Pockets

This document describes how to create and work with pockets for docking and other Deep Origin tools.

The Pocket class is the primary way to represent binding sites in the Drug Discovery toolbox.

Constructing a Pocket¶

Most pocket workflows start from a Protein. For example:

from deeporigin.drug_discovery import Protein, BRD_DATA_DIR

protein = Protein.from_file(BRD_DATA_DIR / "brd.pdb")
protein.remove_water()

Using the PocketFinder tool¶

To discover novel pockets on a protein structure, use the PocketFinder tool. See the PocketFinder documentation for synchronous and asynchronous execution, cost estimates, loading results by record ID, and visualization.

From a residue number¶

Create a pocket centered on a specific residue:

from deeporigin.drug_discovery import Pocket

pocket = Pocket.from_residue_number(
    protein=protein,
    residue_number=123,
    chain_id="A",
    cutoff=5.0,
)

residue_number is the residue index in the loaded structure. chain_id is optional when the number is unique. cutoff (Å) sets how far from that residue to include surrounding residues in the pocket.

From a crystal ligand¶

When a PDB contains a co-crystallized ligand, extract it from the protein and define the pocket from that ligand geometry:

from deeporigin.drug_discovery import Protein, Pocket

protein = Protein.from_pdb_id("1EBY")
ligand = protein.extract_ligand()
pocket = Pocket.from_ligand(ligand, name="crystal_ligand_pocket")

extract_ligand() removes the ligand from the protein structure in place. See Extract crystal ligands for warnings, excluded residue names, and customization.

From a ligand¶

Create a pocket from a ligand structure loaded separately (for example from an SDF file):

from deeporigin.drug_discovery import Pocket, Ligand, BRD_DATA_DIR

ligand = Ligand.from_sdf(BRD_DATA_DIR / "brd-2.sdf")
pocket = Pocket.from_ligand(ligand, name="ligand_pocket")

This uses the ligand coordinates as the pocket definition. For loading ligands from other sources, see Work with Ligands.