Add SONATA point-process cell support by ilkilic · Pull Request #51 · openbraininstitute/BlueCelluLab

ilkilic · 2026-01-09T15:17:01Z

This PR adds support for SONATA point-process cells.

Point-process cells are instantiated from SONATA model_type = "point_process" and model_template, mapping directly to NEURON point mechanisms defined in HOC/MOD files.

What is included

New HocPointProcessCell for SONATA point-process populations
Mechanism instantiation driven by model_template (e.g. nrn:IntFire1, hoc:)
Integration with CircuitSimulation (instantiate_gids, callbacks, deletion)
Spike detection and recording compatible with existing simulation flow
Support for SONATA inputs and SynapseReplay via _add_stimuli
Compatibility with SONATA edge loading through existing SonataCircuitAccess

codecov · 2026-01-09T15:25:56Z

Codecov Report

❌ Patch coverage is 94.73684% with 17 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
bluecellulab/cell/point_process.py	93.05%	10 Missing ⚠️
bluecellulab/circuit_simulation.py	80.95%	4 Missing ⚠️
tests/test_allen_v1/test_ringcells_allen_v1.py	96.77%	2 Missing ⚠️
bluecellulab/point/point_connection.py	96.77%	1 Missing ⚠️

Files with missing lines	Coverage Δ
bluecellulab/cell/core.py	`78.92% <ø> (ø)`
...ab/circuit/circuit_access/sonata_circuit_access.py	`95.92% <100.00%> (+0.11%)`	⬆️
bluecellulab/circuit/synapse_properties.py	`69.23% <100.00%> (+1.23%)`	⬆️
bluecellulab/point/connection_params.py	`100.00% <100.00%> (ø)`
bluecellulab/reports/utils.py	`94.76% <100.00%> (+0.11%)`	⬆️
bluecellulab/synapse/__init__.py	`100.00% <100.00%> (ø)`
bluecellulab/synapse/synapse_factory.py	`91.30% <100.00%> (+0.49%)`	⬆️
bluecellulab/synapse/synapse_types.py	`85.43% <100.00%> (+2.40%)`	⬆️
bluecellulab/point/point_connection.py	`96.77% <96.77%> (ø)`
tests/test_allen_v1/test_ringcells_allen_v1.py	`96.77% <96.77%> (ø)`
... and 2 more

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

The logic for loading from sonata files is to swap out the list of parameters if the BBP TYPE is not present Likewise, when instantiating a synapse object, if no TYPE field is present, assume it is for allen and create a Exp2Syn synapse

…e instantiated. The reporting adds try/except since these don't have a soma

… target point neurons and deliver replay spikes to the proper netcon

…rties

Removed call to finalize for point process type cells since this is not needed

Updated allen tests to collection cell_info and get ExpSyn data as consequence

…re netcon code Remove more unused code sections; can be added when use case example is available

Added additional test for invalid Mechanism given to HocPointProcessCell

darshanmandge

Thank you, @jamesgking, for adding this to bluecellulab! I have left some comments.

Refactor to conform better to existing BlueCellulab structures

darshanmandge

Thanks a lot, @jamesgking!

AurelienJaquier · 2026-06-30T09:45:50Z

+        self,
+        cell_id: Optional[CellId],
+        mechanism_name: str,
+        spike_threshold: float = 1.0,


That is a strange default value for spike threshold. Maybe we could use -30.0, as is done in https://github.com/openbraininstitute/BlueCelluLab/blob/main/bluecellulab/cell/core.py#L526

For point-process cells, spikes are already emitted as events so the threshold is only kept for API compatibility. 0.0 feels like a better neutral default than 1.0 or -30.0.

Ok, I didn't know that it was different for point process cells. Sounds good then!

AurelienJaquier · 2026-06-30T09:51:02Z

+            duration = getattr(stimulus, "duration", np.inf)
+
+            spikes_of_interest = spikes_of_interest[
+                (spikes_of_interest >= delay) & (spikes_of_interest <= duration)


Shouldn't keep spikes between delay and duration + delay? Instead of delay and duration?

Good catch. This logic exists in core Cell.add_synapse_replay() too but I think it is wrong there as well: duration should be interpreted as a length, so the stop time is delay + duration. I will fix both core.py and point_process.py for consistency.

If you haven't started, I can do this now. Or maybe you want the change in Cell as a separate commit to make the change easier to track in git history?

Thank you @jamesgking for fixing it!

AurelienJaquier · 2026-06-30T09:52:49Z

+            vs = h.VecStim()
+            vs.play(vec)
+
+            if self.pointcell is None:


We could put this check before the for loop imo

AurelienJaquier · 2026-06-30T09:57:27Z

+    delay: float     # ms
+    weight: float    # NetCon weight
+
+    # isec: int = -1


Why is this commented out?
This should be removed maybe?

AurelienJaquier

looks good to me!

wip

626e54f

jamesgking self-assigned this Mar 9, 2026

jamesgking added 17 commits March 11, 2026 15:17

Merge branch 'main' into point-process

bd23ea4

Merge remote-tracking branch 'origin/main' into point-process

17195f0

Minor fixes to pass lint checks

988664b

Merge remote-tracking branch 'origin/main' into point-process

c9f0db8

Intermediate steps to allow point process cells (wrapped in hoc) to b…

e3b12e9

…e instantiated. The reporting adds try/except since these don't have a soma

Corrections to load synapse data correctly from sonata edge files for…

7adda77

… target point neurons and deliver replay spikes to the proper netcon

Introduce test for allen point neurons

c203180

Fix lint

3185fbb

To test for allen type edges, instead look for limited count of prope…

63a1b59

…rties

minor: limit use of docformatter to before 1.7.8 until it has bug fixed

dc95223

Instantiate allen biophysical cells in another test, adding support data

8249daa

Forgot updated files in previous commit

cba681e

Improved allen test instantiation

708b303

Removed call to finalize for point process type cells since this is not needed

Removed prototype code ultimately not used for allen use case

45b18aa

Updated allen tests to collection cell_info and get ExpSyn data as consequence

Involve NEURON ParallelContext when testing point cells to trigger mo…

e74a00b

…re netcon code Remove more unused code sections; can be added when use case example is available

Corrected SynapseType to indicate ALLEN_CHEMICAL and removed ALLEN_POINT

f0264f2

Added additional test for invalid Mechanism given to HocPointProcessCell

jamesgking requested review from AurelienJaquier and darshanmandge May 4, 2026 05:23

darshanmandge reviewed May 4, 2026

View reviewed changes

Merge branch 'main' into point-process

7b1551e

WeinaJi reviewed Jun 19, 2026

View reviewed changes

Comment thread bluecellulab/cell/point_process.py Outdated

jamesgking added 2 commits June 29, 2026 07:31

Address code review comments

b6f22e4

Refactor to conform better to existing BlueCellulab structures

Add more Cell members to BasePointProcessCell and handle weight better

83d7a94

darshanmandge reviewed Jun 30, 2026

View reviewed changes

Comment thread bluecellulab/point/point_connection.py

fixes for tests to all point cells to use info_dict

1ae6781

darshanmandge previously approved these changes Jun 30, 2026

View reviewed changes

More safety with point_connection and syn_description

131e853

jamesgking dismissed darshanmandge’s stale review via 131e853 June 30, 2026 09:33

jamesgking marked this pull request as ready for review June 30, 2026 09:35

darshanmandge previously approved these changes Jun 30, 2026

View reviewed changes

AurelienJaquier reviewed Jun 30, 2026

View reviewed changes

Fix for replay when duration is used; address clean up requests

731777e

jamesgking dismissed darshanmandge’s stale review via 731777e June 30, 2026 11:30

AurelienJaquier approved these changes Jun 30, 2026

View reviewed changes

jamesgking merged commit 9df2c2d into main Jun 30, 2026
9 checks passed

Uh oh!

Conversation

ilkilic commented Jan 9, 2026

Uh oh!

codecov Bot commented Jan 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

darshanmandge left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

darshanmandge left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

AurelienJaquier left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

codecov Bot commented Jan 9, 2026 •

edited

Loading