FIxing modeller bug for WT case

[steven-albanese]

Modeller was throwing an error when given WT in the yaml pipeline.

• fixing the way this is being handled
• Adding a test for the WT case

[steven-albanese]

Tests are passing, but I'm not sure if I've got the test for the WT case correct

[andrrizzi]

The fix looks good! For the test, I wouldn't add a new one but just add to test_modeller_mutations something like

# We haven't set the strip_protons options, so this shouldn't do anything
exp_builder._db._setup_molecules(mol_id)
assert not os.path.exists(output_path)

# ADD THIS BLOCK
# Calling modeller with WT creates a file (although the protein is not mutated).
exp_builder._db.molecules[mol_id]['pdbfixer'] = {
    'apply_mutations' : {
        'chain_id' : 'A',
        'mutations': 'WT',
    }
}
setup_molecule_output_check(exp_builder._db, mol_id, output_path)
os.remove(output_path)  # Remove file for next check.
# NEW BLOCK ENDS

# Now we set the strip_protons options and repeat
exp_builder._db.molecules[mol_id]['modeller'] = {
    'apply_mutations': {
        'chain_id': 'A',
        'mutations': 'T85I',
    }
}
setup_molecule_output_check(exp_builder._db, mol_id, output_path)

[andrrizzi]

Whops! Sorry, it looks like the test I suggested fails because when using WT with modeller it doesn't save the pdb file. If that's expected, feel free to substitute

setup_molecule_output_check(exp_builder._db, mol_id, output_path)
os.remove(output_path)  # Remove file for next check.

with

exp_builder._db._setup_molecules(mol_id)
assert not os.path.exists(output_path)

Can you also update the release history (docs/whatsnew.rst) before merging?

[steven-albanese]

Hmm weirdly, it looks like it's actually the second setupt_molecule_output_check that is failing:

yank/Yank/tests/test_experiment.py

Line 1356 in bd9a493

setup_molecule_output_check(exp_builder._db, mol_id, output_path)

Do I need to reinitialize the ExperimentBuilder object in between the wild type and mutant case?

[andrrizzi]

Hmm. I don't think you should! Can you check what exp_builder._db.is_molecule_setup(mol_id) returns after the call to os.remove(output_path)?

[andrrizzi]

Tests finally pass! Should I merge this and go ahead with a bugfix release?

[steven-albanese]

Yep!

…

On Aug 24, 2018, at 4:42 PM, Andrea Rizzi ***@***.***> wrote: Tests finally pass! Should I merge this and go ahead with a bugfix release? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.