2016.01.15 Dev Chat
( Link back to Dev Chat archive )
On Scientist 3D RNA Presentation, Evolution of DPAT Analysis Tools & FMN Probability Prediction, Eterna Scripting Tools, Forum Titles & Brainstorming Cool 3D Nanostructures & Machines
LFP6: News item :) [12:03 AM]
LFP6: http://www.eternagame.org/web/news/6445579/ [12:03 AM]
Jennifer Pearl: dint see it and I was looking at it last night. not sure when it was posted [12:04 AM]
LFP6: While I was at school, but [12:03 AM]
Elves: thanks LFP6! [12:04 AM]
LFP6: Oh, hey Elves! [12:04 AM]
Jennifer Pearl: Hi Elves [12:04 AM]
LFP6: Does this mean there is a chat? [12:04 AM]
RedSpah: Hi Elves [12:04 AM]
jandersonlee: Ther was a countdown so yes [12:04 AM]
Elves: hi LFP6, Jen, Red! [12:04 AM]
Elves: last time there was a monthly meeting, the dev chat was not held. So I'm not sure if devs will be coming back or not. [12:05 AM]
LFP6: @JL it's automated for fridays at this time [12:05 AM]
RedSpah: elves, don't enter my page if you're lightheaded [12:05 AM]
RedSpah: I did some magic to it [12:05 AM]
Elves: lol [12:05 AM]
Elves: thanks for the pro tip Red! [12:05 AM]
Jennifer Pearl: i like Red's page [12:06 AM]
jandersonlee: @LFP6 using the xhttp guts of your submitDesign function and a few others -- very helpful [12:06 AM]
LFP6: Aah [12:06 AM]
LFP6: As long as you saw the params :) [12:06 AM]
LFP6: And formatting, but yeah [12:06 AM]
LFP6: Integrated into your own non-web tool? [12:06 AM]
jandersonlee: @LFP6 - yes also some help from Nando and John (I'm now a bot - or cyborg?) [12:07 AM]
LFP6: Hehe [12:07 AM]
Jennifer Pearl: so I have mostly finished Sara for now and have been moding desings that have not been submitted yet using partition function, pairing probabilities and filteing for specific stacks [12:07 AM]
Jennifer Pearl: @Elves [12:07 AM]
Jennifer Pearl: Its a Smart Remmy [12:07 AM]
LFP6: Eventually we'll make SDKs for various languages, it'll be all pretty-like [12:07 AM]
Elves: yay! jandersonbot5000! [12:07 AM]
RedSpah: inb4 [12:07 AM]
RedSpah: EXTERMINATE [12:07 AM]
Elves: cool Jen, nice work! [12:08 AM]
jandersonlee: I'm trying to think of a good name for a bot account - suggestions welcome [12:08 AM]
LFP6: iRobot... Or wait... [12:08 AM]
jandersonlee: mubot? (mutations) [12:08 AM]
Jennifer Pearl: mubot sounds nice [12:08 AM]
LFP6: Muuuuuubot [12:08 AM]
Jennifer Pearl: mooooobot [12:09 AM]
LFP6: Cowbot confirmed [12:08 AM]
Elves: cyborglee [12:09 AM]
jandersonlee: plus "mu" is very Zen [12:08 AM]
Elves: lol [12:09 AM]
Jennifer Pearl: what doed your bot do JL? [12:09 AM]
Jennifer Pearl: does [12:09 AM]
jandersonlee: right now just batch submissions of single base mutations [12:09 AM]
Jennifer Pearl: nice [12:10 AM]
jandersonlee: collect the whole set... [12:10 AM]
jandersonlee: I got Rhiju's permission for this round [12:10 AM]
Jennifer Pearl: so you have teh submission part integrated with it [12:10 AM]
Jennifer Pearl: cool [12:10 AM]
jandersonlee: no - a web script for generation and a sperate EternaScript for submission [12:10 AM]
Jennifer Pearl: nice [12:11 AM]
Jennifer Pearl: I do all y stuff locally and I will be using LFP6's script for all of Sara's submissions. [12:11 AM]
Jennifer Pearl: my [12:11 AM]
jandersonlee: computes GC GU UA and Energy (punt on melt) [12:11 AM]
Jennifer Pearl: nice [12:12 AM]
Jennifer Pearl: melt is on my list too but havnt got to it. I think it is somehting that is very helpfull [12:12 AM]
Jennifer Pearl: Iv'e noticed that some temps just done ever have any good designs but it is sublab specific [12:13 AM]
jandersonlee: I don't use it, so I don't mind punting [12:12 AM]
jandersonlee: (selfish I know) [12:13 AM]
jandersonlee: does a bot have Buddha nature? [12:14 AM]
jandersonlee: mu [12:14 AM]
Jennifer Pearl: So does anyone want to see some cool stuff about pairing probabilities in stacks at teh binding site for FMN labs [12:14 AM]
jandersonlee: yes please [12:14 AM]
Jennifer Pearl: Ok so i have been looking at minimum predicted paring probabilites in stacks at teh FMN binding site [12:15 AM]
Jennifer Pearl: adn I have found that each sublab has idea pairing probability ranges for the main stack at teh binding site that. The large loop side and not the small loop side [12:16 AM]
Jennifer Pearl: also for the short stack to [12:16 AM]
jandersonlee: looking at FMN- condition I assume? [12:17 AM]
jandersonlee: any estimates for FMN+ pairing? [12:17 AM]
Jennifer Pearl: yes [12:18 AM]
Jennifer Pearl: when in the 2nd state [12:18 AM]
jandersonlee: how do you compute those? [12:18 AM]
Jennifer Pearl: now I have found that stack length is also a factor [12:18 AM]
Jennifer Pearl: running thorough Vienna 2 with Nando's hack and then plotting all teh probabilites for each lab [12:19 AM]
Jennifer Pearl: all 3000+ [12:19 AM]
Jennifer Pearl: for each lab [12:19 AM]
Jennifer Pearl: sublab [12:19 AM]
jandersonlee: (ouch) nice [12:19 AM]
Jennifer Pearl: Usng DPAT [12:19 AM]
jandersonlee: what ranges seem good? [12:20 AM]
LFP6: JL-WHat exactly did you get permission for? [12:20 AM]
LFP6: Sorry, I had to go afk [12:20 AM]
jandersonlee: @LFP6 complete single mutations runs against old designs (150-200 designs per run) in a few selected sublabs [12:22 AM]
rhiju: hey everyone -- sorry i'm late -- had a laptop malfunction right before the chat. just catching up and eager to hear jen's results... [12:22 AM]
Jennifer Pearl: So in general a range of .6 to 1.0 is a good range [12:23 AM]
Jennifer Pearl: a real rough value [12:23 AM]
Jennifer Pearl: more often .8 to 1.0 [12:23 AM]
jandersonlee: is that percent or probability? [12:23 AM]
Jennifer Pearl: probability [12:23 AM]
Jennifer Pearl: so 80 to 100 [12:23 AM]
Jennifer Pearl: for .8 to 1 [12:24 AM]
Elves: hi Rhiju! [12:24 AM]
Jennifer Pearl: here is some plots for Same State NG 2 main stack at teh binding site [12:24 AM]
Jennifer Pearl: https://dl.dropboxusercontent.com/u/87351147/Stack%20Pair%20Probs%202.JPG [12:24 AM]
jandersonlee: For same state I've been seeing under 2% MS2/FMN- and >80% MS2/FMN+ sees OK [12:24 AM]
jandersonlee: *seems OK [12:25 AM]
jandersonlee: For Exclusion it's flipped [12:25 AM]
Jennifer Pearl: would that influence the probabilites? [12:25 AM]
Jennifer Pearl: ive noticed elusions are harder [12:26 AM]
Jennifer Pearl: havnt looked at exlusion for probs yet [12:26 AM]
jandersonlee: <0.02 and > 0.8 [12:26 AM]
Jennifer Pearl: just all teh same sates [12:26 AM]
Jennifer Pearl: states [12:26 AM]
Jennifer Pearl: ok [12:26 AM]
rhiju: can you plot the ratio of the +FMN to -FMN probability? That should provide a good estimate of the ratio of Kd (MS2 binding affinities) in -FMN and +FMN conditions. [12:26 AM]
jandersonlee: That's not based on a close statistical analyis just a hunch from a few selected cases [12:26 AM]
jandersonlee: I don't even have a bulk way to compute it yet - involves manual steps [12:27 AM]
rhiju: maybe what we should do is to simulate the binding curves for these designs and then compute the predicted score based on those simulated binding curves ... [12:27 AM]
jandersonlee: @rhiju that could be good [12:28 AM]
rhiju: some of the eterna devs know the mathemtatical relations to do this -- @jandersonlee i think you carried out similar calculations for the A/B puzzle, so do you know the relations? [12:29 AM]
Jennifer Pearl: so if you look at teh picture i posted you can see the ideal stack is 2ndState-18:68,19:67,20:66 (using DPAT notations) adn that the only deigns taht score 100 are between .6 and 1 with a couple outliers at around .4 [12:30 AM]
jandersonlee: @rhiju I have them coded somewhere [12:30 AM]
Jennifer Pearl: and as teh stacks get longer teh designs do worse [12:31 AM]
jandersonlee: (I think) [12:31 AM]
jandersonlee: @JP which stacks are you referring to? [12:31 AM]
Jennifer Pearl: the stacks for large loop side of the FMN binding site for SSNG2 [12:32 AM]
Jennifer Pearl: here is the link again [12:32 AM]
jandersonlee: @JP - OK - got it [12:32 AM]
Jennifer Pearl: https://dl.dropboxusercontent.com/u/87351147/Stack%20Pair%20Probs%202.JPG [12:32 AM]
Jennifer Pearl: now if you look at the small loop binding side it is different [12:32 AM]
Jennifer Pearl: I have found that stack like to be long or short not oin the middle [12:33 AM]
Jennifer Pearl: with a high probability of forming [12:34 AM]
Jennifer Pearl: here is link to some plots [12:34 AM]
Jennifer Pearl: https://dl.dropboxusercontent.com/u/87351147/Stack%20Pair%20Probs%201.JPG [12:34 AM]
Jennifer Pearl: they are orderied by stack length from longest to shortest left to right [12:34 AM]
jandersonlee: So the best designs have short stacks (3 pairs) and probability of binding > 0.6 [12:34 AM]
jandersonlee: but short loops are different [12:35 AM]
Jennifer Pearl: a minimum nuc pair porbability >.6 and yes short 3 base pair long stack for the large loop side [12:36 AM]
Jennifer Pearl: but different for the short loop side [12:36 AM]
jandersonlee: Does it matter if the MS2 arm is in the short loop or in the dangle? [12:37 AM]
Jennifer Pearl: it needs a solid non-switching short or very long stack. at teh binding site wiht a very high minumum prob [12:37 AM]
Jennifer Pearl: I fdid not look at that specificly but i have foudn teh short loop is better if it does not switch [12:37 AM]
jandersonlee: That's what I'm guessing [12:38 AM]
jandersonlee: (when it's a long stack - non-switching) [12:38 AM]
Jennifer Pearl: trying to get a plot ready to show you but I had a family emergncy the other day and have been distracted [12:40 AM]
Jennifer Pearl: so it is not ready yet but will bein a minute or so [12:40 AM]
LFP6: On an unrelated side note @rhiju (while we're in dev chat) [12:40 AM]
LFP6: I had mentioned a while back about possibly adjusting the roles & titles of staff on the forums, and was hoping if we could move forward with that. I have the info I could find on what needed to be changed here: http://pastebin.com/xkKsv1YC [12:40 AM]
LFP6: I can do the titles myself if you give me the go-ahead (of course please mention any adjustments/additions that need to be made), but the roles would need to be handled by a rep/admin. [12:40 AM]
rhiju: @LFP6 totally! go ahead. for stuff that a rep/admin needs to do, can you message jnicol? [12:41 AM]
LFP6: Sure [12:41 AM]
LFP6: Would you take a look at that list to make sure that I got everyone right? [12:41 AM]
LFP6: Or if you know of any other (past) employees on the forum that I missed [12:42 AM]
rhiju: sure please send me the list via PM or e-mail. [12:44 AM]
LFP6: I'll shoot you a PM now [12:44 AM]
Jennifer Pearl: ok. here is the plots [12:45 AM]
Jennifer Pearl: https://dl.dropboxusercontent.com/u/87351147/Stack%20Pair%20Probs%203.JPG [12:45 AM]
rhiju: i was curious about thoughts on joe's presentation on steps towards eterna-3d, [12:45 AM]
rhiju: link here for those who didn't catch it: https:// www.youtube.com/watch?v=adb4H6a3OII [12:45 AM]
LFP6: I'm just watching that now [12:45 AM]
Guzz: hi [12:46 AM]
rhiju: anyone have a secret structure that they'd want '3D printed' at the nanometer scale through RNA self-assembly? [12:46 AM]
Jennifer Pearl: so using this I have plugged it into Sara adn by filtering for all teh values Vienna 2 gives for the partition functiona dn MFE in teh 1st state adn using the pair probabilites I can filter out about 2/3 of the bad designs [12:46 AM]
Elves: cool jen! [12:46 AM]
Elves: ya medicine!!! [12:46 AM]
rhiju: @jennifer that sounds super cool! [12:46 AM]
LFP6: Ok Rhiju, sent [12:46 AM]
Jennifer Pearl: I have modes all teh designs for SSNG 1 ,2 adn 3 doing so [12:47 AM]
jandersonlee: i@rhiju if I tell you it won't be secret [12:47 AM]
Jennifer Pearl: and for NG 2 adn NG 3 I included stack filtering for my markers theory [12:47 AM]
rhiju: @jal haha [12:47 AM]
jandersonlee: @JP nice [12:47 AM]
Elves: organized motif library will help a lot [12:48 AM]
jandersonlee: @rhiju I was wondering about something that might self-bind to make a room-temperature RNA crystal [12:48 AM]
jandersonlee: should be easy to x-ray verify [12:48 AM]
Elves: we could print a tiny library for feynman [12:49 AM]
Elves: since that was his inspiration for nanotech [12:49 AM]
jandersonlee: I liked the idea of specify a directio/length and it picks teh motifs [12:49 AM]
Astromon: Antman has been folding rna and will need a tiny library as well [12:50 AM]
rhiju: @jandersonlee, crystals were actually the inspiration for DNA nanotechnology in the original paper by N Seeman in the early 1980s [12:50 AM]
Elves: janderson yes! smart library tools ftw [12:50 AM]
Jennifer Pearl: and in closing to finishe my thought i ran SSNG 2 Rounds 1 and 2 (all 2000+) f acnd checked if designs matched the constraints and it only returned just over 200 designs. there is alot of 30 adn 60's but also 80's nd 90's [12:50 AM]
jandersonlee: @JP - looking forward to seeing the results [12:51 AM]
rhiju: its actually been hard to do, as crystal-forming molecules should not have strong contacts between molecules, but weak ones that can be slowly tuned to get stronger. Very little success to date in the DNA field, despite 3 decades of work -- just one hi-res crystal to my knowledge. [12:51 AM]
jandersonlee: start in 2D - sheets? [12:51 AM]
jandersonlee: graphene seems important - maybe RNA sheets might too [12:52 AM]
rhiju: so what's actually interesting is that we want some kind of structure that is not just mind-blowing but also hasn't been tackled yet withother nanometer substrates. so fire away with ideas and i can tell you if its bene done [12:52 AM]
rhiju: *been done [12:52 AM]
Elves: gyroid [12:52 AM]
jandersonlee: toroid? [12:52 AM]
rhiju: @elves, lots of folks are working on long-term information storage with DNA (better than RNA for saving 1's and 0's as it is more stable in the ery long term, millennia...) [12:53 AM]
Elves: platonic solids obviously [12:53 AM]
jandersonlee: smoke rings of RNA? [12:53 AM]
rhiju: @jal, 2D -- lots of 'fabrics' already, including some by codygeary, a fellow eterna player [12:53 AM]
Jennifer Pearl: i like smoke rings [12:53 AM]
Elves: oh yea i forgot about that!!! [12:53 AM]
rhiju: @elves, most solids have been instantiated with DNA, and some with RNA [12:53 AM]
rhiju: though those woul d be good starter puzzles [12:54 AM]
LFP6: ?Can you do interlocking shapes? :) http://www.firstinspires.org/sites/all/themes/first/assets/images/FIRST_logo.png [12:53 AM]
rhiju: @jal, toroids and even moebius strips have been created from DNA [12:54 AM]
Elves: https://en.wikipedia.org/wiki/Gyroid [12:54 AM]
Jennifer Pearl: that is pretty cool [12:54 AM]
rhiju: @LFP6 hmm, now we're talking. [12:54 AM]
jandersonlee: a box that opens with a switch latch? [12:54 AM]
Elves: great question lfp6 [12:55 AM]
Elves: gears [12:55 AM]
LFP6: @JL A switch that make ssense XD [12:55 AM]
rhiju: @jandersonlee, box with latch -- was created by ebbe anderson (who BTW gave an eterna popup talk last january -- lmk if you want the link) [12:55 AM]
LFP6: Ooh, gears are cool [12:55 AM]
rhiju: @elves. GYROID. Wow. [12:55 AM]
Elves: :D [12:55 AM]
LFP6: Heh [12:55 AM]
Jennifer Pearl: that is pretty cool [12:56 AM]
Astromon: http://prntscr.com/9qkllw [12:56 AM]
rhiju: I do think that 'computers' would be new. We are of course tackling those through 2D design, switches, XOR gates, etc., but there may be better ways to make those gates with 3D [12:56 AM]
rhiju: And I like the idea of having robot arms [12:56 AM]
rhiju: its one of thefeynman grand prize challenges: https://www.foresight.org/GrandPrize.1.html [12:57 AM]
Elves: pick n place eh? [12:57 AM]
LFP6: Doing that in tech! [12:57 AM]
Jennifer Pearl: yay robot arms [12:57 AM]
LFP6: (Robotic arm simulations) [12:57 AM]
rhiju: for individual gears, there's been a lot of work with 3D DNA origami, albeit at larger length scales than what RNA could achieve. [12:58 AM]
rhiju: what's an intricate thing that depends on gears? [12:58 AM]
Jennifer Pearl: watch [12:58 AM]
Jennifer Pearl: old one [12:58 AM]
rhiju: yea! [12:58 AM]
Elves: binding molecular switch circuits? [12:58 AM]
Jennifer Pearl: engine [12:58 AM]
LFP6: RNA Hydraulics [12:58 AM]
Jennifer Pearl: but no so much i go with watch [12:59 AM]
Jennifer Pearl: how woud you do hydrolics? [12:59 AM]
Jennifer Pearl: @Rhiju i have a quick question [12:59 AM]
jandersonlee: cylinder - like nanotubes [12:59 AM]
LFP6: I was joking, but that would be cool [1:00 PM]
rhiju: wow cool ideas. [1:00 PM]
rhiju: nanotubes -- already doable. e..g: http://bionano.physik.tu-muenchen.de/Source/funke_NatureNano2015.pdf [1:00 PM]
jandersonlee: is there something nanotubes are used for that RNA tubes could do instead? [1:00 PM]
Jennifer Pearl: do you think it would be good from DOE viewpoint to take on mod of each diesign form a list of 50 to get 50 submissions or get a sample of at least 3 from each mod (by each mod i mean a list of mods for a singel source design) [1:01 PM]
whbob: A switching RNA acting as a turbine in a nanotube. [1:02 PM]
jandersonlee: how about a jig-saw puzzle? [1:01 PM]
Jennifer Pearl: wanted to ask while you were here. looking for advice on best way to do this experiment [1:02 PM]
rhiju: @jal yes certainly, RNA tubes could be instantiated inside a cell -- harder to do that with DNA. [1:02 PM]
jandersonlee: I guess Cosy's work is something like that in 3d [1:02 PM]
rhiju: @jal, jigsaw puzzle -- nice idea. [1:02 PM]
Jennifer Pearl: sorry to go off subject [1:02 PM]
jandersonlee: Cody's [1:02 PM]
rhiju: @jal, hmm 3D jigsaw puzzle. [1:03 PM]
Jennifer Pearl: @rhiju did you see my question? [1:03 PM]
Elves: https://en.wikipedia.org/wiki/Simple_machine [1:03 PM]
rhiju: @jennifer great question about making systematic mods -- i think jandersonlee is actually doing such a study in the current round [1:03 PM]
rhiju: @jal got a wiki page on your current plans? [1:04 PM]
LFP6: Hmmm... What about generating RNA shapes from pcitures? [1:04 PM]
rhiju: @elves, simple machines -- very nice! hand't seen that wikipedia page [1:04 PM]
jandersonlee: @rhiju - not public yet [1:04 PM]
jandersonlee: @JP _ PM me your email and I'll give you access to a Google doc [1:05 PM]
Elves: awesome graphic [1:05 PM]
Jennifer Pearl: ok [1:05 PM]
Jennifer Pearl: sent [1:05 PM]
rhiju: @LFP6 that's a really cool idea. [1:05 PM]
jandersonlee: @JP You could try full runs of a few designs abd selected ones (expected to be good and bad) on some more [1:06 PM]
Guzz: I've only watched about half of the video, is there a way to do 3d switches with the tools? [1:06 PM]
rhiju: what if we had a 3dscan, and then could automatically instantiate it with RNA. [1:06 PM]
rhiju: @Guzz yes, we've been thinking about that -- actually by embedding the same 'aptamers' for the FMN molecule that we use in 2D switches! [1:06 PM]
LFP6: @Rhiju: Oh, and should chat mods have champion w/ management on the forums? Or something to bring up/discuss at some other point? [1:07 PM]
Elves: :) why, i have a high res 3D scanner [1:07 PM]
jandersonlee: you could use 2d-jigsaw puzzle for a base with 3d height for a surface [1:07 PM]
LFP6: Sorry, off topic again but I'm just PMing John [1:07 PM]
rhiju: i think we're in some agreement as a community that the coolest things to strive for would be dynamic -- that's been one very helpful them from the feedback so far. [1:07 PM]
Guzz: Nice. That should be useful for grabbing applications for example. [1:07 PM]
LFP6: That would be neat [1:07 PM]
Astromon: Sry was late Is there a Link to todays devchat monthley video? [1:08 PM]
Elves: hyperna http://biologos.org/uploads/static-content/mathematics_3_5.jpg [1:08 PM]
jandersonlee: or jigsaw peices that bind to different fourescent colors [1:08 PM]
LFP6: Woah Elves [1:08 PM]
Elves: https://en.wikipedia.org/wiki/E8_%28mathematics%29 [1:08 PM]
Jennifer Pearl: @JL thanks. all the designs might do good and tehre is alist of about 50 source designs with betwen 1 adn 50 mods for each desing. like I said I only filter out about 1800 of 2000 designs and there are still some bad scores [1:09 PM]
Elves: E8!!! https://en.wikipedia.org/wiki/E8_%28mathematics%29#/media/File:E8_roots_zome.jpg [1:09 PM]
Elves: astro yes [1:09 PM]
LFP6: Looks complicated [1:09 PM]
rhiju: @Guzz, agreed. I think an all-purpose grabber would be wonderful. Especially if it were triggered -- like a mousetrap. [1:09 PM]
Elves: http://www.eternagame.org/web/news/6445579/ [1:09 PM]
rhiju: hey everyone got to go -- @elves could you post transcript when you get a chance? [1:10 PM]
Elves: yep! thanks :) [1:10 PM]
Astromon: Thanks<> [1:10 PM]
Jennifer Pearl: by Rhiju [1:10 PM]
Guzz: Forgive me if this is a silly question, but do we understand the mechanism whereby the spindles fibers are spun up during prophase of mitosis? [1:10 PM]
rhiju: @guzz, I think we know (most of) the molecule involved, but not the detailed mechanism. E.g., we can't "reconsistute" the process from purified parts. [1:11 PM]
LFP6: Thanks Rhiju, later. Sent a PM back btw [1:11 PM]
rhiju: bye everyone! [1:11 PM]
whbob: Bye [1:13 PM]
Astromon: late Riju [1:12 PM]
Elves: bye! thanks for the cool chat everyone [1:12 PM]
Astromon: thanks you too Elves will look forward to read the chat< [1:14 PM]
Jennifer Pearl: it was good to see you Elves [1:16 PM]
Astromon: @jen-pearl hi jen i was wanting to ask you and make sure i was doing the lab right. The FME we want the level to be between 26 and 30..is that in the firrst state? [1:16 PM]
Jennifer Pearl: yes alwasy 1st state righ tnow [1:17 PM]
