CLI and downloads¶

Install from a source checkout with Python 3.12+ and Poetry:

poetry install
poetry run interop-trace --help
poetry run interop-trace components
poetry run interop-trace experiments

Start with the first experiment, then use configuration for other combinations. poetry run python -m interop_trace exposes the same CLI.

Download cache¶

From the repository root, components reuse downloads in .interop_trace/cache/. They record the source URL, name, type, byte count, retrieval time and available release metadata. Inspect or clear downloads with:

poetry run interop-trace cache list
poetry run interop-trace cache clear

Clearing asks for confirmation; add --yes for scripts. Clear between runs: experiment records refer to cached input files. Clearing preserves experiment outputs and their provenance records, but removes the underlying cached inputs and downloaded Sage installation.

Matching URLs reuse files if their presence and size agree with the record. Reuse does not verify checksums or check for source updates. A live UniProt query can return a different release after clearing; recorded versions identify the stored response, not a permanently pinned source. Archive needed inputs separately when exact long-term reproduction matters.

For a different dedicated cache, use the same directory for execution and inspection:

poetry run interop-trace run erwinia_sage --output results/run-002 --cache-directory /path/to/cache
poetry run interop-trace cache list --directory /path/to/cache
poetry run interop-trace cache clear --directory /path/to/cache

The cache uses Pooch for transfers and extraction. An ownership marker protects unrelated nonempty directories from being used or cleared as caches.

For direct Python use, see the API reference. The experiment runner coordinates component setup automatically.