parse contact matrices from PDBbind and store in chunks

pdbbind.py

@@ -1,26 +1,96 @@
 from mpi4py import MPI
 from mpi4py.futures import MPICommExecutor
 
-from openbabel import pybel
-from Bio import SeqIO
+import warnings
+from Bio.PDB import PDBParser, PPBuilder, CaPPBuilder
+from Bio.PDB.NeighborSearch import NeighborSearch
+from Bio.PDB.Selection import unfold_entities
+
+import numpy as np
+import dask.array as da
+
+from rdkit import Chem
 
 import os
+import re
+
+# all punctuation
+punctuation_regex  = r"""(\(|\)|\.|=|#|-|\+|\\|\/|:|~|@|\?|>>?|\*|\$|\%[0-9]{2}|[0-9])"""
+
+# tokenization regex (Schwaller)
+molecule_regex = r"""(\[[^\]]+]|Br?|Cl?|N|O|S|P|F|I|b|c|n|o|s|p|\(|\)|\.|=|#|-|\+|\\|\/|:|~|@|\?|>>?|\*|\$|\%[0-9]{2}|[0-9])"""
+
+cutoff = 5
+max_seq = 2048
+max_smiles = 512
+chunk_size = '1G'
+
 def parse_complex(fn):
     try:
         name = os.path.basename(fn)
-        seq = str(next(SeqIO.parse(fn+'/'+name+'_protein.pdb', "pdb-seqres")).seq)
-        mol = next(pybel.readfile('sdf',fn+'/'+name+'_ligand.sdf'))
-        smi = mol.write('can').split('\t')[0]
-        return name, seq, smi
-    except:
+
+        # parse protein sequence and coordinates
+        parser = PDBParser()
+        with warnings.catch_warnings():
+            warnings.simplefilter("ignore")
+            structure = parser.get_structure('protein',fn+'/'+name+'_protein.pdb')
+
+#        ppb = PPBuilder()
+        ppb = CaPPBuilder()
+        seq = []
+        for pp in ppb.build_peptides(structure):
+            seq.append(str(pp.get_sequence()))
+        seq = ''.join(seq)
+
+        # parse ligand, convert to SMILES and map atoms
+        suppl = Chem.SDMolSupplier(fn+'/'+name+'_ligand.sdf')
+        mol = next(suppl)
+        smi = Chem.MolToSmiles(mol)
+
+        # position of atoms in SMILES (not counting punctuation)
+        atom_order = mol.GetProp("_smilesAtomOutputOrder")
+        atom_order = [int(s) for s in list(filter(None,re.sub(r'[\[\]]','',mol.GetProp("_smilesAtomOutputOrder")).split(',')))]
+
+        # tokenize the SMILES
+        tokens = list(filter(None, re.split(molecule_regex, smi)))
+
+        # remove punctuation
+        masked_tokens = [re.sub(punctuation_regex,'',s) for s in tokens]
+
+        k = 0
+        token_pos = []
+        token_id = []
+        for i,token in enumerate(masked_tokens):
+            if token != '':
+                token_pos.append(tuple(mol.GetConformer().GetAtomPosition(atom_order[k])))
+                token_id.append(i)
+                k += 1
+
+        # query protein for ligand contacts
+        atoms  = unfold_entities(structure, 'A')
+        neighbor_search = NeighborSearch(atoms)
+
+        close_residues = [neighbor_search.search(center=t, level='R', radius=cutoff) for t in token_pos]
+        residue_id = [[c.get_id()[1]-1 for c in query] for query in close_residues] # zero-based
+
+        # contact map
+        contact_map = np.zeros((max_seq, max_smiles),dtype=np.float32)
+
+        for query,t in zip(residue_id,token_id):
+            for r in query:
+                contact_map[r,t] = 1
+
+        return name, seq, smi, contact_map
+    except Exception as e:
+        print(e)
         return None
 
 
 if __name__ == '__main__':
     import glob
 
-    filenames = glob.glob('pdbbind/data/v2020-other-PL/*')
-    filenames.extend(glob.glob('pdbbind/data/refined-set/*'))
+    filenames = glob.glob('data/pdbbind/v2020-other-PL/*')
+    filenames.extend(glob.glob('data/pdbbind/refined-set/*'))
     comm = MPI.COMM_WORLD
     with MPICommExecutor(comm, root=0) as executor:
         if executor is not None:
@@ -29,7 +99,10 @@ if __name__ == '__main__':
             names = [r[0] for r in result if r is not None]
             seqs = [r[1] for r in result if r is not None]
             all_smiles = [r[2] for r in result if r is not None]
+            all_contacts = [r[3] for r in result if r is not None]
 
             import pandas as pd
             df = pd.DataFrame({'name': names, 'seq': seqs, 'smiles': all_smiles})
+            all_contacts = da.from_array(all_contacts, chunks=chunk_size)
+            da.to_npy_stack('data/pdbbind_contacts/', all_contacts)
             df.to_parquet('data/pdbbind_complex.parquet')

pdbbind.slurm

@@ -1,9 +1,9 @@
 #!/bin/bash
 #SBATCH -J preprocess_pdbbind
-#SBATCH -p batch
+#SBATCH -p gpu
 #SBATCH -A STF006
 #SBATCH -t 3:00:00
-#SBATCH -N 12
-#SBATCH --ntasks-per-node=8
+#SBATCH -N 2
+#SBATCH --ntasks-per-node=16
 
 srun python pdbbind.py