[neuroConstruct] channel units bug in exporting a single neuroml file

P Gleeson p.gleeson at ucl.ac.uk
Thu Nov 10 18:01:38 GMT 2011


Hi Aditya,

That's fixed in the neuroConstruct SVN now. The meta:property tags are 
of course highly non standard. The dt and sim duration belong rightly in 
a SED-ML file, along with details of what to save, plot etc. I'm working 
on an option exporting valid SED-ML beside the NeuroML.

With regards to the temperature, this could be set in the SED-ML too, 
though there will probably have to be some concept of a (global) 
temperature in a NeuroML 2 file (e.g. as a property of a <network>, or 
maybe <population>, as different cells in different parts of the brain 
could in theory have different temps...).

Cheers,
Padraig



On 09/11/11 11:32, Aditya Gilra wrote:
> Thanks Padraig,
>
> I'm sticking to multiple neuroml files for now rather than the single 
> file option.
> However, there's a glitch in the multiple files option. When you 
> export to neuroml as a single file, the temperature for the simulation 
> is set as a meta:property (along with simulation time and dt). 
> However, when exporting as multiple files, the Level3 file does not 
> contain any meta:property tags for these. Could you include that?
>
> I was quite surprised by the difference that 32 deg Celsius vs 35 deg 
> Celsius makes to the action potential in the granulecell example model 
> (KDR tau is temperature dependent) [Took me a while to figure out what 
> was wrong!].
>
> Best,
> Aditya.
>
>     ------------------------------------------------------------------------
>     *From:* P Gleeson <p.gleeson at ucl.ac.uk>
>     *To:* Aditya Gilra <aditya_gilra at yahoo.com>
>     *Cc:* "neuroconstruct at ucl.ac.uk" <neuroconstruct at ucl.ac.uk>;
>     neuroml-technology at lists.sourceforge.net
>     *Sent:* Wednesday, 9 November 2011 4:18 PM
>     *Subject:* Re: [neuroConstruct] channel units bug in exporting a
>     single neuroml file
>
>     Hi Aditya,
>
>
>     On 07/11/11 12:13, Aditya Gilra wrote:
>>     Hi,
>>     Do we have a web interface for bug reports or should I mail to
>>     this list?
>     Posting to the list is probably best, no forum or bug tracker for
>     neuroConstruct at the moment... I've cc'd the NeuroML list too.
>>     Am using neuroconstruct 1.5.1.
>>     I exported a single/unified neuroml file for say the granulecell
>>     in the example models,
>>     i.e. export->neuroml-> select 'level 3' and tick 'generate single
>>     neuroml level 3 file' and then click 'generate all neuroml scripts'.
>>     Now this creates an nml file with <channels units="SI Units"> tag
>>     amongst other tags. However, some of the channels continue to
>>     have physiological units. There is no overriding units attribute
>>     in their <channel_type> tags either. An example is the Gran_KA_98
>>     channel. Of course this problem does not occur if multiple xml
>>     files for each channel, cell, network are created, since each
>>     channel has its own units.
>     Yes, it was a problem in the neuroConstruct single L3 file
>     export... I had code in place to check for a mixture of SI &
>     Physiological units in the ChannelML files but for some reason I
>     took it out.
>
>     The option of adding a flag to the <channel_type> element to
>     specify the units might be the simplest, but I really don't want
>     to go updating the v1.x specs at this stage with so much going on
>     for v2.0. If others feel strongly about it though let me know...
>     This issue won't be a problem in v2, since all quantities must
>     specify their own units (dimensions will be restricted in the
>     component class definition), e.g. v="-60mV"...
>
>     While it would be possible in theory to make a general conversion
>     utility (in Python or XSL) to change an SI v1.8.1 file into
>     Physiological & vice versa, it would need to know what each
>     quantity was, there would be no way to automate the conversion.
>     Unfortunately not something I can spend time on with lots to do
>     for v2.0.
>
>     So I'll put back in the check in nC if a user tries to export to a
>     single L3 file containing a mix of SI & Physiol ChannelML files
>     (and a warning when you press "Validate" that there is a
>     mixture...). For such projects, the options would be to export as
>     individual NeuroML files with one model component in each, or
>     manually convert the ChannelML files to the same units. This
>     latter option is painful I know, but I've just done this
>     conversion for all the standard nC example projects.
>
>     Updating the latest neuroConstruct code and example projects with
>     nCupdate.sh/nCupdate.bat will get you the fix and should be able
>     to export them all to valid single Level 3 files.
>
>     Regards,
>     Padraig
>
>
>>     Let me know when you fix this (after you've verified this is a
>>     bug.) (I suppose Padraig will be following up?)
>>     Thanks,
>>     Aditya Gilra.
>>
>>
>>     _______________________________________________
>>     neuroConstruct mailing list
>>     neuroConstruct at ucl.ac.uk  <mailto:neuroConstruct at ucl.ac.uk>
>>     http://www.mailinglists.ucl.ac.uk/mailman/listinfo/neuroconstruct
>>     Please note that the neuroConstruct mailing list is publicly archived
>
>
>     -- 
>     -----------------------------------------------------
>     Padraig Gleeson
>     Room 321, Anatomy Building
>     Department of Neuroscience, Physiology&  Pharmacology
>     University College London
>     Gower Street
>     London WC1E 6BT
>     United Kingdom
>
>     +44 207 679 3214
>     p.gleeson at ucl.ac.uk  <mailto:p.gleeson at ucl.ac.uk>
>     -----------------------------------------------------
>
>
>


-- 
-----------------------------------------------------
Padraig Gleeson
Room 321, Anatomy Building
Department of Neuroscience, Physiology&  Pharmacology
University College London
Gower Street
London WC1E 6BT
United Kingdom

+44 207 679 3214
p.gleeson at ucl.ac.uk
-----------------------------------------------------

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mailinglists.ucl.ac.uk/pipermail/neuroconstruct/attachments/20111110/068a9df0/attachment.htm>


More information about the neuroConstruct mailing list