2013.05.01 Dev Chat

From Eterna Wiki

( Link back to Dev Chat archive )


On New Lab Interface


jeehyung: hello all [6:01 PM]

jeehyung: welcome to the Eterna dev chat : ] [6:02 PM]

Eli Fisker: hi Jee :) [6:01 PM]

TomoeUzumaki: hey jee! [6:01 PM]

starryjess: hi Jee [6:01 PM]

fluffy3: hello [6:02 PM]

Brourd: Hi Jee, how are you today? [6:02 PM]

jandersonlee: (obligatory) Hi Jee [6:01 PM]

jeehyung: hi Eli, Tomoe, starry, fluffy Brourd, janderson [6:02 PM]

jeehyung: Good here : ] [6:02 PM]

TomoeUzumaki: jee, when is the app coming out? [6:02 PM]

Brourd: That is good :) [6:02 PM]

TomoeUzumaki: just wondering? [6:02 PM]

jeehyung: @Tomoe we are actually hiring few students from UCB to take care of that this summer [6:03 PM]

jeehyung: Hopefully by Fall we'll ahve something [6:03 PM]

TomoeUzumaki: :D [6:03 PM]

TomoeUzumaki: yaay [6:03 PM]

TomoeUzumaki: that would make my life [6:03 PM]

TomoeUzumaki: amazing [6:03 PM]

Brourd: very nice Jee [6:03 PM]

Eli Fisker: @Jee, Mat couldn’t be here, but I have some questions from him [6:03 PM]

jeehyung: @Eli of course : ] [6:04 PM]

Brourd: saw the updates to the eternadev site, looks good [6:04 PM]

Eli Fisker: @Jee, there are overlapping sections in labs. A suggestion for a solution to this problem, is if we can have layers? So the hairpin barcode can be in an underlying layer and the main design above and we can swap between the layers when solving. [6:04 PM]

Eli Fisker: Here is a lab with overlap: http://prntscr.com/12thsg [6:04 PM]

jeehyung: @Brourd oh which one? there are bunch of updates that's going on there [6:04 PM]

Brourd: lab proposal, scripting, etc. [6:04 PM]

jeehyung: @Eli Ah. layer idea is good. Got it. I'm writing it down [6:05 PM]

osiah jay: bla bla bla [6:05 PM]

Eli Fisker: thx [6:05 PM]

Brourd: also saw the ability to add bases to the lab designs, any idea when the riboswitch lab will launch with that? [6:05 PM]

TomoeUzumaki: what bases [6:05 PM]

Eli Fisker: @Jee, one more from Mat: [6:05 PM]

jeehyung: @Brourd probably soon. We'll first have to finish up cloud lab though [6:05 PM]

Eli Fisker: There are overlapping sections in puzzles also. If we can have layers for puzzles too, it will help us deal with and view puzzles that has overlapping sections. [6:05 PM]

Eli Fisker: Examples on overlapping puzzles, several levels of the same shapehttp://eterna.cmu.edu/game/puzzle/422344/, http://eterna.cmu.edu/game/puzzle/412619/,http://eterna.cmu.edu/game/puzzle/412619/ [6:05 PM]

Matt Indykiewicz: hello Eli [6:06 PM]

osiah jay: hello [6:06 PM]

Eli Fisker: hi Matt [6:06 PM]

Eli Fisker: hi Osiah [6:06 PM]

jeehyung: @Eli got it - layed interface for the normal puzzle as well [6:06 PM]

osiah jay: hello [6:06 PM]

jeehyung: hi Matt, osiah [6:06 PM]

Brourd: oh, sounds good Jee, very good [6:06 PM]

Eli Fisker: @Jee, great [6:06 PM]

osiah jay: hello [6:06 PM]

Brourd: hi osiah [6:07 PM]

Brourd: @ Jee - btw, saw the strategy market was added back on to the main site [6:07 PM]

jeehyung: We'll also launch web version of the lab voting browser as well. Hopefully that'll make browsing and voting much eaiser [6:07 PM]

Brourd: yay! [6:07 PM]

TomoeUzumaki: :D\ [6:07 PM]

jeehyung: @Brourd yes although we are not fully ready to re-open strategy market yet [6:08 PM]

Yoruno: heya, what's this "kcal" value? should i be paying attention to it? [6:07 PM]

Brourd: a lot of updates soon then [6:08 PM]

TomoeUzumaki: mmhm [6:08 PM]

ElNando888: any ETA for the results of Cloud Lab switches ? [6:08 PM]

jeehyung: @Yoruno kcal value is the free energy state of the RNA. The lower it is, the more likely that the RNA is stable [6:08 PM]

Yoruno: ah... so negative value is good? [6:08 PM]

TomoeUzumaki: jee, why was eteRNA offline from 3:30 to 4:00 est? [6:08 PM]

starryjess: @jee, are all the designs going to be picked for the next round of labs like they were for the last one? [6:08 PM]

jeehyung: @ElNando, I actually got the results now. We should be able to publish it on Friday without scoring. The scoring will arrive next week. [6:09 PM]

jeehyung: @Yoruno yes negative is good : ] [6:09 PM]

jandersonlee: @jee: any chance of end-user scripting of strategy scripts? [6:09 PM]

ElNando888: thx jee [6:09 PM]

Yoruno: great, thanks [6:09 PM]

fluffy3: I thought there was a range of negative energy for folding? [6:09 PM]

jeehyung: @Tomoe, I'm still figuring that out. It seems like there was a burst of traffic, [6:09 PM]

TomoeUzumaki: ah [6:09 PM]

TomoeUzumaki: because I was worried [6:09 PM]

starryjess: uh oh [6:09 PM]

TomoeUzumaki: thx [6:10 PM]

jeehyung: @starry yes we'll pick 1600 designs and if the # of designs submitted are less than it, then all of them get picked [6:10 PM]

starryjess: ah, ok [6:10 PM]

Eli Fisker: Cool [6:10 PM]

Brourd: weird, denial of service attack? [6:11 PM]

jandersonlee: 160 per lab - or will there be more before the deadline [6:10 PM]

ElNando888: any insider tips on those external lab projects which are supposed to be added to the current ones ? [6:11 PM]

jeehyung: @janderson about the strategy market & scripting, that's the plan. WE hope to ask players to start implementing submitted strategies using the scipting interface [6:12 PM]

Eli Fisker: @Jee, I saw you said that you were not fully ready to re-open the strategy market. Before we had an option to run a strategy in the strategy market. Can we get that option back, so we can see how a strategy scores different lab designs? [6:12 PM]

jeehyung: @janderson, ElNando. Most likely they will be picked with new cloud lab interface. And we'll have "voting" for puzzles there. [6:12 PM]

jandersonlee: @jee - cool [6:12 PM]

Brourd: great! [6:13 PM]

jeehyung: @Eli - Got it. I'll ask the scripting dev to add "run strategy on solutions" function as well as "run puzzle solver on player puzzles" [6:13 PM]

Eli Fisker: Jee, big thx :) [6:13 PM]

Eli Fisker: I like to be able to see what our bot is doing [6:14 PM]

Brourd: Will the scoring for the cloud labs be added to the eternabot? [6:14 PM]

Brourd: Well, I know it will, but how about a date by when that will happen [6:14 PM]

Matt Indykiewicz: Purple [6:15 PM]

fluffy3: May be the wrong forum for this question: what is the desirable range for negative kcal's for folding stability? [6:15 PM]

JR: will we be able to see where the bot failed? [6:15 PM]

TomoeUzumaki: what did I miss? [6:15 PM]

jeehyung: We'll add option to run "strategy" on all designs submitted in the lab puzzles [6:15 PM]

Brourd: @ fluffy - it is a bit of a hard question to answer, it varies based on the puzzle, what you want the constraints to be, etc. [6:16 PM]

TomoeUzumaki: ah [6:16 PM]

jeehyung: @fluffy3 Actually, there is not a specific range,. It's important that your kcal in your current shape is lower than any other kcal in other shape. That grants you stability. [6:16 PM]

TomoeUzumaki: jee, I just got some designs synthesised (they're still getiing synthesised) [6:16 PM]

jeehyung: @JR - all the bots will report specfiically which sequence they used on the puzzles. So you'll be able to see how it did [6:16 PM]

jandersonlee: how about running strategies against lab results? proof in the pudding and all that. [6:16 PM]

Eli Fisker: fluffy, it will depend on the designs size. Big designs have lots of negative energy. But generally designs with max amount of negative energy (usually all GC-pairs) do bad, and same with puzzles lowest amount of negative energy, usually those with all AU-pairs, that will also generally do bad [6:16 PM]

TomoeUzumaki: how long does it take to get results out? [6:16 PM]

JR: good thx [6:17 PM]

fluffy3: *Brourd. Could you provide a range? the highest and lowest for folding stability? [6:17 PM]

Brourd: @ tomoe - from the last batch, probably 3-5 weeks from now [6:17 PM]

jeehyung: @janderson - yes all the lab designs including synthesized ones (results). As you said it'll give you good overview on how well those straegies are doing : ] [6:17 PM]

TomoeUzumaki: Brourd, thanks! [6:17 PM]

Brourd: sorry fluffy, as Jee said, your target kcal just has to be the "lowest" in order to be stable [6:17 PM]

jandersonlee: natural mode == target mode [6:17 PM]

fluffy3: So, the lowest possible for that particular puzzle? [6:18 PM]

jandersonlee: (assuming the model is right) [6:18 PM]

jeehyung: @All, I actually have to leave bit early today around 6:30pm EST (I'm sorry - I got a meeting that I can't get out of). So please throw in as many questions as you can now : ] [6:18 PM]

Eli Fisker: @Jee, Mat is suggesting we get an extra category in our feed. Just like we have categories for all messages, news, blogs, lab, he suggest that we get a category for group messages. [6:18 PM]

Brourd: not necessarily the lowest possible [6:18 PM]

jeehyung: @Eli group category is actually being implemented now by our dev, Diana : ] [6:19 PM]

TomoeUzumaki: ;D [6:19 PM]

fluffy3: Is that information available somewhere for each of the puzzles (the lowest possible kcal)? [6:19 PM]

Eli Fisker: Great :) [6:19 PM]

Eli Fisker: @Jee, another request from Mat: It is getting hard to find messages. Can we get the option to search in messages with a word or sort them after senders? [6:19 PM]

TomoeUzumaki: jee, some of the groups are "broken" [6:19 PM]

jandersonlee: will we be able to "fix" bases in cloud labs? [6:19 PM]

TomoeUzumaki: as in, it is impossible to subscribe to them [6:19 PM]

Yoruno: i just got -47.1 on synthetic lvl0 if that helps indicate... mightve overdone it somehow [6:19 PM]

jeehyung: @jandersonlee - yes you'll be able to lock the bases [6:20 PM]

jeehyung: @Tomoe can you tell me which group it is? [6:20 PM]

Eli Fisker: @Jee, Spacefolder had a suggestion for cloud labs: Please keep numbering them consecutively. It makes it easier to navigate (and refer) to puzzles. [6:20 PM]

Eli Fisker: Just like the first static cloud lab had numbers to all the lab designs. [6:20 PM]

TomoeUzumaki: jee, the group is [6:20 PM]

jeehyung: @Eli - noted.Thanks! [6:20 PM]

TomoeUzumaki: EteRNAddicts Anonymous [6:20 PM]

Eli Fisker: np [6:20 PM]

Brourd: @ Jee - new players, when they register and try to reclaim their points, get stuck with the reclaiming points screen [6:20 PM]

jandersonlee: will teh cloud labs be the same 63 bases? [6:20 PM]

Brourd: it doesn't automatically redirect [6:20 PM]

fluffy3: Yoruno-I have no idea if that kcal amount is good or bad [6:20 PM]

jeehyung: @Brourd ok I'll look into it. Thanks for the pointer. [6:21 PM]

fluffy3: for folding [6:21 PM]

TomoeUzumaki: oop, I have to go [6:21 PM]

TomoeUzumaki: swim practice :D [6:21 PM]

jeehyung: @janderson yes they'll still have 63 bases for now. Although that may change in the future. [6:21 PM]

Brourd: so, some will exit and lose points, etc. if they don't refresh [6:21 PM]

Eli Fisker: @Jee, Zanna requested that we can edit puzzle descriptions (and delete newly made and erroneous puzzles) Quote: “They need to make it so we can change it later.” [6:21 PM]

jandersonlee: +1 [6:21 PM]

JR: are there 10 more cloud labs to be added to the existing 10? [6:21 PM]

hoglahoo: fluffy3: the kcal amount is not good or bad in itself, it is just a calculated result. It is the comparison of that kcal with all other possible folds, which is hard to deduce [6:21 PM]

jeehyung: @Eli got it. Editable puzzle content [6:21 PM]

TomoeUzumaki: bye guys! [6:21 PM]

TomoeUzumaki: swim practice calls [6:21 PM]

Brourd: later tomoe [6:22 PM]

Eli Fisker: Thx, Jee [6:22 PM]

jeehyung: @JR we have 9 now. We'll add 11 from the cloud lab [6:22 PM]

jeehyung: bye Tomoe [6:22 PM]

hoglahoo: fluffy3, you could get the same kcal value two different ways: one way could be terrible and the other way could be a winner. It is the pairs and mutations that matter [6:22 PM]

TomoeUzumaki: later Brourd, jee, Eli, jess, hogla, etc. [6:22 PM]

Eli Fisker: Bye Tomoe [6:22 PM]

fluffy3: @hog so, the most desirable kcal for each puzzle is yet to be determined? [6:22 PM]

hoglahoo: later, Tomoe [6:22 PM]

TomoeUzumaki: and fluffy [6:22 PM]

JR: is barcode included in scoring? [6:22 PM]

TomoeUzumaki: and Matt [6:22 PM]

jandersonlee: how soon will the cloud-lab open? [6:22 PM]

Brourd: so, are we going to stick with regular puzzles or throw some switches in with that 11? [6:22 PM]

fluffy3: bye tom [6:22 PM]

hoglahoo: fluffy3, no, there is no most desirable kcal. There is a most desirable sequence, which in turn yields a kcal [6:22 PM]

hoglahoo: but that kcal can be produced many different ways [6:23 PM]

Brourd: @ JR - all bases with blue/yellow score are added to the ultimate scoring [6:23 PM]

jeehyung: @janderson we are running quality check on Friday. We'll open it for the beta testing next week [6:23 PM]

Jeanny: poop [6:23 PM]

hoglahoo: So you can't say the kcal itself is good [6:23 PM]

JR: ok thx [6:23 PM]

Brourd: so, 79 bases scored in every lab if I remember correctly [6:23 PM]

fluffy3: Is it the kcal that is stabilizing the fold or the sequence or are they one in the same? [6:23 PM]

Brourd: that sounds good Jee! [6:24 PM]

hoglahoo: fluffy3, I'm not even sure excactly. The way I thought is the kcal we are looking at is how much energy the model thinks it would take to destroy the structure? [6:24 PM]

Brourd: so, no scoring for the switches until next week though? Is this because the interface to view multiple mapping methods comes with the new lab browsing? [6:25 PM]

fluffy3: Hog, so , the kcal may be the most important? [6:25 PM]

JR: theo update - new labs? [6:25 PM]

jeehyung: @Brourd not exactly. here was a confusion in the pipeline between analysis & scoring. We have to reset that process [6:25 PM]

Brourd: oh [6:25 PM]

hoglahoo: fluffy3: no, because destroy can mean misfold. You are trying to fold the structure - so you do need negative kcal. But to avoid misfolds, you cannot just shoot for some particular kcal [6:25 PM]

Matt Indykiewicz: purple [6:25 PM]

Brourd: we should probably prevent that from happening in the future :) [6:26 PM]

jeehyung: @JR theo labs will be back once we get the base add/remove functions working in the puzzles. [6:26 PM]

JR: good thx again [6:26 PM]

jeehyung: @Brourd will do : ] [6:26 PM]

Brourd: great! and the results will be for [6:26 PM]

fluffy3: Hog, yes, I see, each puzzle would have it's own desirable kcal for folding properly [6:26 PM]

Brourd: switches, mutate and map experiments, and a bunch of other stuff as well? [6:27 PM]

Brourd: or are we only getting the switch results in this batch? [6:27 PM]

hoglahoo: fluffy3: no, each has its own desirable sequence. But different sequences can produce the same kcal [6:27 PM]

starryjess: @jee, but you're already seen the results? [6:27 PM]

hoglahoo: So the kcal is only desirable if it is the result of that desirable sequence [6:27 PM]

hoglahoo: if it is the result of an undesirable sequence, it is not desirable anymore, even though it may be the exact same value [6:27 PM]

starryjess: *you've [6:27 PM]

fluffy3: hog so the sequence may be more important than the kcal? The same kcal may not guarentee a proper fold? [6:28 PM]

starryjess: of the switch labs [6:28 PM]

hoglahoo: fluffy3: right [6:28 PM]

fluffy3: Thank you for that explanation, Hog, I learned something new today! [6:28 PM]

Brourd: nice! [6:29 PM]

fluffy3: gtg [6:29 PM]

Yoruno: i dont even know the relationships of these colors properly, im just making it as colorful as possible and make it fold properly while at it... [6:29 PM]

hoglahoo: great! [6:29 PM]

jandersonlee: will it just be non-switches for the first cloud lab round? I'm assuming the processing differs [6:29 PM]

Brourd: bye fluffy [6:30 PM]

hoglahoo: I mean, great you learned something not great you have gtg [6:29 PM]

jeehyung: @Brourd just the switch results this round. We are still working on multiple mapping data viewer [6:30 PM]

fluffy3: I understand [6:30 PM]

jnicol: i need to run now, thanks for the info jee :) [6:30 PM]

jeehyung: @starry yes we have the results. We just couldn't score it yet : [ [6:30 PM]

jeehyung: @jnicol - see you! [6:30 PM]

Eli Fisker: bye John and fluffy [6:30 PM]

Brourd: Do you think we got any winners Jee? [6:30 PM]

starryjess: can you say whether they are overall good or overall bad? [6:30 PM]

Brourd: or are we just absolutely terrible switch makers :) [6:31 PM]

jeehyung: @janderson it'll be none switch labs for the cloud lab. We'll add switch function in the next round [6:31 PM]

Brourd: @ starry - no score, no way to know [6:31 PM]

jandersonlee: My guess is that they were a bit "challengiing" [6:31 PM]

jeehyung: @Brourd,starry frankly I don't know yet. We'll ened to get scores first. [6:31 PM]

starryjess: ok [6:31 PM]

Eli Fisker: hehe, that was a fine way to put it, JL [6:31 PM]

jeehyung: I actually have to run now all. Sorry to cut the discussion short. [6:32 PM]

starryjess: hehe [6:32 PM]

starryjess: bye Jee [6:32 PM]

hoglahoo: jee, you meant starryjess not starryfrankly ;) [6:32 PM]

Eli Fisker: Bye Jee [6:32 PM]

jandersonlee: np Jee tanks [6:31 PM]

starryjess: lol hog [6:32 PM]

jeehyung: I'll post the chat log in the news item. Please feel free to message me if you have further questions : ] [6:32 PM]