These are chat archives for openworm/ChannelWorm

19th
Jun 2015
Travis Jacobs
@travs
Jun 19 2015 15:31
Hey @VahidGh !
Vahid Ghayoomie
@VahidGh
Jun 19 2015 15:32
Hi @travs
Travis Jacobs
@travs
Jun 19 2015 15:33
So you were saying the other day after we decided on 7 ion channels
that we could have some numbers of each subtype (K, Na)
I'll bring up the spreadsheet and see how far along we are in each of those categories
Vahid Ghayoomie
@VahidGh
Jun 19 2015 15:35
Right, we have 9 folders
Travis Jacobs
@travs
Jun 19 2015 15:35
Ok, so which two shall we leave out?
Vahid Ghayoomie
@VahidGh
Jun 19 2015 15:36
4 K channls, 2, ca-activated K channels (SLO), and 3 Ca channels
2 Ca channels which were poorly-studied :)
Travis Jacobs
@travs
Jun 19 2015 15:37
Right, such as Cav2_RNPQType_UNC-2
Vahid Ghayoomie
@VahidGh
Jun 19 2015 15:37
Yeah
Travis Jacobs
@travs
Jun 19 2015 15:37
Ok, so we can go with the others then!
There was only one paper left in the Kv1_Shaker_SHK-1 directory
Shall we go through that one together?
It looks to me like fig. 1 a, c, d may be useful
Vahid Ghayoomie
@VahidGh
Jun 19 2015 15:44
It seems today is not a good day for my bandwidth, sorry if I'm a bit slow
Travis Jacobs
@travs
Jun 19 2015 15:45
That's ok! I will continue to work through and add papers to this chart, and leave comments here if I have questions for you
But please join in if your bandwidth allows!
Vahid Ghayoomie
@VahidGh
Jun 19 2015 15:49
I'm actually in but too slow when opening papers :worried:
I guess it's better to look at them locally
Travis Jacobs
@travs
Jun 19 2015 15:50
Ah! Good idea
Vahid Ghayoomie
@VahidGh
Jun 19 2015 15:51
Ok, now it's better
Travis Jacobs
@travs
Jun 19 2015 15:52
Ok perfect
So you're looking at Santi et al. Fig 1?
Vahid Ghayoomie
@VahidGh
Jun 19 2015 15:52
yep
Travis Jacobs
@travs
Jun 19 2015 15:53
Ok, so it looks to me like a,c,d are good
maybe e, I'm not sure. what do you think Vahid?
I'm not sure if we should be including these mutants (ex: SLO-2) or just wild-type (since that's what we're modelling)
Vahid Ghayoomie
@VahidGh
Jun 19 2015 15:57
Yes, I was reading the paper
It seems it's based on two patch clamp, a whole-cell and a single channel
Travis Jacobs
@travs
Jun 19 2015 16:01
Right
So I guess we don't need whole-cell data for channelworm
I have marked them as whole-cell in the chart
Fig 5 looks like it would have been useful, but it is for mutant cells only, so I guess we won't use it
Travis Jacobs
@travs
Jun 19 2015 16:07
Ok, so it looks like that paper did not have too much info, I'll move onto the next ion channel now: Kv2_Shab_EXP-2
And Davis et al. 1999
Vahid Ghayoomie
@VahidGh
Jun 19 2015 16:09
But Fig. 1,2,3 were useful
Travis Jacobs
@travs
Jun 19 2015 16:10
Ok, I will include those. I thought they would not be useful because they are whole-cell
Vahid Ghayoomie
@VahidGh
Jun 19 2015 16:10
I was looking at the paper deeper, it was interesting, as now it's discussing about having 3 K channels in muscle cells, as opposed to two in the Boyle & Cohen model!
Travis Jacobs
@travs
Jun 19 2015 16:10
Hmm!
Vahid Ghayoomie
@VahidGh
Jun 19 2015 16:11
Whole cell patch clamps are also useful for integration purposes
Travis Jacobs
@travs
Jun 19 2015 16:12
That's true, it's good to include them for sure
I've added 2 and 3 as well
Fig 3 in Davis et al looks good
Vahid Ghayoomie
@VahidGh
Jun 19 2015 16:17
@travs, It's a little boring for both of us, with my bandwidth problem :(
Are you OK with having this meeting in a better time?
Travis Jacobs
@travs
Jun 19 2015 16:17
Absolutely!
Vahid Ghayoomie
@VahidGh
Jun 19 2015 16:18
If it was any improvement, I'll mention you here, OK?
Travis Jacobs
@travs
Jun 19 2015 16:19
Ok, sounds good.
If nothing changes, send me a PM and we will schedule another time
Chee Wai Lee
@cheelee
Jun 19 2015 16:19
Hey guys. Didn't say hi earlier ... saw you were working.
Travis Jacobs
@travs
Jun 19 2015 16:20
hey Chee Wai!
What's up?
Chee Wai Lee
@cheelee
Jun 19 2015 16:20
Was trying to get all the logistics in place so I have some hope of doing something useful during the Hackathon Sunday.
Travis Jacobs
@travs
Jun 19 2015 16:20
Haha I hear you there
Chee Wai Lee
@cheelee
Jun 19 2015 16:20
gitter is actually pretty awesome. Sad I had never worked with it before.
Travis Jacobs
@travs
Jun 19 2015 16:21
Yeah I love this app, super useful
What are you thinking about doing for the hackathon?
Chee Wai Lee
@cheelee
Jun 19 2015 16:21
got all 3 chat channels neatly arrayed for my account.
I've got no clue yet, frankly. I've got PyOpenWorm installed (the pip instructions don't work for my Mac, and I'll not worry about installation to my own preferred folder instead of /usr/local/Python2.7
trying to get the other pieces of repository code in place
and then I'll try to wrap my head around how the pieces fit together.
Great job on getting all the community material organized, btw!
Travis Jacobs
@travs
Jun 19 2015 16:24
Thanks! It's a group effort :)
Chee Wai Lee
@cheelee
Jun 19 2015 16:24
:)
Travis Jacobs
@travs
Jun 19 2015 16:24
And yeah that's a good idea. Have you installed any other repos?
Chee Wai Lee
@cheelee
Jun 19 2015 16:24
not yet
Travis Jacobs
@travs
Jun 19 2015 16:25
And the reason pip did not work for PyOW was because we have not pushed it to PyPI yet (but we will very soon!)
Chee Wai Lee
@cheelee
Jun 19 2015 16:25
As I started going, I realized I've not yet fully thought through my professional workflow setup that I wanted.
ah, cool!
Anyway, was updating my own Google docs to note how I ought to be organizing workflows on my machine.
My memory can be amazingly bad.
Travis Jacobs
@travs
Jun 19 2015 16:26
Right, I remember you mentioning those docs in the general meeting
Haha, that's why I love computers so much; they can do my remembering for me :P
Chee Wai Lee
@cheelee
Jun 19 2015 16:27
yup yup
Travis Jacobs
@travs
Jun 19 2015 16:27
If you need a hand setting up any of the repos let me know, I know some of them can be a little difficult to get started with
Chee Wai Lee
@cheelee
Jun 19 2015 16:27
it's great when you're working on just one thing
but when you start working on multiple things, things fly apart without the help of Google Docs
sure thing, I'll check in with you if you're online!
Imho, it's all a matter of trying to follow the instructions
and then noting when the instructions don't quite get it right and documenting those.
When instructions fail, the code speaks :)
Travis Jacobs
@travs
Jun 19 2015 16:30
Haha that's totally true. And if the instructions fail we'd better update those instructions as well :)
Chee Wai Lee
@cheelee
Jun 19 2015 16:30
yup yup, I'll give you guys a heads up if I run into any issues
and document each of them
so far so good
following INSTALL.md for PyOpenWorm
Travis Jacobs
@travs
Jun 19 2015 16:32
Excellent :D
Chee Wai Lee
@cheelee
Jun 19 2015 16:32
aaaannd I got my preferred folder structure sorted out, so it is time to go grab the other 4 repos
Anyway, in the back of my mind I'm always thinking about measuring performance ... so in this case the code base is mostly python, and that's where my mind wanders now.
I'm guessing parallelism is not going to be a major concern until much farther down the road.
Travis Jacobs
@travs
Jun 19 2015 16:35
Right. In terms of performance measurement, the main thing I have used is iPython's %timeit magic function
Chee Wai Lee
@cheelee
Jun 19 2015 16:36
For the most part it should be good enough now.
Travis Jacobs
@travs
Jun 19 2015 16:36
Haha yes I would say so, I don't believe there is any multithreading going on right now
Chee Wai Lee
@cheelee
Jun 19 2015 16:36
I can look at profiling tools for when things get more complex.
Travis Jacobs
@travs
Jun 19 2015 16:36
But there are actually some issues about this
There is this one in PyOpenWorm openworm/PyOpenWorm#114
Chee Wai Lee
@cheelee
Jun 19 2015 16:37
I've not yet taken a look at the codes, are we gonna be using the SciPy library support?
Travis Jacobs
@travs
Jun 19 2015 16:37
Yes, scipy is a dependency for a few things here
e.g. pyNeuroML
Chee Wai Lee
@cheelee
Jun 19 2015 16:37
Nice. I'll see if I can find an off-the-shelf python profiler that will help.
cool
Travis Jacobs
@travs
Jun 19 2015 16:38
That would be cool!
Chee Wai Lee
@cheelee
Jun 19 2015 16:38
if not, I'll take a look into writing a simple profiling tool that's gonna be useful for us.
For Mac users, it is probably as simple as turning on Instruments when the code runs.
but a more general solution (with minimal perturbation) would be helpful.
Joe Bowen
@joebowen
Jun 19 2015 16:39
You might want to check out cProfile.
Travis Jacobs
@travs
Jun 19 2015 16:39
Yeah, it would be cool to have something that was integrated with TravisCI as well
Chee Wai Lee
@cheelee
Jun 19 2015 16:39
Hi Joe! Cool, will check that out!
There are literally thousands of profilers out there lol
Travis Jacobs
@travs
Jun 19 2015 16:39
Hey Joe!
Joe Bowen
@joebowen
Jun 19 2015 16:40
hey. I've used this in the past, https://gist.github.com/fabioz/8314370
I'm working from home today, so I'm kind of "half here"
Chee Wai Lee
@cheelee
Jun 19 2015 16:41
hehe. Anyway here's a blogpost about cProfile: https://ymichael.com/2014/03/08/profiling-python-with-cprofile.html
Looks like your typical gprof output, which should be pretty decent.
I'll dig deeper and see if it supports more sophisticated callpath profiling if we need it.
Instruments will do callpath profiling, but so far I've found Instruments a little clunky to use.
I think cProfile makes for a very good "low-hanging fruit" case.
Chee Wai Lee
@cheelee
Jun 19 2015 16:50
Anyway I apologize for flooding the chat after saying "hi" lol. It's bedtime, and I got all the repos on my machine. Will try to get all the pieces fitted together in my mind tomorrow.
Travis Jacobs
@travs
Jun 19 2015 16:51
Haha, no worries Chee Wai; it's good talking to you!
Joe Bowen
@joebowen
Jun 19 2015 16:52
Have a good one!
Chee Wai Lee
@cheelee
Jun 19 2015 16:52
Good chatting with you guys! Have a great day!
Travis Jacobs
@travs
Jun 19 2015 16:52
Let's chat again tomorrow and see how this stuff is to set up :)
thanks!
Chee Wai Lee
@cheelee
Jun 19 2015 16:52
Sounds good! :)