Developer Chat

betahelix Hi everyone, we'll be starting our Scientist Chat in a few minutes...
auntdeen2 yes - correct, marie :)
timoteusirc bad timing for brow
auntdeen2 hi beta
drumpeterIRC cool. this will be a fun chat.
betahelix Hi auntdeen! Hi gang, thank you all for coming today!
betahelix (we can get started 1 minute early for once )
auntdeen2 o.o
frood66 last chance for ice cream guys
auntdeen2 setting a bad precedent, lol
cassandraberry81897 ill take some ice cream frood
betahelix bkoep is here as well, you might remember him as bk from a previous chat:
auntdeen2 hi bk
betahelix Are you changing usernames for every chat, bkoep, or is this the one? ;)
bkoep haha, I'm sticking with bkoep
bkoep since it's my Foldit username
betahelix On that note... why don't you take it away with some Foldit design results!
bkoep hello everyone
cassandraberry81897 hi bkoep
bkoep so for the people who don't know me, I've been analyzing all your solutions from the design puzzles
bkoep I definitely have the coolest job at the Baker lab
bkoep a lot of the designs you have come up with are really cool, and we can't wait to try some of them out in the lab
tokens have you analyzed the last design puzzle yet?
drumpeterIRC good job at fixing the sticky wiggle
bkoep @tokens yes, the last trimer puzzle has had some of the best results we've seen yet
tokens nice
betahelix @drumpeter we are very happy that the bug was finally found and the fix works!
katfish You guys really started on time! Here now! :)
bkoep but there are still little issues here and there that we are trying to work on
bkoep protein design is hard
spvincent Are these the same puzzles that are appearing inR@h?
drumpeterIRC hey katfish
betahelix Hi katfish!
katfish Hello hello!
bkoep @spvincent yes, Rosetta@home is our main tool for analyzing foldit design solutions
BletchleyParkirc What happened to the designs for the flu and sepsis puzzles ?
betahelix @BP I believe there is an upcoming Foldit podcast about the Flu results, right katfish?
bkoep over the past few weeks, anyone who has been running Rosetta@home has been watching your design solutions fold up from extended peptide chains
katfish Yes indeed there is, beta
katfish BP, we'll have that podcast up in the next week
BletchleyParkirc That os for flu,ok. Sepsis succesful ?
bkoep for those of you that are curious, the reason it takes us so long to analyze your design solutions is that we have to make sure there are no other protein folds for your designed amino acid sequences
bkoep we send your design sequences to Rosetta@home, which makes 100,000s of predictions about possible folds for each sequence
bkoep it gives us back results that look like this:
tokens that looks like a succesful fold, that picture
bkoep every red dot is a prediction by Rosetta@home
bkoep the y-axis is the inverse of the foldit score (so lower points are better scoring structures)
TimovdL And the blob on the left are the likely stable positions?
betahelix @BP It seems like we are running into some hurdles with Sepsis... not only because it's obviously a very difficult problem, but our collaborators (who perform the experimental part of that project) have sadly run into funding issues
we haven't given up on it quite yet... but it is making things a lot more difficult!
bkoep and the x-axis is similarity to design (points on the left are close to the design)
timoteusirc is there any unsuccesful fold pic from rosetta for comparison
bkoep Yes, the green blob on the left is where we perturb the design structure to make sure it doesn't unfold immediately
bkoep those results, by the way, are for a design from puzzle 735 Player-Solution Loop Redesign: Round 2
timoteusirc fascinating
BigMaxPorter wow
bkoep when we looked at that solution here, we noticed there was one problem with the design (actually it was a problem in a lot of solutions for that puzzle)
betahelix @timoteus for example, an unsuccessful fold would be if ALL the lowest red dots were on the right side of the figure
TimovdL That it is unlikely that it will fold in the designed structure?
bkoep we made one mutation to fix that problem, and the new Rosetta@home predictions looked like this:
timoteusirc oh, i can see that pattern now
BigMaxPorter this game is harder than i thaught
spvincent Are the design solutions still too 'regular'? Foldit solutions tend to have very straight sheets, while real world sheeys in proteins seem to have all kinds of warps and kinks in them.
bkoep this graph looks better to us because the Rosetta@home predictions (red dots) reach farther left (closer to design), and the best scoring structures are close to the left
tokens can you explain what the problem was?
karstenw big max, you can get help in puzzle chat
betahelix BigMax is no longer with us...
bkoep @tokens yes, there was a residue that had a really bad backbone
bkoep but it was scoring well in foldit because of another term in the foldit score function
tokens which term? I'm just curious.
TimovdL So maybe that that other term should only count if the backbone is not too bad?
bkoep @spvincent, "regular" is not necessarily bad for us in protein design
bkoep "regular" often means "very stable"
bkoep one of the reasons proteins in nature are less "regular" is because they don't need to be "very stable"
frood66 Ahh - this is troublesome
bkoep they just need to be stable enough to function for a bit, and then they can be degraded by the body when they need to be
drumpeterIRC im back
spvincent so long helices aren't necessarily bad in design puzzles?
bkoep in design, we want "very stable" proteins, because that's the most difficult part to design
bkoep the idea is we can make very stable proteins less stable if we need to
mimi2 so really "unnatural" shapes mght be good?
drumpeterIRC have we talked about the trimer?
bkoep @spvincent, long helices are not bad from a stability perspective, but they're not as interesting to us
drumpeterIRC i use sheets
bkoep @mimi2, yes unnatural shapes are very good
betahelix Although David Baker still enjoys helical bundles :)
drumpeterIRC haha
bkoep in fact, we hope that we can use design to come up with proteins that are not similar to any proteins found in nature
tokens I noticed in the last trimer that there were some high scoring designs that didn't seem to have an individual core
bkoep @TimovdL, that's one way to address the problem with that score term
tokens In the sense that to get the top score you shouldn't focus on having a core in the monomer. What do you think of that?
bkoep For now, we've just down-weighted that score term, and I haven't seen similar problems yet in the more recent trimer designs
TimovdL If you want us to design individual cores, the score function should reflect that
bkoep @tokens and @TimovdL, multimer designs without individual cores can exist in nature
bkoep but they are difficult for us to model in the Baker lab
karstenw i'm pretty sure i don't know a core from my elbow, just sayin'.
bkoep so you're right, we prefer to see multimer designs with individual cores
spmm agree karsten
bkoep @Timo we'd like to include that in the score function, but there are some problems with Core Exists filter as it is right now, and we need to fix that first
TimovdL You answered my next remark before I could make it
marie_s a core is a sign that the protein alone is compact?
spmm so like 3 cytochrome type shapes together?
bkoep @karstenw and @spmm and @marie_s, when I talk about protein cores, I just mean that stable proteins in nature are "globular" and compact in shape because they have hydrophobic residues that they like to shield from water
bkoep so they fold up in a way that the hydrophobic residues are internal, or in the "core" of the protein
Kartourus UGH >.< this cytochrome is really irritating me. i have one open space and it will have no flaws >.<
bkoep protein designs without a core of hydrophobic residues are not likely to fold up into stable structures
karstenw got it, thanks
karstenw my best "cores" tend to score poorly, but if i find a good one, that is the objective?
alicia101 hi
bkoep @karstenw, yes, a well-packed hydrophobic core is essential to a good protein design
drumpeterIRC i like to think of the core where the hydrophobics go
bkoep unfortunately, it is difficult for us to score a hydrophobic core
bkoep (that's why we have to use the Core Exists filter)
betahelix (difficult to score it quickly)
spmm and also hard to visualise for us - quickly unlike hydrogen bond?
drumpeterIRC yes. the filters show what segs should be in the core, right?
phi16 can you speak a little more about why it is difficult to score a hydrophobic core?
bkoep so I would encourage you, if you have a design that's scoring poorly but it has a nice hydrophobic core, use the Share with Scientists button
karstenw right, ok.
TimovdL Cant there be a score term for the hydrophobics that gives points if it is not in open space? (and negative if not?)
spmm it is the grouping of phobes rather than individuals perhaps in the core
bkoep @Timo, curiously enough, that's the score term that caused the problem in solutions for puzzle 735
karstenw oh, can the core be shared among the mirrored proteins, or better to make the live protein have its own core?
bkoep (the score function favored an arginine because it was in an exposed position, but the backbone was only compatible with a glycine)
bkoep @karstenw, yes the core can be shared (and should be shared) between symmetric chains in symmetry puzzles
karstenw ty
betahelix any other design questions for bkoep?
spvincent What is the purpose(s) of designing these proteins that don't exist in nature?
TimovdL So for symetric you want individual cores and shared cores?
karstenw hmmm...this will take some planning
drumpeterIRC good question timo
timoteusirc this might be dump question, but the process in fold it is like: we fold it -> you send it to rosetta -> rosetta gets results -> you create that protein in lab if it folds correctly?
bkoep @spvincent, eventually we'd like to design proteins for new therapeutics, new materials, new science tools
tokens Are you still working on fixing the scoring function so that other amino acids won't be accepted when the backbone bends in a way that only glycine makes sense?
betahelix @tokens yes we are!
phi16 yes, likely 4.5 billion years, @kars
bkoep i.e. things that don't already exist
betahelix @phi16 well the designs that you are all coming back with are already very impressive!
bkoep @timoteus, that is exactly the pipeline we have in place right now
spmm does the science know all of the proteins now ?
betahelix Every time David Baker sees your designs he always gets super-excited and wants to test them all!
bkoep with some added human-analysis, of course
betahelix and he is always amazed at what all of you are able to accomplish starting just from a symmetric extended chain
drumpeterIRC i was going to ask if yall are looking for a mix of helices and sheets.
bkoep @spmm, no, there are many proteins we don't know the structure for
bkoep but there is a database of all known protein structures (the PDB)
bkoep and there has not been a new topology (type of fold) submitted to the PDB since 2009
karstenw wow
timoteusirc this probably have nothing to do with foldit but do you use x-rays in proteing design? guys from chemistry borroved our (physics lab) x-ray gear few weeks ago, or are they just toying with it?
tokens proteins are solved via xray crystallography
betahelix @drumpeter any interesting/new topologies are always very exciting (a mix of helices & sheet are always welcome)
spmm so even if the (designed) protein existed in nature it may just not have been seen (ie be in the pdb) so that is still useful?
TimovdL Where can I find a pointer to what is considered a topologie for proteins?
bkoep @spmm, yes, all successfully designed proteins are useful, because they demonstrate that we understand the physics behind protein folding
bkoep but new designs are especially exciting
spmm :)
betahelix any last questions about design?
drumpeterIRC may i take a pic to see if my trimer is considered interesting?
phi16 do you consider Anfinsen's dogma still true?
bkoep @drumpeterIRC, it's best to use the Share with Scientist button
betahelix @phi16 if not, we're in big trouble!
drumpeterIRC thanks. i would hate to over use it. ive never actually used it
betahelix I want to move to the next topic, which is that I will be starting a faculty position in the CIS Department at the University of Massachusetts Dartmouth this fall
drumpeterIRC congrats
katfish Yes, a big congrats beta!
spmm oh dont liek the sound of this
spvincent congratulations! Will you still be involved with FoldIt?
auntdeen2 congrats beta
auntdeen2 how will this affect the game?
karstenw congrats, very exciting
cassandraberry81897 congrats beta
betahelix Thank you! And yes, my lab will focus on extending the Foldit framework beyond structure prediction by making it part of the experimental process for solving protein structures!
spmm woot!
spvincent Nice!
timoteusirc congrats!
timoteusirc that sound amazing
BletchleyParkirc very good, I hope it will incorporate all the positive sides of foldit. Congratulations.
TimovdL Congrats, and we will miss you
betahelix So I'm not leaving you, on the contrary, the idea is to bring more people in to work on Foldit!
TimovdL spoke to soon
betahelix :)
auntdeen2 so - how exactly will this work?
betahelix This does mean that I will be gone for the entire month of August, however, and will be starting up my lab in September.
betahelix I will hopefully bring in some grad students to work on the project, as well as postdocs, but obviously this will not be instantaneous!
betahelix (Unless any of you are looking for a postdoctoral position )
TimovdL So then there will be 2 labs working on and using Foldit?
betahelix Yes!
cassandraberry81897 thats awsome
timoteusirc too bad its not my field haha
TimovdL Too bad I am a bit too old for changing career now
BletchleyParkirc you never are Timo, if it is your passion
betahelix bkoep will be the lead scientist working on Foldit here in the Baker Lab (so feel free to PM him and katfish while I am away in August
gloverd Is there a name for your new lab?
betahelix Good question, I hadn't thought about that yet... but I should come up with a cool name... the Beta Helix lab might confuse some people there!
auntdeen2 they can deal with it
karstenw perhaps butcher labs?
gloverd Sounds good!
karstenw or candlestick maker?
betahelix As I mentioned, I'm interested in making Foldit part of the experimental process for solving protein structures, and in fact we're going to have an exciting puzzle up next month:
auntdeen2 very punny kars
betahelix The Unsolved Salmonella Bacteriophage puzzles that you have been working on have cryo-EM data available
timoteusirc oh, great! i think salmonella is my nemesis
TimovdL Will there be another shift in focus on foldit?
betahelix So we're going to give you that puzzle with Cryo-electron microscopy data as an Electron Density puzzle.
TimovdL YES
cassandraberry81897 that sounds like fun
karstenw ooh
mimi2 maschochist cass?
spmm cool can you give us a diagram of where the repeats are in the density - a block model of the units
cassandraberry81897 whats that word mean mimi?
mimi2 you like hurting yourself
betahelix The interesting aspect about Cryo-electron microscopy data is that we can easily trim it to one copy (unlike crystallography data)
cassandraberry81897 ..nah i just like chalenges
betahelix so it "should" be easier than the recent Electron Density puzzles in that regard
TimovdL Easy for us karsten
frood66 oh godd
auntdeen2 :)
frood66 *good
betahelix Anyway... I'm very excited about this puzzle, and hopefully it'll be a lot of fun!
mimi2 we'll look forward to it
cassandraberry81897 i cant wait
karstenw i already have ideas zooming through my head. can't wait to see that ed
marie_s can we work a little more on anemia?
betahelix Thank you all very much for joining today's Scientist Chat... I'll stick around if there are any other questions!
spmm excellent - I saw some dodgy proteins with very high scores in the last two rounds
auntdeen2 beta? what will be the structure now of the game in terms of who is responsible for what?
spmm congratulations and have a great break beta - moving house and all that
auntdeen2 ie - if server crashes, who do we contact?
TimovdL I have posted a question in the blog about why idealize does not solve all negative scores
karstenw ooh any chance we can load old solutions to the new ed?
TimovdL (backbone)
katfish You can always PM me about server crashes or anything pressing
auntdeen2 thanks kat
phi16 Thank you, Beta. All the best.
betahelix @marie we are waiting on the Worcester Polytechnic Institute Scientists to finish their analysis of your first round of puzzles, so we can post the next round (if needed, of course
marie_s thx
spmm @beta will you drop in and say hi - and also post pics of your shiny new lab
betahelix Thanks everyone! yes, PMing katfish is always a great first step as she always knows who to contact (and she has our cell numbers
timoteusirc thank you beta, its been great to chat with you. learned lots of new stuff today. tho im not biochemist but its always good to learn new stuff
betahelix @karstenw great suggestion, you should totally be able to load in your previous solutions to the new cryo-EM puzzle
spmm welcome to the pointy end bkoep
karstenw yay!
betahelix Thank you, and thanks to bkoep for joining the Foldit family!
karstenw thanx bk and beta. bbl
TimovdL Tx for this informative cha
TimovdL chat
katfish Thanks again for coming out, everyone!
betahelix the transcript will be up later for the homies that couldn't make it to the party today
bkoep yes, thanks for a great chat
BletchleyParkirc bye !
betahelix Bye everyone, thanks again and keep up the great folding!
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