Run the invariance tests with
python -m <database>.test invariance --result-destination path/to/invariance/results.sqlite
and the regression tests with
python -m <database>.test regression --result-destination path/to/regression/results.sqlite --reference-result path/to/regression/reference.sqlite
Before running the regression tests, compute a regression reference with
python -m <database>.test regression --compute-reference-result --result-destination path/to/regression/reference.sqlite
Tucan invariance tests against 1 PubChem SDF (containing 500,000 compounds),
distributed over 8 processes ran for about 42 minutes (2502 seconds).
Assuming that PubChem contains about 150,000,000 compounds,
the expected runtime is 150,000,000 compounds / 500,000 compounds * 42 minutes = 12600 minutes = 210 hours = 8.75 days.
The runtime could be reduced significantly by distributing the test over more processes.