generatedNeuroML2 directory does not include synapses

[pgleeson]

@clbarnes The NeuroML 2 model here (as well as the neuroConstruct project which generated it) have been pretty much superseded by the networks generated by c302: https://github.com/openworm/c302.

The Readme there should give an overview of the current status and links to examples, including how to run the models and visualise them on OSB.

Let me know if anything is unclear (or there is some documentation that led you here that you think should be updated).

[clbarnes]

I was mainly hoping to find one data source which had both morphologies and connectivity (down to synapse location), ideally staying in one serialisation format. I'm not so interested in running models or anything. It's certainly possible I'm missing something in the c302 repo but I can't see any data source for the locations of the synapses.

[pgleeson]

There is this type of information in the c302 examples, although only the examples with D or D1 feature multicompartmental cells (A, B, C are point neurons), for example:

https://github.com/openworm/c302/blob/master/examples/c302_D_Syns.net.nml

• imports this and other multicomp. cells: https://github.com/openworm/c302/blob/master/examples/cells/AIZL.cell.nml
• specifies the synapse properties, e.g. expTwoSynapse, gapJunction
• specifies the list of explicit connections, in projection -> connectionWD

Note though that all connections are soma to soma in the network, i.e. no preSegmentId/postFractionAlong attributes as used here: https://github.com/NeuroML/NeuroML2/blob/master/examples/NML2_MultiCompCellNetwork.nml#L95.

This is down to the fact that we're just not at the stage that the multicomp cells for the network are tuned to do anything sensible yet, let alone respond correct input at the right point on the dendrites etc. so somatic input for all is just as good as anything else...

We did have a script to generate random/close connections based on proximity: https://github.com/openworm/CElegansNeuroML/blob/master/CElegans/pythonScripts/RegenerateConnectome.py (which does produce a valid NeuroML 2 file with segment locations), but this would just add another confusing factor at the moment. All work for the future.

Hope this helps a bit...