2020.03.18 Dev Chat

From Eterna Wiki

rhiju: ‎hello everyone! ‎[12:00 PM]
rhiju: ‎its all-hands-on-deck for community chat ‎[12:00 PM]
LFP6: ‎🎉 ‎[12:01 PM]
joshm17: ‎hi rhiju! ‎[12:01 PM]
rhiju: ‎To initiate the chat, I thought I’d lay out our scientific plans as we see it now. Things are changing hourly however. ‎[12:01 PM]
rhiju: ‎There are 4 ways we can contribute ‎[12:01 PM]
rhiju: ‎1. We can figure out principles to stabilize mRNA vaccines for deployment and protein expression. ‎[12:02 PM]
rhiju: ‎2. We can identify structured segments of the RNA that might be targeted for antiviral drugs ‎[12:02 PM]
rhiju: ‎3. We can try to predict how the RNA evolves by setting up puzzles that simulate structure/function under known evolutionary constraints ‎[12:03 PM]
rhiju: ‎4. We can teach each other, and engage the broader world, in thinking of additional ways to fight the virus ‎[12:03 PM]
lroppy: ‎–prresent ‎[12:04 PM]
lroppy: ‎and proud to be here ‎[12:04 PM]
rhiju: ‎In terms of #4, a lot of back-and-forth is happening at Stanford that is leading to further ideas on an almost daily basis ‎[12:04 PM]
rhiju: ‎So I want to make sure we don’t give such ‘mutual education’ initatives short shrift ‎[12:04 PM]
rhiju: ‎As an example, there are folks here who are trying to figure out the best segments of the virus to look for in the environment as affected individuals shed it. This would allow for tracking. ‎[12:05 PM]
rhiju: ‎They pointed out to me that the most stable fragments of shed virions are likely the ones with RNA structure – can we help identify those? ‎[12:05 PM]
rhiju: ‎So you’ll see why we are setting up new coronavirus-specific puzzles for the entire community that ideally should include links and videos about the science of the virus ‎[12:06 PM]
rhiju: ‎and the science behind potential medicines ‎[12:06 PM]
rhiju: ‎it is absolutely critical that we learn as much as possible about this pathogen and how to attack it ‎[12:06 PM]
rhiju: ‎and I suspect that our efforts as a community, if we are all deeply informed on the science, will even lead to new ways to take on the virus computationally and experimentally that no one has thought of yet. ‎[12:07 PM]
joshm17: ‎⚔ ‎[12:07 PM]
rhiju: ‎that’s the kind of bottom-up science that Eterna is unusually good at, even compared to other citizen science projects and other extraordinary ‘expert’ research efforts ‎[12:07 PM]
rhiju: ‎OK, that’s the setup. The floor is open for questions ‎[12:08 PM]
rhiju: ‎(As a quick aside, I have to take a phone call in about 22 minutes (9:30am pacific) and will be AFK at that time.) ‎[12:08 PM]
lroppy: ‎hi all ‎[12:09 PM]
rhiju: ‎OK, then I have questions for you ‎[12:10 PM]
lroppy: ‎have we thought about a push-back email ‎[12:10 PM]
lroppy: ‎to everyone who has ever played eterna ‎[12:10 PM]
lroppy: ‎trying to re-engage folks who fell away ‎[12:10 PM]
LFP6: ‎Hasn’t come up, but it’s theoretically doable ‎[12:10 PM]
rhiju: ‎Good idea lroppy! ‎[12:10 PM]
rhiju: ‎Lroppy, cammy, lfp6, and others – thanks for also beginning the crowdfunding initiative ‎[12:10 PM]
rhiju: ‎We have several professional devs ready to join the project to help with buildout, but need to raise funds to support them. ‎[12:11 PM]
joshm17: ‎Would it be more helpful to diversify what we’re trying to solve (#1-4) or focus on one particular puzzle? ‎[12:11 PM]
rhiju: ‎@joshm17 good question. For at least the next week, let’s diversify, and then regroup to see if we should focus. ‎[12:12 PM]
rhiju: ‎One major uncertainty is whether we will be permitted to do lab experiments and if so, how many. The situation is changing back and forth on literally an hourly basis. ‎[12:13 PM]
worseize: ‎When labs are gonna be open ? ‎[12:13 PM]
rhiju: ‎@worseize, great question. We are designing a first lab and hope to deploy later this week. ‎[12:13 PM]
rhiju: ‎IT will be focused on how a particularly promising RNA element is evolving, and whether its rate of evolution is sufficiently restricted that it might be worth going after as an antiviral target and as a segment for diagnostics. ‎[12:14 PM]
rhiju: ‎Those of you who are interested in helping design that puzzle, including associated background material to explain it, please stay tuned to the player developer slack ‎[12:14 PM]
joshm17: ‎@rhiju any thoughts on collaborating with friends at Foldit? ‎[12:15 PM]
rhiju: ‎worth reaching out to them – they are doing important stuff, but so far it is independent of what we can do on the RNA side of things. ‎[12:16 PM]
joshm17: ‎good point - maybe some of their players want to help though? :) ‎[12:16 PM]
LFP6: ‎It’d be a great crossover if we could pull it off ha ‎[12:16 PM]
LFP6: ‎I know they actively have COVID puzzles ‎[12:16 PM]
joshm17: ‎I’ll make a shoutout in their discord :) ‎[12:16 PM]
rhiju: ‎One possibility that @dosoonkim proposed is: if Foldit designs new protein therapeutics or protein vaccine antigens, we can design the mRNAs that code for those proteins and would enable rapid deployment ‎[12:16 PM]
LFP6: ‎ooo ‎[12:17 PM]
rhiju: ‎@joshm17 thanks, that sounds good ‎[12:17 PM]
rhiju: ‎here’s an example of an RNA element: https://www.biorxiv.org/content/10.1101/2020.03.13.991083v1 ‎[12:17 PM]
LFP6: ‎Wish I had thought to retweet them on our Twitter when they first announced ‎[12:17 PM]
rhiju: ‎I have been having some discussions with players about puzzles they’d like to see on a daily basis to ‘train’ ‎[12:19 PM]
rhiju: ‎there is curiosity abuot each of the RNA elements that are thought to be in the virus ‎[12:19 PM]
rhiju: ‎And how they might wrap up into a helical filament when packaged in the virion ‎[12:19 PM]
rhiju: ‎What are you curious about? ‎[12:19 PM]
joshm17: ‎so the question is what should daily training puzzles teach? ‎[12:21 PM]
rhiju: ‎Another question for players – there is a positive feedback loop between getting folks engaged in playing, getting more science done, and getting funding to get more science done by us and players ‎[12:22 PM]
rhiju: ‎what do you think are the most important steps developers can take to engage more players in our efforts? ‎[12:22 PM]
rhiju: ‎@joshm17 yes exactly what should daily training puzzles teach? ‎[12:22 PM]
rhiju: ‎I am preparing puzzles that focus on the various elements of the RNA, such as the 5’ UTR, which controls the translation and bizarre replication of the virus ‎[12:22 PM]
joshm17: ‎Well, what knowledge do expert players use regularly that novices don’t know? (I am a novice, so I don’t know the answer to this) ‎[12:23 PM]
rhiju: ‎as well as the frameshift element that the virus needs to have one sequence code for two proteins ‎[12:23 PM]
rhiju: ‎and there are a number of structured bits known in the so-called 3’ UTR, which in many cases have mysterious functions ‎[12:23 PM]
rhiju: ‎so those are important pieces of knowledge that expert scientists know that players don’t generally ‎[12:24 PM]
rhiju: ‎@joshm17 i like your question – what do expert players use? ‎[12:24 PM]
joshm17: ‎is the goal of training puzzles to teach players what scientists know, or to teach novice players what expert players know? two very different things ‎[12:24 PM]
rhiju: ‎i think both! ‎[12:25 PM]
lroppy: ‎have we thought about which of our energy models might be most helpful? ‎[12:25 PM]
Eli Fisker: ‎@Rhiju, what do you mean when you say “bizarre replication of the virus”. How is the corona virus different to “normal” viruses or normal replication? ‎[12:25 PM]
lroppy: ‎perhaps one aspect of training might be to narrow to whichever we think is best ‎[12:25 PM]
Omei: ‎Rhiju had to leave for his call. He’ll be back. ‎[12:26 PM]
lroppy: ‎also, will we be intentionally de-prioritizing other projects ‎[12:26 PM]
rhiju: ‎@lroppy good idea. my lab has strong evidence that Vienna (our default engine) is not the best. so we almost certainly need puzzles that teach about alternative folding engines ‎[12:26 PM]
rhiju: ‎before I step out (I’ll be back), let me bring up some ideas for what lab puzzles may look like: ‎[12:26 PM]
DigitalEmbrace: ‎So we want to design RNA that would potentially disengage those CV elements by binding with them? ‎[12:26 PM]
rhiju: ‎A. DEsign switches to put into mRNA vaccines that allow them to be turned off in patients who have a bad response ‎[12:27 PM]
rhiju: ‎B. Figure out how to design an mRNA vaccine that is highly structured in its coding region, and with lab feedback figure out what kinds of structures are optimal for getting the mRNA trnaslated into a protein and for keeping it stable ‎[12:28 PM]
Eli Fisker: ‎Perhaps give a molecule alongside the vaccine that turns on the mRNA production, but stops it as soon as this additional molecule is no longer administered. ‎[12:28 PM]
Eli Fisker: ‎For question A ‎[12:28 PM]
rhiju: ‎@eli yes exactly ‎[12:28 PM]
rhiju: ‎@difgitalembrace yes – let me write that out too: ‎[12:29 PM]
worseize: ‎switches mmm… 4 state ? 2 state ? why it differs from puzzle to puzzele how many states there are (also if we use NUPACK - it is kinda slow for 4 states) ‎[12:29 PM]
rhiju: ‎C. Figure out what antisense oligos could be deployed as therapeutics that bind to conserved RNA regions of the virion. ‎[12:29 PM]
LFP6: ‎RE: Energy models - Our entire tutorial sequence is really built around Vienna 1, as is a lot of typical player approaches/teaching (I think). ‎[12:29 PM]
LFP6: ‎I think it would be good to think about both in the short term, what kinds of training would help expound on other models ‎[12:29 PM]
LFP6: ‎And in the long term, how our tutorials could be structured to be less dependant on the “quirks” of a given model (and possibly even talk to designing well for experiments, and how the differences in different models correlated to that) ‎[12:30 PM]
Omei: ‎Can we learn how to design stable RNA using shorter sequences than a full mRA? ‎[12:30 PM]
Eli Fisker: ‎For question C. Can we get an overview of the conserved RNA regions in the virion? Preferably in game. ‎[12:31 PM]
lroppy: ‎need updated glossary for new players and new concepts ‎[12:32 PM]
Eli Fisker: ‎@Omei, do you mean while using sequence from a partial mRNA? ‎[12:33 PM]
rhiju: ‎@eli yes I like that idea. We’ll need your and other players’ help to make these puzzles playable by all. ‎[12:33 PM]
wayment: ‎I’m a das lab grad student studying different energy models – there’s a lot of ideas at the moment for how to improve them ‎[12:34 PM]
wayment: ‎I’ve been analyzing some of the massive eterna experimental datasets that we have and developing metrics to score how well energy models do on them ‎[12:35 PM]
wayment: ‎one thing that I’ve found is that the experimental data – both riboswitches and chemical mapping – suggest that the molecules are more “melted” than vienna or nupack predicts them to be ‎[12:35 PM]
lroppy: ‎what does that mean ‎[12:36 PM]
lroppy: ‎? ‎[12:36 PM]
wayment: ‎vienna and nupack predict stems to be more stabilized than they should be. ‎[12:36 PM]
wayment: ‎this sort of goes along with what LFP was saying about “quirks” of energy models – better energy models sort of “blur out” things that vienna and nupack are more prone to stabilize. ‎[12:37 PM]
wayment: ‎a quick hack that I found is actually running Vienna at 60 degrees celsius does much better than vienna at room temp. ‎[12:37 PM]
Omei: ‎@Eli That could be part of it. But I would expect that thermal stability of RNA has generic principles that might be uncovered faster if player were free to test designs of their choice. ‎[12:37 PM]
Eli Fisker: ‎@Omei, I’m happy to hear. ‎[12:38 PM]
Eli Fisker: ‎This brings me back to what Rhiju said earlier: “As an example, there are folks here who are trying to figure out the best segments of the virus to look for in the environment as affected individuals shed it. This would allow for tracking. ‎[12:39 PM]
Eli Fisker: ‎They pointed out to me that the most stable fragments of shed virions are likely the ones with RNA structure – can we help identify those?” ‎[12:39 PM]
Eli Fisker: ‎My comment: We know a thing or two about making RNA stable in Eterna, so I think so. ‎[12:40 PM]
LFP6: ‎I think new models could potentially be a great help, but as we train players, how do we account for differences in models? There are many strategies specific to different models. ‎[12:40 PM]
LFP6: ‎My perspective: (1) we should try to expound on what the various strategies are for different models ‎[12:41 PM]
LFP6: ‎(2) we should design the overall strcture of the tutorials to present things in a way that presents general design principles, with the specific strategies for different models presented as “extensions” (ie, they should be able to fit into a br ‎[12:41 PM]
LFP6: ‎(3) we should discuss WHY the differences exist from the perspective of “what does this actually mean in the lab - shoud I be using this strategy? What does it mean for the characteristcis of my design?” (ultimately, if we can better present th ‎[12:41 PM]
LFP6: ‎…those within the game, as opposed to relying on player knowledge, that would be better!) ‎[12:41 PM]
Omei: ‎I don’t know if this is crazy, but it seems to me that we could empirically study thermal stability (where the RNA backbone actually gets cleaved, ot just misfolded) using a variation on the Cloud Lab protocol. ‎[12:42 PM]
rhiju: ‎@Omei yes exactly I agree. ‎[12:42 PM]
rhiju: ‎The tricky thing here is whether (and if so when) could we do those experiments. ‎[12:43 PM]
LFP6: ‎Oops point 2 was cut off - ie, they should be able to fit into a broader sense of “how do you go about solving a structure” ‎[12:43 PM]
rhiju: ‎The reason I had to step out was to talk to Stanford’s leadership abuot sustaining expeirments and if so, how to do so safely. ‎[12:43 PM]
rhiju: ‎The conversation was quite promising. ‎[12:44 PM]
lroppy: ‎not sure I heard any response to my question about de-prioritizing any other efforts (& if so, communicating to player community that certain things are now on hold) ‎[12:45 PM]
Korosi: ‎Can this conversation be recorded and available in the forum for those players that arent here? ‎[12:45 PM]
lroppy: ‎yes ‎[12:45 PM]
rhiju: ‎So we should deploy labs as if we can do the experiments, and if the project is sufficiently compelling – i.e., may have an impact on pandemic – and lab experimentalists are available to do them, I will seek permission to do the experiments. ‎[12:45 PM]
lroppy: ‎I have it ‎[12:45 PM]
wayment: ‎@LFP6 here’s an off-the-wall thought – what as far as player education energy models, what about making sort of an energy model “sandbox” where players could mess around with a simple energy model ‎[12:46 PM]
Korosi: ‎Awesome thanks Lro! ‎[12:46 PM]
LFP6: ‎@wayment Mess around as in tweaking parameters? ‎[12:46 PM]
LFP6: ‎@Korosi I’ll make sure it gets in the archive on the wiki ‎[12:47 PM]
wayment: ‎say, I want to make all stems less stable, or I want to increase penalties for all loops ‎[12:47 PM]
rhiju: ‎@lroppy I don’t think we are deprioritizing other efforts, but I may have forgotten something? For example, ribosome challenge experiments have been shutdown but we have data in hand that will likely go out in the next weeks. ‎[12:47 PM]
LFP6: ‎@wayment I think that would be super neat ‎[12:47 PM]
lroppy: ‎ok thx ‎[12:47 PM]
wayment: ‎I would see it as sort of a side-interest thing, not per se on the track of tutorials ‎[12:47 PM]
rhiju: ‎I think we should continue the ribosome stuff because (1) it will let some of us think about non-coronavirus research for 1/4 of our time and maintain sanity, and (2) several of our coronavirus efforts happen to involve the ribosome. ‎[12:48 PM]
lroppy: ‎ok ‎[12:48 PM]
rhiju: ‎@wayment, I think your idea of energy function exploraiton would be feasible. @ElNando888 may have advice. ‎[12:49 PM]
Gerry Smith: ‎mental sanity has long gone ‎[12:49 PM]
rhiju: ‎@Gerry you can say that again. ‎[12:49 PM]
wayment: ‎It would be quick to whip up a toy RNA energy sandbox that wasn’t in eternajs, but in a toy model @rhiju and I have been playing with ‎[12:50 PM]
wayment: ‎but for sure integrating into eterna would be preferable ‎[12:50 PM]
LFP6: ‎I mean the key would be that you wouldn’t need to recompile ‎[12:50 PM]
rhiju: ‎@Gerry But bringing our attention to positive action, at least for me, is helping restore a sense of control. We can do something about this pandemic. ‎[12:50 PM]
rhiju: ‎@wayment there should be a way to ‘inject’ parameters into folding via EternaScript ‎[12:51 PM]
LFP6: ‎Or just via EternaJS ‎[12:51 PM]
Gerry Smith: ‎be the future we want ‎[12:52 PM]
rhiju: ‎@lfp6 ah yes you’re right. that wouldn’t be so hard. ‎[12:52 PM]
LFP6: ‎I mean, the EternaScript bindings would just pass things through EternaJS ‎[12:52 PM]
LFP6: ‎At any rate ‎[12:52 PM]
LFP6: ‎The key is that when its called, we need to be able to pass in parameters as a struct/string/whatever ‎[12:52 PM]
rhiju: ‎OK everyone, we’ve got just a few minutes left. ‎[12:53 PM]
wayment: ‎I could prototype a handful of toggles to vienna parameters that represent interesting changes, would that be useful? ‎[12:53 PM]
LFP6: ‎Again, as long as it doesn’t require recompilation ‎[12:54 PM]
Eli Fisker: ‎Should someone be interested in historic science writing, I can recommend the book Virus Hunter by Greer Williams. ‎[12:54 PM]
Eli Fisker: ‎Should someone be interested in historic science writing, I can recommend the book Virus Hunter by Greer Williams. ‎[12:54 PM]
Eli Fisker: ‎It is the story about how pioneers figured out first what was the viruses that caused disease and then their work in finding a vaccine. ‎[12:54 PM]
rhiju: ‎@Eli can you post that to our getsat thread? ‎[12:54 PM]
rhiju: ‎https://getsatisfaction.com/eternagame/topics/fight-the-covid-19-with-eterna ‎[12:54 PM]
Eli Fisker: ‎@Rhiju, sure. ‎[12:54 PM]
rhiju: ‎Thanks, Eli ‎[12:55 PM]
lroppy: ‎like agile, we should set short, periodic meetings to keep community updated ‎[12:55 PM]
rhiju: ‎@Lroppy I agree ‎[12:55 PM]
lroppy: ‎perhaps this slot every Wednesday ‎[12:56 PM]
lroppy: ‎or Friday ‎[12:56 PM]
rhiju: ‎And perhaps even a daily chat time to come together with updates and ideas. I may not be able to make all of them, but that could also help with sanity ‎[12:56 PM]
rhiju: ‎Let me make an observation – I see lots of folks online right now who I look to for humor, good scientific insight, and cameraderie. ‎[12:57 PM]
rhiju: ‎I think we may all a bit shell-shocked now, but I am finding that by engaging with this pathogen as an active scientist, citizen, and molecule designer, my spirits are high. ‎[12:58 PM]
rhiju: ‎We are now in a position to do the most important research of our lives. I hope you’ll all join me and the other developers and lab scientists of Eterna in this effort. ‎[1:00 PM]
rhiju: ‎Thanks everyone! ‎[1:00 PM]
Gerry Smith: ‎can’t think of a better or more important community to be in ‎[1:00 PM]
rhiju: ‎Stay tuned for news for further info on next chats. Please put thoughts on forums https://getsatisfaction.com/eternagame. Tell your friends about Eterna and our campaign for crowdfunding and for a surge of developers. ‎[1:01 PM]
hoglahoo: ‎thanks! I will make sure to explain it is an online crowd ;D ‎[1:02 PM]