2019.03.05 Dev Chat
everyday847: hey i’m andy watkins, i designed a lot of the ribosome challenges (to be fair, omei provided about 99% of the actual eterna infrastructure know-how, so all credit there) [5:00 PM]
everyday847: whatever the “community chat” is, apparently it’s happening now? and maybe here? [5:00 PM]
everyday847: so i’m excited to talk about those in particular [5:00 PM]
whbob: hi 847 [5:00 PM]
DigitalEmbrace: Hi Andy! [5:00 PM]
everyday847: it’s really wild how many distinct solutions you’ve found [5:01 PM]
everyday847: i realize this is like, par for the course at this point, but still [5:01 PM]
fmdz: Hi there, it’s Feriel! [5:02 PM]
everyday847: hi feriel! [5:02 PM]
whbob: hi fm [5:02 PM]
mgotrik: i am also here [5:02 PM]
fmdz: I am s postdoc at das lab working mainly on OpenTB project [5:02 PM]
LFP6: I’ll likely be lurking :] [5:02 PM]
fmdz: Nice to meet you all [5:02 PM]
sarahsmaga: hi all - i’m a project coordinator at C-GEM, where we’re interested in re-engineering the ribosome to produce cool new materials [5:03 PM]
whbob: same here [5:03 PM]
JR: I’m here [5:04 PM]
whbob: hi JR [5:04 PM]
hwaymentsteele: Hi all! I’m Hannah, I’m a phd student in the das lab working on improving secondary structure prediction. [5:06 PM]
Astromon: o/ [5:04 PM]
whbob: hi hway [5:05 PM]
hoglahoo: wow. is this an organized event or some divine coincidence [5:05 PM]
whbob: haha [5:05 PM]
Astromon: haha [5:05 PM]
everyday847: we were told it was an organized event [5:05 PM]
everyday847: and thus it became one [5:05 PM]
hoglahoo: clever planning [5:05 PM]
whbob: hey astro, hog [5:05 PM]
hoglahoo: hey whbob [5:06 PM]
Astromon: heya bob [5:06 PM]
everyday847: is there anything in particular that remains unclear about the ribosome puzzles? i.e., how eterna-style optimization fits into our big picture, how we’re gonna test stuff, etc.? [5:06 PM]
Astromon: Hi hanna nice to meet you [5:06 PM]
everyday847: i can imagine that there’s all kinds of different slices of data or utilities or… idk, something that would make these very daunting puzzles more “digestible” [5:06 PM]
JR: how are you going to select [5:07 PM]
whbob: yes, if ribosomes are mostly constrained, what is the best method to limit our search for sites that mutate? [5:07 PM]
everyday847: @JR: so my understanding of our plans is: once you all have been warmed up with these pilot challenges, we are going to show you some challenges that are more constrained, so there will naturally be fewer solutions. [5:08 PM]
whbob: is 16s the best area? [5:08 PM]
everyday847: @JR: we might implement some of those constraints with locks, or we might instead have eternascript implemented scoring functions that will also help stratify good and bad solutions [5:08 PM]
everyday847: @JR: i.e., you mutated a base that’s 100% conserved across every ribosome ever found, and also we can see it making a pretty tertiary contact, you get a big penalty [5:09 PM]
JR: any idea how much configuration plays into this [5:09 PM]
everyday847: @whbob: this might be a bit of personal bias, but i don’t think we should limit ourselves too much. if it isn’t violating a lock, and if we haven’t implemented a special scoring function to penalize it, change it! [5:10 PM]
Astromon: good thing is you can try a less contrained made sequence and paste it into a more constained one and some of the moves will still work [5:10 PM]
everyday847: @whbob: not sure what you mean by “is 16s the best area” – we’d love to see designs for all three components [5:11 PM]
everyday847: @JR: what do you mean by configuration exactly [5:11 PM]
everyday847: @Astromon: that’s a really weird ‘feature’, yea [5:11 PM]
rhiju: hi everyone sorry i’m late… just catching up with the conversation so far [5:11 PM]
rhiju: thanks everyone for coming [5:11 PM]
everyday847: (also, I realize I’m monopolizing the chat a little; do please feel free to cross-talk!! i can keep track) [5:12 PM]
Omei: Hi, all. [5:12 PM]
Astromon: Hi rhiju! [5:12 PM]
JR: a 3D view = configuration [5:13 PM]
rhiju: @JR to answer your question about how we are going to select, we won’t select [5:13 PM]
rhiju: we’re going to ask you to vote. =) [5:13 PM]
whbob: I’ve read that the 23s 5s rRNA large subunit is highly constrained. That leaves the 16s, but maybe not? [5:13 PM]
rhiju: @whbob 16s also has strong apparent constraints [5:13 PM]
rhiju: that’s based on what we see in Nature for ribosomes across Life [5:14 PM]
whbob: hi rhiju, Omei [5:14 PM]
rhiju: however, its worth noting that ribosomes in living cells may not be the absolute best guide for our puzzles [5:14 PM]
rhiju: for two reasons [5:14 PM]
Astromon: the trhee new 16s puzzles has more constaints than the first set [5:15 PM]
rhiju: ribosomes in living cells may get ‘help’ while folding from other stuff inside cells; we kind of want ribosomes that can fold better on their own. [5:15 PM]
Astromon: ah [5:15 PM]
everyday847: @Astromon: yeah, those are based off apparent constraints from what bases are also making tertiary contacts [5:16 PM]
Astromon: protein binding sites? [5:16 PM]
rhiju: on the other hand, ribosomes in living cells have to do a lot, including translate proteins that need to be inserted into membranes; we’re looking for molecules that could be more specialized to make particular proteins [5:16 PM]
everyday847: @Astromon: yes, and also just RNA-RNA tertiary contacts [5:16 PM]
Astromon: ah ok, ty [5:17 PM]
rhiju: one question for @astromon @whbob and others is how best to make the information available in the puzzles on what we know about Natural ribosomes [5:17 PM]
rhiju: without forcing players to stick to them [5:17 PM]
rhiju: I know Eli has been leading a conversation in getsat [5:17 PM]
rhiju: and omei of course [5:17 PM]
rhiju: https://getsatisfaction.com/eternagame/topics/ribosome-pilot-challenge [5:17 PM]
Astromon: yeah thats a great read [5:17 PM]
whbob: I think it was JR who has mutated one side of a GC pair at the end of a stem with amazing results. [5:18 PM]
rhiju: that sounds super interesting [5:18 PM]
rhiju: can you point me to that result? [5:18 PM]
Astromon: getting that booster that marks moves would be good as an ect gui booster [5:19 PM]
whbob: I think many of his solutions just concentrated onthe closing pair of hairpins. [5:20 PM]
rhiju: help me understand @astromon – can you post a link to a puzzle that has that booster? [5:20 PM]
Astromon: ill have to find and load it shouldnt bea prob [5:20 PM]
JR: can you list the info you want to dispaly in the puzzle [5:20 PM]
DigitalEmbrace: So we aren’t necessarily trying to improve the E. Coli ribosomes but rather to create new ribosomes? [5:21 PM]
Omei: I think https://eternagame.org/web/script/9262099/ is the booster @astromon is referring to. [5:21 PM]
spvincent: @rhiju: I don’t quite follow what’s meant by “molecules that could be more specialized to make particular proteins”. Can’t you just code the particular protein in mRNA? [5:21 PM]
Gerry Smith: …and if new ribosomes, is that why we dont have to pay attention to matching forms? [5:22 PM]
rhiju: @digitalembrace good question. we are actually hoping to get ribosomal RNAs that will work when they are mixed with ribosomal proteins from E. coli, so in that sense we want better E. coli ribosomes [5:22 PM]
Omei: Right now, you need to make your own copy, as usual. But if there’s sufficient demand, we can add it to the curated booster list. [5:23 PM]
Astromon: ill do that after the chat , i dont want to miss anything [5:23 PM]
rhiju: but we’d probably be OK if the ribosomes were only able to make proteins that were, say, 100 amino acids long and did not have to be inserted into membranes [5:23 PM]
Brourd: @Rhiju something to consider (if the goal is to make a stable ribosome), could be to have a small number of submissions be based on players redesigning all of the helices for like 3-4 submissions and only conserve loop and tertiary contacts. [5:23 PM]
rhiju: eventually we’d like to make weird polymers besides proteins [5:24 PM]
rhiju: it would help to have a ribosome that more stably and effectively assembles when we synthesize it in vitro with its proteins around [5:24 PM]
rhiju: @omei @astromon would be helpful if you could find a puzzle where that booster is in action [5:25 PM]
Omei: It works for any puzzle. [5:25 PM]
everyday847: @spvincent imagine that (this is an arbitrary example) you could make synthetic insulin 100x more effectively if you had a ribosome that was REALLY GOOD at making insulin and awful at making anything else [5:25 PM]
rhiju: @JR when you asked about ‘list the info’, was that a quesrtion for devs or players? [5:25 PM]
Omei: I’ll pick one. [5:25 PM]
everyday847: @spvincent natural ribosomes don’t have the privilege of making just one thing really well; they have to be pretty good at a bajillion tasks [5:25 PM]
everyday847: @spvincent while specialization towards any particular substrate is a little far off, if we can hyper-stabilize the ribosome then phase two can be engineering in hyper-specialized behavior [5:26 PM]
Omei: Say https://eternagame.org/game/puzzle/9169766/ [5:26 PM]
rhiju: @spvincent, hope i answered your question – we will indeed be using different mRNAs for different proteins, but we will be focusing on soe proteins (and also non-protein polymers) where having a robust ribosome variant that works well in vitro [5:27 PM]
rhiju: *would be powerful [5:27 PM]
spvincent: ok, tx [5:27 PM]
Omei: Load booster 9262099 from with thew script executor [5:27 PM]
rhiju: @everyday847 thanks [5:27 PM]
Omei: Respond positively (Done, I think) to the prompt and start mutating bases. [5:28 PM]
xkennag: astro you there [5:28 PM]
Astromon: ok i got the booster loaded and 5s open [5:28 PM]
jandersonlee: OK vs Cancel @Omei [5:28 PM]
Omei: Thanks. [5:29 PM]
Omei: (OK is what you want.) [5:29 PM]
rhiju: @omei i am getting ‘access denied’ [5:30 PM]
jandersonlee: And Cancel when you want to stop following the mutations. [5:30 PM]
whbob: so we want to get (find) an area within our ribosome that can be used to paste in a protein codon and let it do it’s stuff? [5:30 PM]
spvincent: Any thoughts of ultimately extending the amino acid set beyond the normal 20? [5:30 PM]
Omei: Oh. Of course. Duh. [5:30 PM]
LFP6: You need to create your own copy of the script [5:30 PM]
Astromon: @riju http://prntscr.com/mtrxep [5:30 PM]
Astromon: .tools [5:30 PM]
LinkBot: https://docs.google.com/document/d/1W6OiK8EA8XPllKANABI8ujcskdc62ylHDP99F-bjn-c/edit?usp=sharing [5:30 PM]
Omei: That’s the way we “protect” players from arbitrary boosters. [5:30 PM]
Astromon: this link explains how to load boosters [5:30 PM]
LFP6: (Re: We need a massive revamp of scripts) [5:31 PM]
rhiju: @astromon got it. screenshot helped. makes sense. we should have this on by default [5:31 PM]
jandersonlee: That’s why a cutated version would help. [5:31 PM]
jandersonlee: curated [5:31 PM]
rhiju: @everyday847 shall we go through the rest of questions. [5:31 PM]
Astromon: I agree rhiju [5:31 PM]
everyday847: yup! [5:31 PM]
everyday847: @spvincent: that’s exactly one of the goals. [5:32 PM]
Astromon: i agree with JL too [5:32 PM]
spvincent: cool! [5:32 PM]
everyday847: we’d also like to go beyond amino acids themselves – there’s a lot of evidence that the main function of the ribosome isn’t doing any chemically specific catalysis, but rather just brings reagents close together and that’s enough [5:32 PM]
everyday847: (i can go through the chemistry that makes that make sense if you’re interested) [5:32 PM]
rhiju: question for all of you here, re: boosters and feature requests for ribosome puzzles [5:33 PM]
spvincent: My organic chemistry is a little rusty I’m afraid [5:33 PM]
JR: Can you post that to the forum 847 [5:33 PM]
everyday847: so in principle the ribosome could catalyze any (-ish) chemistry to make any (-ish) polymer, but nature has only ever needed it to do one thing, so “in principle” is doing a LOT of work [5:33 PM]
rhiju: suppose we wanted to convey to players which bases are least mutated in Nature [5:33 PM]
everyday847: sure, where? in that big get-sat thread? [5:33 PM]
rhiju: but without locking them [5:33 PM]
rhiju: what’s the best way to do that? [5:34 PM]
JR: do a new one [5:34 PM]
Astromon: what are reagents [5:34 PM]
spvincent: great: somewhere where it could be read at leisure [5:34 PM]
rhiju: like should we have another constraint box? [5:34 PM]
Omei: For players: everyday847 is Andy Watkins, researcher at CGEM [5:34 PM]
jandersonlee: Colored marks would help, but the UI only supports one color now I think. [5:34 PM]
Astromon: yes color marks! [5:35 PM]
rhiju: @jandersonlee have you seen any other way? like via eternascript? [5:35 PM]
jandersonlee: @rhiju I’ve not seen it in Eterna, no [5:35 PM]
rhiju: ok if @astromon says color marks with an exclamation point then we better take it seriously [5:35 PM]
spvincent: Reagents: the chemicals in a reaction [5:35 PM]
rhiju: @LFP6 how hard would it be to activate colored marks over bases [5:36 PM]
Omei: @LFP6 could probably add a color option to the booster API pretty quickly, and then players could figure out the UI aspect. [5:36 PM]
Astromon: cool, tnx [5:36 PM]
rhiju: @omei love that idea [5:36 PM]
Astromon: a dif color for binding spots perhaps [5:36 PM]
Astromon: protein binding* [5:36 PM]
whbob: one of our icon boxes is where we click the icon and it highlights specific areas of the sequence to show where it does not meet the constraints [5:37 PM]
rhiju: there might be lots of disinct information to show – protein binding, parts involved in rna-rna tertiary contacts, [5:37 PM]
rhiju: nucleotides that are conserved in bacteria [5:37 PM]
rhiju: eli fisker’s suggestions (see getsat thread) [5:37 PM]
whbob: Could that be used to show sequence locations? [5:37 PM]
LFP6: Looking into the options… [5:37 PM]
rhiju: so maybe there could be different constraint boxes setting target values [5:38 PM]
rhiju: and if you click on one it would create some coloring of the nucleotides to guide you to what it thinks is mutatable vs. not [5:38 PM]
Astromon: sounds great [5:38 PM]
rhiju: did you all know that if you click the constraint box for nature vs. target you can get red coloring to show up on the main secondary structure? [5:39 PM]
rhiju: i only (re)discovered that recently [5:39 PM]
rhiju: totally forgot about it [5:39 PM]
Omei: Me too. [5:39 PM]
DigitalEmbrace: So have a box we can click to show different pieces of information, like rhiju us saying, that would be great. [5:39 PM]
jandersonlee: I use that a lot [5:39 PM]
whbob: that’s what I was talking about [5:40 PM]
LFP6: I think adjusting the API for setting colored marks might be doable [5:40 PM]
rhiju: but i really like @omei’s idea to expand the booster API so that some of you can create your own things – then share screenshots via getsat or slack to pitch to each other or to devs [5:40 PM]
LFP6: At least as a stopgap [5:40 PM]
LFP6: Would of course mean overriding any existing marks with the new color [5:40 PM]
rhiju: @lfp6 NICE [5:40 PM]
JR: may be a stupid question but is ribosome-protein binding like space ship docking - trying to give me a picture [5:40 PM]
Omei: I think it would be helpful to have a display of the ratio of energy reduction and number of mutations. [5:41 PM]
rhiju: @JR i don’t think its super deeply understood. some proteins probably do dock like a spaceship onto the RNA [5:41 PM]
Omei: … kind of a rough guide to the tradeoff between the two. [5:41 PM]
rhiju: some other ones bind in stages, waiting for the RNA to switch its structure before completing the seal [5:41 PM]
Astromon: (: [5:41 PM]
rhiju: ribosome assembly in vivo is just getting understood. still lots of mysteries [5:42 PM]
LFP6: Yeah, I think API for colored bases should be doable [5:42 PM]
rhiju: example of a cool paper on how ribosomes come together in e. coli: https://www.cell.com/cell/pdf/S0092-8674(16)31592-6.pdf [5:42 PM]
jandersonlee: Muti color at once, or just one color at a time? [5:42 PM]
LFP6: Long as there’s no stupid interoperability things I don’t notice, which is always a catch 22 lol [5:42 PM]
JR: thx [5:42 PM]
LFP6: @jandersonlee Only one color [5:43 PM]
rhiju: @LFP6 I think enabling players to creste color boosters would be transformative. [5:43 PM]
LFP6: Not sure how multicolored markings would work [5:43 PM]
jandersonlee: Multi color [5:43 PM]
LFP6: “Colors other than black” [5:43 PM]
whbob: what about shapes rather than colors? squares, diamonds etc [5:44 PM]
jandersonlee: Doesn’t help as much if it removes/recolors the black marks. [5:44 PM]
LFP6: Would be neat to actually change the base colors, but I’m not that adventurous lol [5:44 PM]
Omei: @jandersonlee That’s all I was thinking would be easy when I suggested it. [5:44 PM]
rhiju: even if only single color, if they could be turned on/off by player, we could see the different kinds of information [5:44 PM]
LFP6: (I don’t think there’s a good way to do that anyways) [5:44 PM]
rhiju: @lfp6 there’s presumably a way to change base colors, since that’s what we used to do for SHAPE data [5:45 PM]
LFP6: I think they’re still separate bitmaps [5:45 PM]
LFP6: Not sure though [5:45 PM]
rhiju: aha [5:45 PM]
LFP6: At any rate, as I said I’m not adventurous enough to touch that area of the code [5:46 PM]
LFP6: Actually no, you [5:46 PM]
LFP6: you’d be right [5:46 PM]
jandersonlee: 0=Normal, 1=Black, 2=Pink, 3=Red? [5:46 PM]
LFP6: Since we have a gradient [5:46 PM]
rhiju: oooh gradient [5:46 PM]
rhiju: just want to echo @whbob’s comment, which captures what’s needed succinctly: “whbob: one of our icon boxes is where we click the icon and it highlights specific areas of the sequence to show where it does not meet the constraints [2:37 PM]” [5:47 PM]
LFP6: Yeah I don’t want to get into that bit of code lol [5:47 PM]
rhiju: ok better answer some other questions now [5:47 PM]
LFP6: Just looked at it [5:47 PM]
rhiju: or listen to more feature requests [5:47 PM]
rhiju: any more questions/suggestions for ribosome puzzles? [5:47 PM]
rhiju: @brourd your idea to only mutaste helices is an interesting one [5:48 PM]
rhiju: we could imagine having a helix only icon which counts up number of non-helix mutations and goes red when that number is above your favorite value. [5:49 PM]
rhiju: and if player clicks on it the RNA is colored by helix vs. loop [5:49 PM]
jandersonlee: target helix? [5:49 PM]
rhiju: i feel like a lot of the ideas in this thread could be instantiated through color boosters, invented by players [5:49 PM]
jandersonlee: (in natural mode?) [5:49 PM]
DigitalEmbrace: I think we would appreciate any background material being posted in a Forum thread where we can read more about the science and discuss it. [5:50 PM]
rhiju: @everyday847 any progress on that getsat thread? [5:50 PM]
rhiju: if you’re stuck, i bet @lfp6 could set it up [5:50 PM]
hi1000: hai [5:50 PM]
jandersonlee: Having scripted access to the outline marker might be an alternative to colored marks [5:51 PM]
hi1000: i need to do some homework [5:51 PM]
JR: He doesn’t have to do it now, i’m going to have to research the parts anyway [5:51 PM]
Astromon: hey hi [5:51 PM]
rhiju: @jandersonlee yes i think @LFP6 will look into that – of course as you pointed out, might be nice to not override outline marker with coloring. [5:51 PM]
JR: to get a better understanding of what goes on. [5:52 PM]
LFP6: @jandersonlee Base marker would be far easier [5:52 PM]
whbob: I made a spreadsheet of the 5s open puzzle last week & put in the the base number/ type of the submissions. Next I want to color code the mutations to show if they are in loops or stems etc. [5:52 PM]
everyday847: FYI: https://getsatisfaction.com/eternagame/topics/redesigning-the-ribosome-chemical-goals?rfm=1&topic_submit=true [5:52 PM]
rhiju: @whbob sounds great [5:52 PM]
everyday847: @rhiju i was writing a lot, unsurprisingly, so i wasn’t alt-tabbing [5:53 PM]
rhiju: for the current ribosome puzzles, what do folks think about changing the usual ‘win condition’ of stabilizing the predicted structure (‘nature mode’) to match the target structure [5:54 PM]
rhiju: instead what if we just ask that the difference in energy is lower than some gap? [5:54 PM]
sarahsmaga: @DigitalEmbrace, 847, others: C-GEM has a blog, and we’d like it to be a resource for EteRNA players. If there’s some background info you had trouble finding or isn’t very clear, let me know sarah@gem-net.net and we can write something up! [5:54 PM]
Omei: Super!!! Thanks. [5:55 PM]
rhiju: @sarahsmaga could you post that info also in the forum thread that andy just set up? [5:55 PM]
rhiju: alright everyone this has been super-helpful already [5:55 PM]
rhiju: 5 mins left [5:55 PM]
JR: change “win” to whatever you think is best for me [5:55 PM]
rhiju: @JR thanks [5:55 PM]
sarahsmaga: previous blog posts summarized developments in the literature, may be of interest http://gem-net.net/blog/ [5:55 PM]
whbob: @rhiju: yes to target structure match. After feedback maybe we will need to adjust, but I like that for now. [5:56 PM]
rhiju: @whbob thanks [5:56 PM]
rhiju: everyone, we also have an interesting list of questions for you about updating the puzzle-maker [5:56 PM]
rhiju: but we’re running out of time [5:56 PM]
rhiju: so we’ll probably have that in a later community chat [5:57 PM]
rhiju: its been nice to focus on ribosome puzzles – as you can see, this is a totally unprecedented challenge [5:57 PM]
Astromon: sounds good, i’ll look out for it< [5:57 PM]
rhiju: yea, we’re going to be returning to switches after ribosomes, and we suspect that we’ll make even more rapid progress if we can get players to create more elaborate lab-like puzzles for switches [5:58 PM]
Astromon: ideed! very exiting project [5:58 PM]
mgotrik: yes ^ [5:58 PM]
rhiju: and maybe after that we will make riboswitch ribosomes [5:59 PM]
Astromon: Great! [5:59 PM]
rhiju: so we can turn on and off ribosomes and make them do arithmetic and stuff [5:59 PM]
rhiju: in addition to being little factories [5:59 PM]
whbob: very cool [6:00 PM]
rhiju: the science fiction future is not that far off! [6:00 PM]
rhiju: thanks everyone for coming – see you in the game [6:00 PM]
Gerry Smith: thank you [6:00 PM]
whbob: ty rhiju [6:00 PM]
Astromon: Thanks! [6:00 PM]
rhiju: everyone, we also have an interesting list of questions for you about updating the puzzle-maker [5:56 PM]
rhiju: but we’re running out of time [5:56 PM]
rhiju: so we’ll probably have that in a later community chat [5:57 PM]
rhiju: its been nice to focus on ribosome puzzles – as you can see, this is a totally unprecedented challenge [5:57 PM]
Astromon: sounds good, i’ll look out for it< [5:57 PM]
rhiju: yea, we’re going to be returning to switches after ribosomes, and we suspect that we’ll make even more rapid progress if we can get players to create more elaborate lab-like puzzles for switches [5:58 PM]
Astromon: ideed! very exiting project [5:58 PM]
mgotrik: yes ^ [5:58 PM]
rhiju: and maybe after that we will make riboswitch ribosomes [5:59 PM]
Astromon: Great! [5:59 PM]
rhiju: so we can turn on and off ribosomes and make them do arithmetic and stuff [5:59 PM]
rhiju: in addition to being little factories [5:59 PM]
whbob: very cool [6:00 PM]
rhiju: the science fiction future is not that far off! [6:00 PM]
rhiju: thanks everyone for coming – see you in the game [6:00 PM]
Gerry Smith: thank you [6:00 PM]
whbob: ty rhiju [6:00 PM]
Astromon: Thanks! [6:00 PM]
