These are chat archives for openworm/ChannelWorm

9th
Aug 2015
Travis Jacobs
@travs
Aug 09 2015 15:53
Hey @/all, the link to the hangout is in this agenda. See you inside!
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:26
Hi everyone
Chee Wai Lee
@cheelee
Aug 09 2015 16:26
Hey Vahid!
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:26
Who's where? :D
Hi @cheelee
Shreejoy Tripathy
@stripathy
Aug 09 2015 16:29
hi @VahidGh
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:30
@travs, the Django test with Pytest was great.
Working well with repo updates (especially data model changes)
Chee Wai Lee
@cheelee
Aug 09 2015 16:30
I'm gonna see if there's something I should be working with @kevcmk on testing and performance regression. If not, I'll try and add the tests for #146
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:30
Hi @stripathy
Shreejoy Tripathy
@stripathy
Aug 09 2015 16:31
is anyone working on #149?
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:33
@cheelee, great
@stripathy, Not sure if anyone has tried this yet!
Shreejoy Tripathy
@stripathy
Aug 09 2015 16:34
yeah, it seems hard and not sure how it'd be done, i'm just curious
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:35
I guess we should consider some logical constraints for this one, at the first place
Shreejoy Tripathy
@stripathy
Aug 09 2015 16:36
hmm, like that data values are in the right ranges
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:37
A bit more practical than such a logic I mean :D
Shreejoy Tripathy
@stripathy
Aug 09 2015 16:37
@VahidGh @travs is the curation interface in a version freeze?
Travis Jacobs
@travs
Aug 09 2015 16:38
@stripathy Yep!
@stripathy And as for #149, we could use Selenium and/or pyAutoGui for such a task. Shouldn't be too hard, going from what people say about Selenium. Of course the actual digitization will not be simulated, else we could just automate the whole process
But yeah, nobody is working on it right now!
Shreejoy Tripathy
@stripathy
Aug 09 2015 16:41
i see, ok, thanks!
Travis Jacobs
@travs
Aug 09 2015 16:41
@VahidGh Thanks, py.test seems really good!
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:42
@travs, I did some changes to the data model
Added protein model, and removed related fields from IonChannel model
Regarding ongoing adaptations via adapter
Travis Jacobs
@travs
Aug 09 2015 16:43
@VahidGh Ok, but these are in dev right?
the changes
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:44
Merged with the master, as needed to be uploaded to the OpenShift
Travis Jacobs
@travs
Aug 09 2015 16:50
Hmm
This defeats the purpose of a version freeze, no?
I guess we have no policy for when to review updates, so maybe we should implement that
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:52
@travs, Also I found some references claiming the value of Cm for xenopus oocytes, muscle cells, etc
Added some needed ones to the db, so could be used to resolve #96
@travs, Yeah, we need to think about cases we need to upload some new data to the OpenShift
Travis Jacobs
@travs
Aug 09 2015 16:54
I agree
But I thought the idea of version freezing was to implement these changes/additions incrementally, so that everything (ex: adapters) can move in lock-step
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:57
Yeah, maybe we can handle this by some test, e.g. Travis CI alert if the adapter is not the same as data model
Then only by resolving this, the dev branch is mergable
Travis Jacobs
@travs
Aug 09 2015 16:58
yeah, that's started in adapter_test.py.
Vahid Ghayoomie
@VahidGh
Aug 09 2015 16:59
@travs, great
So, as a broader case, we can specify a category of tests, to those that need to be in sync
Shreejoy Tripathy
@stripathy
Aug 09 2015 17:04
@VahidGh @travs - what's the process for digitizing an i/t figure where there's multiple traces in the figure?
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:06
@stripathy, I described this under #46
Travis Jacobs
@travs
Aug 09 2015 17:08
@VahidGh Does that talk about multiple traces?
Shreejoy Tripathy
@stripathy
Aug 09 2015 17:08
my example is a little better than that one, thankfully
i think it's digitizable - but you have to know the voltage step each trace corresponds to
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:09
Yes, We need the first, the last, and the trace for initial holding potential
Shreejoy Tripathy
@stripathy
Aug 09 2015 17:09
hmm
i think for now i'll just save a bunch of i/t figures and let someone else deal with digitizing them
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:10
But in this case the first and the last one is sufficient (OF COURSE, IF there is an I/V, G/Gmax/V, etc is also available)
@stripathy, I was going to reply for the curation test case, to be considered as a transaction :D
Shreejoy Tripathy
@stripathy
Aug 09 2015 17:12
?
@VahidGh what do you mean?
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:14
So, the process must start from data and end with a model
The experiment I had with some channels these days, was forcing me to do the whole process for every channel
About every experiment has it's own uniqueness
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:19
For example there is an I/V curve in the experiment with saying -1 pA of current for holding potential of -40 mV
But this is not exactly happening with the I/t curve, so we need some manual adjustments...
Shreejoy Tripathy
@stripathy
Aug 09 2015 17:20
ok i'll curate what i can and let you deal with the inconsistencies
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:21
@stripathy, That's fine, thanks.
@stripathy, I think your best help would be with patch clamp data curation
Chee Wai Lee
@cheelee
Aug 09 2015 17:25
@travs Is there an example in ChannelWorm which accesses the django database I can refer to?
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:28
@stripathy, I was just saying regarding your experience with Neuroelectro
Shreejoy Tripathy
@stripathy
Aug 09 2015 17:29
right, i agree with you
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:29
That's a really cool project, congrats!
And similar queries for other database object will look similar, just with different model types than PatchClamp
Note that if you are running django from the command line you will need to do ./manage.py shell to get an interactive shell
Chee Wai Lee
@cheelee
Aug 09 2015 17:35
@travs gotcha
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:35
@cheelee, Also take a look at this one
Chee Wai Lee
@cheelee
Aug 09 2015 17:35
@VahidGh cool, looking at it!
Vahid Ghayoomie
@VahidGh
Aug 09 2015 17:36
You can either use the djago shell as @travs mentioned or by the configuration I did in the above-mentioned script
Chee Wai Lee
@cheelee
Aug 09 2015 17:36
@VahidGh @travs For #148 my plan is to have a test replicate the database with pre-loaded test data so I can test delete operations.
Travis Jacobs
@travs
Aug 09 2015 17:38
That sounds reasonable!
Also note that when running tests with django, a temp db is created,so any creation/destruction you do is isolated if you're using the default testing db
Chee Wai Lee
@cheelee
Aug 09 2015 17:40
ah ok. For deletes, that wouldn't work, as a new temp db would be dependent on a successful insertion test.
Travis Jacobs
@travs
Aug 09 2015 17:42
Good point!
Travis Jacobs
@travs
Aug 09 2015 17:53
Ok everyone, moving back into the hangout shortly!
Stephen Larson
@slarson
Aug 09 2015 17:56
@kevcmk on this key you are talking about right now; the default travis-ci setup has just involved clicking a button to set up the webhook, and not explicitly dealing with a key
@kevcmk can you say more about what you're trying to do with it? I can dig in & look for it
Travis Jacobs
@travs
Aug 09 2015 17:58
@stripathy You still around?
Shreejoy Tripathy
@stripathy
Aug 09 2015 17:58
yes