ProteinPrep¶

Prepare a Protein for virtual screening or target work with the Deep Origin Protein Prep tool. Protein Prep cleans the structure using the chain, cofactor, water, and ligand lists you pass in, then runs loop modelling and protonation.

ProteinPrep is async only: submit with start(), track with wait() or watch(), then load the prepared structure with get_results(). There is no run(), and this tool does not produce a cost quote.

get_results() returns an in-memory Protein whose remote_path points at the prepared PDB. That object has no platform protein id until you sync() or update() it. The input protein on the ProteinPrep instance is unchanged.

Preparing a protein¶

Create a protein, then a ProteinPrep. The tool needs a 4-character Protein Data Bank (PDB) ID for loop-modelling templates. If the protein already has pdb_id (for example from Protein.from_pdb_id), you can omit it.

from deeporigin.drug_discovery import Protein, ProteinPrep, BRD_DATA_DIR

protein = Protein.from_file(BRD_DATA_DIR / "brd.pdb")
prep = ProteinPrep(protein, pdb_id="1EBY")
prep.start()
prep.wait()
prepared = prep.get_results()

In a notebook, watch progress while the execution runs:

prep = ProteinPrep(protein, pdb_id="1EBY")
prep.start()
task = await prep.watch()
prep.wait()
prepared = prep.get_results()

Optional keep and remove lists match the tool inputs. Empty keep-chain means keep all chains; empty cofactor and water lists keep none of those; empty remove-ligand means no extra ligand names are stripped.

prep = ProteinPrep(
    protein,
    pdb_id="1EBY",
    keep_chain_ids=["A"],
    keep_cofactor_ids=["MG", "ZN"],
    keep_water_residue_names=["HOH"],
    remove_ligand_ids=["LIG"],
)

Working with existing runs¶

Reconnect to a Protein Prep run started earlier, in this or a previous session:

from deeporigin.drug_discovery import ProteinPrep

prep = ProteinPrep.from_id("<executionId>")
# Or the most recently created ProteinPrep run:
prep = ProteinPrep.from_last_run()

prep.sync()
prepared = prep.get_results()