These are chat archives for openworm/ChannelWorm

11th
May 2015
Stephen Larson
@slarson
May 11 2015 16:29
@travs and @VahidGh ready when you are
Vahid Ghayoomie
@VahidGh
May 11 2015 16:30
Hi
I'm ready
Travis Jacobs
@travs
May 11 2015 16:30
Hello!
Stephen Larson
@slarson
May 11 2015 16:31
Hi all; great
OK so, I think the place to start is for Travis to review what’s happened in the last week since we spoke on establishing the roadmap for the summer
With supporting links to resources that have been produced
Travis Jacobs
@travs
May 11 2015 16:33
Sure. I'll link the unified roadmap drawing to start with. This is a high-level overview of how the summer's project is structured.
There is also an updated documentation page that gives a description of this diagram, and goes into a little detail in each of the subprojects.
That page is definitely essential reading for someone getting up to speed on what's going on here.
Stephen Larson
@slarson
May 11 2015 16:35
its looking great
Travis Jacobs
@travs
May 11 2015 16:36
Since then, we have made some progress talking about a more detailed workflow that will be replicated in each of the subprojects. A diagram of the implementation for ChannelWorm can be seen here.
Also, ChannelWorm now has a waffleboard that will be useful in keeping track of where our issues stand over the summer.
Stephen Larson
@slarson
May 11 2015 16:37
Point on that — do we link that in the docs?
I don’t see it..
Travis Jacobs
@travs
May 11 2015 16:39
No, excellent point Stephen. I'll raise an issue on Github for that and then implement it.
Stephen Larson
@slarson
May 11 2015 16:39
Cool — in the main openworm_docs repo
Travis Jacobs
@travs
May 11 2015 16:39
Will do
Stephen Larson
@slarson
May 11 2015 16:40
ok —updates on conversations with @gsarma and @rgerkin ?
Travis Jacobs
@travs
May 11 2015 16:41
Also, if we zoom out a bit again, the summer's projects (without their interactions) are also described in this diagram
Right, both @rgerkin and @gsarma are definitely a great asset the validation/testing we will be doing in each of these subprojects
Stephen Larson
@slarson
May 11 2015 16:42
@VahidGh are you clear on who @rgerkin is and what SciUnit is and what we’re trying to do with the tests on that?
Travis Jacobs
@travs
May 11 2015 16:43
I spoke with Rick ( @rgerkin ) a few days ago, and he gave me an introduction to SciUnit, it's capabilities and how to implement it, including some necessary next steps, which I am in the process of rendering into Github issues.
Vahid Ghayoomie
@VahidGh
May 11 2015 16:43
Yes
Stephen Larson
@slarson
May 11 2015 16:44
@travs since validation/testing is an issue that cross cuts several sections, we should probably have a paragraph or two in the unified roadmap docs above the individual repo discussion that talks about it and sciunit, and makes links out that stuff, within the ‘high-level overview'
Travis Jacobs
@travs
May 11 2015 16:45
On this point, our next meeting may be more code-oriented, as we start to decide how to implement SciUnit in ChannelWorm (and, in parallel, in the other summer projects)
Absolutely @slarson, I will make note of this in the issues I will raise as well.
Stephen Larson
@slarson
May 11 2015 16:47
Cool
Travis Jacobs
@travs
May 11 2015 16:47
I spoke with @gsarma yesterday, and he expressed interest in writing code for the tests as soon as possible.
Stephen Larson
@slarson
May 11 2015 16:47
OK great— maybe he can be assigned to the issues coming out of the session with @rgerkin that you are rendering then?
or at least some of them?
He can also weigh in here on this discussion when he gets a chance :)
@VahidGh — feedback on any of this?
Travis Jacobs
@travs
May 11 2015 16:48
As I raise these issues, I will @mention him in each of them. Yes or assigned, you're taking the words out of my mouth haha :smile:
But that's right, he certainly is interested in writing some code
Vahid Ghayoomie
@VahidGh
May 11 2015 16:48
I think this is a good plan
Also @miladjafary and I had a discussion to start working on the fitting process
Milad will start to work on the digitization
Travis Jacobs
@travs
May 11 2015 16:52
Ok @VahidGh
Are there issues related to digitization in the repo right now? If so I'd like to assign @miladjafary to them so that we know they are covered.
Stephen Larson
@slarson
May 11 2015 16:53
On fitting— couple things
I’m going to get a 1:1 with Padraig next week
he’s out at a meeting for open source brain this week
In that meeting I will press him for some updates on Neurotune
I think we should not wait for him but get started on whatever we can do
The Achard & Deschutter 2006 paper begins to give a roadmap for methods for fitting
I have also heard there are better methods and I am inquiring with some contacts about this
Worst case we should try implementing something in python ourselves and work around neurotune until it is more up to date
we should not let ourselves be roadblocked
Thoughts?
Travis Jacobs
@travs
May 11 2015 16:55
Totally agree on every point.
Vahid Ghayoomie
@VahidGh
May 11 2015 16:55
@travs, yes, #17 that I've assigned Milad befor for that :D
Travis Jacobs
@travs
May 11 2015 16:55
Haha awesome @VahidGh, you're way ahead of me!
@slarson
I will add that paper to my reading list and get more acquainted with the fitting process.
Vahid Ghayoomie
@VahidGh
May 11 2015 16:56
@slarson, I agree
4
Travis Jacobs
@travs
May 11 2015 16:56
My assumption is that we are going to work on fitting in parallel with test writing. Correct?
Vahid Ghayoomie
@VahidGh
May 11 2015 16:57
I'll take a look at the paper mentioned and also considering the same work in Boyle & Cohen approach we will start to develop a fitting procedure
This message was deleted
Stephen Larson
@slarson
May 11 2015 16:59
Hmm yes. Also I’m recalling past work that has been done in this space
Our documentation on this topic is not great
Should be more here
which is basically not documented at all :-/
Let me see in the archives what we have on this; one sec
Travis Jacobs
@travs
May 11 2015 17:01
OK
Can you guys see this folder?
Travis Jacobs
@travs
May 11 2015 17:05
I can
Stephen Larson
@slarson
May 11 2015 17:05
can you see the PDF in there?
Travis Jacobs
@travs
May 11 2015 17:06
Yes, just downloading it now
The pdf is what we are interested in?
Stephen Larson
@slarson
May 11 2015 17:06
well both but yes the PDF has more info
Vahid Ghayoomie
@VahidGh
May 11 2015 17:08
Great! I also found resources especially for fitting HH models, looking to find them...
Travis Jacobs
@travs
May 11 2015 17:08
Ok excellent
Stephen Larson
@slarson
May 11 2015 17:10
OK well that’s some stuff
then there’s the neurotune documentation which you have seen @VahidGh , yes?
Vahid Ghayoomie
@VahidGh
May 11 2015 17:10
Yes in addition to pyelectro
Stephen Larson
@slarson
May 11 2015 17:11
yes ok
so in theory there are examples there that we can run…?
but if i recall you have run into problems?
Vahid Ghayoomie
@VahidGh
May 11 2015 17:11
We have some limitations for Neurotune+pyelectro for this purpose
Stephen Larson
@slarson
May 11 2015 17:12
Ok, what are the current limitations you are running into?
And apologies because I think you’ve said this before but I’m forgetting
if there’s an issue that captures this just link it
Vahid Ghayoomie
@VahidGh
May 11 2015 17:13
Actually we had a hangout with Padraig about this
and we discussed the limitations
Stephen Larson
@slarson
May 11 2015 17:14
OK — so I think getting those into issues on the neurotune repo is important
the main one being, is there an easily run example / tutorial that is the ‘hello world’ of this stuff
i’m noticing we don’t seem to have that yet unfortunately
@travs now that we are getting back into roadmap building, let’s also make a note to add the milestones column diagram into the docs page on the unified roadmap, cool?
Travis Jacobs
@travs
May 11 2015 17:16
Absolutely.
Vahid Ghayoomie
@VahidGh
May 11 2015 17:16
Yes, but that was because we were not sure to develop the fitting process from scratch or customizing Neurotune
Stephen Larson
@slarson
May 11 2015 17:18
@VahidGh Got it. Well it seems to me then that there should be a milestone somewhere to build a ‘Hello World’ example in python of doing the fitting process, and it is up to whomever spends time on it to either reuse Neurotune or build it from scratch. But emphasis should be on having documentation be such that other people can run it. Agreed?
That HeuristicWorm repo is close to that but it is in C++ and has zero documentation
Vahid Ghayoomie
@VahidGh
May 11 2015 17:19
Also I was working these days on the feature set for prediction algorithm, but trying to work on both in parallel
Stephen Larson
@slarson
May 11 2015 17:20
Vahid have you seen this? http://neurofitter.sourceforge.net/
I’m adding that hello world example as a milestone under neurotune for now in the diagram
Travis Jacobs
@travs
May 11 2015 17:23
@VahidGh It occurs to me that there is no optimization/fitting milestone in the Github repo yet. If there isn't somewhere you already track these issues, I will create it now, and we can add issues to it as we go.
Vahid Ghayoomie
@VahidGh
May 11 2015 17:23
Great, I'll take a look at this stuff. Most of them are similar to Neurotune+pyelectro.
Stephen Larson
@slarson
May 11 2015 17:23
Possibly we should fork that neurotune / pyelectro repo and do whatever new work in there
just to not create even more repos
hehe
but we can work independently of that code base if we have to
so that way the milestone is captured under there anyway
Vahid Ghayoomie
@VahidGh
May 11 2015 17:24
Yes
Stephen Larson
@slarson
May 11 2015 17:24
and we can re-integrate the actual code as necessary
Travis Jacobs
@travs
May 11 2015 17:24
Ok sounds good!
Stephen Larson
@slarson
May 11 2015 17:24
OK — let’s talk about digitizing
Vahid Ghayoomie
@VahidGh
May 11 2015 17:25
@travs, I guess it was in the pipeline issue #16 but it could be seperated
Stephen Larson
@slarson
May 11 2015 17:26
Yes you have #17
Travis Jacobs
@travs
May 11 2015 17:26
@VahidGh OK. I will add the milestone and anything that is not specific to NeuroTune can be an issue in there
Stephen Larson
@slarson
May 11 2015 17:27
@travs great
on #17 — i think to get that moving we need to write up instructions on how to use the webplotdigitizer to digitize some data
Travis Jacobs
@travs
May 11 2015 17:27
@slarson For the digitizing side, we can make another detailed diagram
Yes
Stephen Larson
@slarson
May 11 2015 17:28
Seems to me that we need to establish the following things to make that process work
We need a list of papers with figures to be digitized
This probably also requires us to have priorized which papers contain information about channels we are going to start with
Vahid has already started with some
And we have a spreadsheet of ion channels already
but we need to have a focused list of say, 5 top channels we are working on
these need to be the 5 channels with the most information possible, plus those that are the most helpful to muscle and motor neuron modelling
We need to answer the question of which those are.
Vahid Ghayoomie
@VahidGh
May 11 2015 17:30
Yes, they are tagged in Mendeley and linked in the spreadsheet
Stephen Larson
@slarson
May 11 2015 17:30
From there, we make the list of papers / figures to be digitized
ok great; we are close to the right representation for those
but we should just have the focused list that also shows which papers we have
and where the ion channels go
and that will make it obvious they are the right ones to start with for everyone at a glance
Vahid Ghayoomie
@VahidGh
May 11 2015 17:31
#8 is about ion channels in priority
Stephen Larson
@slarson
May 11 2015 17:32
Terrific
OK what that issue is missing is the list of papers
and we need to make the PDFs of those papers available
so that people can extract the figures from them
and put them in the webplot digitizer
I’m thinking a dropbox folder with them
that is taken from a focused list in mendeley
Travis Jacobs
@travs
May 11 2015 17:33
I'm working on a tentative diagram here
Stephen Larson
@slarson
May 11 2015 17:33
@travs great — make it world editable?
don’t have permission
Travis Jacobs
@travs
May 11 2015 17:33
Yes, I'll move it to the OpenWorm drive now
Stephen Larson
@slarson
May 11 2015 17:33
thanks
Travis Jacobs
@travs
May 11 2015 17:34
Stephen Larson
@slarson
May 11 2015 17:35
Looking good
we need to put the webplotdigitizer on there
and then coming out the other end we need to establish a list that has the links to the data that comes out of it
Travis Jacobs
@travs
May 11 2015 17:38
Ok, so for each image that goes in, we will get one set of data out
Stephen Larson
@slarson
May 11 2015 17:38
That’s right
There’s another bit of this for PyOpenWorm btw...
so that should probably be in there
which is that PyOpenWorm should have a structure to capture the data too
Travis Jacobs
@travs
May 11 2015 17:39
Rightt
Also the names are placeholders for now, feel free to make them more semantic haha
Stephen Larson
@slarson
May 11 2015 17:39
yep, that’s fine
ok, so this diagram that you are making should be included in the channelworm docs, with a textual description of this process
and this is a key page i think for recruiting additional volunteers to help
Travis Jacobs
@travs
May 11 2015 17:40
Yes you're right
Stephen Larson
@slarson
May 11 2015 17:40
because if there are like 15-20 figures to digitize, there should be a procedure that folks can follow
it will force us to define how to actually carry out the steps
Vahid Ghayoomie
@VahidGh
May 11 2015 17:41
Yes, good point
Stephen Larson
@slarson
May 11 2015 17:41
We should make an issue under channelworm and under the Modeling Ion Channels from Patch Clamp Experiments milestone to make that page
Vahid Ghayoomie
@VahidGh
May 11 2015 17:41
One problem I had with the GA was choosing min-max parameters
this could be a good starting point
Stephen Larson
@slarson
May 11 2015 17:42
Yeah — Achard / Deschutter could help with that
even just reproducing their results
or at least with their starting point
Vahid Ghayoomie
@VahidGh
May 11 2015 17:43
Is their approach according to HH?
Stephen Larson
@slarson
May 11 2015 17:44
Yes
OK, another issue that we need is to create a tutorial for executing the flow we imagined in the diagram here: http://docs.openworm.org/en/latest/Projects/muscle-neuron-integration/#modeling-validation
Now, the thing is right now we can’t do it all
but we should make a page that has stubs that call out what can’t be done
but that also fills in the parts that can be done right now
The idea is to make the gaps very clear for everyone
even I forget exactly what the gaps are right now sometimes
and have to remember
Travis Jacobs
@travs
May 11 2015 17:45
Right. I'll make an issue for this, I'm writing these down.
Yes that's a great idea
Stephen Larson
@slarson
May 11 2015 17:46
this tutorial should be a detailed walkthrough
Travis Jacobs
@travs
May 11 2015 17:46
Also, regarding the issue you mentioned a couple minutes ago "We should make an issue under channelworm and under the Modeling Ion Channels from Patch Clamp Experiments milestone to make that page", does #17 constitute this?
Oh wait, you mean the documentation page.. gotcha
Stephen Larson
@slarson
May 11 2015 17:46
No, but good question
right
#17 is to do it
what we need is a meta issue to describe a page about doing it
So that page doubles as a strategy document for the steps that need to done
Travis Jacobs
@travs
May 11 2015 17:47
Right
Stephen Larson
@slarson
May 11 2015 17:48
both this walkthrough of that diagram, and the one for the digitizing that you are making should be new pages under channel worm
and they will require thinking :) and testing code :) in order to create them
OK we’ve been going for a while here
I think we’ve addressed the top part of this diagram quite well today; http://docs.openworm.org/en/latest/Projects/muscle-neuron-integration/#muscle-neuron-team
Travis Jacobs
@travs
May 11 2015 17:50
Ok @slarson , the pages are not on the Unifying page? Channelworm will get its own page now is that right?
Vahid Ghayoomie
@VahidGh
May 11 2015 17:50
Yep :D @miladjafary and I will start with the digitization process
Stephen Larson
@slarson
May 11 2015 17:50
yes — they will be linked from the unifying page
Travis Jacobs
@travs
May 11 2015 17:50
Ok perfect
Stephen Larson
@slarson
May 11 2015 17:51
we want to extend it for those two topics
as they are satisfying channelworm milestones
actually we may want to quickly do the rst->md conversion there too — depending on how annoying that is :)
Travis Jacobs
@travs
May 11 2015 17:52
Well @VahidGh I will begin writing the documentation, but it will mostly be stubs until we have a defined workflow. You and @miladjafary and I can add to this as we discover the steps needed
Stephen Larson
@slarson
May 11 2015 17:52
but yeah, Vahid made a great start over there
ok there are additional milestones that we will need to define guys
one is how exactly anybody puts the data from the plots into pyopenworm
and what in pyopenworm needs to be there to make that happen
that’s started with the ‘data modelling’ stuff which is great
we just need to keep going with it
then as you look at the unifying diagram it is quite clear that arrows pointing down from that top third need specifying as well
how do we fit the channel models into muscle and neuron models exactly
And how do we sync with interactions happening on the muscle and neuron modeling
as we have other contributors who have been working there
but we will leave this for the agenda for the next conversation
we’ll keep working in parallel on getting tasks done on the low level and fleshing out the roadmap on the high level
but we should try to close off the roadmap by next week
Travis Jacobs
@travs
May 11 2015 17:56
Ok, this all sounds good to me
Stephen Larson
@slarson
May 11 2015 17:57
So, next steps (and @travs I’m relying on you here to have taken note of all this), we’ll create these issues, we’ll set up another session to do this again in about a week with the three of us at least, and we’ll get going on filling in the blanks on the issues in the mean time
we’ll keep looping in others as we go
Alright great
Travis Jacobs
@travs
May 11 2015 17:58
Yes, I've got notes here :smile:
Stephen Larson
@slarson
May 11 2015 17:58
thanks guys for the time
Travis Jacobs
@travs
May 11 2015 17:58
Awesome
Yes, this was great
So until next time!
Stephen Larson
@slarson
May 11 2015 17:59
@VahidGh thank you — let me know if there are more thoughts
Vahid Ghayoomie
@VahidGh
May 11 2015 17:59
Thank you
These are great stuff to start working on
Travis Jacobs
@travs
May 11 2015 18:04
@VahidGh could you make @gsarma a contributor for the ChannelWorm repo?
Vahid Ghayoomie
@VahidGh
May 11 2015 18:06
Done
Travis Jacobs
@travs
May 11 2015 18:23
@VahidGh Thank you!