def __init__(self, folderName):
self.basePath = os.path.join(FLAGS.workplace, folderName)
self.tempFolderPath = os.path.join(self.basePath, 'temp')
self.dockingFolderPath = os.path.join(self.basePath, 'dock')
self.thread_num = FLAGS.thread_num
self.process_num = FLAGS.process_num
try_create_chain_folder(self.tempFolderPath)
try_create_chain_folder(self.dockingFolderPath)
def __init__(self):
self.parse()
self.tempFolderPath = tempfile.mkdtemp()
self.basePath = os.path.join(self.workplace, 'blast')
self.formsPath = os.path.join(self.basePath, 'forms')
self.err_log_file = os.path.join(self.basePath, 'err.log')
self.mergedPath = os.path.join(self.basePath, 'merged_forms')
try_create_chain_folder(self.tempFolderPath)
try_create_chain_folder(self.formsPath)
def downloads(self,item):
download_address = self.get_address(item)
if os.path.exists(os.path.join(FLAGS.rowdata_folder,item+'.pdb')):
print item," exists"
return None
print 'download ',item
os.system('wget -P {} {}'.format(FLAGS.rowdata_folder,download_address))
pdbname = item.lower()
ligand_folder = os.path.join(FLAGS.splited_ligand_folder,pdbname)
try_create_chain_folder(ligand_folder)
try:
parsed = prody.parsePDB(os.path.join(FLAGS.rowdata_folder,item+'.pdb'))
except:
self.error_log('can not parse {}.\n'.format(item))
return None
hetero = parsed.select('(hetero and not water) or resname ATP or resname ADP or sesname AMP or resname GTP or resname GDP or resname GMP')
receptor = parsed.select('protein or nucleic')
if receptor is None:
self.error_log("{} doesn't have receptor.\n".format(item))
return None
if hetero is None:
self.error_log("{} doesn't have ligand.\n".format(item))
return None
ligand_flags = False
for each in prody.HierView(hetero).iterResidues():
if each.numAtoms() <= 10:
continue
else:
ligand_flags = True
ResId = each.getResindex()
ligand_path = os.path.join(FLAGS.splited_ligand_folder,pdbname,"{}_{}_ligand.pdb".format(pdbname,ResId))
try_create_chain_parent_folder(ligand_path)
prody.writePDB(ligand_path,each)
if ligand_flags:
receptor_path = os.path.join(FLAGS.splited_receptor_folder,pdbname+'.pdb')
prody.writePDB(receptor_path,receptor)
else:
self.error_log("{} doesn't convert, not ligand have more than 10 atoms.\n".format(item))
def __init__(self):
self.parse()
try_create_chain_folder(self.formsPath)
def convert_database_to_av4(database_path,
positives_folder=None,
decoys_folder=None,
receptors_folder=None):
"""Crawls the folder (receptors in this case) and saves every PDB it finds
into .npy array with 1) coordinates 2) mapped to the atom name number """
# make a directory where the av4 form of the output will be written
output_path = str(database_path + '_av4')
if not os.path.exists(output_path):
os.makedirs(output_path)
def save_av4(filepath, labels, elements, multiframe_coords):
labels = np.asarray(labels, dtype=np.int32)
elements = np.asarray(elements, dtype=np.int32)
multiframe_coords = np.asarray(multiframe_coords, dtype=np.float32)
if not (int(len(multiframe_coords[:, 0]) == int(len(elements)))):
raise Exception(
'Number of atom elements is not equal to the number of coordinates'
)
if multiframe_coords.ndim == 2:
if not int(len(labels)) == 1:
raise Exception(
'Number labels is not equal to the number of coordinate frames'
)
else:
if not (int(len(multiframe_coords[0, 0, :]) == int(len(labels)))):
raise Exception(
'Number labels is not equal to the number of coordinate frames'
)
number_of_examples = np.array([len(labels)], dtype=np.int32)
av4_record = number_of_examples.tobytes()
av4_record += labels.tobytes()
av4_record += elements.tobytes()
av4_record += multiframe_coords.tobytes()
f = open(filepath + ".av4", 'w')
f.write(av4_record)
f.close()
count = 0
database_ligand_path = os.path.join(database_path, 'actives')
database_receptor_path = os.path.join(database_path, 'receptors')
for receptor in os.listdir(database_ligand_path):
for ligand_name in os.listdir(
os.path.join(database_ligand_path, receptor)):
count += 1
destFile = os.path.join(output_path, receptor,
ligand_name + ".av4")
if os.path.exists(destFile):
continue
if FLAGS.orchestra_arrayjob and FLAGS.orchestra_jobid % FLAGS.orchestra_jobsize != count % FLAGS.orchestra_jobsize:
continue
ligand_folder = os.path.join(database_ligand_path, receptor,
ligand_name)
splited_ligands = os.listdir(ligand_folder)
if len(splited_ligands) == 0:
with open(os.path.join(database_path, 'empty.txt'),
'a') as fout:
fout.write(ligand_folder + '\n')
continue
path_to_receptor = os.path.join(database_receptor_path,
receptor + '.pdb')
path_to_first_ligand = os.path.join(ligand_folder,
splited_ligands[0])
try:
prody_receptor = prody.parsePDB(path_to_receptor)
prody_first_ligand = prody.parsePDB(path_to_first_ligand)
multiframe_ligand_coords = prody_first_ligand.getCoords()
# for decoys set all the label as 0
#labels = np.array([0])
# for actives set all the label as 1
labels = np.array([1])
# if have more than one ligands, write them as one multiframe ligand
if len(splited_ligands) > 1:
for rest_ligand in splited_ligands[1:]:
prody_rest = prody.parsePDB(
os.path.join(ligand_folder, rest_ligand))
# see if decoy is same as the initial ligand
if not all(
np.asarray(prody_rest.getElements()) ==
np.asarray(prody_first_ligand.getElements())):
raise Exception(
'attempting to add ligand with different order of atoms'
)
multiframe_ligand_coords = np.dstack(
(multiframe_ligand_coords, prody_rest.getCoords()))
# for decoys set all the label as 0
#labels = np.concatenate((labels, [0]))
# for actives set all teh label as 1
labels = np.concatenate((labels, [1]))
except Exception as e:
print e
stats.ligands_failed += 1
print "ligands parsed:", stats.ligands_parsed, "ligands failed:", stats.ligands_failed
continue
stats.ligands_parsed += 1
print "ligands parsed:", stats.ligands_parsed, "ligands failed:", stats.ligands_failed
# create an output path to write binaries for protein and ligands
path_to_pdb_subfolder = os.path.join(output_path, receptor)
try_create_chain_folder(path_to_pdb_subfolder)
# convert atomnames to tags and write the data to disk
def atom_to_number(atomname):
atomic_tag_number = atom_dictionary.ATM[atomname.lower()]
return atomic_tag_number
print prody_receptor.getElements()
receptor_elements = map(atom_to_number,
prody_receptor.getElements())
ligand_elements = map(atom_to_number,
prody_first_ligand.getElements())
receptor_output_path = os.path.join(path_to_pdb_subfolder,
receptor)
save_av4(receptor_output_path, [0], receptor_elements,
prody_receptor.getCoords())
ligand_output_path = os.path.join(path_to_pdb_subfolder,
ligand_name)
save_av4(ligand_output_path, labels, ligand_elements,
multiframe_ligand_coords)