Skip to content

Tutorials

One runnable script per anchor workflow (jupytext percent format; open as a notebook with jupytext --to ipynb examples/<name>.py or run with python examples/<name>.py after flyconn data pull):

  • W1 circuit tracing with signs and confidence: examples/w1_pathways.py
  • W4 version drift: examples/w4_version_drift.py
  • W2 in-silico experiment from one YAML file: flyconn run examples/specs/w2_malecns_lb3_silence_gng232.yaml --out runs/w2 (controls on by default; the report states that MaleCNS runs are uncalibrated)
  • W2 silencing screen (sweep) from one YAML file: flyconn run examples/specs/w2_sweep_silence_gng.yaml --out runs/w2_sweep silences each of the 10 most active GNG cell types one at a time and writes sweep.parquet plus one ranked report. A spec holds exactly one sweep: kind:

    ```yaml sweep: silence_each: {select: {cell_type: "GNG*"}, group_by: cell_type, max_items: 10}

    or: group_by: neuron (one variant per neuron)

    or, for readout-vs-rate curves:

    sweep: {rate_hz: [10, 50, 100, 150, 200]}

    ```

    Shared conditions run once; BH correction spans the whole sweep; see the sweep caveats. - W3 male vs female per cell type: examples/w3_male_vs_female.py - W5 genetic access (exploratory): examples/w5_driver_lines.py