Online chat with the EteRNA Team - switch puzzle strategies!

Gameplay and Strategies
1

Now that we have been working on switch puzzles for a while, the EteRNA team wants to have a meeting with the players to discuss your strategies for solving the puzzles. We are hoping to write up some of the results and ideas so far in a formal paper for the scientific community - and we want your help with it! It has been very exciting to see the flow of great ideas on the forum, and we want to help you further develop and test your ideas.

To do this, we have created a starting template for submitting an idea. You can post your comments directly on the forum or perhaps post a GoogleDoc link. Whatever works for you!

Title - what is your idea? Is it a general method for solving any switch, or is it a way to work around tricky elements like a complicated junction?

Precedent - how did you get your idea? What was your inspiration and why did you choose this problem to focus on?

Description of method - how can we implement your strategy? Does your strategy involve ratios of different nucleotides? Orientation of nucleotides in loops, hairpins, etc? Does it depend on other strategies? Try to keep it concise, though! If you have trouble fitting it under about 100 words, that may indicate that it should be split up into several related strategies (totally ok!).

Testing your method - what are your suggestions for how to test the method? Provide up to three - it’s good to have several possibilities. Feel free to upload secondary structures or game images to help explain your ideas.

Since this is a new direction for us, we want to meet with you in the game to discuss it! We’ll set up a time for the EteRNA team to be in the chat so that we can answer questions, and plan what future meetings may be like. We hope to do several more after this one. Please give us your input on these meetings - we want them to be fun and productive! Can people meet on Wednesday, 10/31 at 3:00 pm Pacific Standard Time? We look forward to talking about everyone’s ideas!

Regards,
The EteRNA team

2

Title :

Change scoring of switch lab challenges

Precedent :

For one, there’s another thread about that issue on this forum. Second, data shows that none of the synthesized designs has been able to fold (even remotely) properly in the unbound shape. Third : round 2 saw no improvements at all. Fourth : the only 2 new better scoring designs came from 2 perfect newbies.

Description of method :

If my understanding is correct, EteRNA is a directed stochastic search. Human brains generate candidates, quite randomly actually, since none of us really knows what he’s doing, those candidates get scored, and the crowd tries to improve on those results. So far, it has been very successful. Could be the human brains, but I know one thing for sure : if the scoring function had not worked properly, it would have failed.

I believe that the current scoring function for switches is flawed (see above in ‘Precedent’). I would propose a scheme like :

  • 30% for proper shape 1
  • 30% for proper shape 2
  • 40% for switching (the current 100%)

Testing the method :

Before changing it definitely, alternative scoring (like my idea) could simply be made available. The players would be free to follow their instinct on which scoring function is the better one.

3

Switch strategy - GC-pairs pr. length of strings

Precedent: I was just looking for patterns in GC-content distribution. The idea date back to start october according to my Switch diary.

Strategy:
2 basepairs – 1-2 GC pairs
3 basepairs – 2 GC pairs
4 basepairs – 2 GC-pairs for shapes that have more than two 4 basepair strings. 25% of the strings are allowed having only 1 GC-pair
7 basepairs – 3 GC-pairs

For anything below and above this number of GC-pairs mentioned give -1 pr. each

4

I am very interested in the area of ‘ideas for testing all of your theories’. There are 2 posts that capture my thoughts 1) https://getsatisfaction.com/eternagam… and 2) https://getsatisfaction.com/eternagam…

I think we should concentrate on getting some tools that will help us with these switches. Also, an effort needs to be made to get the energy model to a better level of simulation, anything to get a better model should be first priority. Maybe a backend supercomputer, for a ‘finalized’ design.

Thanks, John

5

Chatlog from scientist chat 31-10-2012

(Did a quick spellcheck and only moved for what Google docs told on big time)

jeehyung: hi all [11:00 PM]
Brourd: Hi Jee [11:00 PM]
starryjess: hi jee [11:00 PM]
rhiju: hey everybody! [11:00 PM]
Brourd: replied to your message, and how are you today? [11:00 PM]
Brourd: Hi rhiu, how are you today as well? [11:00 PM]
jeehyung: great : ] we’ll have your puzzle mod up in challenges soon [11:01 PM]
kws4679: hello! [11:01 PM]
Eli Fisker: Hi Rhiju :slight_smile: [11:00 PM]
Brourd: thanks, I made another in my spare time as well [11:01 PM]
starryjess: hi rhiju [11:01 PM]
Eli Fisker: Hi Jee [11:01 PM]
rhiju: excellent, glad to see that many of you could make it to this chat. [11:01 PM]
Eli Fisker: :slight_smile: [11:01 PM]
Brourd: Hi kws4679 [11:01 PM]
kws4679: hi Brourd [11:01 PM]
Eli Fisker: Hi kws too [11:01 PM]
kws4679: Hello Eli Fisker [11:02 PM]
Brourd: Well, I shall start it off, I guess [11:02 PM]
Eli Fisker: :slight_smile: [11:02 PM]
Toughguy: hey everybody! (i’m Tom from the forum posts) [11:06 PM]
Brourd: So, this paper about switches you are writing [11:02 PM]
Eli Fisker: Hi Tom :slight_smile: [11:02 PM]
Brourd: Hey toughguy Tom [11:02 PM]
tsuname: Pablo here [11:02 PM]
Eli Fisker: Hi Pablo :slight_smile: [11:02 PM]
Brourd: Hi Pablo [11:03 PM]
Toughguy: so, i saw that we already have a few posts in response to the template i put up earlier in the week [11:07 PM]
Eli Fisker: Yes, I have one more switch strategy from Jieux, who couldn’t be here [11:03 PM]
Eli Fisker: Mat can’t be here as well [11:03 PM]
Toughguy: oh, well then that would be great to share on the forum too [11:08 PM]
Adrien Treuille: Hi all. [11:04 PM]
Eli Fisker: Hi Adrien :slight_smile: [11:04 PM]
Brourd: HI adrien [11:04 PM]
jnicol: hi all [11:04 PM]
Eli Fisker: Tom, agree. He just sent it to me and asked to share it here, so I have put it up in a document for now [11:04 PM]
Eli Fisker: https://docs.google.com/document/d/1U… [11:05 PM]
Eli Fisker: And I will ask him about what to do with it later [11:05 PM]
Eli Fisker: And suggest he put it up [11:05 PM]
jeehyung: Here’s a link to the forum post about today’s meeting topic [11:06 PM]
jeehyung: https://getsatisfaction.com/eternagam… [11:06 PM]
Toughguy: so, today is mainly so we can answer questions about the template, your ideas for switch strategies, and ideas for testing all of your theories [11:10 PM]
Eli Fisker: ok [11:06 PM]
Brourd: Is this about switch puzzles in general? [11:07 PM]
Brourd: or more specifically FMN switches? [11:07 PM]
Toughguy: switch puzzles in general [11:11 PM]
Toughguy: so far, we have only done fmn switches [11:11 PM]
Toughguy: but we have other switching molecules available that we may use in the near future [11:11 PM]
tsuname: Although we could also have strategies for simple puzzles, i.e. no switching [11:08 PM]
tsuname: I read jieux’s strategy, looks good, but to preserve modularity, I would write it up as different strategies that would depend on each other [11:09 PM]
rhiju: well, perhaps one question for players … [11:09 PM]
Brourd: Is the plan to use these strategies in a new algorithm, like EteRNA bot? [11:09 PM]
Eli Fisker: I will send him the chatlog afterwards, with your suggestion in, Tsuname [11:09 PM]
rhiju: brourd, that’s the question [11:09 PM]
rhiju: to solve switch puzzles, do you thnk it will be possible to collate a bunch of ‘features’ of good switch designs? [11:10 PM]
Brourd: Hmmm [11:10 PM]
rhiju: or do we need something like jieux’s strategy, which is more like a full protocol or ‘recipe’ [11:10 PM]
jnicol: my opinion, its too early to be thinking about bots to solve switches [11:11 PM]
rhiju: interesting, jnicol [11:11 PM]
rhiju: maybe we can break this into 2 questions [11:11 PM]
Eli Fisker: I kind of like the recipe idea, as I think that the puzzle will be needed to be solved in steps [11:11 PM]
jnicol: were still feeling our way through solving these ‘easy’ ones [11:11 PM]
rhiju: there are actually two ‘fronts’ on switches [11:12 PM]
Eli Fisker: fronts - schools on how to solve them? [11:12 PM]
rhiju: in one ‘front’, there’s the question of why switches that ‘work’ in silico, don’t work as well in real life [11:13 PM]
rhiju: that’s analogous to what we tried to solve for non-switch puzzles [11:13 PM]
rhiju: and there we got an impressive list of features that players ‘saw’ in designs that ‘worked’ expeimentally [11:13 PM]
Brourd: yep [11:13 PM]
rhiju: so one question is wlll we be able to apply the same ‘strategy market’ idea [11:14 PM]
rhiju: a second ‘front’ (coming soon) is a purely in silico question [11:14 PM]
rhiju: there are no algorithms to really effectively solve switch puzzles that involve transitions between more than two states [11:14 PM]
Adrien Treuille: are there efficient (polynomial time) algorithms for two state switches? [11:15 PM]
rhiju: for example, now we have +/- FMN. we can also create puzzles with four states, and you would be asked o toggle between them by adding FMN or not, and by adding theophylline or not (4 conditions) [11:15 PM]
Brourd: any time frame for these awesome 3+ state problems? [11:16 PM]
rhiju: @adrien, no there are not polynomial time algorithms for designing switches [11:16 PM]
rhiju: well, i think we could put up 3+ state problems as challenges soon [11:16 PM]
rhiju: would you be interested? [11:16 PM]
Brourd: Sweet [11:16 PM]
Eli Fisker: Yep [11:16 PM]
starryjess: wow [11:16 PM]
Toughguy: yeah, i can design the challenge puzzles soon for the new ligand, theophylline [11:21 PM]
rhiju: Great, toughguy [11:17 PM]
Brourd: I like that idea [11:17 PM]
jnicol: are these modeled off what nature currently does, or we think it does? [11:17 PM]
rhiju: (FYI, tom and I are sitting in the same room) :wink: [11:17 PM]
Toughguy: so i’ll make some basic ones with the aptamer for just theophylline [11:21 PM]
Adrien Treuille: (FYI, Tom = Toughguy) ! [11:17 PM]
Toughguy: and also some for theophylline and fmn that have 4 states [11:21 PM]
jnicol: and is a 3 state switch useful in nature? [11:17 PM]
Brourd: so, you just gave tom a high five, right? [11:18 PM]
rhiju: @jnicol, there are some natural RNA switches that have different functions depending on the presence or absence of more than one small molecule [11:18 PM]
rhiju: Some are called ‘tandem riboswitches’ [11:18 PM]
rhiju: ANd moreover, many engineers are trying to make RNA switches that turn on or off only when they see molecules X, Y, and Z [11:19 PM]
jnicol: but, wouldn’t it be better to concentrate on the 2 state switches? [11:20 PM]
rhiju: Folks at caltech and MIT. One application is in ‘gene therapy’, where you ask a gene to be delirvered only to cells that are expressing the right four other mRNAs [11:20 PM]
Toughguy: yes, jnicol, in the short term, i think our lab puzzles will definitely focus on the 2 state switches [11:24 PM]
Brourd: @ jnicol - for now, we would be dealing with them in silico, if I am following right [11:20 PM]
jnicol: I think we have a better chance at wowing the field if we came up with some great 2 state swicthes [11:20 PM]
rhiju: @jnicol, I agree too, for the ‘experimental’ front. [11:20 PM]
Adrien Treuille: So… an important topic for discussion today is “what should the Switch Strategy Market look like?” [11:21 PM]
Eli Fisker: Rhiju, that MIT thing sounds cool [11:21 PM]
Adrien Treuille: – what worked last time? [11:21 PM]
Eli Fisker: But we could play with 3 state puzzles in the meantime [11:21 PM]
Adrien Treuille: – what could we be doing better? [11:21 PM]
jnicol: I like the sound of multi state switches, but I would ratyher try to get you guys more funding for what we are doing :slight_smile: [11:22 PM]
Toughguy: (yes, eli! should be fun) [11:26 PM]
Adrien Treuille: – how are (in vitro) switches different from previous in vitro puzzles? [11:22 PM]
Eli Fisker: Tom :slight_smile: [11:22 PM]
rhiju: right, to go back to adrien’s thread and the strategy market – what do we need? [11:22 PM]
Brourd: @ Adrien, using something like the previous strategy market should work [11:23 PM]
Brourd: using specific features [11:23 PM]
Adrien Treuille: @brourd – I basically agree [11:23 PM]
rhiju: @brourd, do we need to give players access to a scripting language to formalize their algorithms and ideas? [11:23 PM]
rhiju: or would that be too burdensome? [11:23 PM]
Adrien Treuille: but there are many things we could do differently [11:23 PM]
Quasispecies: hello [11:23 PM]
Adrien Treuille: e.g. (1) : scripting languages [11:23 PM]
Eli Fisker: Rhiju, I know Mat would say yes to that [11:23 PM]
bigbloto: we want scripting [11:24 PM]
Adrien Treuille: e.g. (2) : tighter integration with the game [11:23 PM]
Eli Fisker: Hi Quasispecies :slight_smile: [11:23 PM]
JR: scripting - yes [11:24 PM]
rhiju: @adrien, what do you mean tighter integration? [11:24 PM]
Toughguy: i think the scripting language could be useful at some point, but i think it could be burdensome to newer folks [11:28 PM]
tsuname: how complex scripting? you mean full scripting power or writing sort of pseudo-code algorithms? [11:24 PM]
Brourd: Apparently everybody wants scripts [11:24 PM]
Adrien Treuille: I think that we need to balance scripting with written descriptions [11:24 PM]
Eli Fisker: As long as those who can’t program can get their strategies programmed too [11:24 PM]
Adrien Treuille: for example, think of Eli’s rule from the last Strategy Market – about GC pairs in multiloops [11:25 PM]
Adrien Treuille: it can be described very simply, but it takes a fair amount of code to formalize [11:25 PM]
Eli Fisker: and I can’t program :slight_smile: [11:25 PM]
Adrien Treuille: we don’t want to lose those beautiful insights and simply turn this into a scripting competition [11:25 PM]
Toughguy: totally agreed, adrien [11:29 PM]
Eli Fisker: But I will like the option for those who can [11:25 PM]
jeehyung: Maybe we can ask players to “describe in English” first [11:26 PM]
Adrien Treuille: at the same time, we had a bunch of people working really hard over here at CMU typing up all the player ideas, and that can’t scale [11:26 PM]
Toughguy: perhaps we could offer each option [11:30 PM]
jeehyung: And then those can be picked up by devs and players who can program [11:26 PM]
jeehyung: We do need a scripting system (and we have almost working one) to make the system scalable [11:26 PM]
Adrien Treuille: Ideally the solution would be to have two lists, one of ideas and one of scripts – with a 1-many relationship [11:26 PM]
rhiju: how many of you on the chat do have programming experience? [11:26 PM]
jeehyung: @Adrein yes - proposal & implementation [11:27 PM]
jnicol: me [11:27 PM]
Quasispecies: me [11:27 PM]
JR: me [11:27 PM]
Adrien Treuille: So that players can propose beautiful simple ideas, and others can express them in code [11:27 PM]
ElNando888: 30+ years [11:27 PM]
rhiju: what if we gave some reward for writing programs for other players? e.g., more votes [11:27 PM]
Eli Fisker: Nice [11:28 PM]
Adrien Treuille: yes [11:28 PM]
Adrien Treuille: I think that’s really nice [11:28 PM]
jnicol: I’m new to the scripts. Would they fill in the puzzle according to some written rules, ie alternate all the AU’s? [11:28 PM]
rhiju: ok, scripting + rewards for writing scripts (esp. helping other players) seems like a win [11:29 PM]
Toughguy: i like that rhiju [11:33 PM]
JR: scripts = series of sweeps over puzzle - do this , do this [11:29 PM]
rhiju: it will be a lot of coding… what does jee think? [11:29 PM]
Adrien Treuille: I think the other thing that was missing from the previous strategy market was better integration in the game [11:30 PM]
tsuname: hmmmm maybe this could turn into some sort of a toolbox in-game? [11:30 PM]
Toughguy: what do you mean, adrien? [11:34 PM]
Adrien Treuille: @tsuname – that’s just what I’m thinking! [11:30 PM]
rhiju: i think i know what adrien meant [11:30 PM]
Adrien Treuille: let players call “strategies” from within the game [11:30 PM]
Adrien Treuille: let players call “EteRNAbot” from within the design interface [11:30 PM]
tsuname: yea, that would be awesome [11:30 PM]
Adrien Treuille: add columns to the experiment viewer for different player strategies [11:30 PM]
Eli Fisker: Totally agree, Adrien. [11:30 PM]
Eli Fisker: Mat will be so happy [11:30 PM]
Adrien Treuille: :slight_smile: [11:31 PM]
Brourd: Sounds great [11:31 PM]
Adrien Treuille: I have another… kind of weird question… [11:31 PM]
jeehyung: The scripting will be an open programming language. You should be able to program a script that will run specific sequence coloring, or make scripts that scores RNA (strategy market) [11:31 PM]
Toughguy: so players can submit strategies, have them coded up, and then those strategies are available to help solve puzzles? [11:35 PM]
jeehyung: So with the toolbox, you should be able to call other people’s script to score/solve puzzles, etc [11:31 PM]
Adrien Treuille: do the players who submitted strategies to the last strategy market feel that their efforts were “rewarded” enough in game? [11:31 PM]
kws4679: scripting sounds great [11:31 PM]
Adrien Treuille: if not, how could we better reward this extremely valuable scientific work? [11:32 PM]
rhiju: to the programmers out there (elnando888), any suggestions on what the script language should be? [11:32 PM]
JR: as reality check - forum post as to what players think they want? [11:32 PM]
jeehyung: @rhiju we are using javascript [11:32 PM]
rhiju: @jee, aha. [11:32 PM]
ElNando888: no preference, I have coded with every possible language (almost) :slight_smile: [11:33 PM]
tsuname: hey, that would be interesting, how about putting popularity on strategies, and getting something in return for how much a strategy is used [11:33 PM]
JR: need simple stupid - java no good [11:33 PM]
jnicol: javascript, yes [11:33 PM]
ElNando888: simple guideline, work with what you’re already using [11:33 PM]
Quasispecies: i’ve got experience with python and fortran, but probably for accessibility something like lua at foldit [11:33 PM]
rhiju: the other good thing about javascript is that there are websites like codeacademy.org to teach it [11:34 PM]
rhiju: ok, going back to adrien’s question on rewarding scientific work… [11:34 PM]
Adrien Treuille: We’ve already invested in Javascript, and I think that’s a good choice especially as everyone’s browser can execute it, which means people can perform computation locally rather than burdening the server. [11:34 PM]
Brourd: It’s scientific work, our only reward needs to be recognition and a portion of the $1,000,000 from the nobel prize [11:35 PM]
Toughguy: pablo (tsuname) had a nice idea, i thought - to reward based on the frequency at which someone’s strategy is used [11:39 PM]
Adrien Treuille: Lol [11:35 PM]
Adrien Treuille: @tom/tsuname - that is a very interesting idea! [11:35 PM]
JR: sure [11:36 PM]
tsuname: @ Brourd lolz [11:36 PM]
rhiju: how to count when a strategy is ‘used’? [11:36 PM]
xaner: Oh hi. What’s the point of the rewards? Abusing our brain’s reward system? [11:36 PM]
Toughguy: maybe you get the reward if your strategy is used in a successful lab design [11:40 PM]
tsuname: @rhiju when it;s executed [11:36 PM]
Eli Fisker: HI Xaner [11:36 PM]
rhiju: what if its jus a google doc like jeiux’s switch strategy? [11:37 PM]
Adrien Treuille: So regarding recognition… we can pour effort into more ways to “reward players” … e.g. badges if you submit a strategy, badges if your strategy makes it into EteRNAbot, etc… but believe it or not, this infrastructure takes time to build [11:37 PM]
xaner: hi again [11:37 PM]
jeehyung: @rhiju for the implemented ones, we can count number of executions [11:37 PM]
Adrien Treuille: I’m wondering if you felt that the existing Strategy Market infrastructure did enough to reward players hard work or if we should work on this [11:37 PM]
Brourd: Oh, I had forgotten about the entire “time” thing :slight_smile: [11:37 PM]
jeehyung: or we can have voting system like in the lab [11:38 PM]
rhiju: what if we could let players register that they ‘liked’ someone else’s strategy? [11:38 PM]
Brourd: There wasn’t particularly any reward, and I came in after the strategy market was made [11:38 PM]
ribonucleic: jee, when will my strategy market idea be ready, like when it is tested, published, etc. [11:38 PM]
mat747: back hi and re-hi [11:38 PM]
Brourd: the “liked” idea sounds better, prevents the abuse of just using a strategy over and over again in order to get rewards [11:38 PM]
Quasispecies: points… achievements… anything to give that extra squirt of dopamine [11:38 PM]
JR: forget about rewards - get it up and reuning first [11:39 PM]
xaner: that’s what I thought [11:39 PM]
Eli Fisker: wb, Mat :slight_smile: [11:38 PM]
jnicol: For your paper, what are the main points you will be writing about? [11:39 PM]
Quasispecies: but yes, ultimately participation should be its own reward [11:39 PM]
rhiju: @JR and @xaner, yes – let’s stop talking about rewards [11:39 PM]
bigbloto: “like” risks rewarding more for an author’s perceived visibility in the community rather than actual value, imo [11:39 PM]
jandersonlee: for me the reward is synthesis. getting the cloud lab up so more designs can be run is important in that regard [11:39 PM]
xaner: Shouldn’t there be some kind of market study here? [11:39 PM]
ribonucleic: bigbloto, who are you [11:39 PM]
JR: need driver pgm and subroutines that [11:40 PM]
xaner: identifying which rewards would work best on the players would improve their motivation and efficiency [11:40 PM]
mat747: i agree with janderson [11:40 PM]
JR: sweep over file [11:40 PM]
rhiju: back to @jnicol, we don’t yet have a fixed plan for the switch paper [11:40 PM]
Brourd: btw, how is progress on the cloud lab and the updated site doing? [11:41 PM]
jnicol: ok [11:40 PM]
jeehyung: how about reward is gamepoints, but you can buy extra synthesis slots with those points? [11:41 PM]
jeehyung: @Brourd, updated site is going beta today - I’ll post a news item [11:41 PM]
rhiju: in fact, one thing I was hoping to discuss during this chat was the possibility of having more player input and even player writing on such a paper (or set of papers) [11:41 PM]
Brourd: I am on board with that Jee [11:41 PM]
Eli Fisker: Jee, cool [11:41 PM]
Adrien Treuille: my ideal reward structure would be (special badges + the ability to sort players by thier score at creating strategies) [11:41 PM]
Brourd: I’m on board with that as well Rhiju [11:42 PM]
JR: second score? [11:42 PM]
xaner: I think rewards need to be tailored to the players [11:42 PM]
Brourd: @ Adrien, too many boats, but I guess I like that idea as well [11:42 PM]
rhiju: [@brourd, cloud lab has been progressing slower than hoped, partly because our main technical expert has had sevral family-related emergencies over the last year.] [11:42 PM]
Brourd: Aww, that’s terrible [11:42 PM]
jandersonlee: foldit has separate scores for solo and evolver as different people have different skilsets [11:42 PM]
rhiju: [@brourd, but its a top priority, and we have pushed forward on illumina-basd readouts, and have even been able to make libraries of 4000 RNAs, albeit not yet on eterna player designs …] [11:43 PM]
rhiju: back to papers… [11:43 PM]
jnicol: Any chance on getting a high end server to get better simulation results? [11:44 PM]
rhiju: do any of the players on chat have experience in writing articles for publication in any kind of field (e.g. science, journalism)? [11:44 PM]
Toughguy: so, rhiju, i guess for the paper you are looking to have players contribute strategies, and then discuss some of the most successful strategies in the eventual paper? see how they compare to existing switch design algorithms? [11:48 PM]
jandersonlee: CS - years ago [11:44 PM]
xaner: what’s up with all the old people? [11:44 PM]
rhiju: @jandersonlee, did you find it difficult to write the articles? [11:45 PM]
jandersonlee: not reall - time consuming :slight_smile: [11:45 PM]
Quasispecies: @ rhiju - yes. sorry i’ve been away for so long, but i’ve been busy with my own work and had less time for hobbies [11:45 PM]
rhiju: yes, i totally agree. [11:45 PM]
rhiju: abut time consuming. [11:45 PM]
Quasispecies: but to answer your question earlier, yes i have experience in that respect [11:45 PM]
rhiju: @quasispecies, cool. [11:46 PM]
jandersonlee: it helped to have a good grounding in the scinetific/experimental method in highschool chemistry/college courses [11:46 PM]
Adrien Treuille: @rhiju – were you thinking one paper / switch strategy or coarser granularity [11:46 PM]
rhiju: ok, now the challenge is – what would it take to get players like jandersonlee or quasispecies to help us and others write up this scientific knowledge? [11:47 PM]
rhiju: perhaps as multiple papers? [11:47 PM]
jandersonlee: no doubt there is a white-paper or tutorial online somewhere [11:47 PM]
rhiju: @jandersonlee, a white-paper for what? writing papers? [11:47 PM]
jandersonlee: (on how to write a scientific paper) [11:47 PM]
Toughguy: @adrien, i personally think the best way to get the field’s attention would be to produce something like EteRNA bot Switch that we can compare as a final product to other existing tools [11:51 PM]
Brourd: How many existing switch algorithms are out there? [11:48 PM]
rhiju: there are almost none… nupack has something on their site. [11:48 PM]
Brourd: Hmm, not much to compare to then [11:49 PM]
rhiju: back to papers, toughguy/tom has some thoughts… [11:49 PM]
Toughguy: so, we could do several potential things - we could shoot for one mega paper, in which lots of players contribute little pieces [11:54 PM]
Toughguy: or, we could practice on shorter papers, focusing on just one or two strategies at a time [11:54 PM]
rhiju: well, look, some players have already written lots of ‘little’ papers [11:50 PM]
Brourd: Hmmm [11:50 PM]
Brourd: Well, that depends, each little paper has to have a focus, will it be just on individual strategies? [11:51 PM]
xaner: what’s the goal of this? Advertisement? Funding? [11:51 PM]
Brourd: and then, will we eventually compile them all into one big paper anyway? [11:51 PM]
jandersonlee: What might help is some guidance on picking a research focus for a paper that would be interesting to the field and advance it by some small delta [11:51 PM]
Eli Fisker: science is the goal [11:51 PM]
rhiju: for example, here’s a start from some folks thinking abot barriers: https://docs.google.com/a/stanford.ed… [11:51 PM]
tsuname: although just the fact of having a community document the process of discovery/engineering would be an awesome demonstration as well [11:52 PM]
rhiju: and here’s one from eli on what makes an RNA design problem difficult: https://docs.google.com/a/stanford.ed… [11:52 PM]
rhiju: These contain interesting insights, and if they could be formalized with rigorous experimental tests, and written up in a ‘standard’ scientific format, they could be useful to a wide audience [11:52 PM]
Quasispecies: that’s really interesting. who did that work with barriers? [11:52 PM]
Brourd: ElNando [11:53 PM]
ElNando888: my “thing” about energy barriers is very far from a scientific paper, IMO [11:53 PM]
jandersonlee: hypothesis-experimenta_design-test-analysis-results? [11:53 PM]
xaner: Shouldn’t the people who initiated this also submit the papers? [11:53 PM]
rhiju: a bunch of people thought about barriers, and elnando888 put it together [11:53 PM]
rhiju: @xaner, yes certainly the people who put it together should be authors and submitters [11:53 PM]
jandersonlee: but where is the test and results? [11:53 PM]
rhiju: well, should we create a mechanism where some group of people get synthesis slots to test their ideas? [11:54 PM]
rhiju: even before the ‘cloud lab’? [11:54 PM]
Toughguy: yes, rhiju, i like that [11:58 PM]
jandersonlee: or improvement over standard practice? [11:54 PM]
Toughguy: perhaps we could do “small clouds” [11:58 PM]
Brourd: So, we have several micro papers, is the goal to eventually compile? [11:54 PM]
Brourd: small clouds sound nice [11:55 PM]
Toughguy: where we have a few ideas we test at a time [11:58 PM]
Toughguy: locally on a smaller deep sequencer [11:59 PM]
rhiju: well, i’d say that we didn’t have to compile the papers [11:55 PM]
jandersonlee: sounds promising [11:55 PM]
xaner: Do the players have full access to the program’s source and what exactly they are “playing” with? [11:55 PM]
Brourd: @ Xaner - we are currently using Vienna version 1. something [11:55 PM]
Brourd: available online [11:55 PM]
rhiju: we’ve ben in talks with folks at plos currents to create a new ‘current’ for ‘micropapers’: http://www.plos.org/ [11:55 PM]
jandersonlee: nice forum then [11:56 PM]
rhiju: the barrier now is getting the stuff tests and written up. [11:56 PM]
Quasispecies: i think before we talk about papers, we need very specific problems to tackle and a plant to attack them [11:56 PM]
xaner: Hum. Maybe eterna could get plos to lend them a space to drop their micro papers at or simply host their papers themselves [11:56 PM]
Brourd: yep [11:57 PM]
rhiju: @jandersonlee, what if we start with papers that are purely ‘in silico’, written by players [11:57 PM]
jandersonlee: forum and google docs is a starting proxy for ramping up [11:57 PM]
rhiju: for example, there isn’t any paper on ‘what makes an RNA design hard’ [11:57 PM]
rhiju: but eli fisker and others have thought about this [11:57 PM]
jandersonlee: @rhiju - if you think that would be of interest to the target audience - you’re the expert there [11:58 PM]
Brourd: Hard in silico, or hard in vitro? [11:58 PM]
rhiju: and arguably, the existing ‘player puzzles’ in eterna can test how hard a puzzle is for bots and for humans [11:58 PM]
xaner: Players should be able to test other players’ results [11:58 PM]
rhiju: @brourd (here i’m talking about in silico just o get started) [11:58 PM]
Brourd: Hmmm, what about other in silico models? [11:59 PM]
rhiju: so, in some sense we already have interesting scientific content that is not available elsewhere and could be tested by in silico ‘experiments’ (and later with the cloud lab with wet-lab experiments ) [11:59 PM]
Brourd: one design may be easy in one model, but impossible in another [11:59 PM]
xaner: It would be cool if it were possible to organically obtain results without any single player writing down anything [11:59 PM]
rhiju: @brourd, if we get scripting going, we could ask players to create bots that can solve more puzzles than existing bots – there’s another idea. [0:00 AM]
Brourd: yes, that sounds like an interesting idea Rhiju [0:00 AM]
rhiju: ok, so now the problem is … how do we formalize that knowledge into a reasonably paper? [0:00 AM]
mat747: timeframe for scripting ? [0:00 AM]
jandersonlee: or puzzles that are solved by one bot and not another to be tested in vitro later to see if they form [0:00 AM]
Brourd: @ Rhiju - for example, I believe RNASSD bot uses the Constraint Generation model, which allowed it to solve a puzzle unsolvable in the Vienna engine, [0:01 AM]
xaner: I don’t know what data is collected yet on the player’s puzzles. Anyone knows? [0:01 AM]
jandersonlee: (i.e. checking the veracity of the bots) [0:01 AM]
Brourd: @ Xaner - a few forum posts about the current limits of the bots are out there [0:01 AM]
rhiju: in principle, and academic scientist like me could polish and write up what the players put together, but that’s hard for me. [@brourd and jandersonlee, great ideas] [0:01 AM]
Adrien Treuille: Unfortunately, I have to go now… so I want to quickly propose a thought … given that we have so few synthesis slots, and don’t yet know how to write a community paper, I propose that we start by writing – as a community – a single (mega) Switch Design paper. We can start writing the paper now. As we continue to write the paper, we’ll figure out which experiments we need to do, a [0:02 AM]
rhiju: what if we ‘gamify’ paper writing to teach players how to write their own ideas up? [0:02 AM]
rhiju: i started writing some ideas on this here: [0:02 AM]
rhiju: https://docs.google.com/a/stanford.ed… [0:02 AM]
xaner: I mean, if we quantify how fast X players solve a puzzle, how many different strategies they come up with, etc, then we can quantify things like “how hard it is to solve”. [0:02 AM]
rhiju: this was meant to be for my lab [0:02 AM]
Eli Fisker: Rhiju, cool thanks [0:02 AM]
rhiju: for example, one thing that i do when i writ papers is to set ‘persnal goals’, like … i want to write the introduction by 5pm [0:03 AM]
Quasispecies: lol @ the transition state. dear god. too true. [0:03 AM]
Brourd: Very nice Rhiju [0:03 AM]
Eli Fisker: Rhiju, can we get that template up in the forum? [0:03 AM]
rhiju: that could be fun to have as a ‘timer’ in a game [0:03 AM]
Brourd: @ Adrien, one mega paper by the whole community could work [0:04 AM]
rhiju: @ brourd, what would be the best subject for that mega paper? [0:04 AM]
rhiju: the ‘what makes an rna design problem hard’? [0:04 AM]
xaner: that’s the question [0:04 AM]
xaner: I say submit a poll [0:04 AM]
LFP6: Happy halloween everyone! How did the dev chat go? [0:04 AM]
rhiju: why not have several potential papers/projects? [0:05 AM]
Brourd: We could focus on the in silico problem, point out flaws in the models, and how similar soluion in vitro fail [0:05 AM]
jnicol: limitations of current energy models [0:05 AM]
Eli Fisker: Hi LFP, most of them are still around [0:05 AM]
xaner: here we have a few propositions already [0:05 AM]
jandersonlee: probably 1-5 open topics/papers at any one time? [0:05 AM]
Adrien Treuille: both in silico and in vitro papers are possible [0:05 AM]
rhiju: yea, that’s very interesting! [0:05 AM]

6

GDoc of chat log: https://docs.google.com/document/d/1c…

7

We had an extra scientist chat with Jee today. (25 nov, 2012) I have added a chatlog here for all interested players. Lots of exciting news about what the near future will bring. Enjoy!

jeehyung: hi all [10:23 AM]
eternacac: So, it’s nice to reawaken some of those skills [10:23 AM]
Eli Fisker: hi Jee :slight_smile: [10:23 AM]
jeehyung: hi Eli : ] [10:23 AM]
Eli Fisker: Eternac, happy to hear [10:23 AM]
Brourd: hey jee [10:24 AM]
Eli Fisker: Jee, how is things? [10:23 AM]
jeehyung: hi Brourd [10:23 AM]
eternacac: hi jee [10:24 AM]
jeehyung: hi eternacac [10:24 AM]
jeehyung: good here.we have a new wiki [without those ads now!] and a synthesis team blog coming soon,… [10:24 AM]
jeehyung: and I think we’ll have a prototype for scripting interface tomorrow. [10:24 AM]
Brourd: oh, cool [10:25 AM]
eternacac: yeah, I’m a non-diplomaed former chemist [10:24 AM]
Eli Fisker: Adfree, great bonus [10:24 AM]
Eli Fisker: Synth team blog, awesome! [10:25 AM]
jeehyung: It should be much cleaner now. Hopefully we can organize all the great stuff players write in the wiki : ] [10:25 AM]
eternacac: sounds great [10:25 AM]
Brourd: how is the switch puzzle maker coming along? [10:25 AM]
Eli Fisker: Wow, you guys have been busy, Jee [10:25 AM]
jeehyung: On the synthesis team blog, we are thinking about having our players write posts too. that should be fun : ] [10:26 AM]
Brourd: nice [10:26 AM]
jeehyung: @Brourd, I finally allocated 2 full days to finish up the interface. [tomorrow and the day after] [10:26 AM]
wateronthemoon: hi. so here’s who’s awake at 2:30am [10:26 AM]
jeehyung: so we can probably testing it in the beta site in 2-3 days [10:26 AM]
Eli Fisker: Jee, I love that with the organizing. Should make it so much easier for people to find what they need to move on [10:26 AM]
jeehyung: hi wateronthemoon [10:26 AM]
Brourd: great Jee, the Stanford lab will be ecstatic [10:27 AM]
jeehyung: @Eli exactly. Things will be so much easier for new players [10:27 AM]
Eli Fisker: :slight_smile: [10:27 AM]
jeehyung: @Brourd I’m sure they will. Also, I’ll discuss the “puzzle solving volunteer” idea with the team in this week’s meeting. [10:27 AM]
wateronthemoon: am a newbie but i love this so thanks all [10:27 AM]
jeehyung: I’m sure they’ll love the idea [10:27 AM]
jeehyung: @wateronthemoon good to know : ] [10:27 AM]
mat747: hi jee, brourd [10:28 AM]
jeehyung: hi mat [10:28 AM]
Eli Fisker: Jee, Mat and I will looks forward to what they say for the puzzle solving volunteer idea. [10:28 AM]
Eli Fisker: say to [10:28 AM]
Brourd: hi mat, welcome wateronthemoon [10:29 AM]
eternacac: Brourd–Stanford has a stat based linguistics parser—wonder what it would do with RNA… [10:29 AM]
jeehyung: No doubt they will love the idea : ] [10:29 AM]
Eli Fisker: funny idea, Eternac :slight_smile: [10:29 AM]
Eli Fisker: Who knows what will come out. It just might work for something with RNA, if set up right [10:30 AM]
Brourd: @ jee - how was the vacation? [10:30 AM]
jeehyung: @eternaac Actually stat based parsing and recomposition of sequences is a hot topic in protein/RNA design : ] [10:31 AM]
jeehyung: @Brourd it has been bit busy, but relaxing : ] [10:31 AM]
Eli Fisker: hehe, awesome [10:31 AM]
Brourd: that is good [10:32 AM]
eternacac: @wateronthemoon–I’m only 10 days old here [10:33 AM]
Eli Fisker: I really love how people from a lot of different work fields combine to solve the RNA folding problem. [10:33 AM]
Brourd: Sounds like everything is going great with the site Jee, but Eli did mention something, what happened to the strategy market? [10:33 AM]
mat747: jee - have you got info/links to stat based parsing and recomposition of sequences [10:33 AM]
eternacac: Yes–way multi disiplinary [10:33 AM]
Eli Fisker: Eternac :slight_smile: [10:33 AM]
jeehyung: @Brourd it’s being remodeled with the scripting interface. It’ll be part of the Eterna scripting. [10:34 AM]
jeehyung: @mat, I don’t exactly remember the name of the field. Actually give me a few minutes… [10:34 AM]
mat747: sure no rush [10:35 AM]
jeehyung: Now this is sort of related. “RNA tectonics”. It’s about assembling “working elements” of different RNA sequences to make a completely new shape [10:35 AM]
jeehyung: http://www.sciencedirect.com/science/… [10:35 AM]
eternacac: @jee that would be cool if you had some parsing links [10:35 AM]
jeehyung: In fact, if eterna 3d comes along, RNA tectonics is likely the first approach we take. [10:35 AM]
eternacac: Soo much to investigate [10:36 AM]
Eli Fisker: Thanks for the link to the article, Je [10:36 AM]
mat747: Jee - thanks, but we have done that here [10:36 AM]
eternacac: Grats to Jee [10:36 AM]
jeehyung: @mat yes what our players do in the lab is very like tectonics, and I think that’s why they work : ] [10:37 AM]
Eli Fisker: :slight_smile: [10:37 AM]
jeehyung: But if this moves to 3d, it’ll be a completely different game [10:37 AM]
mat747: in what way [10:38 AM]
jeehyung: the game will be more like cutting parts from natural RNAs, assembling them together, and running a simulation see if it folds into the target shape [10:38 AM]
rlmarchal: 3D is almost reality [10:38 AM]
jeehyung: We can’t simulate RNA in real time with every single nucleotide change as we do now. [10:38 AM]
jeehyung: So the process will be more like a batch process where you replace a batch sequences [parts] instead of single nucleotide [10:39 AM]
eternacac: would we lose 2D to get 3D? [10:39 AM]
Eli Fisker: I want to keep eterna 2d. But I won’t mind an extra game [10:39 AM]
jeehyung: @eternaac of course no : ] Unless our players completely solves the 2d design problem [10:39 AM]
eternacac: for moi 2D/3D very different mind set [10:40 AM]
mat747: Jee - Ok thx - I have played with cut parts in foldit, but getting to do it with dna will be new for me [10:40 AM]
Eli Fisker: Hehe, first time I consider slowing down. :wink: [10:40 AM]
mat747: rna [10:40 AM]
eternacac: anybody else here love origami? [10:41 AM]
rlmarchal: Oh, yes [10:41 AM]
Eli Fisker: Eternac, scientists love to do origami with RNA and DNA [10:41 AM]
wateronthemoon: me 2 [10:41 AM]
jeehyung: dna origami http://en.wikipedia.org/wiki/DNA_origami [10:42 AM]
eternacac: Yes, but I meant the structural flat planar type–paper & such :wink: [10:42 AM]
Eli Fisker: hehe, I bet scientist mathematicians love it too [10:42 AM]
mat747: jee - what is the timeframe for 3D [10:43 AM]
eternacac: it’s a great field in math too [10:43 AM]
Eli Fisker: cool, eternac [10:43 AM]
jeehyung: @mat that is hard to say. We want to nail down switch first then probably more in-vivo rna designs [10:43 AM]
wateronthemoon: folding pattern texture color space [10:43 AM]
jeehyung: I’ll have to say later part of next year. [10:43 AM]
Eli Fisker: water, I’m almost thinking RNA haiku [10:44 AM]
jeehyung: Or if we get another full time grad student, maybe he can be in charge of Eterna 3D [10:44 AM]
eternacac: RNA haiku–that’s what the parser is for [10:44 AM]
Eli Fisker: lol [10:44 AM]
mat747: jee - np - we will still have fun in the meantime [10:45 AM]
wateronthemoon: yes lol [10:45 AM]
Eli Fisker: Yep. Lots of fun [10:45 AM]
mat747: oh and useful science [10:45 AM]
Eli Fisker: That too [10:45 AM]
rlmarchal: @jee - what field would the new grad be in? [10:46 AM]
jeehyung: @rimarchal well, he’ll probably have to be in CS to deal with all the game/web development and computational biology side [10:46 AM]
jeehyung: or someone in RNA science who can still handle the technical stuff : ] [10:46 AM]
rlmarchal: excellent [10:47 AM]
Brourd: @ Jee - a few players have asked where the lab scoring points and messages we gain after each lab round have gone to. Has that been removed in preparation for the cloud lab? [10:47 AM]
jeehyung: @Brourd oh I saw the message. Sorry I didn’t reply. That is a bug, and it’ll be fixed before the next round results. [10:48 AM]
Brourd: thx [10:48 AM]
Brourd: and I saw something for a link to a sequence browser when on the beta site a few weeks ago, when can we expect to see that :wink: [10:49 AM]
jeehyung: @Brourd, actually that was for the cloud lab, but it got pushed behind the other stuff now : [We’ll get back to it once the cloud lab is ready to kick off [10:49 AM]
eternacac: Gee give Jee a break! [10:50 AM]
mat747: jee - talking of the cloud lab, is any news on results for the players puzzles submitted before [10:50 AM]
jeehyung: @mat Rhiju’s team is still working on those 1500 sequences from player projects. It’s their first time to even try that many, and I think they are going over a lot of trial and error now. [10:51 AM]
Brourd: thanks Jee, let’s hope they have it all figured out in a few more months [10:52 AM]
jeehyung: Unfortunately I can’t really tell ETA. They are still working on it. [10:51 AM]
eternacac: newbie question–how do I see the solution I made for a puzzle after I’ve cleared it? Clicking on my profile link seems to take back to play the puzzle. [10:52 AM]
jeehyung: @Brourd That’d be best. synthesizing sequences in such scale will really turn the game around [10:52 AM]
jeehyung: @eternaac. When you re-enter the puzzle you cleared, the cleared solution will be loaded by default [10:52 AM]
eternacac: Oh!! [10:53 AM]
Eli Fisker: That question and answer will go straight to the FAQ :slight_smile: [10:53 AM]
jeehyung: However, if you try to re-solve the puzzle, then it would show that autosave instead of cleared solution. In that case, just reset the puzzle and reload to bring up the cleared solution again. [10:53 AM]
eternacac: great to know, thanx [10:54 AM]
Brourd: so, the server pulls the latest cleared solution? [10:55 AM]
jeehyung: @Brourd exactly [10:55 AM]
eternacac: do you get more points for a new solution resolve? [10:55 AM]
mat747: jee - it would be nice if we could have all our solution in a list [10:55 AM]
Eli Fisker: Eternac, no same numbers of points, no matter how you solve [10:55 AM]
jeehyung: @eternacac Unfortunately no : [you only get points for the first time solving it [10:55 AM]
jeehyung: @mat Ah, kind of like your own repository? That would be cool [10:55 AM]
mat747: jee - yes [10:55 AM]
Eli Fisker: Jee, I agree with Mat [10:55 AM]
eternacac: that’s why at this stage I didn’t want to resolve [10:55 AM]
Eli Fisker: Sometimes we just look for how a particular structure can be solved [10:56 AM]
Eli Fisker: and then we scan for that structure [10:56 AM]
Eli Fisker: That is what I do in my collection of screenshot puzzles [10:56 AM]
rlmarchal: @jee - any thoughts on making other players’ solutions available to those who have clearded a puzzle? [10:57 AM]
jeehyung: @rlmarchal, that’s interesting. We could combine this with mat and Eli’s idea… [10:57 AM]
eternacac: @mat, own repository–I’d been thinking along same lines–therefore my question above [10:57 AM]
Eli Fisker: That would be just beautiful, Jee [10:57 AM]
eternacac: Seems like making all known solutions available would hasten learning [10:58 AM]
rlmarchal: EteRNA is great for sharing [10:58 AM]
mat747: jee - do i need to post the idea for players to access to our own repository [10:58 AM]
Eli Fisker: Eternace, agree [10:58 AM]
eternacac: yes, how would I access eterna repository? [10:59 AM]
jeehyung: @mat I already wrote down your and rlmarchal’s idea in our issue list [10:59 AM]
mat747: cool thx [10:59 AM]
rlmarchal: thx [11:00 AM]
jeehyung: Actually the sequence browser idea is kind of like this too. It would be a big repo where everyone could see everyone else’s solutions for every puzzle… [with some restrictions] [11:00 AM]
eternacac: started my own collection, in self interest and self protection [11:01 AM]
eternacac: discovered I’m collecting some stuff much like J Pearl [11:01 AM]
jeehyung: @eternaac what are you collecting? sequences? [11:02 AM]
ElNando888: hello folks [11:03 AM]
Eli Fisker: Hi Nando :slight_smile: [11:03 AM]
eternacac: screen shots of lab high scores with dot plots, melts & stats [11:03 AM]
mat747: hi nando [11:03 AM]
jeehyung: @eternaac that’s neat [11:03 AM]
jeehyung: hi Elnando [11:03 AM]
eternacac: whence my asking for a way to discriminate very similar dot plots for any insights that might produce [11:04 AM]
Brourd: hey Elnando [11:04 AM]
ElNando888: is it me or some people are up quite late ? [11:04 AM]
Eli Fisker: I’m early - I don’t know about the rest :slight_smile: [11:05 AM]
eternacac: I’ll send ,post or publish—if I figure out how [11:05 AM]
jeehyung: @eternaac we hope to add that feature when we have pair-wise comparison interface : ] [11:05 AM]
Brourd: Only 2 am here? [11:05 AM]
ElNando888: :wink: [11:05 AM]
Eli Fisker: Great eternac, send me a link when you do [11:05 AM]
Eli Fisker: hehe [11:05 AM]
eternacac: I asked for a difference chart of A/B dot plots [11:05 AM]
Brourd: In another 3 hours, I’ll be up early [11:06 AM]
Eli Fisker: lol [11:06 AM]
jeehyung: sorry to cut the discussion. I have to run now : [Thanks for the chat today everyone [11:06 AM]
Eli Fisker: Bye Jee [11:07 AM]
ElNando888: ciao jee [11:07 AM]
rlmarchal: bye jee [11:07 AM]
Brourd: Take care Jee [11:07 AM]

8

Don’t start out by boosting your loops, hairpins, and bulges. As the player
validates the pairs back and forth between the two switches, unless the puzzle is symmetrical, the process outputs to one of the above loops types. So, reversing the logic, you can say the loop bases control the pairs patterns. Does that offer up it’s own stability within the loop types patterns, or are the loops the dumping grounds for the pairing process, don’t know.

9

Some Switch Design rules of thumb that I use:

Roughly equal numbers of GC bonds in each state. (Of course that depends on roughly equal number of bonds in each state.) Likewise roughly equal GUs.

Determine pairs that involve fixed NTs. Fixed As require Us to match. Prefer Cs over Us when matching fixed Gs. Prefer As over Gs when matching fixed Us. (So far no fixed Cs, but they would require Gs.) Switch state, and propagate dependencies outwards, preferring A’s over Gs to match Us and Cs over Us to match Gs, but not being 100% exclusive (exceptions allowed).

Typically don’t G-boost *any* loops unless the energy balance of the two modes is way off, then sometimes use a G-boost to compensate.

Try to avoid using CC and GG when GG occurs in the fixed bases. Likewise try to minimize the use of GU and UG and especially avoid UGU, GUU and UUG when AACAA occurs in the hook. (These sequences seem to introduce mismatches.)

11

Also, once I get a stable configuration I often play with it to try and improve its characteristics (e.g. dot plot). In rare cases you can flip a pair in one state (if there are no dependencies on the other shape). Sometimes you can flip a set of dependent pairs in one state in a way that also fixes all dependencies in the other state. More often you can change out one or more purines (A vs G) or pyrimidines (C vs U). It helps in this case to know what the dependencies are in the alternating states though. Discussed more in: “Color-Coded Hand and Finger and other tricks”

12

Adding chatlog from 28 nov, 2012:

jeehyung: hi all [6:22 PM]
jandersonlee: no News posting about it, and no response to messages, so it was a maybe at best for today [6:22 PM]
Eli Fisker: Hi Jee :slight_smile: [6:23 PM]
jandersonlee: ah there’s the man! [6:23 PM]
jeehyung: sorry our dev meeting got long today [6:23 PM]
jandersonlee: Hi Jee [6:23 PM]
Brourd: Hi Jee [6:23 PM]
Brourd: Everything go well with the dev meeting? [6:24 PM]
Eli Fisker: Gary, you might be safe with that extra gc-pair, as there are many short strings in this lab [6:23 PM]
jandersonlee: no problem, hopefully becuase you were reporting on all teh new features! [6:23 PM]
jandersonlee: how goes switch maker and scripting? [6:24 PM]
jeehyung: @Brourd yes lot of exciting sutff today : ] [6:24 PM]
jeehyung: @jandersonlee I’m working on the switch maker as we talk now. I think it’ll take few more hours…but the beta will be ready soon [6:25 PM]
Eli Fisker: cool [6:25 PM]
jandersonlee: nice jee! [6:25 PM]
Brourd: Great Jee, I saw that a few of the new In Vivo switch puzzles are up [6:26 PM]
Brourd: as well [6:26 PM]
jeehyung: @Brourd yes - it was solved faster than I could have imagined : P [6:26 PM]
Brourd: hehe [6:27 PM]
jeehyung: and here we have really primitive version of web scripting [6:26 PM]
jeehyung: http://kws.eternadev.org/web/script/ [6:26 PM]
jeehyung: hey all sorry - I got another brief dev meeting just now. Will be back soon! [6:27 PM]
Brourd: See ya then Jee [6:28 PM]
jandersonlee: ciao Jee [6:28 PM]
jandersonlee: @Brourd: the solution for FIX 6-11 also works for VAR 6-11 [6:30 PM]
Brourd: :smiley: [6:30 PM]
Brourd: I would hope so [6:30 PM]
jeehyung: hi all back : ] [6:30 PM]
Eli Fisker: wb [6:30 PM]
Brourd: welcome back Jee [6:31 PM]
Eli Fisker: Jee, any instructions on how to use the new script interface? [6:31 PM]
Brourd: So, with these new in vivo switches, is there going to be a separate lab where we submit solutions, or are all the solutions automatically sent to the server going to be used? [6:32 PM]
mat747: wb [6:32 PM]
jmf028: hello jeeyung [6:38 PM]
jmf028: jeehyung* [6:38 PM]
Brourd: You can answer Eli’s question first [6:32 PM]
jmf028: never met you before lol [6:39 PM]
jeehyung: @Eli Actually, Justin who is in charge of the interface is creating the documentation now. It’ll be far better documented when it’s actually out. [it’s very primitive version!] [6:32 PM]
jeehyung: @jmf good to meet you : ] [6:32 PM]
Eli Fisker: Jee, I’m happy to hear [6:32 PM]
jeehyung: @Brourd - that’ll be a separate lab [6:32 PM]
jeehyung: and in the new lab it’s likely that we are not going to have structure constraint [6:33 PM]
Brourd: Tell Justin Thaks [6:33 PM]
Brourd: Ahh [6:33 PM]
jeehyung: meaning that you can submit solutions even if it dosnt exactly fold into the target shape. [6:33 PM]
Brourd: yep, understood [6:34 PM]
jmf028: wait jee [6:40 PM]
Brourd: for both states? [6:34 PM]
Eli Fisker: That will be interesting. [6:33 PM]
jeehyung: Brent [who provided us the swithces] would love to see variety of designs [6:33 PM]
mat747: Yes thx justin [6:33 PM]
jmf028: how does that make it better [6:40 PM]
jeehyung: @Brourd - yes for both states [6:34 PM]
Eli Fisker: Will allow us to see how nature behaves without the restraints of the energy model, but might also make it a bit harder to hit right [6:34 PM]
Eli Fisker: as the energy model searves as sort of a guide line [6:34 PM]
jeehyung: @jmf For this time, we want to test how this goes. There is no guarantee it’ll get better, but it’s worth trying : ] [6:34 PM]
jmf028: got that right eli [6:41 PM]
jmf028: lol alright [6:41 PM]
jmf028: your the expert [6:41 PM]
jmf028: :smiley: [6:41 PM]
Brourd: sounds great Jee, and we shall give Brent as much variety as we possibly can [6:35 PM]
jandersonlee: so design to the guidelines then tweak and submit? [6:35 PM]
jandersonlee: At least for the bound target :slight_smile: [6:36 PM]
jmf028: i would same thing janderson [6:43 PM]
jeehyung: Also, one nice thing about the puzzle is that Brent already has a lot of experimental results [6:36 PM]
jmf028: i wouldn’t know how to know when to submit without following energy model [6:43 PM]
Eli Fisker: Oh, I like that [6:37 PM]
jeehyung: So we may start with a lot of feedbacks from the beginning [6:37 PM]
Brourd: so, what is their target for glowing? [6:38 PM]
jeehyung: @Brourd I’m not so sure of glowing, but what I heard is the switch will be a trigger for certin type of protein production [6:39 PM]
Brourd: ahh [6:39 PM]
jnicol: hi everyone [6:39 PM]
Eli Fisker: Hi John :slight_smile: [6:39 PM]
jmf028: hey jnicol [6:46 PM]
Eli Fisker: So soon we will not only be makers of RNA, but protein too. In one go :slight_smile: [6:40 PM]
jmf028: i keep thinking brourd is yelling [6:47 PM]
Jieux: Hello everyone, just getting in myself for a minute… Eli… will you have a transcript available, I’m just catching the end of the conversation. [6:41 PM]
jmf028: but that would be agh [6:47 PM]
jeehyung: @jnicol, Jieux hi [6:40 PM]
jeehyung: We’ll have the chat log posted as a news : ] [6:40 PM]
Brourd: sounds good Jee, when will these new labs start? [6:41 PM]
Eli Fisker: Hi Jieux, yes, coming [6:40 PM]
Brourd: hi jnicol [6:41 PM]
jmf028: hey jieuz [6:47 PM]
jmf028: x* [6:47 PM]
jeehyung: Well we’ll first have to finish “Hands and Finger” first. It’ll also give our new synthesis team member time to get used to switches [6:41 PM]
jmf028: finish as in get through all the rounds [6:48 PM]
jmf028: ? [6:48 PM]
jandersonlee: and our new players :slight_smile: [6:42 PM]
jeehyung: @jmf yes finish as in finding a successful design, or maybe going through 4-5 rounds [6:42 PM]
jeehyung: @jandersonlee that too : ] [6:42 PM]
jmf028: okay [6:48 PM]
jandersonlee: how is the new machine behaving? [6:42 PM]
Brourd: Well, I suppose we will have the switch maker and scripts to hold us over those 2 1/2 months [6:43 PM]
Brourd: so, that’s fine with me :slight_smile: [6:43 PM]
jmf028: we can make switches? [6:49 PM]
jandersonlee: (Small and Easy Switch :slight_smile: [6:43 PM]
jmf028: :smiley: [6:49 PM]
bigbloto: we have switch maker? [6:43 PM]
jeehyung: @jandersonlee I think it’s going ok. They are trying to redo last few rounds [which had lot of failures] [6:43 PM]
jmf028: lol bigb ima make the first one [6:50 PM]
Eli Fisker: JL, lol [6:43 PM]
jeehyung: @jmf, bigbloto yes - the switch maker is on it’s way : ] [6:43 PM]
bigbloto: race [6:43 PM]
bigbloto: ce [6:43 PM]
Brourd: btw Jee, will we get the new scoring protocol and a thorough for this coming round of switch results? [6:44 PM]
jmf028: whens it due Jeehyung [6:51 PM]
jmf028: range of time? [6:51 PM]
Brourd: *thorough explanation [6:44 PM]
jnicol: can we also get which bases are being scored? [6:44 PM]
jnicol: even for older labs? [6:45 PM]
jmf028: what does that mean jnicol? [6:52 PM]
jmf028: just asking? [6:52 PM]
jeehyung: @Brourd they did a new scoring system and they are proofreading it now. If everything goes well we’ll get new scoring before enxt round [6:45 PM]
jnicol: i’m working with the older labs, and would like to correlate the numbers i get with the scored values [6:46 PM]
jmf028: oh i see [6:52 PM]
jmf028: like correlating with a 95 lab score? [6:53 PM]
jeehyung: @jnicol Yes that’s possible. We can add that as a feature. Let me add that to the task list [6:47 PM]
jnicol: yes, i want to make sure i can reproduce the overall scoring [6:47 PM]
jmf028: ok [6:54 PM]
jmf028: cool [6:54 PM]
jnicol: also, should the data be normalized by scaling linearly from min to threshold and from threshold to max (for older labs) [6:47 PM]
jnicol: sometimes this scaling is not symmetrical [6:47 PM]
jeehyung: @jnicol yes they are normalized exactly in the way you described [6:48 PM]
jnicol: nice, thx [6:48 PM]
jeehyung: It’s correct they are not symmetrical [6:48 PM]
Ns52: rank 525 now YAY!! [6:48 PM]
jnicol: and you scale linearly, correct [6:48 PM]
jeehyung: that’s right [6:49 PM]
jeehyung: @Ns52 congrats : ] [6:49 PM]
jmf028: gj Ns [6:55 PM]
jnicol: elnando and I were talking this morning, would it possible to see the RNASSD source code for the bot? [6:49 PM]
Ns52: jmf working on any new puzzles lately [6:49 PM]
jmf028: i just finished the squid [6:56 PM]
Eli Fisker: s52 = 552 Nice letter number match [6:50 PM]
jmf028: haha [6:57 PM]
jeehyung: @jnicol we do have the source code. But I think we’ll have to ask the authors to see if they want the code in public. [6:50 PM]
Ns52: ok i will try that puzzle jmf [6:50 PM]
jeehyung: Let me query them first [6:50 PM]
jnicol: if you need a non-disclosure or some kind of waver, we understand [6:50 PM]
jnicol: im willing to sign [6:50 PM]
jmf028: Ns ive got Dire wolf, mantaray, anteater, and squid made [6:58 PM]
jmf028: at least since i talked to you last [6:58 PM]
jnicol: i wasnt thinking public, perhaps we can be considered as part of the dev team? [6:51 PM]
Ns52: yeah i saw all those [6:51 PM]
jmf028: okay cool [6:58 PM]
Ns52: they all looked nice [6:52 PM]
jnicol: public would be better though if thats possible [6:52 PM]
jmf028: thanks [6:59 PM]
jmf028: what is this waver we are signing? [6:59 PM]
jmf028: Am i to be included or dont worry? [6:59 PM]
Ns52: what waver are you talking about [6:52 PM]
jmf028: im pretty sure we aren’t included [6:59 PM]
jmf028: me and you Ns [7:00 PM]
jnicol: no need to worry jmf, it would be to view proprietary source code [6:53 PM]
jmf028: oh I see [7:00 PM]
Ns52: k [6:53 PM]
jeehyung: @jnicol I don’t think it’ll get very complicated and I’m sure they’ll agree to share the code with you [and some other players]. But I just want to make sure they are ok with it : ] [6:54 PM]
dmc_22: http://eterna.cmu.edu/sites/default/f… [6:54 PM]
Paisana: hii guys!(: [6:54 PM]
dmc_22: i cant get the 2nd loop to bond appropriately :confused: [6:54 PM]
Eli Fisker: Hi Paisana [6:54 PM]
jeehyung: Hi dmc_22 and Paisana [6:54 PM]
jnicol: great, that could really help us [6:55 PM]
jmf028: hi pals [7:01 PM]
Eli Fisker: Dmc, take out the blue from that loop [6:55 PM]
jandersonlee: hmm maybe we need chat rooms :slight_smile: [6:55 PM]
Eli Fisker: and put in one red at each opening of that loop [6:55 PM]
jmf028: OOOOOHHHH [7:02 PM]
jmf028: i like jandersons idea [7:02 PM]
dmc_22: eli i tried that first and it wouldnt go and ive tried swapping paired bases [6:56 PM]
Brourd: @ Jee - does RNASSD use the constraint generation energy model? [6:56 PM]
jeehyung: @Brourd Do you mean if they are subject to the pair constraints as players do? [6:56 PM]
Eli Fisker: Dmc, you will propably also need to stabilize that 2-2 loop. [6:56 PM]
Eli Fisker: Check this loop guide: [6:56 PM]
Brourd: no, the CG energy model Andronescu published [6:57 PM]
Eli Fisker: https://docs.google.com/document/d/1r… [6:57 PM]
dmc_22: ahh got it the red at the bottom of the first loop did the trick, thanks! [6:57 PM]
jeehyung: @Brourd I’m not faimiliar with the work, let me check [6:57 PM]
Eli Fisker: np [6:57 PM]
Brourd: As it was able to solve a puzzle that is unsolvable using the Turner 99 parameters, so, just curious :slight_smile: [6:58 PM]
jeehyung: Ah it’s 2010 paper. Then no. RNASSD code was created in 2004 using the same Turner 99 parameters as ViennaRNA [6:58 PM]
jeehyung: It may be using slightly different setting and that might be where the difference is coming from [6:58 PM]
Brourd: okay, thanks :slight_smile: [6:59 PM]
jmf028: doing your training bigb gonna be first to solve lol [7:06 PM]
Brourd: So Jee, run into any problems with the switch puzzle maker? [7:02 PM]
hotcreek: hello [7:02 PM]
Brourd: hey hotcreek [7:03 PM]
jmf028: hello hotcreek :smiley: [7:09 PM]
hotcreek: hi brourd [7:03 PM]
hotcreek: hi jmf [7:03 PM]
Eli Fisker: HI Hotcreek :slight_smile: [7:03 PM]
hotcreek: hi eli [7:03 PM]
hotcreek: how are you guys [7:03 PM]
jeehyung: @Brourd a few : [Many things can go wrong while creating switches so I’m spending some time to put safety checks [7:03 PM]
jeehyung: hi hotcreek [7:04 PM]
hotcreek: hi jee [7:04 PM]
jandersonlee: @jee: just the two in vivo challenges so far? and why were you surprised how quickly they were solved? [7:04 PM]
Eli Fisker: Hotcreek, good here [7:04 PM]
jeehyung: @jandersonlee just two for now, but more will come. [7:04 PM]
jeehyung: And I was surprised because I was almost sure those 2 were unsolvable [7:04 PM]
jeehyung: But clearly I was wrong : P [7:04 PM]
Eli Fisker: :slight_smile: [7:05 PM]
jandersonlee: What you don’t have to solve to submit :wink: [7:05 PM]
Brourd: Ahh, yeah, there were a few things I was wondering about Jee, as both shapes have to have an equal number of bases, but when you add one onto one, where will one added on to the other go [7:05 PM]
jeehyung: @jandersonlee, well we try. But we know we are not as good as our players. So we still publish them in the hope. [7:05 PM]
jeehyung: And it usually works : ] [7:05 PM]
Brourd: :smiley: [7:06 PM]
ElNando888: hello everyone [7:06 PM]
jeehyung: @Brourd the new base gets attached at the end [7:06 PM]
jandersonlee: give it to Brourd. if he cannot solve it… [7:06 PM]
Eli Fisker: Hi Nando :slight_smile: [7:06 PM]
jeehyung: hi ElNando [7:06 PM]
jmf028: Nando i just passed you in rank [7:13 PM]
jmf028: i think 2 ahead buddy :smiley: [7:13 PM]
jmf028: no big deal [7:13 PM]
Brourd: That will work as well Jee :slight_smile: [7:07 PM]
jmf028: by the way hi Nando lol [7:14 PM]
jandersonlee: If we resolve the puzzle a different way, does it get looged (I know we don’t get extra credit) [7:07 PM]
ElNando888: congrats jmf, but I stopped solving puzzles quite a while already [7:07 PM]
jmf028: :frowning: [7:14 PM]
ElNando888: got bored with those GU things [7:07 PM]
jeehyung: @janderson yes every player solution gets logged [7:07 PM]
Eli Fisker: Nando, how is your bot? [7:07 PM]
ElNando888: making progress, slowly [7:08 PM]
Eli Fisker: Happy to hear [7:08 PM]
jnicol: jmf, you have to add elnando points with his bots points, he is still ahead! [7:08 PM]
hotcreek: hi jnicol [7:08 PM]
Eli Fisker: john, :slight_smile: [7:08 PM]
jmf028: well jnicol if im only two ranks ahead then of course im not ahead with his combined total [7:15 PM]
jmf028: :smiley: [7:16 PM]
jnicol: hello, but i’ve been here awhile :wink: [7:09 PM]
hotcreek: well i havent been on for a wwhile [7:09 PM]
jeehyung: I have to go back to deveopment now : [I’ll post a chat log later. If you have any question, please message me. [7:09 PM]
jeehyung: Thanks for the chat everyone : ] [7:10 PM]

13

From Design of Multistable RNA:
http://rnajournal.cshlp.org/content/7…

Using an intersecting graph theory algorithm–

"The procedure defined…was implemented using the scripting language Perl. This allows easy modifications…
On the other hand, the program makes use of the C routines of the Vienna RNA Package (via a Perl extension module)
, and thus has access to fast routines for computation of RNA secondary structures and base-pairing probabilities.

For the small example [33-nt]…the program takes about 1.3 s per sequence on a 333 MHz Pentium II.
While manual design is not too hard for such an example, the optimization procedure yields significantly
better sequences. For the 115-nt SV11 example…it designs one sequence in about 10 min."

So, a 3 GHz dual or quad processor should do our switch labs in 15-60 seconds by my estimation.

Got my attention!

14

Stumbled upon this forum post again and thought I would add a little. Several eterna players have shared their switch puzzle solving strategies in the meantime. There is a collection of them in here.

Switch puzzle guide

For others having a switch puzzle solving strategy, feel free to share.