Script for generating images/web views of model ancestry

[pgleeson]

Brief description

Models in computational neuroscience (particularly detailed models with ion channels & multicompartmental cells) are generally based on previous work, with minor or major alterations to the code. Recording the history/relationship of these can be useful for finding out which contributor added which parameters/features and why.

This task aims to make a standard way to record these relationships in structured files and generate graphical representations to elucidate the relationships.

Information

Work on this has started here: https://github.com/OpenSourceBrain/CommunityModellingCA1. One of the aims of this project is to record the various cell/network models of hippocampal region CA1. Manual images of the relationships between modles in this area have been generated by @andrisecker, e.g.

However, it would be better if such diagrams could be generated automatically from structured text descriptions, see here. This would encourage updates to the ancestry description; a researcher would only have to add a line to the text file and the image can be updated.

A script has been started to generate an image like the above from this, see OpenSourceBrain/CommunityModellingCA1#4. This needs to be improved and other output formats supported:

• update python script for generating (nicely formatted) png
• generate interactive/clickable/expandable web page from this using javascript (maybe https://d3js.org/)

Once the format for these csv files is stable and the images/views can be generated other model ancestry files can be requested from the community (cerebellum, cortex, basal gangli, olfactory cortex, etc.)

Knowledge prerequisites

Python
JSON
JavaScript (optional)
Computational neuroscience knowledge useful, but not required.

Contact

Contact p.gleeson -at- ucl.ac.uk if you are interested in contributing to this task, or fork https://github.com/OpenSourceBrain/CommunityModellingCA1 and start working.

[borismarin]

Might be relevant https://icg.neurotheory.ox.ac.uk/

On 29 June 2016 at 13:29, Padraig Gleeson [email protected] wrote:

Brief description

Models in computational neuroscience (particularly detailed models with
ion channels & multicompartmental cells) are generally based on previous
work, with minor or major alterations to the code. Recording the
history/relationship of these can be useful for finding out which
contributor added which parameters/features and why.

This task aims to make a standard way to record these relationships in
structured files and generate graphical representations to elucidate the
relationships.
Information

Work on this has started here:
https://github.com/OpenSourceBrain/CommunityModellingCA1. One of the aims
of this project is to record the various cell/network models of hippocampal
region CA1. Manual images of the relationships between modles in this area
have been generated by @andrisecker https://github.com/andrisecker,
e.g.

https://raw.githubusercontent.com/OpenSourceBrain/CommunityModellingCA1/master/ca1_interneurons.jpg

However, it would be better if such diagrams could be generated
automatically from structured text descriptions, see here
https://github.com/OpenSourceBrain/CommunityModellingCA1/blob/master/CA1Models.csv.
This would encourage updates to the ancestry description; a researcher
would only have to add a line to the text file and the image can be updated.

A script has been started to generate an image like the above from this,
see OpenSourceBrain/CommunityModellingCA1#4
OpenSourceBrain/CommunityModellingCA1#4. This
needs to be improved and other output formats supported:

• update python script for generating (nicely formatted) png
• generate interactive/clickable/expandable web page from this using
  javascript (maybe https://d3js.org/)

Once the format for these csv files is stable and the images/views can be
generated other model ancestry files can be requested from the community
(cerebellum, cortex, basal gangli, olfactory cortex, etc.)
Knowledge prerequisites

Python
JSON
JavaScript (optional)
Computational neuroscience knowledge useful, but not required.
Contact

Contact p.gleeson -at- ucl.ac.uk if you are interested in contributing to
this task, or fork
https://github.com/OpenSourceBrain/CommunityModellingCA1 and start
working.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#11, or mute the
thread
https://github.com/notifications/unsubscribe/ADSvb8DHaOqyUI-Bs_rYXeGOAj3f1w2_ks5qQmUMgaJpZM4JBE-6
.

[BiaDarkia]

I don't have enough javascript knowledge to generate an interactive website from the python script, but I have quite a bit of experience with python and would be happy to have a go on improving the python script.

[pgleeson]

Thanks @BiaDarkia. The CSV file should be fairly stable, so have a look at what the options are for generating GraphViz, or there may be other Python packages you could use. NetworkX might be an option, and it seems to be possible to integrate that with GraphViz and JavaScript.

[BiaDarkia]

@pgleeson I will submit a pull request shortly. There is still some work to be done, but the PNG file(s) that are generated by Parse.py already look much nicer. To give a short summary, I changed the following:

• As in the JPG files models are now distinguished based on cell types rather than year of publication
• As for the JPG files Parse.py now generates two .gv and two .png files, one for pyramidal cell models and one for inhibitory interneuron models
• PNG files are now automatically generated when running Parse.py
• I added more space in between nodes and changed the appearance of nodes to get nicer looking PNG files as output