def get_secstruct(pose):
""" Uses DSSP to get a secondary structure string for the given pose """
sec_struct = Dssp(pose)
sec_struct.insert_ss_into_pose(pose)
ss = str(pose.secstruct())
return ss
def __init__(self, pdbid, dist=8.0):
"""'dist' is the maximum distance between two residues to
consider them as part of an interface"""
self.dist = dist
self.pdbid = pdbid
pdb_prefix = '/wynton/home/database/pdb/remediated/pdb/'
self.path = os.path.join(pdb_prefix, self.pdbid[1:3], 'pdb{}.ent.gz'.format(
self.pdbid
))
# self.path = pdbid
self.pose = pose_from_file(self.path)
ss_str = Dssp(self.pose).get_dssp_secstruct()
self.secstruct = contiguous_secstruct(ss_str)
def bblock(self, pdbkey):
if isinstance(pdbkey, list):
return [self.bblock(f) for f in pdbkey]
if isinstance(pdbkey, (str, bytes)):
pdbkey = hash_str_to_int(pdbkey)
assert isinstance(pdbkey, int)
if not pdbkey in self._bblock_cache:
pdbfile = self._key_to_pdbfile[pdbkey]
pose = self.pose(pdbfile)
entry = self._dictdb[pdbfile]
ss = Dssp(pose).get_dssp_secstruct()
bblock = BBlock(entry, pdbfile, pdbkey, pose, ss)
self._bblock_cache[pdbkey] = bblock
return self._bblock_cache[pdbkey]
def build_pdb_data(self, entry, uselock=True):
"""return Nnew, Nmissing"""
pdbfile = entry['file']
pdbkey = hash_str_to_int(pdbfile)
cachefile = self.bblockfile(pdbkey)
posefile = self.posefile(pdbfile)
if os.path.exists(cachefile):
if not self.load_cached_bblock_into_memory(pdbkey):
if os.path.exists(cachefile):
raise ValueError(
f'cachefile {cachefile} exists, but cant load data from associated key {pdbkey}'
)
raise ValueError(
f'cachefile {cachefile} was removed, cant load data from associated key {pdbkey}'
)
if self.load_poses:
if not self.load_cached_pose_into_memory(pdbfile):
print('warning, not saved:', pdbfile)
return None, None # new, missing
elif self.read_new_pdbs:
if uselock: self.check_lock_cachedir()
read_pdb = False
# info('CachingBBlockDB.build_pdb_data reading %s' % pdbfile)
pose = self.pose(pdbfile)
ss = Dssp(pose).get_dssp_secstruct()
bblock = BBlock(entry, pdbfile, pdbkey, pose, ss)
self._bblock_cache[pdbkey] = bblock
# print(cachefile)
with open(cachefile, 'wb') as f:
pickle.dump(bblock._state, f)
# print('saved new bblock cache file', cachefile)
if not os.path.exists(posefile):
try:
with open(posefile, 'wb') as f:
pickle.dump(pose, f)
info('dumped _bblock_cache files for %s' % pdbfile)
except OSError as e:
print('not saving', posefile)
if self.load_poses:
self._poses_cache[pdbfile] = pose
return pdbfile, None # new, missing
else:
warning('no cached data for: ' + pdbfile)
return None, pdbfile # new, missing
def build_pdb_data(self, entry):
"""return Nnew, Nmissing
Args:
entry (TYPE): Description
Returns:
TYPE: Description
"""
pdbfile = entry['file']
cachefile = os.path.join(self.cachedir, 'bblock',
flatten_path(pdbfile))
posefile = self.posefile(pdbfile)
if os.path.exists(cachefile):
assert self.load_cached_bblock_into_memory(pdbfile)
if self.load_poses:
assert self.load_cached_pose_into_memory(pdbfile)
return None, None # new, missing
elif self.read_new_pdbs:
self.check_lock_cachedir()
read_pdb = False
# info('BBlockDB.build_pdb_data reading %s' % pdbfile)
pose = self.pose(pdbfile)
ss = Dssp(pose).get_dssp_secstruct()
bblock = BBlock(entry, pdbfile, pose, ss)
self._bblock_cache[pdbfile] = bblock
with open(cachefile, 'wb') as f:
pickle.dump(bblock._state, f)
if not os.path.exists(posefile):
with open(posefile, 'wb') as f:
pickle.dump(pose, f)
info('dumped _bblock_cache files for %s' % pdbfile)
if self.load_poses:
self._poses_cache[pdbfile] = pose
return pdbfile, None # new, missing
else:
warning('no cached data for: ' + pdbfile)
return None, pdbfile # new, missing
def __getitem__(self, index):
result = pose_from_file(self.data[index])
transformed = self.transform(result)
secondary = Dssp(transformed)
return (transformed, result.sequence(),
secondary.get_dssp_unreduced_secstruct())