Stage-by-stage batch runs and statistics¶

run_all.py runs the numbered stages of the original study for the batches of one configuration file. Use it when you want the genotype statistics, or several cohorts in one run. It expects videos named Day<N>_<mouse>_Trial<T> and a reward table.

For a single recording set with no statistics, stride tmaze is the shorter way; it writes the same decision and gait tables.

Try it on mock data¶

About a minute, from the repository root:

python tests/mock/make_mock_dataset.py --out tests/mock/_data --only core
python run_all.py --config tests/mock/_data/core/config.readme.yaml --output mock_output
stride status --output mock_output          # what was run, when, with which code

On your own recordings¶

1. Write the config¶

cp config.example.yaml my_config.yaml         # edit the `batches:` section

batches:
  my_batch:
    name: "My batch"
    video_dir: "/path/to/videos/"                # Day<N>_<mouse>_Trial<T>.mp4, searched recursively
    meta_trials_csv: "/path/to/meta_trials.csv"  # day,mouse,reward
    metadata_csv: "/path/to/metadata.csv"        # Mouse,Gender,Genotype
    fps: 120
    column_map:
      mouse_id: ["mouse"]
      mouse_id_sep: ""
      sex: "Gender"

Add pose_model_paths and roi_model_paths to run inference, or leave them out to use existing pose and ROI files. Every key: Configuration reference.

2. Run¶

python run_all.py --config my_config.yaml --output output
python run_all.py --config my_config.yaml --output output --batch my_batch
python run_all.py --config my_config.yaml --output output --stages 3,4-6,7-9    # selected stages

stage key	does
`1-2`	distortion check, pose and ROI inference, ROI region files
`3`	decisions, events, metrics, learning cycles
`4-6`	stride detection, keypoint confidence, stride filtering
`7-9`	trial-phase tagging, mixed models, heatmaps
`10`	WT vs Tau classifier

Each stage reads what the previous one wrote under <output>/<batch>/; an explicit path in the batch wins. A stage that fails stops the run with a non-zero exit code and says why.

3. Read the results¶

Per batch: <output>/<batch>/decisions/, the gait tables, learning_cycle_1.csv.
Across batches: lmm_results_all.csv, lmm_fit_log.csv, heatmaps, the classifier tables and figures (Outputs).
stride status --output output lists the run manifests.

Differences from `stride tmaze` to know¶

A trial with no reward row is scored not correct here; stride tmaze leaves it empty.
The decision rules are not applied by run_all.py. Run stride apply-decision-rules on a decisions.csv to get decisions_clean.csv.
The scale is taken per video from the ROI corridor width unless the batch has px_per_cm_override.

The statistics inside `stride tmaze`¶

# in the T-maze config
steps:
  mixed_models: true
  classifier: true
stats_config: config.example.yaml

The metadata must carry a genotype column. Results are written under stats/ of the output folder.