[BBPBGLIB-1137] Account for nonzero total current when using SEClamp; allow nonzero total current when simulating physical electrodes

[joseph-tharayil]

Context

NEURON reports SEClamp currents at different time points than all other currents, thus giving inaccurate results in lfp calculations, in cases where the SEClamp represents synaptic inputs.

This can be fixed by replacing the SEClamp with a new point process mechanism that is identical, except for producing a NONSPECIFIC_CURRENT instead of an ELECTRODE_CURRENT.

However, in cases where SEClamp or IClamp are used to represent physical electrodes, rather than synaptic inputs, the currents should not be included in the LFP calcualtion

Scope

When a current_source or conductance_source is used, we check if the alternative mod file, which is expected to be named 'MembraneCurrentSource.mod' or 'ConductanceSource.mod' is available. If so, it is used, with nothing else changed, unless the key-value pair "represents_physical_electrode" is set to True in the Stimulus Config file. Otherwise the IClamp or SEClamp is used as before.

Testing

Review

• PR description is complete
• Coding style (imports, function length, New functions, classes or files) are good
• Unit/Scientific test added
• Updated Readme, in-code, developer documentation

[joseph-tharayil]

One suggestion we recently got was to add an explicit test that, for a small network, the behavior is identical using the old and the new mod files. I'm not really sure how to implement this. Does anyone have any suggestions?

Maybe once the new parameter is deployed in libsonata, a scientific test could be added to run a small simulation twice with 'represents_physical_electrode' = true and false in order to check the report (and see that the first with i_membrane+IClamp and the second with only i_membrane gives the same results)

I've created the tests, and they run successfully with pytest, but when htey are run through the CI pipeline they cause a segfault. Any idea why this is?

[jorblancoa]

One suggestion we recently got was to add an explicit test that, for a small network, the behavior is identical using the old and the new mod files. I'm not really sure how to implement this. Does anyone have any suggestions?

Maybe once the new parameter is deployed in libsonata, a scientific test could be added to run a small simulation twice with 'represents_physical_electrode' = true and false in order to check the report (and see that the first with i_membrane+IClamp and the second with only i_membrane gives the same results)

I've created the tests, and they run successfully with pytest, but when htey are run through the CI pipeline they cause a segfault. Any idea why this is?

To be honest i didnt check in detail, but in any case it wont work properly until the change in represents_physical_electrode is added to libsonata and there is a new release, so i would wait for that before investigating more.

[bbpbuildbot]

Logfiles from GitLab pipeline #210151 (:white_check_mark:) have been uploaded here!

Status and direct links:

• ✅ docs
• ✅ test-unit
• ✅ test-integration
• ✅ test-integration-e2e
• ✅ test-scientific
• ✅ test-scientific-ngv
• ✅ set_alt_branches
• ✅ build-package
• ⚙️ publish-package
• ⚙️ publish-docs