2015.11.20 Dev Chat
( Link back to Dev Chat archive )
On New Fusion Table Data Browser, A/B Lab Puzzles Due Nov. 30, #DPAT #EnsembleDiversity Experiment Instructions, & Musical RNA
Elves: :D [2:54 PM]
RedSpah: Hi Elves [2:55 PM]
Elves: hi Red! :) [2:55 PM]
LFP6: Hey hey hey! [2:55 PM]
LFP6: Elves, you see the new data browser? [2:55 PM]
Elves: hello LFP6! [2:56 PM]
Elves: nope what is it? [2:56 PM]
LFP6: Something Omei and a grad student (I think) have been working on [2:57 PM]
LFP6: Mucho data [2:57 PM]
Elves: cool! [2:57 PM]
Elves: oooo [2:57 PM]
Omei: Speaking of data browsers ... [2:57 PM]
Omei: Hi, all. [2:58 PM]
Elves: yay Omei! [2:58 PM]
LFP6: Should've shortened that :P [2:57 PM]
LFP6: Thanks for the work Omei!!! [2:58 PM]
LFP6: :D [2:58 PM]
Omei: Hi Elves. Long time, no chat. [2:58 PM]
Elves: glad to see you Omei :D [2:58 PM]
LFP6: And I made a characteristic post of mine in that topic with some UI suggestions and such [2:58 PM]
Jennifer Pearl: hi [2:59 PM]
LFP6: Small things, asthetics, but I think they'd help [2:58 PM]
Omei: Thank you for your post, LPF6. I've been in a meeting, but I will respond as soon as I can. [2:59 PM]
Elves: hi Jen!! [2:59 PM]
LFP6: Hey Jen [2:59 PM]
Omei: Hi Jennfer [2:59 PM]
LFP6: @Omei: Great, thanks [2:59 PM]
Omei: @LFP6: Absolutely. I only quickly scanned it so far, but didn't see anything I wouldn't agree with. I appreciate you taking the time to write it all down. [3:01 PM]
LFP6: Happy to help [3:01 PM]
Jennifer Pearl: hi omei [3:07 PM]
Jennifer Pearl: hi LFP6 [3:07 PM]
rhiju: hi everyone [3:07 PM]
LFP6: Eening Rhiju [3:07 PM]
LFP6: *Evening [3:07 PM]
LFP6: Though afternoon for you I suppose! [3:07 PM]
Elves: hi Rhiju! :) [3:08 PM]
rhiju: @LFP6 thanks for the helpful comments on the data browser prototype -- exactly the kind of feedback that omei & siqi can use at this early stage [3:08 PM]
LFP6: That is good to hear, wasn't sure if I was being premature [3:08 PM]
LFP6: IE "Here's a mockup, all the visuals will be replaed" [3:08 PM]
Omei: Not at all. I'm beating the bushes for any and all early feedback. [3:09 PM]
LFP6: I've experienced that [3:09 PM]
LFP6: Yeah, I just found it to be a little clunky, personally it seems to just be some relatively little things [3:09 PM]
LFP6: But some great work [3:09 PM]
Omei: It really is an early release, not near production quality yet. [3:10 PM]
rhiju: Anyone out there making progress with A/B lab puzzles? Are the sub-labs helping? http://www.eternagame.org/web/lab/6204762/ [3:11 PM]
Jennifer Pearl: i havnt looked at it yet. cant wait [3:11 PM]
LFP6: And I also had some questions about accesing that data, per the info that Jen wants to have access to [3:11 PM]
LFP6: At the end of the post [3:11 PM]
whbob: Hi. [3:13 PM]
Jennifer Pearl: hi whbob [3:12 PM]
Jennifer Pearl: ill be sure to read over teh post later [3:13 PM]
Jennifer Pearl: @rhiju I looked at teh labs and will be playing around later. Lookes less daunting thatn teh full labs [3:13 PM]
Omei: @lFP6: You'll certainly be able to get the data, but we haven't formalized the API yet. Right now, we're just running off of fusion tables. [3:13 PM]
Omei: Hi, whbob! [3:14 PM]
Elves: this community chat provides a walkthrough on making lab mods, for anyone wanting to jump in and get an introduction to how labs work [3:14 PM]
Elves: http://eternawiki.org/wiki/index.php5/2015.10.30_Community_Chat [3:14 PM]
rhiju: @jen if you come up with ideas that you are testing (e.g., DPAT-related), make sure to include those in the comments. You might even include hashtags like #dpat. Omei & Siqi are going to make it easy to quickly filter on tags in the new browser. [3:14 PM]
whbob: @ Omei: Great work! [3:15 PM]
Omei: Thanks. [3:15 PM]
Elves: @rhiju we are discussing the hashtags for that now [3:15 PM]
Elves: will they accept special characters like + [3:15 PM]
Elves: for example #EnsembleDefect5+ [3:15 PM]
Jennifer Pearl: we have some stuff I would like to have eterna community look at [3:15 PM]
Jennifer Pearl: try that is [3:15 PM]
LFP6: @Omei Great. Any chance a similar query interface could still be used, to reduce the overhead from multiple queries and extra unneeded data? [3:15 PM]
LFP6: Or is that even under your control? [3:15 PM]
Elves: To explain to everyone, the lab submissions are a great opportunity to test specific hypotheses, and functionality of tools like DPAT [3:16 PM]
Elves: so when you make a design, if you want to contribute to research for the DPAT tool, you can tag your design with #DPAT [3:17 PM]
Omei: The current plan is that there will gernerall be three server queries, precisely for the reaon of minimizing the data that gets downloaded from the server. [3:17 PM]
Jennifer Pearl: I did some investigating and found that designs that soom desigs that had a high enemble diversity in Vienna 2 using DPAT also had high Ensemble Defect using NUPACK [3:17 PM]
Omei: The first is foe the bowser to find all the labs that are available. [3:17 PM]
Elves: and then if you have a specific target you are aiming for, you can include a tag for that goal, like having #EnsembleDefect5+ ( low ensemble defect ) [3:18 PM]
Omei: Once the player selects the set of labs, there will be a second query for what columns are available in each of the labs. (It won't be the same for all labs.) [3:18 PM]
Brourd: Rhiju - Off topic, but what is the current progress on the synthesis and sequencing of the round 98 designs? [2:19 PM]
Elves: then when we get lab data back, we can easily parse out the designs that are relevant to testing that hypothesis [3:20 PM]
Jennifer Pearl: I would like to have the community make designs to a single constraint. That constraint being the normalized ensemble defect from NUPACK's free online server [3:20 PM]
Omei: There may be ~100 columns for some labs, so the player chooses the columns they are actually interested in. The third server query, the potentially big one, gets only the data the player has asked for. [3:20 PM]
whbob: Is there a glossary to explain the many column headers? [3:22 PM]
Omei: good question, whbob. I don't think there is one now. [3:22 PM]
John Cook: Hi.. This is my first time attending this chat, so forgive my annoying questions. Does this chat deal with software development? Eterna site development? Tool development? [3:23 PM]
Elves: hi John, welcome to eterna :D [3:23 PM]
LFP6: All of the above would be the answer [3:23 PM]
Elves: the dev chat deals with all of the above [3:24 PM]
LFP6: @Omei Cool. I take it that all the lab ata would be accessable this way, including everything in the archives? [3:24 PM]
John Cook: Is there a git or other repository were works in progress are reposed? [3:25 PM]
LFP6: Not really [3:25 PM]
rhiju: @brourd data are all collected, and are being processed. Briefly, there are some exciting results in the next-generation puzzles, which appear to confirm player hypotheses about which structure 'blueprints' lead to more successes or failures. We will release soon (and before the deployment of the next puzzle set). [3:25 PM]
Elves: internally [3:25 PM]
LFP6: Nothing consolidated that's used [3:25 PM]
LFP6: Well, as Elves said, there is but we can't see it. :) [3:25 PM]
JR: Looks like we need more people doing labs. Might think of pushing mods as an acceptable alternative to creating from cratch, save quite a bit of time. [3:26 PM]
Elves: forums are the location for community to log bugs and feature requests [3:26 PM]
Omei: I'm not sue what "all" would be. The first order of business is the currennt series of chip-based experiments. But we're trying to code the browser with a minimum number of assumptions about what data might be displayable. [3:26 PM]
Elves: @JR good point :D I'll link the mod walkthrough again [3:26 PM]
rhiju: @john welcome! the eterna site code is indeed mostly in a github codebase, accessible to the development team. [3:26 PM]
Elves: Make Mods in Labs http://eternawiki.org/wiki/index.php5/2015.10.30_Community_Chat [3:27 PM]
LFP6: @Omei: I'm speaking of everything you can look at through the archives section, I'm saying this specifically with the thought of keeping with one (improved!) API for grabbing the data, and the various little helpful tools that have been added like side-by-side comparrisons [3:27 PM]
John Cook: Ok. In that case, is there a documented (hopefully Restful) API which a tool developer can use to interact with the site? [3:28 PM]
rhiju: @JR that's a great point. we also think the lab's will be less mysterious when we roll out the new puzzle progression (hope you've had a chance to check those out and give back comments to bekeep). we're also hopeful that sublabs will help, particularly in january when we roll out A*B/C*C puzzles. [3:29 PM]
LFP6: @John Yep [3:28 PM]
LFP6: EteRNAScript [3:29 PM]
LFP6: Under puzzles -> EteRNA Scripts [3:29 PM]
Elves: here is a link to bekeep's puzzle progression work [3:29 PM]
Elves: https://getsatisfaction.com/eternagame/topics/puzzle-progression-puzzle-requests [3:29 PM]
LFP6: Docs for the API are in the wiki, and not complete yet [3:29 PM]
Omei: @LFP6: I believe there is an intent to eventually make all that available to the new browser, yes. It's kind of outside the browser development itself. [3:30 PM]
LFP6: But not hard to figure out [3:29 PM]
LFP6: Must be within EteRNAScript though, as there is no cross-domain allowances [3:30 PM]
LFP6: @Omei Cool, thanks [3:30 PM]
Omei: @John Cook: FWIW, I have used YAML in the past to get around the XSS constraints. PM if you're interested in knowing more. [3:32 PM]
whbob: @ JR: I can't think of anything to contribute to the two sub labs that have oligo constraints. [3:33 PM]
whbob: @ JR: I think you covered them all:) [3:33 PM]
John Cook: @Omei thanks, I will do that [3:33 PM]
Elves: @whbob Jen and I have some ideas on what to contribute for sublabs :D [3:34 PM]
John Cook: So is the primary concept about this chat for benders to talk with the actual developers about nice to haves and such? [3:34 PM]
Elves: if anyone wants to help with DPAT research, we're wondering which designs will produce which level of Ensemble Diversity & Ensemble Defect [3:34 PM]
JR: @rhiju - New labs are not strange but different in that one has to re-learn how to manipulate the strings to do what you want, they are not normal puzzles. You have to learn how to push a string apart without affecting the rest of the puzzle. [3:35 PM]
Omei: @JC: Or anything else they want to talk with devs about. [3:35 PM]
Elves: @JC yes This chat is where the community interfaces directly with the dev team to make requests [3:36 PM]
whbob: @ Elves: I would like to help:) [3:36 PM]
Elves: Hi whbob! and thanks :D [3:36 PM]
Jennifer Pearl: yay [3:36 PM]
Elves: we can explain the basic workflow for submitting labs to help with DPAT [3:36 PM]
whbob: OK [3:37 PM]
Elves: and then it's also fun just to have a target mission for lab submissions [3:36 PM]
rhiju: @JR i'm intrigued by your comment. by manipulating the strings do you mean getting used to the oligo sequences (like A and B)? or something else? [3:36 PM]
John Cook: @JR, I have been copying my designs (via copy) and pasting them back in (via paste) after editing them in the paste field or in a text editor. [3:36 PM]
Elves: so DPAT is dot plot analysis that evaluates for what structures have the highest or lowest probability to pair [3:37 PM]
Jennifer Pearl: currently DPAT is looking at things associated with the secondary structure [3:37 PM]
Elves: and ( I think, correct me on the finer points everyone... :D ) Ensemble Defect is a value that says whether a particular design has a low or high variety of structures that could form [3:38 PM]
Elves: in other words, our goal is to test for whether a lab design with particular structures is likely to have a low Ensemble Defect value ( only a few options of structure formation ) or high value ( many... [3:39 PM]
...structure / base pairing options )
Jennifer Pearl: and if you followed last weeks devchat I spoke about Ensemble diversity and how I observed some design had a high value and som had a low value. NUPACK uses Ensemble Defect and I want to see if this same trend is obeserved using that model. [3:39 PM]
Elves: here's a link to the Ensemble Diversity / Defect caht [3:40 PM]
Elves: chat [3:40 PM]
Elves: http://eternawiki.org/wiki/index.php5/2015.11.13_Community_Chat [3:40 PM]
John Cook: One thing that might be nice would be to have better descripts of labs. For example, in the current lab (second to the last puzzle) there are three states... [3:40 PM]
JR: I've cranked out over 1000 lab puzzles by manipulating around the oligo sequences instead of directly trying to control the sequence. It takes some time to leard what works and what doesn't. [3:40 PM]
John Cook: States two and three look exactly the same, but they behave differently in some undocumented way. It would be nice to have additional information. [3:40 PM]
Omei: @rhiju, I think JR and JC are talking about reposiitioning segments, like moving all the "unused" bases to one end, where they can be tied up into a static hairpin. [3:41 PM]
Omei: (I do this as JC described.) [3:41 PM]
John Cook: @Omei exactly, I was thinking about writing a tool that could grab my design so I could do this on my android and push it back. Hence the desire for an API. [3:42 PM]
rhiju: @JR thanks for the clarification. If we change the oligo sequences (which we will in puzzles throughout 2016) does that mean you have to relearn what works and what doesnt? or do you think you've got a 'system' of manipulations to try? (we want puzzles that motivate the development of the latter.) [3:42 PM]
Omei: @JC, there's no existing API to get the sequence before it has been submitted. [3:43 PM]
John Cook: I have a sequence to QR code algorythm that I have been working on if you folks are interested. That way a bender could scan the code with an app. Not sure how they would get it back though :) [3:45 PM]
Omei: There _might_ be a way to get it if you wanted to code it up as a Greasemonkey type extension. [3:45 PM]
JR: Yes, I find. giving the position of the sequences, oligo, I have to "teach" myself what I can do to pull parts of the puzzle away while keeping other parts of the puzzle together. It is hit and miss right now. [3:46 PM]
LFP6: I'm off to eat dinner. Will check in later [3:46 PM]
Elves: bye LFP6 good to see you! [3:47 PM]
Elves: @JC sequence to QR code, now that's interesting! [3:48 PM]
Elves: we have been wishing for a sequence to music output, I saw a web app the other day that converts letters to musical notes, and I used it to play a sequence in music tones. but it is a limited app. [3:48 PM]
Elves: but it would be cool in case you can find patterns in the auditory output [3:49 PM]
Elves: by listening to RNA music :D [3:49 PM]
Omei: @Elves, I like that. Did you try encoding the structure along with the sequence? [3:50 PM]
Elves: @Omei just raw sequence no structure [3:50 PM]
Elves: including structure somehow would be cool [3:51 PM]
Elves: great idea! [3:51 PM]
Elves: it is an alternative to data visualization - data audiolization? lol [3:51 PM]
Omei: Like maybe playing an unpaired base an octave lower than a paired one. [3:51 PM]
Elves: to allow those with listening aptitude to pick out patterns in the sequence [3:51 PM]
Elves: ooo [3:51 PM]
Elves: SHAPE data audio output [3:51 PM]
whbob: Stems could be the back beat behind the loops:) [3:52 PM]
Elves: layered would be interesting too [3:52 PM]
Elves: hahaha [3:52 PM]
John Cook: @Elves, I like the music idea. Perhaps good RNA sounds good. [3:52 PM]
Elves: yeah it would be fun to find out [3:52 PM]
Elves: the QR code idea is interesting as well since it is another kind of direct visualization in a way [3:53 PM]
Elves: you might like Jen Pearl's work with dot plots because of that actually [3:53 PM]
John Cook: Suppose that whales are actually tlaking about folding RNA as they swim around. [3:53 PM]
Elves: lol i bet some scientist whale has a deep sea RNA lab [3:53 PM]
Omei: What could possibly be more interesting to a whale? [3:53 PM]
Elves: DPAT stands for dot plot analysis tool [3:53 PM]
Elves: and it's a little graph you can look at while designing that shows the probability of structures to pair up in your sequence [3:54 PM]
Elves: we would like to use DPAT and NUPACK to predict whether a design will score well, and you can put #DPAT and #EnsembleDefect in your lab submission comment to participate in the experiment, and go to the link... [3:54 PM]
...from Jen below:
John Cook: Deos anyone play "Two Dots"? I think that it is actually improving my folding :) [3:54 PM]
Elves: oh lol Jen just dropped connection [3:54 PM]
Elves: no what is that? [3:55 PM]
Jennifer Pearl: http://www.nupack.org/utilities/ [3:55 PM]
Elves: thanks Jen! [3:55 PM]
Jennifer Pearl: here is alink to apdf on how to use it for this. Really rough just threw together in 3 minutes [3:56 PM]
Jennifer Pearl: https://dl.dropboxusercontent.com/u/87351147/How%20so%20find%20ensemble%20defect%20with%20NUPACK.pdf [3:56 PM]
Elves: so to participate in the DPAT lab experiment, submit a lab design with the hashtags above, and the Ensemble Defect value that is output from the link above, after you copy and paste in your nucleotide sequence [3:56 PM]
John Cook: It is a dot connecting game available on most smartphones. Warning: Highly adictive. [3:56 PM]
Elves: according to the instructions Jen typed up in the pdf link above [3:56 PM]
Elves: cool! [3:56 PM]
Elves: yes eterna is all about pattern recognition [3:57 PM]
Omei: I just installed it and will take it for a spin. But if it is too addictive, it has to go. :-) [3:57 PM]
Elves: which humans excel at, so if you like those games then designing RNA is probably something you will excel at :D [3:57 PM]
Elves: so to summarize the process: 1. check out the Lab Mods walkthrough to learn how to make lab submissions [3:58 PM]
Elves: http://eternawiki.org/wiki/index.php5/2015.10.30_Community_Chat [3:58 PM]
Elves: 2. when you create a lab design and get ready to submit, put these hashtags in the comments: #DPAT #EnsembleDefect [3:59 PM]
John Cook: Let me know how long you keep it. Thankfully, they limit the amount of time you can play, but I think that contributes to the addiction. There was an interesting discussion on the "Note to Self" Podcast. That is where I found out about it. [3:59 PM]
Elves: 3. Also put the Ensemble Defect value in the comments, that you can get from the NUPACK tool: [3:59 PM]
Elves: http://www.nupack.org/utilities/ [3:59 PM]
Elves: following these instructions: [3:59 PM]
Elves: https://dl.dropboxusercontent.com/u/87351147/How%20so%20find%20ensemble%20defect%20with%20NUPACK.pdf [4:00 PM]
Jennifer Pearl: Thanks Elves for helping me with getting this info out there. I have to leave for an appointment I forgot about. [4:00 PM]
Elves: and then when we get lab results back, we will be able to see which designs actually met the target Ensemble Defect values, and eventually correlate that to DPAT output on structure probability [4:01 PM]
Elves: with the end goal being able to predict which structures are most likely to produce a high or low Ensemble Defect [4:01 PM]
Elves: for anyone wondering what Ensemble Diversity and Ensemble Defect are, check out this chat: [4:02 PM]
Elves: http://eternawiki.org/wiki/index.php5/2015.11.13_Community_Chat [4:02 PM]
Elves: thanks Jen! [4:02 PM]
Elves: Great progress on everything, have a fun appointment! [4:03 PM]
Elves: finally here's a link to active labs where you can follow the instructions above to participate and have your designs synthesized in a real lab at Stanford http://www.eternagame.org/web/lab/6204762/ [4:04 PM]
Elves: Thanks everyone for your participation and questions! [4:05 PM]
Elves: Lab deadline is November 30th to get in all submissions [4:05 PM]
Elves: We'll be back next Friday for Community Chat in case you have any questions on the #DPAT experiment [4:05 PM]
John Cook: @Elves In the grand scheme of things, I am still a newby, at least relative to everyone else here. Do codons have any interesting effects or impacts on puzzles. I just focus on the nucleotides. Getting back to your music idea. Only have four notes would be kind of boring. But giving a note to each codon would provide many more notes. [4:05 PM]
Elves: @JC if I understand what a codon is, it's up to you whether you insert specific codons into your sequence. as for effects and impacts, that's exactly what we are studying [4:06 PM]
Elves: you have just wandered into a live research experiment [4:06 PM]
Elves: basically asking a similar question - what are the consequences and behaviors of xyz nucleotide sequence in design [4:07 PM]
Elves: in the grand scheme of things, we are all new to some aspect of the platform as there is a wide range of things you can work on here [4:07 PM]
Elves: so don't worry about that, just find what you like to work on and I am sure you will find a way to contribute to the community [4:08 PM]
Elves: the best thing you can do to help is get geared up to participate in labs [4:08 PM]
Elves: since that is what drives the research output [4:08 PM]
John Cook: Cool. Thanks. [4:08 PM]
Elves: and then whatever your specific interests or background may be, you can focus on what you like [4:08 PM]
Elves: np! happy to have you :) [4:09 PM]
Elves: re: your codon music idea [4:09 PM]
Elves: that makes it a full on musical language now doesn't it :D [4:09 PM]
John Cook: A/B is my first lab. I have completed the first two puzzles. But I have no idea what makes a good deisgn or something which is likely to actually fold in the real world. [4:09 PM]
Elves: @JC you have just described basically the reason why we have the lab in the first place [4:10 PM]
Elves: we also do not know what makes something "good" [4:10 PM]
Elves: and so to gather data from the lab is our best way of validating which sequences or designs create which structures [4:10 PM]
Elves: so you can't go wrong with your submissions, if it's a novel sequence, it will output data saying whether it formed the target shape or not [4:10 PM]
Elves: and from there we can draw conclusions [4:10 PM]
John Cook: Good music makes everything good. music is my life.That is why this sounds good to me: no pun intended. [4:11 PM]
Elves: for DPAT and Ensemble Defect experiment outlined above, you can play around with specific goals in your designs by seeing if you can find structures that have a consistently high or low Ensemble Defect [4:11 PM]
Elves: hahaha [4:11 PM]
Elves: that's super cool [4:11 PM]
Elves: puns are cherished around these parts... :D [4:12 PM]
Elves: well we would love any and all methods for visualizing data output from the lab, and raw sequence data, so please feel free to play around. [4:12 PM]
John Cook: Time to eat. Thanks for the chat. I will connect and we can share ideas about musical RNA. [4:13 PM]
Elves: great thanks! and shortly I will post the transcript for review here, in case anyone wants those instructions for #DPAT experiment [4:14 PM]
Elves: http://eternawiki.org/wiki/index.php5/Dev_Chat [4:14 PM]
Elves: enjoy! [4:14 PM]
