def getRemotePDBHandle(self, id, rcsb_url=settings.rcsb_url):
"""
Get the coordinate file remotely from the RCSB.
@param id: pdb code, 4 characters
@type id: str
@param rcsb_url: template url for pdb download
(default: L{settings.rcsb_url})
@type rcsb_url: str
@return: the requested pdb file as a file handle
@rtype: open file handle
@raise PDBParserError: if couldn't retrieve PDB file
"""
try:
from Bio import File
except:
raise PDBParserError('Could not find Biopython - ' + \
'remote fetching of PDBs is not supported.')
handle = urllib.urlopen(rcsb_url % (id, id))
uhandle = File.UndoHandle(handle)
if not uhandle.peekline():
raise PDBParserError("Couldn't retrieve ", rcsb_url)
return uhandle
def parsePdbFromHandle(self, handle, first_model_only=True):
"""
Parse PDB from file/socket or string handle into memory.
@param handle: fresh open file/socket handle to PDB ressource or string
@type handle: open file-like object or str
@param first_model_only: only take first of many NMR models [True]
@type first_model_only: bool
@return: pdb file as list of strings, dictionary with resolution
@rtype: [str], {'resolution':float }
@raise PDBParserError: if passed in string is too short
"""
lines = []
res_match = None
infos = {}
if type(handle) is str:
if len(handle) < 5000:
raise PDBParserError("Couldn't extract PDB Info.")
handle = cStringIO.StringIO(handle)
## if handle.peekline()[:6] != 'TITLE':
## raise PDBParserError, 'Ressource does not seem to be a PDB:\n%r' %\
## handle.peekline()
for l in handle:
lines += [l]
res_match = res_match or self.ex_resolution.search(l)
if first_model_only and l[:6] == 'ENDMDL':
break
if len(lines) < 10 and '<div>' in lines[0]:
raise PDBParserError, 'No PDB found with this ID.'
if res_match:
if res_match.groups()[0] == 'NOT APPLICABLE':
infos['resolution'] = self.NMR_RESOLUTION
else:
infos['resolution'] = float(res_match.groups()[0])
else:
raise PDBParserError, 'No resolution record found in PDB.'
return lines, infos
def getLocalPDBHandle(self, id, db_path=settings.pdb_path):
"""
Get the coordinate file from a local pdb database.
@param id: pdb code, 4 characters
@type id: str
@param db_path: path to local pdb database
(default: L{settings.pdb_path})
@type db_path: str
@return: the requested pdb file as a file handle
@rtype: open file handle
@raise PDBParserError: if couldn't find PDB file
"""
id = str.lower(id)
filenames = [
os.path.join(db_path, '%s.pdb' % id), db_path + '/pdb%s.ent' % id,
db_path + '/%s/pdb%s.ent.Z' % (id[1:3], id)
]
for f in filenames:
if os.path.exists(f):
## gzipped pdb file
if f[-3:] == '.gz':
return gzip.open(f)
## the gzip module doesn't handle .Z files
## doesn't return open file handle
elif f[-2:] == '.Z':
p = subprocess.Popen(['gunzip', '-c', f],
stdout=subprocess.PIPE)
return p.communicate()[0]
## uncompressed
else:
return open(f)
raise PDBParserError("Couldn't find PDB file locally.")