No replies
Joined: 11/14/2009
Groups: None
Time Name Message
14:02 infjamc My main concern is that there is no way to manually upload a solution yet
14:03 CharlieFortsConscience the lottery is, you dont know how walked the solns are... you assume walked to death
14:03 Aotearoa yes, we need pure ones and walked ones seperate huh
14:03 Seth_Cooper what do you mean about not being able to manually upload a solution?
14:04 Seth_Cooper can you be more specific?
14:04 zoran what we're seeing on our end is that the best candidate solution gets worked more, but there's little exploration with merging insights from multiple solution
14:04 infjamc My understanding is that the top solutions are uploaded automatically periodically
14:04 Aotearoa i agree zoran
14:04 infjamc Allowing for manual uploads could speed up the process
14:05 infjamc Zoran: my experience with the latest all hands puzzle is that the top solutions are sufficiently different that direct copying isn't easy
14:05 Seth_Cooper manual uploads that are made publicly available to everyone, you mean?
14:05 infjamc Seth: Yes
14:05 infjamc Zoran: (continued from last message) The ability to do side-by-side, residue-by-residue comparisons between two solutions would be nice
14:06 zoran i think we can do that. do you think it would help to avoid all people working exactly on the same (currently best) soln
14:06 Aotearoa YES
14:06 infjamc I'm thinking about something more advanced than the current method for using one solution to improve another (loading B as a guide to improve A)
14:07 infjamc The ability to work on two structures simultaneously might be nice
14:07 infjamc ...with a table of side-by-side comparison of the scores of each residue
14:07 zoran this is good. btw you can use one structure as a guide for the other. do you use that?
14:08 infjamc That's what I currently do, but sometimes I feel that it's not enough when two solutions have similar scores at the margin
14:08 infjamc (but very different backbone structures)
14:08 zoran so to recap:
14:09 infjamc If the side-by-side comparison is too much to implement, I can see one simplification that would work
14:09 zoran ability to share saved solutions with all
14:09 Aotearoa YES
14:09 infjamc Namely, displaying the difference of score at this residue relative to the guide when we use tab to see residue scores
14:09 zoran side by side comparison with per-residue score difference
14:10 infjamc Even better would be the ability to have the guide show up in full-atom mode, too
14:10 infjamc It matters when you're working with hydrogen bonds
14:11 zoran ok, i'm sure you have a list of things that would help casp in general. now is the time to make your voice heard
14:12 infjamc I'll let other people speak first-- it seems that I'm being the vocal minority here
14:12 mat747 dev - will you be adding this chat log to the website
14:13 zoran yes, including the summary of the ideas
14:13 aap2 minor request: split show sidchains clashes or exposed in two
14:13 mat747 thanks - back 50 mins
14:13 HillObserver uh, this might sound obvious but it's hard to focus on improving any given puzzle when there are a lot of open puzzles
14:13 Aotearoa me too hillobserv
14:14 Judecca does anybody else think it would be useful to take say, chunks from different alignments and use them together.
14:14 spvincent I imagine everyone has their wish list:unfortunately its different for each player
14:14 Judecca lets say alignment #1 is missing the first 20 or so residues but #2 has the first 20. would it work to mix and match
14:14 infjamc And while we're on the topic of alignments...
14:14 Seth_Cooper Judecca, we have implemented just that
14:14 zoran with casp we don't know what we're getting every day. but we realize that having 3 puzzles for the same target is not good. we will not do this from now on
14:15 CharlieFortsConscience can you give an idea of the range of puzzles for this casp? my first, but is it generally alignmment as we've seen in this first week?
14:15 Judecca Seth_Cooper: sweet :)
14:15 Seth_Cooper allowing you to only thread a part from each alignment
14:15 infjamc Dev: (I've mentioned this in an Email to Firas earlier) How's the progress on cutpoints?
14:16 infjamc The ability to fold a protein as several smaller pieces and then sew them back together is sorely needed when there are large gaps in the alignment
14:16 Aotearoa I would like to see the threading aminos have some kind of numbering system for easier location on 3d protein, showing also orange phobic groups and scores, Blue amino groups and scores and even if possible make the threading possible from the actual screen as well as the threading alignment box.
14:16 zoran except for the very few ab initio (from scratch) targets all will be alignment.
14:17 Aotearoa *then shuffling them back and forth would way more constructive instead of shooting first and what ever we hit calling it the target.
14:17 Seth_Cooper infjamc: cutpoints is something we've looked into from time to time but has never made it all the way to a full feature
14:17 infjamc Okay
14:18 Aotearoa thanks Seth. yeah sometimes its easier to see the protien when all of it is present
14:18 infjamc Again, I really hate to ask so much just because I have the "insider information," but since it's a recurring problem I just want to make the alignment tool easier for everyone to use
14:18 infjamc ...partly because a common complaint I've heard is that the clashing residues often caused by forcing a bad template can be difficult to fix, especially for less experienced players
14:18 Seth_Cooper Aotearoa: you'd like the alignment panel to be numbered, and also show hydrophobiciy as well?
14:18 zoran we won't be able to do cutpoints for casp. seth has 2 prototypes but they never made it up to a feature.
14:19 aap2 could it be possible have an option to leave aminos strechted while/after threading?
14:19 rwillett The alignment already does show hydrophobicity.
14:19 Seth_Cooper Aotearoa: as in, color the letters by hydrophobicity rather than the current color scheme?
14:19 Aotearoa that would just make it easier for us to see before we thread, where they are in relation to the protein shape, as ive seen proline in a helix which is not normal
14:20 rwillett The coloring in the alignment is better how it is.
14:20 Aotearoa *no the colours they are are fine, but to add a small extra feature to show just how many are present, for making sheets easier
14:20 Aotearoa *diving sheets and helix's*
14:20 rwillett Ah.
14:20 zoran spvincent: we'd still like to hear everyones wishlist and understand it better. we'll have to prioritize what can be done later
14:20 rwillett LIke to click between the two.
14:21 infjamc zoran: regarding prioritization-- there's an issue that seems to be present since the alignment tool was introduced
14:21 infjamc Often, at least one residue will have a misplaced backbone atom after applying an alignment
14:21 zoran aap2: they take the coordinate structure from template
14:21 Seth_Cooper infjamc: yes
14:21 Aotearoa *dividing sheets
14:21 spvincent Written documentattion is something I'd very much like to see,particularly for the alignment tool
14:21 Seth_Cooper infjamc: I think we have found why the oxygens were out of place
14:22 infjamc Okay
14:22 Aotearoa great news Seth !
14:22 infjamc Is there a way to fix it via the rebuild tool, though?
14:22 infjamc It really hurts the protein score if that problem means losing a hydrogen bond
14:22 zoran Aotearoa: can you provide more details with each bullet. we're not quite sure what you mean
14:22 Seth_Cooper no I don't think the rebuild tool will take care of that
14:22 Seth_Cooper we have a fix for threading that will prevent it
14:22 spvincent And there are a whole bunch of things related to scripting but its more convenient to prepare a wish list for these rather than try and enumerate them here.
14:22 infjamc Seth: thanks
14:23 aap2 zoran: yes, but now strechted segments are automticaly corrected, i would to fix it myself because it often does things i do not want
14:23 zoran spvincent: (and others) send the list to me, just make sure it is clear what you want and how that would help you
14:24 spvincent Will do: thanks
14:24 infjamc Actually, no to think of it... if aligning part of the template is allowed, there is one step further that might help
14:24 zoran aap2: this is the same as cutpoints request then. it will be hard to get that done for this casp
14:25 infjamc ...allowing the user to set a custom location to force each residue into
14:25 infjamc A self-applied template, essentially
14:25 infjamc This could be useful for large gaps
14:25 infjamc Again, that's for the long term
14:26 zoran we can create a better documentation for the alignment
14:26 Aotearoa While we are shuffling the aminos back and forth to match up to templete, its hard for me to know what groups of aminos are being moved, if i would like to seperate them, its hard to know where they will end up on the template. so by just including numbers would mean for a 200 reside protein, i could divide out 20 per sheet, and counting them is hard on the screen. if once grouped they totaled how many aminos were present in the
14:26 Aotearoa grouping, *perhaps plus a score* then i would quickly see where to move them to make the protein better set out with the right aminos in their favorite places. perhaps just an orange and blue addition.
14:28 Aotearoa do you know what i mean by a score of the total grouped aminos?
14:28 zoran Aotearoa: i think the selective threading that we should have out soon might help with this
14:28 Seth_Cooper a score of just the ones' youve selected?
14:28 CharlieFortsConscience Occasionally, puzzles 'feel' different with wig and shake, some feel 'sticky' and lethargic. Does each puz have wig shk 'thresholds? Rebuilding prolines is currently soul-destroying. Rebuild was jiggered a few updates ago, it's not the power it used to be
14:29 Aotearoa yes, between the breaks in the threading box (black area) and compaired to the template
14:30 Seth_Cooper no, wiggle and shake use the same parameters for the competitions puzzles
14:30 Renton IMAGE:
14:31 zoran CharlieFortsConscience: some if this may be due to the initial state of the puzzles. sometimes they are more optimal than other times. i don't think we changed the functionality
14:32 Seth_Cooper Renton, what should we look at in that image?
14:32 Aotearoa right now i can look at the thread from one of the templates, do a protein_hide_score and see what protein alignment brings back the better score
14:32 infjamc zoran: a problem that I've been encountering a lot recently is similar to what CFC has mentioned
14:33 infjamc Often, I have to reload a puzzle just to get wiggle to work normally again
14:33 Seth_Cooper are there any puzzles in particular?
14:33 infjamc the alternative would be getting a slower wiggle that doesn't go to completion
14:33 infjamc Seth: EVERY puzzle!
14:33 CharlieFortsConscience yes
14:33 infjamc I'm not sure if it's because I'm running Foldit on single core
14:34 zoran and when you reload the game the wiggle works differently?
14:34 CharlieFortsConscience yes
14:34 infjamc (running it with both cores would overheat my cpu)
14:34 HillObserver inf - I run on a single core and don't seem to have this problem
14:34 infjamc zoran: I don't have to reload the game-- I just have to reload the puzzle by going back to the main menu
14:34 Renton *i just wanted to show off my puzzle and teams work*
14:35 zoran sounds like a clear bug. if you can have a saved solution that you can replicate it with, pls send
14:35 infjamc The problem, though, is that it's easily solved by reloading the puzzle
14:35 infjamc That's the coping mechanism I have been working with since March
14:35 infjamc (or possibly earlier)
14:36 infjamc Basically, it's not a major bug that would break the game-- it's just a bit annoying
14:36 Seth_Cooper that is odd, i'm not sure what might cause it
14:36 infjamc My guess is that it might have something to do with the wiggling procedure
14:36 zoran can it be repeatedly generated, or is it random?
14:36 Seth_Cooper if there are some steps to reproduce it you can send them to me
14:36 infjamc During those slower wiggles, I often get residues with much worse backbone scores
14:36 Aotearoa now, my biggest question is: are those templates giving us the right protein shapes possible ? do we know they are Good to work with ?
14:37 infjamc So, my guess is that the problem is that the wiggle procedure changes once a threshold is met
14:37 Aotearoa *Threading Templates*
14:37 infjamc And when that happens, only the local (closer to the worst clash) residues are considered rather than the entire protein
14:37 Seth_Cooper since we don't know the answer, we can't be sure of how good the templates are
14:37 zoran Aotearoa: yes these the best matches. sometimes there are more, so we give you a b and c puzzles with different set of templates
14:37 Seth_Cooper but, we are trying to use the best methods for finding them
14:38 aap2 i would like to keep tab pressed without the information screen flickering on and off
14:38 Aotearoa I am using them cause I believe it it the future, but on the otherhand, I also want to fold like I used too without them, the random shapes they bring up, do not always look like they would if we folded them ourselves.
14:39 zoran aap2: this will have a big effect on the casp outcome :)
14:39 Aotearoa *it is the Future* sorry my spelling is terrible today.
14:39 spvincent And please turn off the pulsing graphics, or provide an option to do so.
14:39 Seth_Cooper the templates are real protein shapes
14:39 HillObserver @spvincent: yes, turn off pulsing graphics
14:39 aap2 zoran: yes! :P
14:39 Seth_Cooper so they are realistic for what a protein would fold into
14:39 Aotearoa are the highest scoring templates so far, far from the original thread from the alignment
14:40 zoran we can turn off the pulsing (or provide an option)
14:40 Aotearoa or are those highest scoring proteins, one's we've made big changes too ?
14:41 spvincent Thanks!
14:41 Seth_Cooper yes, changes from the initial threading have to be made to get closer to the native
14:41 Seth_Cooper that's very important, to improve on the template
14:41 Aotearoa Thanks Seth !
14:42 aap2 i dont know if this is fixed yet but i had some issues with threading upon an already threaded protien, it caused unfixable clashes/the segments got very odd
14:42 Seth_Cooper otherwise, we could just guess that the template is the answer
14:42 Aotearoa *moon walks back to protein 1b*
14:42 HillObserver I've had the same double threading problem aap's had
14:42 Seth_Cooper This may be related to the problem infjamc brought up
14:43 Seth_Cooper which should be fixed in a update soon
14:43 Seth_Cooper do you often change the alignments, or just keep the initial alignment?
14:44 SUCHARD keep the initial
14:44 HillObserver I discovered the problem by accident...usually I don't rethread over an existing thread
14:44 infjamc I tend to change them to get a better match, even if that means leaving more gaps
14:44 aap2 change it a little bit
14:44 aap2 i go for less gaps
14:44 Aotearoa just one more quick question, when threading - if i leave two black spaces between my groupings of aminos, what does the alignment treat that as? the spaces on top line.
14:45 Madde modify it, thread and forget
14:46 Seth_Cooper if you leave spaces in the top line, that is a gap in the alignment that will need to be fixed when threading
14:46 HillObserver dev: I've noticed that some rebuilds take a long time but if I band the area to rebuild then the rebuild works very quickly and comes up with very strange solutions (that it wouldn't do if not banded)
14:46 Seth_Cooper when it threads, it will have to move those segments a bit to close the gap
14:46 Seth_Cooper HillObserver: that's correct
14:46 Aotearoa ahhhh ok
14:47 HillObserver is this a quirk?
14:47 Aotearoa here I am thinking they related to where they are positioned in the proteins shape on the screen. duh.
14:47 CharlieFortsConscience Coming back to the notion of no change to functionality of wiggle, why are local and global so different in their 'aggression'? it never used to be that way. Agan, seems skewed towards difficult hydrophobes like pro and met, but it used to be better. Tied to the rebuild amendments in the last few updates?
14:48 Seth_Cooper HillObserver:
14:49 Seth_Cooper CFC: the two wiggles are different, because to prevent the whole protein from moving, local has to 'cut' the protein chain and reconnect it
14:49 Seth_Cooper CFC: global wiggle doesn't need to do that
14:50 HillObserver thanks Seth, but it's not the speed of the rebuilds, it's that it rebuilds into really weird shapes and will sometimes embed backbone into other parts of the backbone
14:51 HillObserver it doesn't do this if I just run rebuild
14:51 Seth_Cooper HillObserver: well, when bands are present it mostly ignores things like the clashing score, in favor of the bands
14:51 HillObserver ah, I see, ok
14:51 Seth_Cooper HillObserver: perhaps it shoudl take that into account more
14:51 Aotearoa any more casp questions ?
14:52 Seth_Cooper is it difficult to move the alignments when there are more than two lines?
14:52 Seth_Cooper ie several template sequences ?
14:53 HillObserver seth, yes. Sometimes I'll band sheets on a rebuilding script to prevent the protein from exploding and it defeats the purpose if it ignores clashing
14:53 Seth_Cooper HillObserver: I'll look into that
14:53 HillObserver thank you
14:53 Seth_Cooper as far as multiple templates go,
14:54 mimi going off alignment for a moment, something I would like to see would be an option to do a reset of the absolute best score. So that when you restart a puzzle you don't have to be constrained by a previous high score. It can affect the use of certain scripts
14:54 infjamc mimi: you can do this manually
14:54 Renton YAY I agree Mimi
14:54 Seth_Cooper one reason we have multiple puzzles (like 1a 1b 1c) is that we want to give lots of templates but as we add more to a single puzzle the alignment gets harder and harder to use
14:54 infjamc Go to open/share solutions
14:55 mat747 mimi - me too
14:55 infjamc Check show auto and quick saves
14:55 infjamc Then delete your absolute best solution and restart the program
14:55 infjamc Problem: too much hassle?
14:55 mimi but you may want to keep that solution
14:55 aap2 i might still want to use it
14:55 Renton looks back at infjam
14:55 spvincent Can you keep your best but call it something else
14:56 infjamc In that case, load the solution, save a copy, then delete it
14:56 spvincent mimim beat me too it
14:56 Renton pin a tail on it.
14:56 HillObserver seth: I would be in favor of more alignments than (a), (b) and (c) puzzles...don't know about everyone else
14:56 Seth_Cooper perhaps "restore absolute best" from a script should really just call "restore recent best"?
14:56 infjamc But again, that's another extra step!
14:56 CharlieFortsConscience OH.... yes. Development of LUA - is it possible to get reset_best to include quick_save_slots? It'll give SO much more room to manouevre
14:56 infjamc And while we're at it: what about the ability to manually set dihedral angles through LUA?
14:57 spvincent But then you can't use recent best in the script should you ish to do so.
14:57 aap2 also an actualt local wiggle in lua
14:57 infjamc Again, that's just me wanting to run Foldit like PyRosetta, so if that's too much I'll shut up now
14:57 TheGUmmer and freeze segments in LUA
14:57 spvincent right : Quicksave or an equivalent is defintely on the wish list
14:58 CharlieFortsConscience It took so long to get LUA in and working, but we need a 2nd level evolution to fulfil our goal of automating us out of the equation...
14:59 HillObserver one last bit from me: in the selection interface, the behavior option is not adjustable when a script is running (it hides behind the script "cancel/show output" box. Can you make this box transparent so we can click behind it?
15:00 Madde you can press the B key
15:00 TheGUmmer you can move it can't you
15:00 HillObserver thanks Madde.
15:00 Madde I found out accidentally
15:00 CharlieFortsConscience ...and and and.... We used to be able to click set recent best mid-recipe or script, as things were happening
15:01 aap2 more importance sliders, like an sidechain subscore importance slider
15:01 infjamc Oh, and before I forget...
15:01 Seth_Cooper okay, so for LUA: quicksave, local wiggle, and freeze?
15:01 infjamc It would be nice if there were a quick way to force hydrogen bonds
15:01 infjamc Currently, I have to do this 30+ times after applying an alignment:
15:01 infjamc Middle click on hydrogen, drag to oxygen
15:02 infjamc Right click on band, choose change band length
15:02 infjamc Decrease to zero, then click on okay
15:02 infjamc Repeat for every hydrogen band until the entire beta sheet is secured
15:02 infjamc It would be nice if there's a two-click method for doing this
15:02 Seth_Cooper i see, like a 'hydrogen bond band'
15:02 infjamc Namely, click atom or residue A and then atom or residue B
15:03 infjamc The result is a zero-lengthed band at normal strengthen between them
15:03 Seth_Cooper or even just two sheets, and it could find the appropriate backbone atoms
15:03 infjamc Again, it's only 2 clicks vs 5, but it really adds up
15:03 Christoph automatic lining up of sheets would be very cool, if something like that is possible
15:04 Aotearoa A sheet maker too please
15:04 Madde
15:04 mat747 seth - we need change band strength in Lua
15:04 CharlieFortsConscience One more thing.... cys-cys bonding as routine? given their importance in vivo, could it be reflected in Foldo?
15:04 infjamc Aot: same for helices?
15:05 Aotearoa for the newbies yes.
15:05 Seth_Cooper tweak has methods for making sheets and helices as well
15:05 TheGUmmer tweak in |L|UA would be nice
15:05 infjamc The problem is that it works by messing up the loops next to them
15:05 Aotearoa its just hard to align them using the good old U bend
15:05 infjamc And the loops have to be long enoguh
15:06 infjamc *enough
15:06 Seth_Cooper i see
15:06 Seth_Cooper okay, I wanted to ask about the multiple templates, because beta_helix and I were talking about how to make it more usable to have many templates in one puzzle
15:06 Renton Im watching you CFC !
15:06 CharlieFortsConscience :)
15:07 HillObserver I'm all for more multiple templates
15:07 infjamc Renton: what happened? Oh, 299?
15:07 Renton come on give us ONE puzzle !
15:07 aap2 2 puzzles (a and b) with more templates sounds good
15:07 Seth_Cooper basically what we were thinking was instead of aligning them all at once, basically align 1 template at a time
15:07 Christoph maybe only show one line in the aligments window and have the ability to exchange that line with the different aligments
15:07 Aotearoa I agree aap2
15:08 Seth_Cooper and the rest go into a "reserve" that you can swap in
15:08 Aotearoa nice idea !!!
15:08 infjamc It's like forming an all-star team in sports :-)
15:08 Seth_Cooper something like this:
15:08 HillObserver the problem with multiple puzzles is that the really good players will eventually use the best alignment from one of the two puzzles and just replicate it in the other puzzle
15:09 TheGUmmer like pop up with a list of alignments to choose from
15:09 infjamc Seth: I figured out the URL on my own, but it's cut off
15:09 Renton omg 12264 finally. *thank you "Aotearoa's Romance"*
15:09 CharlieFortsConscience Well... I've learned that alignment is basically a head-start, the end-score derives from your 'processing' I dont waste time threading a better score unless there are obvious mismatches
15:09 infjamc Everyone else: the file extension is png
15:09 Christoph seth: yup, looks good
15:09 Seth_Cooper basically the blue arrows would cycle through templates
15:10 Seth_Cooper the green would move them up into what is now the current alignment window
15:10 Seth_Cooper and the red move them out
15:10 Aotearoa looks good to me Seth !
15:11 aap2 if it remembers the player made changes it would be good
15:12 Seth_Cooper yes, the idea is if you take a template out then put it back in it will remember where it was
15:12 Seth_Cooper by where it was, i mean the alignment
15:12 Aotearoa YAY !
15:12 HillObserver btw: we were sent the fold it table to document our puzzle moves. Is it not possible for the program to just simply create a puzzle log file that documents each script run?
15:12 CharlieFortsConscience ...yes - extend the remember cache, it's very short atm
15:13 Renton I have to go, thanks Zoran and Seth ! look forward to seeing you again soon.
15:13 Aotearoa See you guys !
15:13 HillObserver yes, must run too. Thank you folks for listening.
15:13 Seth_Cooper HillObserver: you should ask beta_helix about that
15:14 HillObserver @seth: ok, will do
15:14 Seth_Cooper okay, I think that's good feedback for today! thanks for chatting
15:14 TheGUmmer thanks you for listening
15:14 Seth_Cooper and thanks for playing!
15:15 CharlieFortsConscience Cheers Seth
15:15 Madde will the next chat start an hour earlier, too?
15:15 infjamc Madde: technically, PST on the website should read PDT
15:15 Aotearoa my goat makes three stomps
15:15 infjamc The UTC times are correct, though
15:16 Madde I was looking at the GMT times
15:16 infjamc Please correct me if I'm wrong, but I thought it started on time?
15:16 Seth_Cooper yes, we mean for them to start at 2pm today
15:16 mat747 no
15:16 Seth_Cooper and 11 am tomorrow
15:17 Seth_Cooper it looks like the UTC was off by an hour
15:17 CharlieFortsConscience Madde, where are you?
15:17 Madde Germany
15:17 mimi that is so Seth
15:17 infjamc Madde: summer time?
15:17 Seth_Cooper I'll fix that in the blog post
15:17 Madde yes
15:17 Madde it's GMT+2 here
15:18 Madde and GMT+1 in winter
15:18 infjamc So it started at midnight for you?
15:18 infjamc (the chat)
15:18 Madde it should have
15:18 mat747 GMT +9.5 here
15:19 CharlieFortsConscience .5???
15:19 Madde but it did at 11pm
15:19 infjamc Hmmm
15:19 infjamc That's strange...
15:19 Seth_Cooper Does the blog post look correct now?
15:20 infjamc The time offset was correct for my time zone (U.S. Eastern Daylight time = -4)
15:20 CharlieFortsConscience *you're an extra 30 mins in the future mat?*
15:20 TheGUmmer like newfoundland
15:20 mat747
15:21 infjamc Ah, I see what's going on
15:21 infjamc Pacific daylight time = -7
15:21 infjamc Not -8!!!
15:22 Seth_Cooper You can always check the clock on the top of the Foldit page
15:22 Madde Seth: yes, it looks correct now
15:22 Seth_Cooper well I should get started implementing this feedback
15:22 mat747 cfc - ?
15:23 Seth_Cooper talk to everyone tomorrow!
15:23 aap2 cya
15:23 Madde bye
15:23 BletchleyPark whoa did I miss the chat ?
15:24 infjamc The UTC time on the website was off by one hour... it's fixed now
15:24 aap2 yep
15:24 BletchleyPark yeah great
15:25 BletchleyPark Is the quake bug being worked on ?
15:25 Renton you missed out on seth and zorans autographs on an exclusive card with their photos on it
15:25 BletchleyPark I could care less
15:25 Renton collectors items !
15:25 BletchleyPark is the quake bug being worked oon
15:25 aap2 i dont remember it being mentioned in the chat
15:26 Madde it seems you are the only one with this bug
15:26 Renton not unless you feedback it
15:26 infjamc That depends whether the Irani clerics agree (sorry, bad joke)
15:26 BletchleyPark no, I seem to be the only one mentioning it
15:26 mat747 infjamc - forward or back ?
15:26 infjamc mat747: my guess is that someone got PST and PDT confused
15:27 infjamc The former is UTC-8, while the latter (which is what the US west coast is running on right now) is -7
15:27 Madde I'm using Quake a lot and I never had this problem
15:27 Renton omg you guys are geeks
15:27 infjamc So 2 PM on the west coast is UTC 2100
15:27 infjamc Which is when it actually started
15:27 zoran ok, i'm back. are there more questions/suggestions?
15:28 Renton bletchypark go man gooo
15:28 aap2 lol
15:28 zoran the time being off was actually my fault
15:28 BletchleyPark off global, ciao
15:28 Renton ZORAN is here BP
15:28 --- findow is away (Auto away)
15:29 zoran it looks like the quake has been discussed already. is it deterministically reproducable?
15:30 BletchleyPark Hi zoran, yes it is
15:30 infjamc zoran: I think I might have mentioned this last year
15:30 BletchleyPark on two different machines
15:30 BletchleyPark qith quake with a lot of bands
15:30 BletchleyPark with
15:30 infjamc If you try to band every residue when the protein is more than ~80 residues long, the program crashes
15:30 BletchleyPark this problem did not exist before
15:30 infjamc So the upper limit is around 3000 or 4000 bands?
15:31 BletchleyPark I don't think it uses even that many
15:31 zoran we can do that easly, but is that what BP wants just a bound on the # of bands?
15:31 BletchleyPark the problem was not in the debug version
15:31 BletchleyPark I tried to revert back to the one on the server but I got a 404
15:31 BletchleyPark NO ! :)
15:32 Renton only one hour left cfc
15:32 BletchleyPark no bound on those bands, I just do not want the program to crash on me
15:32 BletchleyPark please
15:32 zoran seth said he looked into it but he couldn't reproduce it on 2 machines he tried
15:33 BletchleyPark well, I can positively reproduce it every time on two machines I tried it on
15:33 zoran windows?
15:33 BletchleyPark yes, XP64 bit and W2K 32 bit
15:34 zoran any way you can try it on the more recent os?
15:35 zoran we don't have these oses
15:35 BletchleyPark it worked before on both OS'es so it is not an OS issue
15:35 BletchleyPark and it also worked fine on XP32
15:35 zoran it now does not work on xp32?
15:36 BletchleyPark that machine is currently not available
15:36 Madde I'm using Quake on XP SP3 32bit and Win7 64bit w/o problems
15:36 BletchleyPark I am referring to a variant of quake with many bands
15:36 BletchleyPark original quake does not use many bands
15:37 Madde oh, ok. I'm using the original
15:37 zoran is there a way others can try to do exactly what you're doing and see if they can replicate the crash?
15:37 zoran do we have that exact script. we thought you were talking about the quake script
15:37 BletchleyPark I could share the recipe with seth ?
15:37 zoran plese do
15:37 BletchleyPark it is a variant on quake
15:38 zoran as well as a saved puzzle on which it verifyably crashes
15:38 BletchleyPark is there a simple way to share it with seth only ?
15:39 zoran message it to him on the web site
15:39 Madde I could try our group's mega quake
15:39 BletchleyPark it is a recipe, how can I message a recipe ?
15:39 zoran it seems something specific to that script.
15:40 BletchleyPark Please try it Madde, see what happens
15:40 BletchleyPark I can make screenshots and save them as jpgs ?
15:41 zoran message seth, and he'll give you instructions.
15:41 BletchleyPark ok, thanks
15:42 zoran anything else we should know regarding improving casp outcomes?
15:43 BletchleyPark not that I can think of now, thanks
15:44 vakobo variant of overnight script i use works fine on XP32 and crashes on W2K and Linux 64bit (ubuntu 10.4). On W2K with error message and w/o any message on linux
15:44 BletchleyPark interesting
15:44 BletchleyPark which error message ?
15:45 zoran vakobo: pls message seth on site with the script. and the description of failure
15:45 vakobo ok
15:46 vakobo simple message or feedback?
15:46 zoran message
15:47 mat747 zoran - i would like Steven Pletsch suggestion for Lua
15:47 mat747
15:51 zoran all of it?
15:51 mat747 if you like
15:52 mat747 "sidechain to the specified position " would help via lua
15:53 CharlieFortsConscience ...but we'll settle for quicksave slot options when running scripts...
15:53 zoran you mean to one of the discrete things it can snap to?
15:54 zoran we'll have quicksave slots
15:54 CharlieFortsConscience more than 3?
15:55 zoran how many?
15:55 Renton as many as amino acids on protein?
15:55 mat747 zoran - yep
15:56 CharlieFortsConscience aim for 10. I'd settle for 5.
15:57 zoran ok we'll have 5 quicksaves in lua
15:58 CharlieFortsConscience :) 6 would be better...
15:59 zoran we can have 20 and only the first 5 have key shortcuts. way too many could be a memory consumption issue
16:00 Madde who needs shortcuts in LUA?
16:00 smith92clone define shortcut
16:00 zoran are people sharing their recipes or is it the case that most people work with private ones?
16:00 zoran hotkey
16:01 smith92clone I run recipes with high vote count, read it, then start making changes.
16:01 CharlieFortsConscience If you implement 5, scripting will reach another level. We desperately need a way to circumvent the gulf between abs_best and recent
16:01 SUCHARD I work with others
16:01 aap2 share with team
16:01 Renton 1b 12265 omg wooohooo
16:02 vakobo it wold be nice if script execution terminates automaticaly on puzzle expiration
16:02 Madde I share some with public, some with team and use slightly modified ones private
16:02 vakobo *would
16:02 --- makedon is away (Auto away)
16:02 Renton 30 stinking minutes left before I can summersault through the house ending in a splits.
16:03 keypad5 <-- 30 minutes to find 60 points... hmmm
16:03 Renton come and fold with us then silly. hahaha
16:04 Renton hard to find the last points though
16:04 zoran ok, must leave now. thank you all. we'll be here tomorrow as well.
16:04 mat747 zoran - a replay to sidechain to the specified position - we can move a selected segment sidechain to different positions in the newinterface now with Shift + arrow keys , it would good if we could do that via LUA
16:05 BletchleyPark night Zoran
16:05 BletchleyPark BP is off global
16:05 vakobo bye, Zoran
16:05 Renton hi Vakobo !
16:06 vakobo hi
16:08 Renton great team work CG
16:08 Renton oh gone already

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, Boehringer Ingelheim, RosettaCommons