OpenWorm Office Hours 11/21/12

gistfile1.txt

[09:35] <+openworm> who has a question i can answer?
[09:36] <+openworm> It has been an interesting few weeks for the project
[09:37] <+openworm> we are reaching the end of our third release
[09:37] <+openworm> meaning we will have some new downloads by the beginning of december
[09:37] <+xrazybud> where will they be available?
[09:37] <+openworm> we will post on twitter, tumblr, g+
[09:38] <+openworm> i'm thinking we may need to add a download section to our website
[09:38] <+openworm> right now our downloads are a bit scattered
[09:38] <+openworm> here is where they are right now
[09:38] <+openworm> https://code.google.com/p/openworm/downloads/list
[09:39] <+srd|> how much new electrophysiological data will the project need to achieve its goals?
[09:39] <+openworm> as we have been transitioning to github, we probably will seek to do github downloads
[09:40] <+openworm> well the more the better :)
[09:40] <+openworm> more specifically though
[09:41] <+openworm> we are hoping that we get neuron by neuron fast calcium imaging of a lot of neurons
[09:41] <+openworm> currently this is what David Dalrymple has proposed to do and has gotten some funding to do if I understand correctly
[09:41] <+openworm> he's proposed to do this comprehensively
[09:42] <+openworm> but there are other labs that are doing this work too, not focused on building a comprehensive data set (as far as I know)
[09:44] <+openworm> beyond that we'd be happy to have more data where someone has stuck an electrode inside c. elegans.  this isn't done much because it is hard and because the worm loses its bodily integrity when you do it
[09:44] <+openworm> we got a little bit of this data for muscle cells from the francis lab at umassmed
[09:49] <+openworm> other questions?
[09:50] == mode/#openworm-office [+o openworm] by ChanServ
[09:50] == mode/#openworm-office [-i] by openworm
[09:50] == john_idol [6cd86e7a@gateway/web/freenode/ip.108.216.110.122] has joined #openworm-office
[09:50] == john_idol [6cd86e7a@gateway/web/freenode/ip.108.216.110.122] has quit [Quit: Page closed]
[09:52] == mode/#openworm-office [-m] by openworm
[09:53] == mariusz__ [46a809bf@gateway/web/freenode/ip.70.168.9.191] has joined #openworm-office
[09:53] <@openworm> Hi mariusz___
[09:53] <mariusz__> Hi
[09:54] <@openworm> A couple of other interesting things that happened this week
[09:54] <@openworm> we had our weekly developers meeting
[09:54] <@openworm> and we had one of our members show off some really neat code
[09:55] <@openworm> http://www.youtube.com/watch?feature=player_detailpage&v=QYi1cpQQ-7E#t=1794s
[09:55] <@openworm> if you haven't seen this bit yet i'd recommend it
[09:55] <@openworm> i can explain more about what it is
[09:56] <@openworm> basically, we've gone from our worm browser (http://browser.openworm.org) to some code that works with the neuron models as electrical models rather than just shapes
[09:56] <@openworm> and in this case the model also understands the connectivity graph
[09:56] <@openworm> so you can both browse the electrical properties of the model and you can see what neurons are connected to what in 3d
[09:57] <slarson> Now there is a really nice existing system for browsing c. elegans neuronal connectivity on the web already: http://www.youtube.com/watch?feature=player_detailpage&v=QYi1cpQQ-7E#t=1794s
[09:57] <slarson> oops
[09:57] <@openworm> http://wormweb.org/neuralnet#c=BAG&m=1
[09:57] <@openworm> tjat
[09:57] <@openworm> that's the link i was shooting for
[09:58] <@openworm> difference is that what we're showing on the video is a system that goes beyond the 2d graph into 3d
[09:58] <@openworm> both have their place
[10:01] <@openworm> questions about that stuff?
[10:02] == john_idol [6cd86e7a@gateway/web/freenode/ip.108.216.110.122] has joined #openworm-office
[10:03] == john_idol [6cd86e7a@gateway/web/freenode/ip.108.216.110.122] has quit [Client Quit]
[10:03] <mariusz__> too many to count:)
[10:03] <+srd|> openworm: how will the parameters of the neurons be inferred from calcium imaging?
[10:04] <@openworm> a few articles / blog posts have come out recently that may be of interest
[10:04] <@openworm> oh
[10:04] <@openworm> great mariusz__ go for it
[10:05] <@openworm> srd| -- we've got a preliminary result on that
[10:05] <mariusz__> well, right now I'm just browsing through the website, learning about that stuff. Hopefully soon I will be able to help with some small stuff, if any help is needed
[10:05] <@openworm> https://docs.google.com/file/d/0B_t3mQaA-HaMZTE1ODA1MjgtNjQxYi00YWU0LWE5MjUtNTQyZDYyOTMxNzYy/edit
[10:05] <@openworm> mariusz__ we can always use help for sure
[10:06] <@openworm> srd| check that link
[10:06] <@openworm> basically we will use model optimization / genetic algorithms to search the parameter space for parameters that are unknown]\
[10:07] <+srd|> what will the fitness function be?
[10:08] <@openworm> for the experiment you see there it was fitting the output of the model to an existing trace
[10:08] <@openworm> but there are many ways to do that
[10:09] <@openworm> you should talk to Mike Vella about his strategy for this
[10:09] <@openworm> for my part, there's a paper that i think is the gold standard for fitness functions for this
[10:09] <@openworm> give me a sec i'll dig it up
[10:10] <@openworm> Achard and De Schutter 2006: http://www.ncbi.nlm.nih.gov/pubmed/16848639
[10:11] <@openworm> Their fitness function involves putting electrophysiology traces into a phase space
[10:11] <@openworm> rather than just fitting to the time points
[10:11] <@openworm> i think this makes much more sense
[10:11] <@openworm> and they get good results
[10:11] <+srd|> that does look like a useful paper
[10:12] <@openworm> we fit to the time points in the example i linked to.  but we realized we were just bootstrapping the existing method given by Boyle and Cohen http://www.comp.leeds.ac.uk/jboyle/JordanBoyle_files/extended.pdf
[10:12] <+srd|> how do you plan to extend its methods from single neurons to multiple neurons?
[10:12] <@openworm> This project is all about biting off small workable pieces of the problem
[10:15] <@openworm> the plan there is to chain this method.  We are starting from a muscle cell whose example electrophysiology we have.  then we will approximate the six motor neurons synapsing onto it based on what we know about its ion channels and whatever more we can gather based on calcium imaging
[10:16] <@openworm> then we will be exploring how to tune the combined system of the single muscle cell with the 6 motor neurons connected to it
[10:16] <@openworm> as a network
[10:16] <@openworm> and radiate outwards from there
[10:18] <@openworm> i have had some preliminary talks with a grad student we think can help us put GCAMP3 in the motor neurons we are interested in
[10:18] <@openworm> so that we can solicit the collection of these data
[10:19] <@openworm> Make sense?
[10:19] <+srd|> sounds like a good plan
[10:19] <+srd|> what's GCAMP3?
[10:20] <@openworm> that's a calcium sensor that fluoresces inside cells
[10:20] <@openworm> http://en.wikipedia.org/wiki/GCaMP
[10:21] <+srd|> ok
[10:21] <+xrazybud> what is the best resource to know what you guys are talking about? :P
[10:21] <@openworm> What brings you all here?
[10:22] <@openworm> Well
[10:22] <@openworm> :)
[10:22] <@openworm> I can try to unpack anything that doesn't make sense xrazybud :)
[10:22] <@openworm> what is confusing?
[10:22] <+xrazybud> i was trying to do research to figure out how i could live forever and found out about openworm
[10:22] <@openworm> OK
[10:23] <+xrazybud> well like, i don't know what a fitness function is for example
[10:23] <@openworm> Well i'd say if openworm results in you living forever that would be beyond our wildest dreams of what this project could accomplish....
[10:23] <@openworm> We're interested in pushing in a very specific direction of AI / computational neuroscience that we haven't seen a lot of energy put into.
[10:24] <+xrazybud> i'm also interested in AI
[10:24] <@openworm> Reverse engineering nervous systems is definitely what we are doing here
[10:24] <@openworm> Fitness function
[10:25] <@openworm> Refers to an equation used in a genetic algorithm that is used to measure when an individual is good
[10:26] <@openworm> http://en.wikipedia.org/wiki/Genetic_algorithm
[10:27] <@openworm> In real life, you can think of it as an equation that could measure how likely you are to successfully pass your genetic material into the next generation
[10:27] <+xrazybud> what are you using genetic algorithms in openworm for?
[10:28] <+xrazybud> does openworm breed?
[10:28] <@openworm> Because there are a lot of unknowns in the model, we use genetic algorithms to help us generate thousands of possible models and then pick the ones most likely to be correct
[10:28] <+xrazybud> ah cool
[10:28] <@openworm> That's the simple way to explain it
[10:29] <@openworm> In fact we're using that approach for the components of the model, not the whole model itself
[10:29] <@openworm> but that's a reasonable way to think about it to start off
[10:30] <@openworm> The paper I linked to above, Achard and DeSchutter 2006, used this approach to infer the right model of a specific neuron in the brain of a mammal
[10:32] <@openworm> Good question, xrazybud -- any thing else i can help explain?
[10:32] <@openworm> sajmn do you have any questions?
[10:33] <@openworm> BTW another recent article we linked to on twitter yesterday is this: http://www.newyorker.com/online/blogs/newsdesk/2012/11/ibm-brain-simulation-compass.html
[10:33] <@openworm> its a little pessimistic but there are a lot of great points the author makes.
[10:35] <@openworm> okay everyone
[10:36] <@openworm> i think it is time
[10:36] <mariusz__> @openworm: currently, what are your biggest problems or needs?
[10:36] <@openworm> thanks very much for coming
[10:36] <mariusz__> to make this project move faster:)
[10:36] <@openworm> oh
[10:36] <@openworm> well we'd love more help from motivated folks.  both programmers and experimentalists
[10:37] <@openworm> we have a lot we want to do and not enough hands to do it
[10:37] <+srd|> openworm: people who are sceptical about mammal whole-brain simulations are prime candidates to be enthusiastic about whole-worm simulations :)
[10:38] <@openworm> it would also be great to host a face to face meeting of all the developers and contributors a couple times a year but this is something that we need funding for
[10:38] <mariusz__> do you have a list, or place where I could read about your 'todos'?
[10:38] <@openworm> srd| yes we have found this too :)
[10:39] <@openworm> ironically there are even people skeptical about whole-worm simulations!  It is interesting the skepticism out there
[10:40] <@openworm> Yes, mariusz__ -- our roadmap is here: https://code.google.com/p/openworm/wiki/Roadmap  We will update this in mid-december to reflect the progress made in the last 6 months and new directions
[10:40] <mariusz__> @openworm - that's why it's critical that this project succeeds
[10:40] <+xrazybud> will i ever be able to have a worm simulation pet? :P
[10:41] <@openworm> That's important for everyone to understand -- we will be charting our course for the first half of 2013 and its a good time to volunteer even if you don't know specifically what you can help on.
[10:41] <mariusz__> well I'm in. just need to know where to report to :)
[10:41] <@openworm> xrazybud -- the closest thing we have now is the demo that works on windows only here: https://code.google.com/p/openworm/downloads/detail?name=c.%20elegans%202.087%2B%5Bupd.%202012%5D%20full.zip&can=2&q=#makechanges
[10:41] <+xrazybud> i want to help too, not sure what i can do though
[10:42] <@openworm> but we are working on the web based version now that will run in any browser
[10:42] <+srd|> openworm: perhaps the demo would make a good screensaver
[10:42] <@openworm> mariusz__ great!  send us an email at info@openworm.org and i'll coordinate with you to find something interesting for you.
[10:42] <@openworm> that goes for anyone else
[10:42] <@openworm> we are always looking for people who want to help
[10:44] <@openworm> And as always, you can follow what's new on our various channels: http://www.openworm.org/#/contacts
[10:45] <@openworm> Alright thanks everyone for coming
[10:45] <@openworm> We'll do this again in two weeks
[10:46] <+xrazybud> thanks for the info