def preprocess_pdbbind(pdbbind_dir):
"""Preprocess pdbbind files for Binana."""
assert os.path.isdir(pdbbind_dir)
# Extract the subdirectories in pdbbind_dir
subdirs = [d for d in os.listdir(pdbbind_dir) if
os.path.isdir(os.path.join(pdbbind_dir, d))]
print "About to preprocess following subdirectories:"
print subdirs
for count, dirname in enumerate(subdirs):
print "Processing %d-th entry %s" % (count, dirname)
subdir = os.path.join(pdbbind_dir, dirname)
ligand, protein = None, None
for molfile in os.listdir(subdir):
if "_ligand.mol2" in molfile:
print "Input ligand: %s" % molfile
ligand = molfile
elif "_protein.pdb" in molfile:
print "Input protein: %s" % molfile
protein = molfile
if not ligand or not protein:
raise ValueError("Ligand or Protein missing in %s" % dirname)
ligand_file = os.path.join(subdir, ligand)
protein_file = os.path.join(subdir, protein)
print "About to preprocess ligand."
hydrogenate_and_compute_partial_charges(ligand_file, "mol2", subdir)
print "About to preprocess protein."
hydrogenate_and_compute_partial_charges(protein_file, "pdb", subdir)
def launch_preprocessing_jobs(dude_dir, script_dir,
script_template, num_jobs):
"""Launch jobs to perform required preprocessing jobs."""
assert os.path.isdir(dude_dir)
target_dir = os.path.join(dude_dir, target)
# Extract list of actives and decoys
actives_dir = os.path.join(target_dir, "actives")
decoys_dir = os.path.join(target_dir, "decoys")
actives = [a for a in os.listdir(actives_dir)]
decoys = [a for a in os.listdir(decoys_dir)]
# Just for debugging purposes
actives = actives[:1]
decoys = decoys[:1]
# Hydrogenate and process the receptor
receptor = os.path.join(target_dir, "receptor.pdb")
hydrogenate_and_compute_partial_charges(
receptor, "pdb", target_dir)
receptor_pdbqt = os.path.join(target_dir, "receptor_hyd.pdbqt")
# Compute work per job
sources = ["active"] * len(actives) + ["decoy"] * len(decoys)
compounds = actives + decoys
num_per_job = int(math.ceil(len(compounds)/float(num_jobs)))
print "Number per job: %d" % num_per_job
def pdbbind_job(pdb_directories):
"""Processes all pdbs in provided directories."""
for count, dirname in enumerate(pdb_directories):
print "Processing %d-th entry %s" % (count, dirname)
ligand, protein = None, None
for molfile in os.listdir(dirname):
if "_ligand.mol2" in molfile:
print "Input ligand: %s" % molfile
ligand = molfile
elif "_protein.pdb" in molfile:
print "Input protein: %s" % molfile
protein = molfile
if not ligand or not protein:
raise ValueError("Ligand or Protein missing in %s" % dirname)
ligand_file = os.path.join(dirname, ligand)
protein_file = os.path.join(dirname, protein)
print "About to preprocess ligand."
hydrogenate_and_compute_partial_charges(ligand_file, "mol2", dirname)
print "About to preprocess protein."
hydrogenate_and_compute_partial_charges(protein_file, "pdb", dirname)
def pdbbind_job(pdb_directories):
"""Processes all pdbs in provided directories."""
for count, dirname in enumerate(pdb_directories):
print "Processing %d-th entry %s" % (count, dirname)
ligand, protein = None, None
for molfile in os.listdir(dirname):
if "_ligand.mol2" in molfile:
print "Input ligand: %s" % molfile
ligand = molfile
elif "_protein.pdb" in molfile:
print "Input protein: %s" % molfile
protein = molfile
if not ligand or not protein:
raise ValueError("Ligand or Protein missing in %s" % dirname)
ligand_file = os.path.join(dirname, ligand)
protein_file = os.path.join(dirname, protein)
print "About to preprocess ligand."
hydrogenate_and_compute_partial_charges(ligand_file, "mol2", dirname)
print "About to preprocess protein."
hydrogenate_and_compute_partial_charges(protein_file, "pdb", dirname)
def preprocess_dude(compounds):
"""Preprocess dude compounds for docking."""
with open(conf_template, "rb") as f:
conf = f.read()
for active in actives:
active_file = os.path.join(actives_dir, active)
active_base = active.split(".")[0]
active_output = os.path.join(actives_dir, "%s_hyd.pdbqt" % active_base)
conf_active_file = os.path.join(actives_dir,
"conf_%s.txt" % active_base)
print active_file
print active_output
print conf_active_file
hydrogenate_and_compute_partial_charges(active_file,
"pdb",
actives_dir,
rigid=False)
active_conf = conf % (receptor_pdbqt, active_output)
with open(conf_active_file, "wb") as f:
f.write(active_conf)
print active_conf
for decoy in decoys:
decoy_file = os.path.join(decoys_dir, decoy)
decoy_base = os.path.basename(decoy_file).split(".")[0]
decoy_output = os.path.join(decoys_dir, "%s_hyd.pdbqt" % decoy_base)
conf_decoy_file = os.path.join(decoys_dir, "conf_%s.txt" % decoy_base)
print decoy_file
print decoy_output
print conf_decoy_file
hydrogenate_and_compute_partial_charges(decoy_file,
"pdb",
decoys_dir,
rigid=False)
decoy_conf = conf % (receptor_pdbqt, decoy_output)
with open(conf_decoy_file, "wb") as f:
f.write(decoy_conf)
print decoy_conf
def preprocess_dude(compounds):
"""Preprocess dude compounds for docking."""
with open(conf_template, "rb") as f:
conf = f.read()
for active in actives:
active_file = os.path.join(actives_dir, active)
active_base = active.split(".")[0]
active_output = os.path.join(actives_dir, "%s_hyd.pdbqt" %
active_base)
conf_active_file = os.path.join(actives_dir,
"conf_%s.txt" % active_base)
print active_file
print active_output
print conf_active_file
hydrogenate_and_compute_partial_charges(
active_file, "pdb", actives_dir, rigid=False)
active_conf = conf % (receptor_pdbqt, active_output)
with open(conf_active_file, "wb") as f:
f.write(active_conf)
print active_conf
for decoy in decoys:
decoy_file = os.path.join(decoys_dir, decoy)
decoy_base = os.path.basename(decoy_file).split(".")[0]
decoy_output = os.path.join(decoys_dir, "%s_hyd.pdbqt" %
decoy_base)
conf_decoy_file = os.path.join(decoys_dir,
"conf_%s.txt" % decoy_base)
print decoy_file
print decoy_output
print conf_decoy_file
hydrogenate_and_compute_partial_charges(
decoy_file, "pdb", decoys_dir, rigid=False)
decoy_conf = conf % (receptor_pdbqt, decoy_output)
with open(conf_decoy_file, "wb") as f:
f.write(decoy_conf)
print decoy_conf
def _featurize_complex(self, mol_pdb, protein_pdb):
"""
Compute Binana fingerprint for complex.
"""
### OPEN TEMPDIR
tempdir = tempfile.mkdtemp()
mol_pdb_file = os.path.join(tempdir, "mol.pdb")
with open(mol_pdb_file, "w") as mol_f:
mol_f.writelines(mol_pdb)
protein_pdb_file = os.path.join(tempdir, "protein.pdb")
with open(protein_pdb_file, "w") as protein_f:
protein_f.writelines(protein_pdb)
mol_hyd_file = os.path.join(tempdir, "mol_hyd.pdb")
mol_pdbqt_file = os.path.join(tempdir, "mol_hyd.pdbqt")
hydrogenate_and_compute_partial_charges(
mol_pdb_file, "pdb", tempdir, mol_hyd_file, mol_pdbqt_file)
protein_hyd_file = os.path.join(tempdir, "protein_hyd.pdb")
protein_pdbqt_file = os.path.join(tempdir, "protein_hyd.pdbqt")
hydrogenate_and_compute_partial_charges(
protein_pdb_file, "pdb", tempdir, protein_hyd_file, protein_pdbqt_file)
mol_pdb_obj = PDB()
mol_pdb_obj.load_from_files(mol_pdb_file, mol_pdbqt_file)
protein_pdb_obj = PDB()
protein_pdb_obj.load_from_files(protein_pdb_file, protein_pdbqt_file)
features = self.binana.compute_input_vector(mol_pdb_obj, protein_pdb_obj)
### CLOSE TEMPDIR
shutil.rmtree(tempdir)
return features
def launch_preprocessing_jobs(dude_dir, script_dir, script_template, num_jobs):
"""Launch jobs to perform required preprocessing jobs."""
assert os.path.isdir(dude_dir)
target_dir = os.path.join(dude_dir, target)
# Extract list of actives and decoys
actives_dir = os.path.join(target_dir, "actives")
decoys_dir = os.path.join(target_dir, "decoys")
actives = [a for a in os.listdir(actives_dir)]
decoys = [a for a in os.listdir(decoys_dir)]
# Just for debugging purposes
actives = actives[:1]
decoys = decoys[:1]
# Hydrogenate and process the receptor
receptor = os.path.join(target_dir, "receptor.pdb")
hydrogenate_and_compute_partial_charges(receptor, "pdb", target_dir)
receptor_pdbqt = os.path.join(target_dir, "receptor_hyd.pdbqt")
# Compute work per job
sources = ["active"] * len(actives) + ["decoy"] * len(decoys)
compounds = actives + decoys
num_per_job = int(math.ceil(len(compounds) / float(num_jobs)))
print "Number per job: %d" % num_per_job
def main(input_file, input_format, output_directory):
"""Preprocesses provided files."""
hydrogenate_and_compute_partial_charges(input_file, input_format,
output_directory)
