schrodinger.seam.testing.stu_tests.ml_train_integration module

Integration tests for TrainModel that require ligand_ml (Smasher).

These tests train a Smasher model and verify the output.

Usage:

$SCHRODINGER/run python3 -m schrodinger.seam.testing.stu_tests.ml_train_integration

schrodinger.seam.testing.stu_tests.ml_train_integration.test_train_produces_model(structures)

Verify that _train produces a valid MLModel.

schrodinger.seam.testing.stu_tests.ml_train_integration.test_model_blob_extracts(structures)

Verify that the model blob can be extracted to disk.

schrodinger.seam.testing.stu_tests.ml_train_integration.test_model_pickle_roundtrip(structures)

Verify that a trained model survives pickle round-trip.

schrodinger.seam.testing.stu_tests.ml_train_integration.test_pipeline_produces_model(structures, options)

Verify that the TrainModel beam transform produces a model.

schrodinger.seam.testing.stu_tests.ml_train_integration.test_multi_target_train()

Train with two properties as prediction targets.

schrodinger.seam.testing.stu_tests.ml_train_integration.test_predict_properties(structures, options)

Train a model then predict properties on the same structures.

schrodinger.seam.testing.stu_tests.ml_train_integration.main()