8s we assume so.. it is nice to see that all that work on monomer design being applied. What has surprised you about what is turning out to be useful in the last couple months
[10:07 AM] beta_helix:
Honestly... it hasn't been a surprise to me. After the amazing results all of you showed the world last year with a Nature paper full of novel designs that folded up just like you predicted in the game... this is the logical next step!
[10:08 AM] beta_helix:
What makes this new task harder, of course, is that we are no longer just asking you to make sure your novel design can fold on its own...
[10:10 AM] beta_helix:
That is the next difficult challenge: making sure it folds up properly AND binds to the COVID-19 spike protein.
[10:10 AM]
Susume2 seems like in theory we could take soem of those stable folds and fit them into the target groove
[10:11 AM] Skippysk8s:
is there any way we could load some of the prior proven folds into the current puzzles and try to fit them? I'm the want my Lego folder
[10:11 AM] beta_helix:
That's a great point, Susume... the only main difference would be ensuring that the side facing covid can bind to the spike (as opposed to being exposed to water).
[10:12 AM]
Susume2 yes, definitely need to mutate and adjust (and repeat ad nauseam)
[10:12 AM] Skippysk8s:
agreed-- I'm looking for less lysine and arg on at least one edge so it will fit snugly
[10:13 AM] beta_helix:
That might be a series of fun puzzles, though! Everyone starts from the same Foldit Nature design, and tries to fix the binding sides
[10:13 AM] beta_helix:
(I'm writing this down!)
[10:13 AM] Skippysk8s:
sort of like our contest... fit the block into the space
[10:14 AM] beta_helix:
We always say: key in the lock!
[10:14 AM]
Susume2 is it possible to evaluate shape complementarity based on just the backbone, without knowing which sidechains will be selected (well we know the ones on the virus)?
[10:15 AM] Skippysk8s:
does the tumbler have to be perfect? our team got some but never all of the potential bonds to meet along the edges. And can we get a filter to make the score improve the better the fit (2 questions)
[10:16 AM] beta_helix:
Susume, you can get a very basic idea... but you would still want to represent the sidechains somehow, you wouldn't want to just strip them off.
[10:16 AM]
Susume2 maybe make them all generic medium sized centroids
[10:16 AM] beta_helix:
Many programs use centroids for example (and I know that word gives you the chills if you were around when we tried centroid mode in Foldit!)
[10:17 AM] beta_helix:
exactly
[10:17 AM]
Susume2 I'm just wondering if there is a way you guys could preselect a foldit design from the nature paper that will likely fit if we pick good sidechains
[10:18 AM] beta_helix:
Yes, I think that would be the way to do, because some of the topologies will just fit better than others.
[10:18 AM]
Susume2 auntdeen and I had lots of fun breaking centroids as a team project ;)
[10:18 AM] beta_helix:
hahahahhahahahhahahaha
[10:19 AM] beta_helix:
Skippysk8 for question 2, that could be possible... depending on how fast those calculations would end up being. It might just be easier to screen the folds initially on our end. But indeed, I can see how that would be a useful filter!
[10:20 AM]
Susume2 as long as we know the filter takes x minutes to run, we can plan how often to apply it
[10:20 AM] Skippysk8s:
we were frustrated as a team with Bruno's fold... we felt like we could go for score but didn't believe the result would be useful for science
[10:21 AM] Skippysk8s:
many of us started trying to link those few unmet bonds, but ended up just going for score
[10:22 AM] beta_helix:
Yeah, I think it would be very useful when you are trying to figure out if your topology is going to fit or if it's a waste of time. (I'll ask the team)
[10:22 AM] beta_helix:
Indeed Skippysk8, I know how frustrating that is...
[10:22 AM] beta_helix:
I remember the giant pi-helices from back in the day
[10:23 AM] Skippysk8s:
well, it can be fun lol...
[10:23 AM] beta_helix:
I kept looking at those solutions and going "why are these scoring so high!?!!"
[10:23 AM] beta_helix:
But that really highlights the amazing work bkoep and the rest of the Foldit Team has done with Foldit scoring in design
[10:24 AM] beta_helix:
It's such an amazing journey to see the difference in high scoring monomer designs in the early days of Foldit, compared to the Nature paper.
[10:24 AM] Skippysk8s:
we love to game. we want to try to keep the game fun and at the same time useful
[10:25 AM] beta_helix:
That is the reason for BUNS, h-bond filters, and the new filters for these coronavirus puzzles.
[10:25 AM] beta_helix:
Exactly, Skippysk8... it is a fine balance, right?
[10:25 AM] Skippysk8s:
agreed
[10:25 AM]
RockOn It seems my PC memory usage for FoldIt is at 1,000 MB's!
[10:26 AM] beta_helix:
We could make the game painful accurate (like an MD simulation with waters) but I promise you it would not be much fun!
[10:26 AM]
RockOn What has changed Windows or FoldIt
[10:26 AM] beta_helix:
Rock On!
[10:26 AM] beta_helix:
oh, that's a bad thing... right. My bad
[10:26 AM]
RockOn How bad?
[10:26 AM] beta_helix:
It's always easy to blame Windows (and fun too!) but indeed Foldit is constantly changing.
[10:27 AM] beta_helix:
Have you noticed a difference between Foldit updates?
[10:27 AM]
RockOn 33% of the CPU
[10:27 AM]
Susume2 RockOn do you mean RAM footprint? or something else?
[10:27 AM]
RockOn Yes, running taskmgr
[10:28 AM]
RockOn Just 1 client and nothing else (app related)
[10:28 AM] beta_helix:
Did you have this problem with previous versions of Foldit?
[10:28 AM]
RockOn No, not unless 2 clients wee running for hours
[10:29 AM]
RockOn One crash today running helix twister
[10:30 AM] Skippysk8s:
my old computer was at 100% with one client -- ran alternating clients with 2 open... it was about 12 years old. the biggest problem is that all the "ligand" cells count towards puzzle size, so it is getting harder with small computers
[10:30 AM] beta_helix:
Hmm... I quickly scanned our feedback and don't see any other Windows reports. https://fold.it/portal/feedback Was there a recent Windows update that was installed?
[10:30 AM] Skippysk8s:
it couldn't do today's puzzles at all
[10:30 AM]
Susume2 devprev got pushed to main about a week ago
[10:31 AM] beta_helix:
Skippysk8, you mean 1838?
[10:31 AM]
Susume2 I suspect one thing that pushes RAM per client up is using all the quicksave slots - some scripts use very many of them
[10:32 AM] Skippysk8s:
I couldn't do any of the COVID puzzle on it
[10:32 AM] Skippysk8s:
puzzles
[10:32 AM] beta_helix:
Undo graph as well, right Susume?
[10:32 AM]
Susume2 yes probably
[10:34 AM]
Susume2 changing tracks and changing back should free the undo graph memory, idk that frees the quicksave slots as well (have not tried it)
[10:34 AM]
RockOn I see the graph at 100 for length AND Memory Usage is at 100%
[10:34 AM] beta_helix:
I'll bring these issues up to the rest of the team...
[10:35 AM] Skippysk8s:
I live at 25 undos Rockon... not ideal. I just save before I start a new recipe each time
[10:36 AM]
RockOn I set Memory Usage to 25% and Taskmgr shows mem down to 700 or so
[10:36 AM] Skippysk8s:
anyhow, it might help our smaller computer owners if the ligand shows everything, but only "counts" the cells near the binding area
[10:36 AM] beta_helix:
Yes, that is a good point.
[10:37 AM]
RockOn CPU still bouncing around from 29% to 36%
[10:37 AM]
Susume2 I keep undo graph at 85 and undo mem 75%, and 800MB is a common RAM footprint for me
[10:37 AM] beta_helix:
I've often wanted Foldit to only visually show the very far off frozen sections of spike (for example) and have Rosetta just ignore those atoms entirely.
[10:38 AM]
RockOn I'm still getting No Response from time to time
[10:38 AM] Skippysk8s:
it does help to try and fit the key into the lock. For example, JoannaH found a great 3 helix log with a spike that got an extra hook. it was pretty cool
[10:38 AM] beta_helix:
It turns out that this is non-trivial to do.
[10:39 AM] beta_helix:
Indeed, Skippysk8, I just meant the regions that are so far away they would have no interaction at all.
[10:39 AM] Skippysk8s:
we appreciate that...
[10:39 AM] Skippysk8s:
what can we do better?
[10:40 AM] Skippysk8s:
I'd be fine if one of my bad folds was presented as "what not to do"
[10:40 AM] beta_helix:
That is a tough question, because you are all doing great! Any deficiencies are on Rosetta and our end!
[10:41 AM] beta_helix:
Skippysk8, I think bkoep does that very well in his Lab Reports: showing what works and what doesn't.
[10:41 AM]
Susume2 I loved the TIM barrel challenge - if you guys ever see a cool player fold and want us to try more of that shape, you could play it that way - a suggested but completely optional shape
[10:41 AM] Skippysk8s:
I know you are all working flat out, but feedback on the good and bad is highly appreciated. Lexie was great -- maybe she has some things she'd like us to try as a challenge
[10:41 AM] beta_helix:
It really is a question of us finding the balance between filters that are scientifically accurate... but don't kill your game. (h memory/speed/fun-wise)
[10:42 AM] beta_helix:
Those are h great ideas! I'm adding them to the novel of excellent suggestions today
[10:42 AM]
RockOn I'm doing a reboot to see if things change...will report in a few mins.
[10:43 AM] beta_helix:
thanks RockOn!
[10:45 AM]
Susume2 in fact a slightly perturbed foldit player design could be an alternative to revisits for beginner-friendly puzzles - I know you need some ongoing data from revisits, but maybe alternate with player designs - pull them apart a little like you do the revisits
[10:47 AM]
RockOn I tried Susume2's idea track change and CPU went down and Ram use down 50%
[10:47 AM]
Susume2 it would let us all play hands on with some designs from other people/teams, but in a category where you're not super hungry for diversity
[10:47 AM] Skippysk8s:
beta-helix do thank the team for getting a beginner puzzle up on covid... maybe we can give them a few things to dock and perhaps one to try and work on shape conformance.... It is good to have the change to expand our folding community
[10:47 AM]
Susume2 great!
[10:48 AM] beta_helix:
Good call, Susume. We'd just have to mess them up enough so that they wouldn't just go back down the their global mins
[10:48 AM]
RockOn So no reboot right now
[10:48 AM] Skippysk8s:
yes -- please bring up all hands.... or at least let all hands be a round 2 if we can't share across groups. I worked a couple puzzles with players not on my team. It would be good to share
[10:48 AM] beta_helix:
Will do, Skippysk8!
[10:49 AM] beta_helix:
All Hands in the past didn't really work well, but that was before design. I think it would be worth just trying it as a new puzzle (rather than the All Hands mechanic we had)
[10:49 AM] alcor29:
Is there anything that can be done to speed up the feedback cycle. It is frustrating to be on round 9 of the C19 not knowing what seemed to work better than others.
[10:50 AM] Skippysk8s:
I understand that first round in design should get as many ideas as possible.... but then working together would help
[10:50 AM] beta_helix:
alcor29... we feel you on that one! Every Foldit meeting we are begging for feedback on previous puzzles.
[10:50 AM] beta_helix:
But science is slow, and covid-19 has slowed things down a lot more
[10:51 AM] beta_helix:
Luckily, things seem to have finally been sorted out... so fingers crossed that we'll get results soon!
[10:52 AM] beta_helix:
Skippysk8, I completely agree with you.
[10:52 AM] beta_helix:
The key is to make sure that any starting puzzles aren't so trapped in their local minimum that there is no way to improve the score... that has often been the issue with starting from high-scoring structures.
[10:53 AM] Skippysk8s:
maybe pull it away and make us redock lol.... we are getting to be good at that
[10:53 AM] beta_helix:
hahahahahhahahaa
[10:56 AM]
RockOn Sure wish it were a 2-way chat that is going on!
[10:57 AM] alcor29:
Has foldit ever tried the cooperative model instead of the competitive gamesmanship model?
[10:57 AM] beta_helix:
alcor29, w.r.t what works better than others... that has been one of the main elements of the Lab Reports
[10:57 AM] beta_helix:
We participated in WeFold
[10:58 AM] beta_helix:
We did, though, alcor29: https://fold.it/portal/node/986517
We messed up the All-Hands puzzle yet again!
We messed up the All-Hands puzzle yet again!
[10:58 AM] beta_helix:
but it did not go particularly well.
[10:59 AM] beta_helix:
We have somewhat resorted to using the "Share with Scientists" feature, in order to get models that might not be the top scoring ones.
[11:00 AM] beta_helix:
@RockOn, do you not see my replies to you?
[11:01 AM] alcor29:
There are many smart and capable people here but I've always wondered if they only concentrated of the results instead of the score and the group competition what the results would be. Not as a partial experiment but as a total citizen science project.
[11:02 AM]
Dhalion I have never share a low scoring design
[11:02 AM] Skippysk8s:
ah, defintely do... I forgot to max out segments about a year ago and got a beautiful pringles fold. It scored 5th in the contest
[11:03 AM] beta_helix:
It's a valid concern, and a tough balance. I honestly feel that if we had started Foldit as just a Citizen Science project, where you just participate without rankings, leaderboards, etc... we would have still gotten a lot of promising results. But I think the human competitive element adds a lot to the motivation and to the results. There is a figure in our first Nature paper's Supp Section that shows this... let me try to find it
[11:04 AM] alcor29:
When only a partial experiment, game players still retain their 'secrets' and certain 'questionable' behaviors. But I get you.
[11:04 AM] beta_helix:
Nature 2009

[11:05 AM] beta_helix:
So this ugly figure shows the lowest Rosetta energy PER TEAM over time (in days)
[11:05 AM] alcor29:
Interesting.
[11:05 AM] beta_helix:
You can see that many teams flattened out (hit their high score) around days 2-3
[11:05 AM]
Susume2 (lower is better, right?)
[11:05 AM] beta_helix:
...except for the cyan team! (Yes, lowest energy = high Foldit score. Thanks!)
[11:06 AM] beta_helix:
Once the Cyan team suddenly was at the top of the leaderboard, bam! Everyone started working overtime
[11:06 AM] beta_helix:
(shown as the yellow rectangle)
[11:06 AM] alcor29:
So maybe only 2-3 days should be allowed to avoid overrefining?
[11:06 AM] beta_helix:
We have indeed discussed this very often...
[11:06 AM] beta_helix:
but not everyone can play Foldit 24/7
[11:07 AM]
eaglgenes101 At 2-3 days, I'd stop polishing existing solutions and start trying other possible avenues
[11:07 AM]
eaglgenes101 Which may or may not be effective
[11:07 AM] beta_helix:
The 7 day window (or longer in certain cases, like the Monkey Virus one) is to ensure that everyone has the opportunity to play at least a few days.
[11:07 AM] beta_helix:
eaglgenes101, that is exactly what you want to do!
[11:07 AM] Skippysk8s:
actually, our team enjoys the end game...we cross from the Ural mountains through the US with an occasional Australian. So we take the last few days for fun
[11:08 AM] alcor29:
Good discussion. Many thanks beta_helix.
[11:08 AM] beta_helix:
That was actually the motivation for our short-lived Exploration puzzles, and the Move Limit.
[11:08 AM] alcor29:
Hope you do it again sometime.
[11:08 AM] Skippysk8s:
thanks beta helix
[11:08 AM] beta_helix:
To force you to try a completely different fold, because it's very likely that your top score is stuck in it's local minimum.
[11:08 AM] beta_helix:
Me too! Thank you all!
[11:09 AM] Skippysk8s:
well, just leave us some play time lol
[11:09 AM]
Susume2 thanks beta!
[11:10 AM] beta_helix:
Thank YOU are for your great suggestions, I'll bring them to Monday's Foldit Team meeting! Take care everyone, and keep up the great folding...

jeff101's picture
User offline. Last seen 8 hours 47 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
Can you post the transcripts directly on the Foldit site?

I don't have a Discord account and don't use an external IRC program.
Would it be possible for you to post the Office Hour transcripts
directly on the Foldit site so players like me can read them?
This seemed to be the norm in the past when Foldit had Science
and Developer Chats.

Thanks!

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