These are chat archives for openworm/ChannelWorm

23rd
Jun 2015
Travis Jacobs
@travs
Jun 23 2015 16:18
@VahidGh @stripathy
Hi guys, I'm a few minutes early but I'll just dump a couple agenda points here for us to start on.
Feel free to add to it as well.
Agenda:
  • Shreejoy's two comments on #8
  • Should data model be written first, or should we filter data afterward?
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:20
@travs are we meeting here in the gitter chat or on google hangouts?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:22
@travs, @stripathy, Hi. I did some improvements to the data model.
Travis Jacobs
@travs
Jun 23 2015 16:22
Historically there have been connection issues with Vahid's ISP and google hangouts such that it is difficult to carry on a conversation, so we adopted Gitter. Sorry if I forgot to mention this earlier
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:22
Tried to cover everything we need
Travis Jacobs
@travs
Jun 23 2015 16:22
@VahidGh awesome! Where?
Main branch?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:23
@stripathy, Yeah, sorry, Google hangout, Youtube, G+, etc are banned in Iran :)
We have to connect via proxies and VPNs with much lower speed
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:24
oh my gosh that's crazy
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:24
@travs, yeah the main branch, it's working online now
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:25
can i get a link to the models.py file?
Travis Jacobs
@travs
Jun 23 2015 16:26
@VahidGh Ok checking it out now. Looks good! So I'll adopt these changes when I define the schema in PyOW
@stripathy models.py
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:27
ok cool
Travis Jacobs
@travs
Jun 23 2015 16:27
So yeah, the second of those points I listed before the meeting is something we could discuss
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:27
yeah, a lot of this looks good
@travs how should we organize this?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:28
@travs, yes, the only thing I guess is left, is the capability to save figures with multiple ion channels. We need to include this somewhere in the web app, would be useful
Travis Jacobs
@travs
Jun 23 2015 16:28
@VahidGh sweet! That would be useful for sure
Travis Jacobs
@travs
Jun 23 2015 16:29
@stripathy in terms of organization, the idea is that I will port these attributes over to PyOpenWorm where we will define the schema, but we should all agree on what defines the schema first.
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:30
These two links now could be used to start information gathering and digitization
Travis Jacobs
@travs
Jun 23 2015 16:30
In other words, what parameters do we need for an ion channel model to be run with the current simulation infrastructure we have @VahidGh do you think the list of parameters in models.py is comprehensive right now?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:31
@travs, yeah, I think so
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:31
my experience is that these lists of parameters will never be comprehensive
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:31
If anything else is needed I can add any time
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:32
right - i think extensibility is important
Travis Jacobs
@travs
Jun 23 2015 16:32
@stripathy Ok, and you mentioned that schema migration is a problem later on right?
Ah ok
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:32
schema migration isn't a problem, it's just a headache
Travis Jacobs
@travs
Jun 23 2015 16:32
Ok, so the suggestion there is to make the schema as comprehensive as possible at the start then?
Even if some fields have to be nulled?
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:33
hm, it's a tradeoff between comprehensiveness vs ease of adding data using the UI
Travis Jacobs
@travs
Jun 23 2015 16:34
@stripathy Good point
We could "gray out" some fields (or make them invisible in the UI) if they are not yet implemented in the data model
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:35
Yes, we need also to work more on the UI
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:35
basically, you need to decide which are the parameters you think you'll absolutely need later for analysis purposes
vs parameters that may be nice to have
are both of you kind of familiar with neuroelectro?
it's similar in that we're trying to extract from the literature electrical data about neurons
Travis Jacobs
@travs
Jun 23 2015 16:36
The basic premise - yes
Right
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:36
One thing I didn't include was some parameter values publishing as the result of channel kinetics analysis
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:37
we started just like you guys 1) first focused on getting out the electrical data; 2) then tried to get a limited set of experimental conditions; 3) get more experimental conditions (ongoing)
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:37
such as V half, etc
these are usually reproducible via figures
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:37
right, I think that's right @VahidGh
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:37
So I decided to ignore those one
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:37
if you're focusing on mining figures, I don't think you need Vhalf
my suggestion to you guys is to define the kinds of figures/data that you're most hoping to find
i think you'll have enough problems getting data from just traditional I/V curves that it's ok to focus on just taht
Travis Jacobs
@travs
Jun 23 2015 16:38
@stripathy good point
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:39
Here I've defined what type of plots we will capture
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:39
oh, another question is: "who is doing the curation?"
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:40
@stripathy, yes, the main focus is on VClamp and/or IClamp, and I/V
@stripathy, Not yet decided
Travis Jacobs
@travs
Jun 23 2015 16:40
@VahidGh Our enterprise here is really to have something to optimize against, no? So maybe if we start with those three plot types only, get a path developed for the optimization using those and a few models, then iterate on top of that later with more plot types
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:41
the simpler you make the curation task, the easier it'll be to train people how to do it
alternatively, if reasonably the only people who will curate is us 3, then it doesn't need to be spelled out that well
hopefully i'm making my point that the curation UI should be driven by who is the expected user
@travs - I agree with you, i think it's a good idea to start simple and focus on just I/V curves
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:42
@travs, But I was thinking about getting data while looking at an article as much as possible (and of course useful) and then decide on how to use during optimization
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:43
but also have a clear path in mind for how to extend to other kinds of data
@VahidGh you're talking about the problem of re-curation. my experience is that re-curating the same articles is almost always guarenteed because in the early stages its really impossible to know what you're looking for
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:44
@stripathy, One problem with just I/V curves is the inactivation ignorance
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:44
@VahidGh can you spell that you?
that out*
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:44
So, I was thinking about having at least one patch clamp figure
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:45
patch clamp = I/T; V/T ?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:45
yes
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:45
that's not a bad idea
Travis Jacobs
@travs
Jun 23 2015 16:45
@VahidGh I can see your point about going deeply into an article since there is necessarily more information, and this will certainly eventually happen. Maybe my concern is just that we will spend too much time on each paper and defining data models for all sorts of plots
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:46
@travs precisely
collect data today for the immediate problems of today
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:46
I/V curves are usually steady state currents so does not consider inactivation as it's time course is usually much slower that activation gates
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:47
I completely agree
Travis Jacobs
@travs
Jun 23 2015 16:47
@VahidGh I agree that we should go with I/V and one of I/T or V/T because of inactivation ignorance. But that is probably enough for now. When we go through these papers we can always come back to them later and look for more esoteric plots (G, G/Gmax, etc)
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:48
@travs, but it's not so hard to capture also those info while reading the paper :)
Travis Jacobs
@travs
Jun 23 2015 16:48
@VahidGh I agree. I think a reasonable approach is to skip those plots for now (or note them but not go any deeper)
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:48
The UI is now capable of that
Travis Jacobs
@travs
Jun 23 2015 16:48
Ok, right, we can note them, but maybe add them later?
I just don't want to get bogged down in so many plots haha :)
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:49
right
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:49
Ok, we can start with at least one I/t, V/t + I/V
Travis Jacobs
@travs
Jun 23 2015 16:49
@VahidGh Agreed
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:49
my experience is that the simpler and quicker the curation task, the happier the curators will be
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:50
@stripathy, right :)
Travis Jacobs
@travs
Jun 23 2015 16:50
@stripathy yes!
And since the curators are us right now, that's just more motivation :)
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:50
we had success trying to curate separate kinds of info separately
hahaha, right!
for example, get out all the I/V plots from a few papers
then go back to those papers and try getting out the experimental conditions
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:51
@stripathy, How you were digitizing V- or I- clamp figures?
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:51
i guess that's just a workflow thing though
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:51
trace by trace or whole the figure in one time
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:51
@VahidGh - we weren't. our focus was digitizing html data tables and then annotating those: http://www.neuroelectro.org/data_table/2820/
@VahidGh one of the main reasons i'm interested in working with you guys is adapting your figure extraction stuff to the kinds of data I care about for neuroelectro
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:53
cool
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:54
ok. so.
does github allow me to make comments on specific lines of code?
Travis Jacobs
@travs
Jun 23 2015 16:54
Yes, so we agree on the plots we should be getting from the literature
@stripathy yes, there should be a "plus sign" on the line when you hover
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:55
ok
Travis Jacobs
@travs
Jun 23 2015 16:55
Hmm, I'm not seeing it right now actually, let me take a look
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:56
thanks
@VahidGh - what do you think about splitting out the experimental conditions into their own class from PatchClamp?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:57
@stripathy, Yes, I have this in my TODOs
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:57
i think that'll help with simplicity
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:57
I did this today to have the big picture
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:57
ok cool
Travis Jacobs
@travs
Jun 23 2015 16:57
@stripathy unfortunately the only way to do this is when you view the actual commit (ex: VahidGh/ChannelWorm@b05be9e)
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:57
ok i see
Travis Jacobs
@travs
Jun 23 2015 16:58
but if it is in regard to an issue you can quote the line in the issue by clicking the line number; not really the same though
Shreejoy Tripathy
@stripathy
Jun 23 2015 16:58
@VahidGh 2 other comments - keep in mind that there's going to be an intracellular molecule concentration and an extracellular
meaning that there's ions like K+ and Ca2+ both inside and outside of the cell
and you have to use the Nernst equation to figure out the balance of their effect
@VahidGh also, i think the MOST important experimental condition is temperature, and the rest can probably be filled in later
Vahid Ghayoomie
@VahidGh
Jun 23 2015 16:59
@stripathy, Yes, I can access to them considering the time. I though it would be enough to have the initial concentration
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:00
initial = ? extracelllar or intra?
Travis Jacobs
@travs
Jun 23 2015 17:00
@stripathy If you are commenting on some of the fields we could track this in openworm/PyOpenWorm#154, since that is where we decided to define the schema
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:00
(part of the reason why I'm asking is so it's clear what the curator is supposed to look for)
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:00
It depends on the channel type as we just considering one channel in a patch clamp experiment
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:01
what do you mean?
i'm saying that there's a molecular concentration that the cell sees; 1) outside the cell - extracellular; 2) inside the cell - intracellular, based on the solution inside the patch pipette
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:02
I mean if we are simulating a patch clamp experiment with Cav1 then the initial concentration is the extracellular Ca
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:02
ok cool - so you mean extracellular
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:02
Yes, for Ca channels
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:02
ohh, right, there's basically 0 intracellular Ca2+
i also had the question about cell types - which experiments are more ideal
when someone inserts a c elegans channel into a xenopus oocyte?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:04
A more comprehensive set of values is then accessible in the higher level simulations using NeuroML when integrating different models in a cell
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:05
my qustion isn't about simulation, it's about representing the data from each paper's methods section
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:06
As I commented under #8, we will keep storing any cell type info and will filter them later
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:07
...my preference is to outline what kinds of cells/experiments are most ideal and then focus on those in the first curation pass
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:09
Data for C. elegans patch clamp experiments is usually less than other mammalian models, so we have to be more stingy on any type of data :D
But we will have priorities of course
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:10
sure, you have a better sense of this than me
my sense is that the "simplest to use data" will be those where someone expressed a c elegans ion channel into a naive cell, like a xenopus oocyte
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:12
@stripathy, Do we have any model out of NeuroElectro right now?
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:12
?
meaning?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:12
I mean a kinetic model in saying NeuroML format
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:13
yeah, i think so, i need to double check with @rgerkin
people have taken the data from neuroelectro and used it to help constrain their model parameters, whether neuroml or not
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:14
@stripathy, Great, so would be useful for us too
Travis Jacobs
@travs
Jun 23 2015 17:15
OK, @stripathy how have you organized your data when manually mining the literature? Spreadsheets as well?
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:16
we have a few stages, the first stage is exploratory where we're like "let's see if the literature contains this kind of data" and we see how people represent certain things
Travis Jacobs
@travs
Jun 23 2015 17:16
I feel like what we have here is probably not structured enough, given what we've decided on today
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:17
agreed
i think we're ready to move to the UI tool
and have a place to make comments on UI changes/improvements/proposed schema changes
maybe... a curat-athon ?
Travis Jacobs
@travs
Jun 23 2015 17:18
Hey! There's an idea! :)
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:18
:D
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:20
cool
Travis Jacobs
@travs
Jun 23 2015 17:20
Yeah, I think the comments can be rendered as github issues on the channelworm repo. We may organize them into their own milestone if we end up having more than a small number
As for the UI tool, @VahidGh do you think we are ready for that? The data is not yet exportable to PyOpenWorm but as long as we can represent it in the channelworm db we can always push it out later
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:22
we (as curators) should appreciate that we'll likely be curating the same content a few times while the DB / curation task stabilizes
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:22
@travs, yes, I think it's ready now. And I have a scheduled job to backup the DB automatically in addition to manually
Travis Jacobs
@travs
Jun 23 2015 17:23
@stripathy true, I'll keep that in mind as we do this
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:24
brb
Travis Jacobs
@travs
Jun 23 2015 17:24
Alright, well why don't we try doing this right now? At least to do one plto from the literature we've identified in the sheet
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:24
Ok, we can start
Travis Jacobs
@travs
Jun 23 2015 17:25
@VahidGh do you know of any screen broadcasting software we can use for a future longer session? Is skype banned?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:27
@travs, Yes, Skype is also banned :( let me see if I can find an alternative to hangout
Travis Jacobs
@travs
Jun 23 2015 17:27
There are also these browser alternatives 1 2 we can try. but this may not be necessary today
@vahid check those two out, they seem cool and web 2.0 :thumbsup:
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:28
those look cool
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:29
Great, they are not banned
Travis Jacobs
@travs
Jun 23 2015 17:29
Sweeeeet
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:29
let's try one? i'd prefer to screenshare this
er... prefer @travs to screenshare
Travis Jacobs
@travs
Jun 23 2015 17:29
Haha, yep give me 2 minutes
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:30
@VahidGh - what i was saying earlier about cell types - i think the kind of cell/experiment is going to dictate how you can use the data
i was noticing that when people were culturing c-elegans cells, they were studying the effect of a single channel by knocking it out
so it's like a "subtractive" approach
vs. adding just that cell to a xenopus oocyte, which would be an "additive" approach
Travis Jacobs
@travs
Jun 23 2015 17:32
I like the auto-generating name
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:32
@stripathy, agreed. Once I was thinking about saving info about mutants regarding the subtractive one
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:33
yeah, i think you need to
@travs where are you?
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:39
@travs @VahidGh btw this pubmed query looks like a good place to look for channels: xenopus oocyte electrophysiology c elegans : http://www.ncbi.nlm.nih.gov/pubmed/?term=xenopus+oocyte+electrophysiology+c+elegans
Vahid Ghayoomie
@VahidGh
Jun 23 2015 17:42
@stripathy, yes, actually lots of papers in the folder using Xenopous oocyte as medium expression
Travis Jacobs
@travs
Jun 23 2015 17:42
Over here instead @VahidGh https://room.co/#/ojjt2jzgcuhj8aor
Travis Jacobs
@travs
Jun 23 2015 17:56
  1. Opened a paper like this one and got the DOI
  1. entered it in the channelworm "new experiment" section
  1. Channelworm -> create new patch clamp
  1. fill out meta-data
Shreejoy Tripathy
@stripathy
Jun 23 2015 17:58
When filling out meta-data: guidelines can help the curator identify when something is "whole-cell" or "voltage-clamp"
Help organize stuff on the patch-clamp sheet
Travis Jacobs
@travs
Jun 23 2015 18:00
And the contributor section is required but I can't select one :(
Shreejoy Tripathy
@stripathy
Jun 23 2015 18:00
try to separate out details from the patch clamp experiemnt (voltage step amplitude is required), from experimental conditions which are not always required (age, solution concentration)
Vahid Ghayoomie
@VahidGh
Jun 23 2015 18:00
yeah, I have to fix this
Travis Jacobs
@travs
Jun 23 2015 18:00
For figures:
  1. add graph
  2. fill meta-data
  3. cut graph from pdf
  4. upload
Also, the text field char limit is too short for captions
Vahid Ghayoomie
@VahidGh
Jun 23 2015 18:03
We don't need all the caption
just some words about the figure
Shreejoy Tripathy
@stripathy
Jun 23 2015 18:04
the user interface needs to make it clear what the curator is supposed to add
Vahid Ghayoomie
@VahidGh
Jun 23 2015 18:05
Yes, we have to work more on ux
@travs, the hardest figure you chose :)
normalized current!
Shreejoy Tripathy
@stripathy
Jun 23 2015 18:06
alright
Vahid Ghayoomie
@VahidGh
Jun 23 2015 18:07
I first should find a better room
:)
Shreejoy Tripathy
@stripathy
Jun 23 2015 18:07
close some browser tabs
and we can try again later
@VahidGh find a better room
Vahid Ghayoomie
@VahidGh
Jun 23 2015 18:07
My apologize again
Shreejoy Tripathy
@stripathy
Jun 23 2015 18:07
i'm going to go - nice working with you both!!
I think we made a ton of progress! yay!!
Vahid Ghayoomie
@VahidGh
Jun 23 2015 18:07
Ok, thanks @stripathy
@travs , the first room was good
let us try again one more time
Travis Jacobs
@travs
Jun 23 2015 18:08
@stripathy thanks again, we'll schedule another one
@VahidGh I have to jump to another meeting right now, but let's talk about this again shortly
@VahidGh are you around in an hour?
or is it too late ??
Vahid Ghayoomie
@VahidGh
Jun 23 2015 18:09
@travs, no problem
Travis Jacobs
@travs
Jun 23 2015 19:03
still around @VahidGh ?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 19:04
@travs, yep
Travis Jacobs
@travs
Jun 23 2015 19:05
Ok
so it looks like we have a couple of changes to make to the UI; I will raise some issues for them and we'll see what we can do
Vahid Ghayoomie
@VahidGh
Jun 23 2015 19:06
Can we try another room? It seems like my government had no problem with the first room you linked, but had no screen sharing capability
And the second one was the same as Hangout for me
Travis Jacobs
@travs
Jun 23 2015 19:07
Ok, the first one was https://appear.in/
You will share your screen? Or me?
Vahid Ghayoomie
@VahidGh
Jun 23 2015 19:07
I should find one similar to appear.in with screen sharing feature
Travis Jacobs
@travs
Jun 23 2015 19:07
Right. Can you see if this one works? https://vline.com/
Because appear.in works if you create the room remember
so if you want to share you can make a room
let's try this
Travis Jacobs
@travs
Jun 23 2015 19:09
ok
Is this the correct link?!
Travis Jacobs
@travs
Jun 23 2015 19:12
haha let us see