How to save each ligand from a PDB file separately with Bio.PDB?

I have a list of PDB files. I want to extract the ligands of all the files (so, heteroatoms) and save each one separately into PDB files, by using the Bio.PDB module from BioPython.

I tried some solutions, like this one: Remove heteroatoms from PDB , that I tried to adapt to keep the heteroatoms. But all I obtain is files with all the ligand in the same file.

I also tried a thing like this :

def accept_residue(residue):
    """ Recognition of heteroatoms - Remove water molecules """ 
    res = residue.id[0]
    if res != " ": # Heteroatoms have some flags, that's why we keep only residue with id != " "
        if res != "W": # Don't take in consideration the water molecules
            return True


def extract_ligands(path):
    """ Extraction of the heteroatoms of .pdb files """
    for element in os.listdir(path+'/data/pdb'):
        i=1
        if element.endswith('.pdb'):
            if not element.startswith("lig_"):
                pdb = PDBParser().get_structure(element[:-4], path+'/data/pdb/'+element)
                io = PDBIO()
                io.set_structure(pdb)
                for model in pdb:
                    for chain in model:
                        for residue in chain:
                            if accept_residue(residue):
                                io.save("lig_"+element[:-4]+"_"+str(i)+".pdb", accept_residue(residue))
                                i += 1 # Counter for the result filename

            

# Main
path = mypath

extract_ligands(path)

Obviously, it raised an error :

AttributeError: 'bool' object has no attribute 'accept_model'

I know that's because of the "accept_residue()" in my "io.save". But I didn't find any logical solution to do what I want to...

At last, I tried a solution like this one, with chain.detach_child() :

                    ...
                    for chain in model:
                        for residue in chain:
                            res = residue.id[0]
                            if res == " " or res == "W": 
                                chain.detach_child(residue.id)
                        if len(chain) == 0:
                            model.detach_child(chain.id)
                     ...

In my mind, it would "detach" all the residues that are not heteroatoms ( res.id[0] == " " ) and all the water ( res.id[0] == "W"). But in fine, all the residues and water are still there and buggy.

So, is it possible to do what I need? ( extract all ligands from all my files and save it one by one separately in PDB files)

You were quite close.

But you have to provide a Select class as second argument to io.save. Have a look at the doc comment. It says that this argument should provide accept_model, accept_chain, accept_residue and accept_atom.

I created a class ResidueSelect that inherits from Bio.PDB.PDBIO.Select. That way I only have to override the methods we need. In our case for chain and residues.

Because we only want to save the current residue in the current chain, I provide two respective arguments for the constructor.

import os

from Bio.PDB import PDBParser, PDBIO, Select


def is_het(residue):
    res = residue.id[0]
    return res != " " and res != "W"


class ResidueSelect(Select):
    def __init__(self, chain, residue):
        self.chain = chain
        self.residue = residue

    def accept_chain(self, chain):
        return chain.id == self.chain.id

    def accept_residue(self, residue):
        """ Recognition of heteroatoms - Remove water molecules """
        return residue == self.residue and is_het(residue)


def extract_ligands(path):
    """ Extraction of the heteroatoms of .pdb files """

    for pfb_file in os.listdir(path + '/data/pdb'):
        i = 1
        if pfb_file.endswith('.pdb') and not pfb_file.startswith("lig_"):
            pdb_code = pfb_file[:-4]
            pdb = PDBParser().get_structure(pdb_code, path + '/data/pdb/' + pfb_file)
            io = PDBIO()
            io.set_structure(pdb)
            for model in pdb:
                for chain in model:
                    for residue in chain:
                        if not is_het(residue):
                            continue
                        print(f"saving {chain} {residue}")
                        io.save(f"lig_{pdb_code}_{i}.pdb", ResidueSelect(chain, residue))
                        i += 1


# Main
path = mypath

extract_ligands(path)

Btw: I tried to improve the readability of your code a little bit in the process...