2014.06.25 Dev Chat

From Eterna Wiki

( Link back to Dev Chat archive )

 

On Barcodes & NOVA 


jnicol: Hi everyone, Welcome to dev chat! [3:00 PM]

jnicol: Hi and byr Drake [3:00 PM]

Drake: The irony xD [5:00 PM]

machinelves: yay! hello :D [3:00 PM]

Brourd: Hello jnicol. Nice to see you. [3:00 PM]

machinelves: ( and bye fire! ) [3:00 PM]

jnicol: and to see you also [3:01 PM]

Brourd: Why, it only feels like we were chatting a few minutes ago. [3:01 PM]

Eli Fisker: Hi John :) [3:01 PM]

jnicol: Hi Eli, elves [3:01 PM]

jnicol: Brourd, when do you want the Mimic 2 lab to go live? [3:01 PM]

Brourd: Whenever I put the finishing touches on it. I hope to have that done soon. [3:01 PM]

jnicol: ok, just want to make sure I'm not the holdup :) [3:02 PM]

jnicol: on this anyway [3:02 PM]

jnicol: I've been busy lately learning the database structure and writing some backend scripts to gather the sequences for the Pilot labs synthesis. [3:02 PM]

Brourd: excellent [3:03 PM]

jnicol: does everyone know how the pilot lbs work now? [3:03 PM]

Brourd: I've been writing up results and taking screenshots, fun. [3:03 PM]

machinelves: John, I PMd these things to you when I thought dev chat was last week, so here's a quick refresher in case they got buried since then:  [3:03 PM]

Eli Fisker: John, no  [3:03 PM]

jnicol: when we have excess slots, the labs with the most votes are also included in the synthesis  [3:03 PM]

machinelves: here's a good interface suggestion on autosave from Mat747 & Eli, I PMd it to Walt, but figured it might help to send it to you as well:  [3:04 PM]

jnicol: we can go up to 1000 slots  [3:04 PM]

Eli Fisker: John, thx, I will add in the WIKI  [3:04 PM]

machinelves: https://getsatisfaction.com/eternagame/topics/autosave_undo_improvement_idea  [3:04 PM]

jnicol: the pilot labs do not have a barcode attached until they are selected for synthesis  [3:04 PM]

Brourd: Yes  [3:05 PM]

jnicol: then the barcode is added automatically by jee's program  [3:05 PM]

machinelves: ah sorry, my chat froze didn't mean to interrupt  [3:05 PM]

Brourd: Right  [3:05 PM]

Eli Fisker: That will give us problems with bad barcodes  [3:05 PM]

jnicol: and labs that have 89 bases are eligible  [3:05 PM]

Brourd: I'm not too sure if Jee's barcodes are bad by default, lol  [3:05 PM]

Eli Fisker: Ah, that limit are new  [3:05 PM]

Brourd: the length limit, yes.  [3:06 PM]

Eli Fisker: Barcodes need to somehow fit to the design also  [3:06 PM]

jnicol: 89 freeform + 41 for the head barcode and tail  [3:06 PM]

Eli Fisker: That an random addition can't do  [3:06 PM]

jnicol: 130 max for the synthesis sequence [3:06 PM]

Brourd: I'm not too sure on a random barcode being bad. [3:06 PM]

Eli Fisker: Some barcodes will be overall great for most design, but interfere with specific lab shapes [3:06 PM]

Brourd: Jee's script for generating barcodes requires, what, 2G-C pairs at least, right? [3:07 PM]

jnicol: yes, thats what your lab Eli is determining :) [3:07 PM]

Eli Fisker: John, yes :) [3:07 PM]

jnicol: but for now the barcode is auto generated [3:07 PM]

Eli Fisker: Ok [3:07 PM]

jnicol: on the selected pilots [3:07 PM]

machinelves: so after you add the random barcode, it will check to see if the molecule is still likely to be stable? [3:07 PM]

Brourd: Don't forget, Eli, all sequences are synthesized twice, you will need to check the RMDB for the barcodes that are auto generated. [3:08 PM]

Brourd: when you look at R1 results, and compare. [3:08 PM]

jnicol: elves, no added for synthesis to make it unique [3:08 PM]

Brourd: Right [3:08 PM]

Eli Fisker: Thx for the idea, Brourd [3:08 PM]

machinelves: right, but i mean, the barcode could potentially destabilize the molecule? [3:09 PM]

jnicol: we are looking into getting those second sets of barcodes into the game format [3:09 PM]

Brourd: That's what I am (not) paid to do. :) [3:09 PM]

Eli Fisker: Yes, I believe so [3:09 PM]

Brourd: Depends [3:09 PM]

machinelves: when a player creates a barcode by hand, they can eyeball the whole molecule for nstability. by tacking on a random barcode after the fact without regard to the whole design, could it be destabilized? [3:09 PM]

Eli Fisker: I think some lab shapes are more vulnerable than others [3:09 PM]

machinelves: *instability [3:09 PM]

Brourd: It's actually a question of how many players eyeball anyway. [3:10 PM]

Eli Fisker: I think it can change which part of the designs are stable [3:10 PM]

machinelves: yes i agree, i think it depends on the design [3:10 PM]

Eli Fisker: https://getsatisfaction.com/eternagame/topics/dgxtuhpmcakn9 [3:10 PM]

machinelves: some may be able to take any barcode, others may be more vulnerable  [3:11 PM]

Eli Fisker: Yes [3:11 PM]

Eli Fisker: I think especially designs with short necks are vulnerable [3:11 PM]

machinelves: thanks, good example! [3:11 PM]

Brourd: Define short? [3:11 PM]

Eli Fisker: 4 bp and below [3:11 PM]

Brourd: At that point, you may just be getting a variety of structures in the ensemble. [3:12 PM]

Eli Fisker: What do you mean with the ensemble? [3:12 PM]

Brourd: hence, the little hot spots in SHAPE. [3:12 PM]

Nando: https://getsatisfaction.com/eternagame/topics/computer_generate_the_barcode_for_lab_puzzles#reply_11366865 [3:13 PM]

machinelves: thanks Nando! [3:13 PM]

Brourd: 10,000 RNA molecules in the solution. The shorter the helix, the more conformations it can potentially fold into. [3:13 PM]

Eli Fisker: Neat example Nando [3:13 PM]

Eli Fisker: Brourd, what do you mean with little hot spots in the SHAPE? [3:14 PM]

jnicol: Nando, just curious, why use the 3.4 loop instead of the 0.8 one. Was it just for this test? [3:14 PM]

Brourd: Exposure to chemical modification. [3:14 PM]

LFP6: Back. I'll read up on what I missed then I'll jump in [3:15 PM]

Brourd: How did the results for those two barcodes come out? [3:15 PM]

Eli Fisker: Which means? [3:15 PM]

Brourd: @ Nando [3:15 PM]

Nando: I think it was quite a random choice when I wrote this [3:15 PM]

Brourd: More yellow in the game :) [3:15 PM]

Nando: @B: that was only a simulation [3:15 PM]

machinelves: can we validate the random barcode after it is chosen with the algorithm to screen out bad matches?  [3:15 PM]

Brourd: Ah, we all know how those simulations turn out ;) [3:15 PM]

Brourd: like mimics, hehe [3:16 PM]

Nando: @jnicol: the good barcodes with 0.8 tend to disappear quickly in a given round... [3:16 PM]

machinelves: i think this relates to the general question of how much barcodes affect folding, and by tacking on an unvalidated random barcode this problem has the potential to be increased [3:16 PM]

jnicol: @N, yes I've noticed, since I usually wait until the last minute [3:17 PM]

Nando: @B: doesn't change the fact that randomness can screw up randomly a couple designs which could have otherwise worked fine [3:17 PM]

Brourd: Could Jee's script be modified to check for this? [3:17 PM]

jnicol: At this point, no [3:18 PM]

LFP6: @All: And if the designer meant for it to e unstable in the first place...? [3:18 PM]

machinelves: lol LFP6 [3:18 PM]

jnicol: perhaps we can revisit [3:18 PM]

Eli Fisker: Lol, LFP [3:18 PM]

Nando: not a simple problem, it's CPU intensive... [3:18 PM]

Eli Fisker: Then a perfect barcode could spoil that plan [3:18 PM]

Brourd: Indeed, Nando. [3:18 PM]

Eli Fisker: By the way, I got a minor bug to report [3:19 PM]

jnicol: ok [3:19 PM]

Eli Fisker: In the Road map [3:19 PM]

Eli Fisker: http://prntscr.com/3whopp [3:19 PM]

LFP6: Oh, and random question that my dad got me to think about. As EteRNA is run from two top computer schools (CMU and Stanford), why are we so low on devs? Or is there a large number who we never see? [3:19 PM]

keith.hernandez: This game is hard yo. [3:20 PM]

jnicol: thanks Eli [3:20 PM]

Eli Fisker: Np John [3:20 PM]

Brourd: However, admins would not be able to choose their barcodes anyway, since pilot projects are chosen after the fact, and if it is required to be a randomly generated barcode, the argument becomes null, yes? [3:20 PM]

LFP6: ...And I know I've followed players. Granted that doesn't do anything, but... http://prntscr.com/3wi36v [3:20 PM]

jnicol: LFP6, usually college students are pretty busy and can only donate a little of their time [3:20 PM]

Brourd: Since no script modifications can be made. [3:20 PM]

LFP6: @J: Right, right [3:21 PM]

LFP6: Just supprised we only have a couple [3:21 PM]

LFP6: This would be good for a research project or somethin' ;) :P [3:21 PM]

jnicol: correct brourd, I may be able to tweak the code to try to get CG closing and such [3:21 PM]

Eli Fisker: That would be good [3:22 PM]

machinelves: at what point is the barcode being randomly generated & attached? if it is done right before the player submits the design, and they just click a button that says 'generate random barcode & validate against algorithm' could the processing be done player side? [3:23 PM]

Brourd: If that is what other admins want, then sure. [3:23 PM]

jnicol: maybe have a library of 'best' barcodes to use first, then if that is exhausted, use the jee's random ones [3:23 PM]

machinelves: barcode library ftw!!! [3:23 PM]

Eli Fisker: Yes [3:23 PM]

Eli Fisker: And keep GU count low in them [3:23 PM]

Eli Fisker: One GU is helpful if the design has long stems [3:24 PM]

LFP6: Nothing should be too high, really. [3:24 PM]

Eli Fisker: (often) [3:24 PM]

Brourd: It is more of an issue with RT, and including GU pairs. [3:24 PM]

jnicol: elves, the paper labs are selected first, then if there are additional slots the most voted for pilot labs are chosen [3:24 PM]

LFP6: I mean, GC could easily create a nice mispair in the wrong place [3:24 PM]

jnicol: when the pilots are selected, thats when barcodes are auto generated to be unique [3:25 PM]

Eli Fisker: Actually Mat noted that Barcodes often have a relative high GC count. They often do well with 2 GC pairs at both ends [3:25 PM]

machinelves: okay, so no barcodes for unselected designs? [3:25 PM]

jnicol: correct [3:25 PM]

machinelves: is there a reason why this must be the workflow? [3:25 PM]

jnicol: we dont want anything to interfere with the paper labs, which are more importatnt [3:26 PM]

machinelves: to decrease pool size of barcodes? [3:26 PM]

Brourd: Players did not want to work as much :( [3:26 PM]

machinelves: to prevent players from having to design barcodes? sorry, i am confused about the reason [3:26 PM]

jnicol: since we cant tell when the pilots will be added to the 'paper' lab batch, we cant allow the player to prechoose a barcode [3:27 PM]

Brourd: No, pilot projects are not "active" labs. [3:27 PM]

LFP6: If all the pilot labs had barcodes, it's possible they could run out [3:27 PM]

machinelves: so it is the pool size? [3:27 PM]

LFP6: Put it this way: If there were 500 paper labs open, but only 500 slots, all of them couldn't use unique barcodes [3:28 PM]

LFP6: Simplified, of course. :P [3:28 PM]

LFP6: But, yeah [3:28 PM]

machinelves: and algorithmic validation has processing issues? [3:28 PM]

jnicol: no, a paper lab barcode may be the same as a pilot lab barcode if we allowed players to design it, then there is a problem since we cant have 2 of the same barcodes [3:28 PM]

LFP6: ^^ [3:28 PM]

machinelves: ahh thank you [3:29 PM]

machinelves: so no way to cross ref those data sets in real time? [3:29 PM]

LFP6: You would then have an increasingly smaller pool of barcodes to choose from [3:30 PM]

LFP6: Possibly running out after a while [3:30 PM]

LFP6: If I understand correctly [3:30 PM]

jnicol: and do what, elimante the pilot lab if it has a duplicate with a paper lab? [3:31 PM]

jnicol: *eliminate [3:31 PM]

machinelves: eliminate that barcode from the pool [3:31 PM]

machinelves: for both [3:31 PM]

machinelves: excuse me for the other one [3:31 PM]

Brourd: Pilot labs wil stack up [3:31 PM]

jnicol: but each round is separate [3:31 PM]

jnicol: @B, exactly [3:31 PM]

Eli Fisker: I have another request. To have the NOVA tutorials show under the Puzzle menu:  [3:31 PM]

Eli Fisker: http://prntscr.com/3wi79b [3:31 PM]

Brourd: Currently, there are currently thousands of designs in the pilot labs, if we assign them all a barcode now, well, that is goign to suck. [3:32 PM]

LFP6: Good idea Eli [3:32 PM]

machinelves: yes, NOVA would be very helpful to have a link from main eterna site! [3:32 PM]

jnicol: I like it Eli [3:32 PM]

machinelves: @jnicol, fyi we emailed jee with a list of issues to forward to walt, can you please touch base & get this through? [3:32 PM]

Eli Fisker: Thx, John [3:32 PM]

Eli Fisker: I have one more [3:33 PM]

Eli Fisker: I have noticed that quite many of our new seem to not find our NOVA tutorials [3:33 PM]

jnicol: Was I included in the email? [3:33 PM]

Eli Fisker: http://prntscr.com/3wi81p [3:33 PM]

machinelves: jnicol no, i don't have your email ;) [3:33 PM]

Eli Fisker: I think they first see the big tempting green button that says play [3:34 PM]

Brourd: Let's remove the tempting green button? [3:34 PM]

jnicol: just PMed you [3:34 PM]

machinelves: kk thx, will send to you post dev chat [3:34 PM]

Eli Fisker: I don't know if moving them around will help [3:34 PM]

Brourd: Although, points are not transferring for current players via NOVA. [3:35 PM]

Eli Fisker: Or changing text [3:35 PM]

jnicol: @B, is this new? [3:35 PM]

LFP6: Or pushing to get NOVA tutorials to replace the existing ones [3:35 PM]

jnicol: I fixed that 2 weeks ago [3:35 PM]

LFP6: @J: Not really [3:35 PM]

Brourd: It's more of an issue with workflow. Players have to be pushed towards NOVA by the site. [3:35 PM]

LFP6: Oh [3:35 PM]

Brourd: INdeed, Starryjess PM'd me about it, and I tried to transfer myself. [3:36 PM]

Eli Fisker: LFP the old ones are well enough if the players first meet them after the nova ones.  [3:36 PM]

LFP6: @B: As I said. I think the NOVA tutorials are supposed to be put on the main site. If not, they should be, as they make much more sense [3:36 PM]

Eli Fisker: LFP, but I like your idea [3:36 PM]

Eli Fisker: That would work the trick [3:36 PM]

LFP6: I was told that it would happen... [3:36 PM]

Eli Fisker: And then let the old tutorials be free of choice, if players seek them out by themselves in the menu [3:37 PM]

jnicol: @B, do you get to the Eterna transfer page from NOVA? [3:37 PM]

Brourd: Yes, Jnicol. And while you are taking traffic away from PBS, update the flash GUI as well. [3:37 PM]

machinelves: @B thanks for the context on how assigning barcodes to thousands of paper labs will not work, you guys know better than I do how labs work.  [3:37 PM]

LFP6: lol B' [3:37 PM]

Brourd: Essentially, NOVA tutorials are a "partnership" [3:38 PM]

jnicol: Yes, my understanding is that we should have the GUI code available to integrate into Eterna [3:38 PM]

Brourd: So, we can't just take them away from NOVA, technically. [3:39 PM]

machinelves: @jnicol since i think it got buried earlier, just wanted to voice support for Mat & Eli's request for autosave https://getsatisfaction.com/eternagame/topics/autosave_undo_improvement_idea [3:39 PM]

Brourd: I like half of the plan being able to work :) [3:39 PM]

Brourd: @ Jnicol - I have to do it the hard way? :P [3:40 PM]

Brourd: for the NOVA point transfer [3:40 PM]

jnicol: @B, how were you trying to do it? [3:41 PM]

jnicol: thanks elves, I just read it [3:41 PM]

machinelves: yay! :D [3:41 PM]

Brourd: Via the tranfer page link. Is there a database crosscheck? [3:42 PM]

jnicol: a cookie is set after you complete the nova tutorials [3:42 PM]

Brourd: Ah [3:42 PM]

Brourd: No clue what happened with starryjess :) [3:42 PM]

jnicol: when you transfer, the eterna page reads the cookie [3:42 PM]

Eli Fisker: For the transfer I just did like this afterwards:  [3:42 PM]

Brourd: Yeah, that would explain my end. [3:43 PM]

Eli Fisker: http://prntscr.com/3wibpp [3:43 PM]

Eli Fisker: http://prntscr.com/3wibj8 [3:43 PM]

Eli Fisker: After logging into my nova account [3:44 PM]

jnicol: Maybe I should display something on the eterna transfer page if the cookie is not detected [3:44 PM]

Eli Fisker: That could work the trick [3:44 PM]

Brourd: Perhaps. Make it witty as well. [3:44 PM]

Eli Fisker: Perhaps on the road map [3:44 PM]

machinelves: i did nova as guest and then cleared my cookies, so ... *solemnly waves goodbye to points, lol* [3:44 PM]

jnicol: Like, you need to complete the Nova tutorials first before collecting your points, you cheater :) [3:44 PM]

Eli Fisker: hehe [3:44 PM]

Brourd: "You are a bad player. No cookies for Santa, no presents :(" [3:45 PM]

Brourd: Ha [3:45 PM]

Eli Fisker: lol, John [3:45 PM]

Brourd: I cleared them once, so my conscience is 95% clear. [3:45 PM]

Brourd: 5% not so clear due to other malevolent deeds, mwahahaha. [3:46 PM]

jnicol: ok, I'm going to declare the end of dev chat now, unless there are more questions [3:47 PM]

Brourd: Yes, what kind of questions would you like to hear? [3:47 PM]

jnicol: dev-like questions :) [3:47 PM]

machinelves: @jnicol a quick followup to code of conduct progress, one option instead of or in addition to an email address for complaints, is from mat, to have a player profile for moderator access, and use the PM system to track complaints https://getsatisfaction.com/eternagame/topics/chat_trolls#reply_13395891 [3:47 PM]

Brourd: I'm sure somebody has one of those! [3:47 PM]

Brourd: Ah, the code of conduct. [3:48 PM]

machinelves: yes thou shalt not be a turkey :-P [3:48 PM]

Brourd: The first rule of Eterna chat, is that we do not talk about Eterna chat. [3:48 PM]

machinelves: hehe [3:48 PM]

jnicol: Thats a hot topic for me :) I want to cleanup some spam/ads [3:49 PM]

hoglahoo: We don't talk about what? [3:49 PM]

RedSpah: second rule: there is no second rule [3:49 PM]

machinelves: exactly. [3:49 PM]

Eli Fisker: John, great [3:49 PM]

Brourd: True, jnicol. There is a lot of hidden spam. [3:49 PM]

machinelves: thanks John! [3:49 PM]

jnicol: think hog has a spam puzzle also :) [3:49 PM]

hoglahoo: three of them at least :D [3:49 PM]

Eli Fisker: lol [3:49 PM]

jnicol: Thanks everyone, talk to you later [3:50 PM]

Eli Fisker: Bye John [3:50 PM]

RedSpah: Brourd, don't forget about the cheap stuff for Diablo 3... woooo.... [3:50 PM]

machinelves: thanks very much!! [3:50 PM]

hoglahoo: later [3:50 PM]

RedSpah: bye John [3:50 PM]

Brourd: Has anyone brought up the uneven number of comments displayed, and the number recorded? [3:50 PM]

machinelves: B I saw that mentioned, not sure if fixed [3:50 PM]

jnicol: In the News side bar? [3:50 PM]

Brourd: Right. [3:51 PM]

jnicol: Meechl mentioned it first [3:51 PM]

LFP6: Just saw that [3:51 PM]

Brourd: Ah, so, somebody noticed the hidden spam. [3:51 PM]

LFP6: Btw John, what's in queue for the devs at this point? [3:51 PM]

LFP6: Out of curiosity [3:51 PM]

Brourd: one of the news items has over 700 spam comments, an older one if I recall. [3:51 PM]

hoglahoo: we wouldn't be talking about it if nobody noticed it : [3:51 PM]

hoglahoo: :) [3:51 PM]

LFP6: Yikes!! [3:51 PM]

Brourd: all hidden, of course. [3:51 PM]

LFP6: Comments reflect that? [3:53 PM]

Brourd: Heavens, no [3:55 PM]

Brourd: http://prntscr.com/3wigtb [3:55 PM]