# Full-network pilot: protocol fixed before simulation

Question: does silencing GABA-consensus mAL output change the response of a broad Fru+/Dsx+ pC1 (P1-related) cohort to stimulation of two candidate sensory populations?

- Dataset: MaleCNS v1.0, all neurons with a nonempty superclass not containing `tbc`, following the authors' counting notebook. Retain all connections between them, no strength threshold. Include isolated neurons. This is a full annotated neuron graph, not every fragment in the segmentation.
- Inputs: foreleg (`ProLN`) LgLG6/7 (candidate male-responsive; 37 cells) and LgLG5/8 (candidate female-responsive; 27 cells). Mapping is tentative and transferred from v0.9: Ryba et al. 2026, DOI 10.1016/j.cub.2026.01.071. This is artificial stimulation, not reconstructed natural pheromone encoding.
- Readout: all `pC1_` cells annotated `fruDsx=coexpress_high`, not the precise P1 genetic driver. No courtship threshold or target-choice rule.
- Stimulus: independent time-discretized Poisson (Bernoulli per timestep) voltage impulses, total 5,550 events/s shared across each input population (150 Hz/cell for 37 male-candidate cells; 205.56 Hz/cell for 27 female-candidate cells). This matches expected total external events, not physiological effect or synaptic output. Trials paired by identical seed and population.
- Trial: 50 ms rest, 200 ms stimulation, 50 ms washout; three seeds (1, 2, 3). Zero initial activity and no background input. Count readout spikes from 50 through 300 ms and report spikes/neuron/second over that 250 ms window.
- Intervention: suppress outgoing transmission of all GABA-consensus `mAL_m*` neurons. Everything else, including generated external input, is identical. Cells may still spike; output block models silencing transmission.
- Dynamics: homogeneous current-based leaky integrate-and-fire model, independently implemented with Numba; Shiu reference constants: resting/reset -52 mV, threshold -45 mV, membrane 20 ms, synaptic decay 5 ms, refractory 2.2 ms, delay 1.8 ms, weight 0.275 mV per anatomical synapse, external impulse 68.75 mV, zero refractory for stimulated cells. Exact linear subthreshold update, timestep 0.1 ms. Synaptic arrivals during a cell’s refractory interval are discarded. Not an exact reproduction of the Brian2 scheduling or a validated MaleCNS parameter fit.
- Fast transmitter assumptions: acetylcholine +1, GABA -1, glutamate -1; other/unclear transmitters zero fast weight. All anatomical connections are retained, but zero-sign cells have no modeled fast output. Neuromodulation, receptors, development, and body feedback are omitted.
- Essential prespecified sensitivity: LgLG5 is glutamatergic and LgLG8 unclear, so additionally run both candidate inputs/conditions with ONLY LgLG5/8 output assumed excitatory. This bracket is not a measured receptor model. Neither sign setting will be selected to obtain a desired result.
- Controls: no-input zero activity; seed-1 restoration equals intact; PPN1 (11431, 12286; explicit Kallman2015 annotation) stimulation at 150 Hz/cell, intact and mAL blocked, as a mechanistic control. PPN1 stimulation is not guaranteed to activate the broad readout in this model.
- Primary comparison: paired mAL-blocked minus intact P1-related firing rate, per input and sign setting. Show all three seeds. No significance claim from three synthetic trials. Also record mAL, PPN1, network recruitment, and binned per-neuron spike counts.
- Visualization: seed 1 is chosen in advance for playback, with all conditions available. Binned spike flashes have a fixed scale and timing; silent cells stay dim. Playback is slowed; no artificial propagation paths or preference animation.
- Stop rule: 30-minute implementation/run budget. If runtime, mapping, or stability prevents completion, report the incomplete experiment rather than inventing results. A null/unstable response is retained. Full-network inclusion is not full biological fidelity.

Sources: https://male-cns.janelia.org/download/ ; https://github.com/flyconnectome/2025malecns/blob/main/supplemental_data/quantify-neuron-connections.ipynb ; https://github.com/philshiu/Drosophila_brain_model/blob/main/model.py ; https://elifesciences.org/articles/11188 ; https://pubmed.ncbi.nlm.nih.gov/41763206/