def load_from_one_cath_pml_file(pml_file, scratch_path, superfamilies,
dssp_path):
'''Load data from a .pml file of superposed
homologous superfamilies from the CATH database.
'''
superfamilies.append([])
candidate_proteins = []
with open(pml_file, 'r') as f:
while True:
line = f.readline()
if not line: break
# Read one structure
if line.strip().startswith('cmd.read_pdbstr'):
pdb_lines = [line.strip()[19:].strip('\\')]
pdb_id = ''
while True:
line = f.readline()
if line.strip().startswith('"""'):
pdb_id = line.strip()[5:12]
break
pdb_line = line.strip().strip('\\')
if len(pdb_line) > 17:
pdb_line = pdb_line[0:16] + ' ' + pdb_line[
17:] # Remove all altLoc flags
pdb_lines.append(pdb_line) # Remove all altLoc flags
# Make a pdb file of the structure for DSSP analysis
structure = structure_from_pdb_string('\n'.join(pdb_lines),
pdb_id)
# Store structures without chain breaks
if len(topology.find_structure_chain_breaks(structure)) == 0:
structure_path = os.path.join(scratch_path,
pdb_id + '.pdb')
io = PDB.PDBIO()
io.set_structure(structure)
io.save(structure_path)
candidate_proteins.append({
'structure': structure,
'path': structure_path
})
for p in candidate_proteins:
try:
find_secondary_structures(p, dssp_path)
except:
continue
superfamilies[-1].append(
p) # Add a protein to a superfamily if there's no exception
def preparePdb(pdb_fname, out_pdb_fname):
''' Prepare the PDB file with only first model and redundancies cut out '''
# 'Absolutize' the path names - rest is done in the temporary dir
pdb_fname = os.path.abspath(pdb_fname)
if not os.path.exists(pdb_fname):
raise IOError('%s does not exist' % pdb_fname)
out_pdb_fname = os.path.abspath(out_pdb_fname)
# Inside the temporary dir
with tempDir() as tmp_dir:
# Temporary names for curated input and output files
new_pdb_fname = 'query.pdb'
out_tmp_fname = 'out.pdb'
# If the original PDB is packed with gzip - unpack it into a new file
if pdb_fname.endswith('.gz'):
rfh = gzip.open(pdb_fname, 'r')
else:
rfh = open(pdb_fname, 'r')
try:
with open(new_pdb_fname, 'w') as wfh:
wfh.write(rfh.read())
finally:
rfh.close()
# Parse structure
# Redirect standard output/error to a cStringIO,
#so that PDBParser stops messing the output
parser = Bio.PDB.PDBParser()
err_fh = io.StringIO()
sys.stdout = err_fh
sys.stderr = err_fh
struct = parser.get_structure('query', new_pdb_fname)
sys.stdout = sys.__stdout__
sys.stderr = sys.__stderr__
# Output formatted info about PDBParser's work to a log
s = err_fh.getvalue()
if s.strip():
logging.info(
"Structure parsing generated following error message(s): \n%s\n%s\n%s"
% ('-' * 120, s, '-' * 120))
# By default use only first model
model = struct[0]
del struct.child_list[1:]
# Check for discontinuities greater than 5 residues - warn about this _specifically_
for chain in model:
chid = chain.id
last_rid = None
for residue in chain:
if last_rid is not None and rid > last_rid + 5:
rid = residue.id[1]
logging.warn(
"Residues %s:%s-%s:%s. Results might be inaccurate, as the break in a protein chain numbers more than 5 residues."
% (last_id, chain, rid, chain))
last_rid = rid
# Save structure without hydrogens
io = Bio.PDB.PDBIO()
io.set_structure(struct)
io.save(new_pdb_fname)
shutil.move(new_pdb_fname, out_pdb_fname)
return out_pdb_fname
def save_structure(struct, name):
file = '{}.pdb'.format(name)
io = PDBIO()
io.set_structure(struct)
io.save(file)
del io
with open(file, 'r') as f:
atoms = f.read()
data = header() + atoms
with open(file, 'w') as f:
f.write(data)
def get_sequence(pdb, chain):
pdb_parser = PDBParser(PERMISSIVE=0) # The PERMISSIVE instruction allows PDBs presenting errors.
pdb_structure = pdb_parser.get_structure(pdb,pdb+".pdb")
pdb_chain = pdb_structure[0][chain]
i = 1
lista=[]
for residue in pdb_chain:
if i < int(sys.argv[3]) or i > int(sys.argv[4]):
lista.append(residue.get_id())
#pdb_chain.detach_child(residue.get_id())
i+=1
for x in lista:
pdb_chain.detach_child(x)
io = PDBIO()
io.set_structure(pdb_chain)
output = sys.argv[5]+"_segment.pdb"
io.save(output)
def extract_ligands(path):
""" Extraction of the heteroatoms of .pdb files """
for pfb_file in os.listdir(path + 'pdbs/'):
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 + 'pdbs/' + pfb_file)
io = PDBIO()
io.set_structure(pdb)
model_selected = pdb[0]
# for model in pdb:
for chain in model_selected:
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
def download_pdb(self, info):
pdb_id, chain_id = info
## Check if atom has alternative position, if so, keep 'A' position and remove the flag
## but somehow this class doesn't seem to function well
class NotDisordered(Select):
def accept_atom(self, atom):
if not atom.is_disordered() or atom.get_altloc() == 'A':
atom.set_altloc(' ')
return True
else:
return False
## BioPython downloads PDB but it gives a lowercase name in pdb{}.ent format
biopdb_name = '{0}/pdb{1}.ent'.format(self.work_dir, pdb_id.lower())
biopdb_modf = '{0}/pdb{1}.mod.ent'.format(self.work_dir,
pdb_id.lower())
if not os.path.isfile(biopdb_modf):
try:
PDB.PDBList(verbose=False).retrieve_pdb_file(
pdb_id,
pdir=self.work_dir,
obsolete=False,
file_format='pdb')
except FileNotFoundError:
print(
' \033[31m> ERROR: BioPython cannot download PDB: \033[0m'
+ pdb_id)
return None
## Replace modified AA to avoid mis-recognition in biopython readin
## Replace disordered atoms and keep only the "A" variant
ReplacePDBModifiedAA(biopdb_name, biopdb_modf)
os.system('grep "REMARK " {0} > {0}.remark'.format(biopdb_modf))
with open(biopdb_modf, 'r') as fi:
remarks = [l for l in fi if re.search('REMARK HET ', l)]
## Read the PDB file and extract the chain from structure[0]
try:
model = PDB.PDBParser(PERMISSIVE=1,
QUIET=1).get_structure(pdb_id,
biopdb_modf)[0]
except KeyError:
print(' \033[31m> ERROR: BioPython cannot read in PDB: \033[0m' +
biopdb_modf)
return None
except ValueError:
print(' \033[31m> ERROR: PDB file is empty: \033[0m' +
biopdb_modf)
return None
### Bug alert: as of 20.02.18, Biopython dev hasn't come up with good
### strategy to fix the 'atom.disordered_get_list()' issue with alternative
### position of residue side chains. To go around this, will physically
### remove "B" variant and keep only "A" variant in
io = PDB.PDBIO()
io.set_structure(model[chain_id])
io.save('{0}/{1}_{2}.pdb'.format(self.work_dir, pdb_id, chain_id),
select=NotDisordered())
# Attach REMARK to end of PDB as safekeeping
os.system('cat {0}/{1}_{2}.pdb {3}.remark > {1}.temp'.format(
self.work_dir, pdb_id, chain_id, biopdb_modf))
os.system('mv {1}.temp {0}/{1}_{2}.pdb'.format(self.work_dir, pdb_id,
chain_id))
# os.system('mv {1} {0}/{2}.ent'.format(self.work_dir, biopdb_name, pdb_id))
# os.system('bzip2 -f {0}/{1}.ent'.format(self.work_dir, pdb_id))
# os.system('rm {0} {0}.remark'.format(biopdb_modf))
return '{0}/{1}_{2}.pdb'.format(self.work_dir, pdb_id, chain_id)
def prepareWithHydrogens(pdb_fname, out_pdb_fname="wth_hydro.pdb"):
''' Prepare the PDB file with hydrogen data (clean up and create a new one). '''
# 'Absolutize' the path names - rest is done in the temporary dir
pdb_fname = os.path.abspath(pdb_fname)
if not os.path.exists(pdb_fname) or not os.path.isfile(pdb_fname):
raise IOError('%s does not exist or is not a file.' % pdb_fname)
out_pdb_fname = os.path.abspath(out_pdb_fname)
if pdb_fname.endswith('.gz'):
rfh = gzip.open(pdb_fname, 'r')
#print pdb_fname
else:
rfh = open(pdb_fname, 'r')
try:
# Parse structure
parser = Bio.PDB.PDBParser()
# Redirect standard output/error to a cStringIO,
# so that PDBParser stops messing the output
err_fh = io.StringIO()
sys.stdout = err_fh
sys.stderr = err_fh
struct = parser.get_structure('query', rfh)
finally:
# Restore streams
sys.stdout = sys.__stdout__
sys.stderr = sys.__stderr__
# ... and close up
rfh.close()
# Output formatted info about PDBParser's work to a logger
s = err_fh.getvalue()
if s.strip():
logging.info(
"Structure parsing generated following error message(s): \n%s\n%s\n%s"
% ('-' * 120, s, '-' * 120))
# By default use only first model
# ... delete the rest
model = struct[0]
del struct.child_list[1:]
# Check for discontinuities greater than 5 residues - warn about this _specifically_ (into the logger, again)
for chain in model:
chid = chain.id
last_rid = None
for residue in chain:
if last_rid is not None and rid > last_rid + 5:
rid = residue.id[1]
logging.warn(
"Residues %s:%s-%s:%s. Results might be inaccurate, as the break in a protein chain numbers more than 5 residues."
% (last_id, chain, rid, chain))
last_rid = rid
# Prepare the remade hydrogens
remakeHydrogens(struct)
# Save structure
if out_pdb_fname.endswith('.gz'):
with closing(gzip.open(out_pdb_fname, 'w')) as wfh:
io = Bio.PDB.PDBIO()
io.set_structure(struct)
io.save(wfh)
else:
io = Bio.PDB.PDBIO()
io.set_structure(struct)
io.save(out_pdb_fname)
return out_pdb_fname
def prepareWithHydrogensPrep23(pdb_fname, out_pdb_fname="wth_hydro.pdb"):
''' Prepare the PDB file with hydrogen data (clean up and create a new one). '''
# 'Absolutize' the path names - rest is done in the temporary dir
pdb_fname = os.path.abspath(pdb_fname)
if not os.path.exists(pdb_fname) or not os.path.isfile(pdb_fname):
raise IOError('%s does not exist or is not a file.' % pdb_fname)
out_pdb_fname = os.path.abspath(out_pdb_fname)
# Inside the temporary dir
with tempDir() as tmp_dir:
# Temporary names for curated input and output files
new_pdb_fname = 'query.pdb'
out_tmp_fname = 'out.pdb'
# Prepare the sources
prep_exec = _preparePrepExec()
# Copy the original file into our temporary directory
# If the original PDB is packed with gzip - unpack it into a new file
if pdb_fname.endswith('.gz'):
rfh = gzip.open(pdb_fname, 'r')
else:
rfh = open(pdb_fname, 'r')
try:
with open(new_pdb_fname, 'w') as wfh:
wfh.write(rfh.read())
finally:
rfh.close()
# Parse structure
# Redirect standard output/error to a cStringIO,
#so that PDBParser stops messing the output
parser = Bio.PDB.PDBParser()
err_fh = io.StringIO()
sys.stdout = err_fh
sys.stderr = err_fh
with open(new_pdb_fname, 'r') as rfh:
struct = parser.get_structure('query', rfh)
sys.stdout = sys.__stdout__
sys.stderr = sys.__stderr__
# Output formatted info about PDBParser's work to a log
s = err_fh.getvalue()
if s.strip():
logging.info(
"Structure parsing generated following error message(s): \n%s\n%s\n%s"
% ('-' * 120, s, '-' * 120))
# By default use only first model
model = struct[0]
del struct.child_list[1:]
# Check for discontinuities greater than 5 residues - warn about this _specifically_
for chain in model:
chid = chain.id
last_rid = None
# Curate disordered residues keeping only the last
chain.child_list = [residue for residue in chain]
chain.child_dict = dict((residue.id, residue) for residue in chain)
for residue in chain:
# Curate disordered atoms keeeping only the last
residue.child_list = [a for a in residue]
residue.child_dict = dict((a.id, a) for a in residue)
if last_rid is not None and rid > last_rid + 5:
rid = residue.id[1]
logging.warn(
"Residues %s:%s-%s:%s. Results might be inaccurate, as the break in a protein chain numbers more than 5 residues."
% (last_id, chain, rid, chain))
last_rid = rid
# Save structure without hydrogens
io = Bio.PDB.PDBIO()
io.set_structure(struct)
io.save(new_pdb_fname, NoHydroSelect())
# Run the preparation executable on the newly created PDB file
if (subprocess.call("%s %s %s 1>tmp.out 2>tmp.err" %
(prep_exec, new_pdb_fname, out_tmp_fname),
shell=True) != 0):
raise RuntimeError(
'Could not prepare corrected structure file for %s' %
pdb_fname)
# Fix the occupancies (creating the last and final temporary PDB file)
final_fn = "final.pdb"
#raw_input('WAITING...')
with open(out_tmp_fname, 'r') as rfh:
with open(final_fn, 'w') as wfh:
for line in rfh:
if line.startswith('ATOM'):
print >> wfh, line[:-1] + " 0.00 0.00 C"
else:
print >> wfh, line,
# Move the output file to the desired location
shutil.move(final_fn, out_pdb_fname)
return out_pdb_fname