amorehead
commented
3 months ago In case anyone else needs a solution to this issue, one can (for the time being) use the following BioPandas function to extract all hetero-atoms uniquely for each ligand "residue".
import numpy as np
from biopandas.pdb import PandasPdb
def insert_unique_ligand_atom_name(input_pdb_file: str, output_pdb_file: str):
"""
Add unique identifier to ligand atoms with the same residue ID and chain ID.
:param input_pdb_file: Path to the input PDB file.
:param output_pdb_file: Path to the output PDB file.
"""
pdb = PandasPdb().read_pdb(input_pdb_file)
ligand_atoms = pdb.df["HETATM"]
ligand_atoms["atom_name"] = ligand_atoms["atom_name"] + np.arange(1, len(ligand_atoms) + 1).astype(str)
pdb.df["HETATM"] = ligand_atoms
pdb.to_pdb(output_pdb_file)
Hello.
After inspecting the formatting of RFAA's structure prediction PDB files for (multi-ligand) protein-ligand complexes, it seems that the formatting for ligand (HETATM) atoms associated with each ligand "residue" (in PDB verbiage) is not compatible with well-known Python PDB file parsing libraries such as BioPython. In particular, it looks like this is because RFAA's output PDB files (for ligand predictions) do not follow PDB file formatting standards (as discussed in https://stackoverflow.com/questions/64032005/is-there-a-way-to-obtain-a-full-list-of-atoms-from-a-ligand-residue).
Is there by chance a simple modification one can make to the RFAA source code to assign unique (alt loc) IDs to each ligand atom (with respect to a given residue) so that the output PDB files can be properly parsed?
Example PDB file output showing "duplicated" ligand atoms for a particular ligand "residue" (e.g., LG1, Chain B):