def backupFile(filename, backup=None, backup_ext='.BAK', **kwargs):
"""Rename *filename* with *backup_ext* appended to its name for backup
purposes, if *backup* is **True** or if automatic backups is turned on
using :func:`.confProDy`. Default extension :file:`.BAK` is used when
one is not set using :func:`.confProDy`. If *filename* does not exist,
no action will be taken and *filename* will be returned. If file is
successfully renamed, new filename will be returned."""
try:
exists = isfile(filename)
except Exception as err:
raise TypeError('filename must be a string ({0})'.format(str(err)))
from prody import SETTINGS
if exists and (backup or SETTINGS.get('backup', False)):
if backup_ext == '.BAK':
backup_ext = SETTINGS.get('backup_ext', '.BAK')
bak = filename + backup_ext
if isfile(bak):
try:
os.remove(bak)
except Exception as err:
pass
try:
os.rename(filename, bak)
except Exception as err:
pass
return bak
else:
return filename
def pathPDBMirror(path=None, format=None):
"""Returns or specify PDB mirror path to be used by :func:`.fetchPDB`.
To release the current mirror, pass an invalid path, e.g. ``path=''``.
If you are keeping a partial mirror, such as PDB files in
:file:`/data/structures/divided/pdb/` folder, specify *format*, which is
``'pdb'`` in this case."""
if path is None:
path = SETTINGS.get('pdb_mirror_path')
format = SETTINGS.get('pdb_mirror_format', None)
if path:
if isdir(path):
if format is None:
return path
else:
return path, format
else:
LOGGER.warning('PDB mirror path {0} is not a accessible.'
.format(repr(path)))
else:
if isdir(path):
path = abspath(path)
LOGGER.info('Local PDB mirror path is set: {0}'
.format(repr(path)))
SETTINGS['pdb_mirror_path'] = path
SETTINGS['pdb_mirror_format'] = format
SETTINGS.save()
else:
current = SETTINGS.pop('pdb_mirror_path')
if current:
LOGGER.info('PDB mirror {0} is released.'
.format(repr(current)))
SETTINGS.save()
else:
raise IOError('{0} is not a valid path.'.format(repr(path)))
def backupFile(filename, backup=None, backup_ext='.BAK', **kwargs):
"""Rename *filename* with *backup_ext* appended to its name for backup
purposes, if *backup* is **True** or if automatic backups is turned on
using :func:`.confProDy`. Default extension :file:`.BAK` is used when
one is not set using :func:`.confProDy`. If *filename* does not exist,
no action will be taken and *filename* will be returned. If file is
successfully renamed, new filename will be returned."""
try:
exists = isfile(filename)
except Exception as err:
raise TypeError('filename must be a string ({0})'.format(str(err)))
from prody import SETTINGS
if exists and (backup or SETTINGS.get('backup', False)):
if backup_ext == '.BAK':
backup_ext = SETTINGS.get('backup_ext', '.BAK')
bak = filename + backup_ext
if isfile(bak):
try:
os.remove(bak)
except Exception as err:
pass
try:
os.rename(filename, bak)
except Exception as err:
pass
return bak
else:
return filename
def pathPDBMirror(path=None, format=None):
"""Returns or specify PDB mirror path to be used by :func:`.fetchPDB`.
To release the current mirror, pass an invalid path, e.g. ``path=''``.
If you are keeping a partial mirror, such as PDB files in
:file:`/data/structures/divided/pdb/` folder, specify *format*, which is
``'pdb'`` in this case."""
if path is None:
path = SETTINGS.get('pdb_mirror_path')
format = SETTINGS.get('pdb_mirror_format', None)
if path:
if isdir(path):
if format is None:
return path
else:
return path, format
else:
LOGGER.warning(
'PDB mirror path {0} is not a accessible.'.format(
repr(path)))
else:
if isdir(path):
path = abspath(path)
LOGGER.info('Local PDB mirror path is set: {0}'.format(repr(path)))
SETTINGS['pdb_mirror_path'] = path
SETTINGS['pdb_mirror_format'] = format
SETTINGS.save()
else:
current = SETTINGS.pop('pdb_mirror_path')
if current:
LOGGER.info('PDB mirror {0} is released.'.format(
repr(current)))
SETTINGS.save()
else:
raise IOError('{0} is not a valid path.'.format(repr(path)))
def updateDefinitions():
"""Update definitions and set some global variables. This function must be
called at the end of the module."""
global DEFINITIONS, AMINOACIDS, BACKBONE, TIMESTAMP
DEFINITIONS = {}
user = SETTINGS.get('flag_definitions', {})
# nucleics
nucleic = set()
for key in ['nucleobase', 'nucleoside', 'nucleotide']:
aset = set(user.get(key, DEFAULTS[key]))
nucleic.update(aset)
DEFINITIONS[key] = aset
DEFINITIONS['nucleic'] = nucleic
# heteros
for key in ['water', 'lipid', 'ion', 'sugar', 'heme',
'at', 'cg', 'purine', 'pyrimidine',]:
DEFINITIONS[key] = set(user.get(key, DEFAULTS[key]))
DEFINITIONS['backbone'] = DEFINITIONS['bb'] = set(user.get(key,
DEFAULTS['bb']))
DEFINITIONS['backbonefull'] = DEFINITIONS['bbfull'] = set(user.get(key,
DEFAULTS['bbfull']))
# element regex
for key in ['hydrogen', 'carbon', 'nitrogen', 'oxygen', 'sulfur']:
DEFINITIONS[key] = recompile(user.get(key, DEFAULTS[key]))
try:
nonstd = SETTINGS[NONSTANDARD_KEY]
except KeyError:
nonstd = NONSTANDARD
DEFINITIONS.update(CATEGORIZED)
else:
for cat in CATEGORIES:
for key in CATEGORIES[cat]:
DEFINITIONS[key] = set(DEFAULTS[key])
DEFINITIONS['charged'] = set(DEFINITIONS['acidic'])
DEFINITIONS['charged'].update(DEFINITIONS['basic'])
for resi, props in nonstd.iteritems():
for prop in props:
DEFINITIONS[prop].add(resi)
DEFINITIONS['stdaa'] = DEFAULTS['stdaa']
DEFINITIONS['nonstdaa'] = set(nonstd)
AMINOACIDS = set(DEFINITIONS['stdaa'])
AMINOACIDS.update(DEFINITIONS['nonstdaa'])
DEFINITIONS['protein'] = DEFINITIONS['aminoacid'] = AMINOACIDS
BACKBONE = DEFINITIONS['bb']
global TIMESTAMP
TIMESTAMP = SETTINGS.get('flag_timestamp', 0)
def getPDBLocalFolder():
"""Return the path to a local PDB folder and folder structure specifier.
If a local folder is not set, **None** will be returned."""
folder = SETTINGS.get('pdb_local_folder')
if folder:
if isdir(folder):
return folder, SETTINGS.get('pdb_local_divided', True)
else:
LOGGER.warning('PDB local folder {0:s} is not a accessible.'
.format(repr(folder)))
def addNonstdAminoacid(resname, *properties):
"""Add non-standard amino acid *resname* with *properties* selected from:
* {props}
>>> addNonstdAminoacid('PTR', 'acidic', 'aromatic', 'cyclic', 'large',
... 'polar', 'surface')
Default set of non-standard amino acids can be restored as follows:
>>> flagDefinition(reset='nonstdaa')"""
resname = str(resname)
if len(resname) > 4:
LOGGER.warn('Residue name {0:s} is unusually long.'
.format(repr(resname)))
propset = set(properties)
for cat, val in CATEGORIES.items():
intersection = val.intersection(propset)
if intersection:
if len(intersection) > 1:
raise ValueError('amino acid properties {0:s} cannot be '
'present together'
.format(', '.join([repr(prp) for prp in intersection])))
for prop in intersection:
propset.remove(prop)
if propset:
raise ValueError('amino acid property {0:s} is not valid'
.format(repr(propset.pop())))
nonstd = SETTINGS.get(NONSTANDARD_KEY, NONSTANDARD)
nonstd[resname] = set(properties)
updateNonstandard(nonstd)
def wwPDBServer(*key):
"""Set/get `wwPDB`_ FTP/HTTP server location used for downloading PDB
structures. Use one of the following keywords for setting a server:
+---------------------------+-----------------------------+
| wwPDB FTP server | *Key* (case insensitive) |
+===========================+=============================+
| RCSB PDB (USA) (default) | RCSB, USA, US |
+---------------------------+-----------------------------+
| PDBe (Europe) | PDBe, Europe, Euro, EU |
+---------------------------+-----------------------------+
| PDBj (Japan) | PDBj, Japan, Jp |
+---------------------------+-----------------------------+
.. _wwPDB: http://www.wwpdb.org/"""
if not key:
return SETTINGS.get('wwpdb', None)
elif len(key) == 1:
try:
key = key[0].lower()
except AttributeError:
raise TypeError('key must be a string')
if key in WWPDB_FTP_SERVERS:
SETTINGS['wwpdb'] = key
SETTINGS.save()
LOGGER.info('wwPDB server is set to {}.'
.format(WWPDB_FTP_SERVERS[key][0]))
else:
raise ValueError('{0} is not a valid wwPDB server identifier'
.format(repr(key)))
else:
raise TypeError('one wwPDB server identifier is expected, {0} given'
.format(len(key)))
def pathRhapsodyFolder(folder=None):
"""Returns or sets path of local folder where files and pickles necessary
to run Rhapsody will be stored. To release the current folder, pass an
invalid path, e.g. ``folder=''``.
"""
if folder is None:
folder = SETTINGS.get('rhapsody_local_folder')
if folder:
if isdir(folder):
return folder
else:
LOGGER.warn('Local folder {} is not accessible.'.format(
repr(folder)))
else:
if isdir(folder):
folder = abspath(folder)
LOGGER.info('Local Rhapsody folder is set: {}'.format(
repr(folder)))
SETTINGS['rhapsody_local_folder'] = folder
SETTINGS.save()
else:
current = SETTINGS.pop('rhapsody_local_folder')
if current:
LOGGER.info('Rhapsody folder {0} is released.'.format(
repr(current)))
SETTINGS.save()
else:
raise IOError('{} is not a valid path.'.format(repr(folder)))
def savePickle(self, folder=None, filename=None):
if folder is None:
# define folder where to look for pickles
folder = SETTINGS.get('rhapsody_local_folder', '.')
if filename is None:
# use the default filename, if possible
if self.PDBID is None:
# when a custom structure is used, there is no
# default filename: the user should provide it
raise ValueError('Please provide a filename.')
filename = 'PDBfeatures-' + self.PDBID + '.pkl'
pickle_path = os.path.join(folder, filename)
# do not store GNM and ANM instances.
# If a valid PDBID is present, do not store parsed PDB
# as well, since it can be easily fetched again
cache = (self._pdb, self._gnm, self._anm)
if self.PDBID is not None:
self._pdb = None
self._gnm = {}
self._anm = {}
for env in ['chain', 'reduced', 'sliced']:
self._gnm[env] = {chID: None for chID in self.chids}
self._anm[env] = {chID: None for chID in self.chids}
# write pickle
pickle.dump(self, open(pickle_path, "wb"))
# restore temporarily cached data
self._pdb, self._gnm, self._anm = cache
LOGGER.info("Pickle '{}' saved.".format(filename))
return pickle_path
def pathVMD(*path):
"""Returns VMD path, or set it to be a user specified *path*."""
if not path:
path = SETTINGS.get('vmd', None)
if isExecutable(path):
return path
else:
LOGGER.warning('VMD path is not set by user, looking for it.')
vmdbin = None
vmddir = None
if PLATFORM == 'Windows':
if PY3K:
import winreg
else:
import _winreg as winreg # PY3K: OK
for vmdversion in ('1.8.7', '1.9', '1.9.1'):
try:
key = winreg.OpenKey(
winreg.HKEY_LOCAL_MACHINE,
'Software\\University of Illinois\\VMD\\' +
vmdversion)
vmddir = winreg.QueryValueEx(key, 'VMDDIR')[0]
vmdbin = join(vmddir, 'vmd.exe')
except:
pass
try:
key = winreg.OpenKey(
winreg.HKEY_LOCAL_MACHINE,
'Software\\WOW6432node\\University of Illinois\\VMD\\'
+ vmdversion)
vmddir = winreg.QueryValueEx(key, 'VMDDIR')[0]
vmdbin = join(vmddir, 'vmd.exe')
except:
pass
else:
vmdbin = which('vmd')
if False:
pipe = os.popen('which vmd')
vmdbin = pipe.next().strip()
vmdfile = open(vmdbin)
for line in vmdfile:
if line.startswith('defaultvmddir='):
vmddir = line.split('=')[1].replace('"', '')
break
vmdfile.close()
if isExecutable(vmdbin):
setVMDpath(vmdbin)
return vmdbin
elif len(path) == 1:
path = path[0]
if isExecutable(path):
SETTINGS['vmd'] = path
SETTINGS.save()
LOGGER.info("VMD path is set to '{0}'.".format(path))
else:
raise OSError('{0} is not executable.'.format(str(path)))
else:
raise ValueError('specify a single path string')
def pathEVmutationFolder(folder=None):
"""Returns or sets path of local folder where EVmutation data are stored.
To release the current folder, pass an invalid path, e.g.
``folder=''``.
"""
if folder is None:
folder = SETTINGS.get('EVmutation_local_folder')
if folder:
if isdir(folder):
return folder
else:
LOGGER.warn('Local folder {} is not accessible.'.format(
repr(folder)))
else:
if isdir(folder):
folder = abspath(folder)
LOGGER.info('Local EVmutation folder is set: {}'.format(
repr(folder)))
SETTINGS['EVmutation_local_folder'] = folder
SETTINGS.save()
else:
current = SETTINGS.pop('EVmutation_local_folder')
if current:
LOGGER.info('EVmutation folder {0} is released.'.format(
repr(current)))
SETTINGS.save()
else:
raise IOError('{} is not a valid path.'.format(repr(folder)))
def changeDefinitions(**kwargs):
defs = SETTINGS.get(DEFINITIONS_KEY, {})
defs.update(kwargs)
SETTINGS[DEFINITIONS_KEY] = defs
SETTINGS[TIMESTAMP_KEY] = int(time())
SETTINGS.save()
updateDefinitions()
def pathPDBFolder(folder=None, divided=False):
"""Returns or specify local PDB folder for storing PDB files downloaded from
`wwPDB <http://www.wwpdb.org/>`_ servers. Files stored in this folder can
be accessed via :func:`.fetchPDB` from any working directory. To release
the current folder, pass an invalid path, e.g. ``folder=''``.
If *divided* is **True**, the divided folder structure of wwPDB servers
will be assumed when reading from and writing to the local folder. For
example, a structure with identifier **1XYZ** will be present as
:file:`pdblocalfolder/yz/pdb1xyz.pdb.gz`.
If *divided* is **False**, a plain folder structure will be expected and
adopted when saving files. For example, the same structure will be
present as :file:`pdblocalfolder/1xyz.pdb.gz`.
Finally, in either case, lower case letters will be used and compressed
files will be stored."""
if folder is None:
folder = SETTINGS.get('pdb_local_folder')
if folder:
if isdir(folder):
return folder, SETTINGS.get('pdb_local_divided', True)
else:
LOGGER.warn('PDB local folder {0} is not a accessible.'
.format(repr(folder)))
else:
if isdir(folder):
folder = abspath(folder)
LOGGER.info('Local PDB folder is set: {0}'.format(repr(folder)))
if divided:
LOGGER.info('wwPDB divided folder structure will be assumed.')
else:
LOGGER.info('A plain folder structure will be assumed.')
SETTINGS['pdb_local_folder'] = folder
SETTINGS['pdb_local_divided'] = bool(divided)
SETTINGS.save()
else:
current = SETTINGS.pop('pdb_local_folder')
if current:
LOGGER.info('PDB folder {0} is released.'
.format(repr(current)))
SETTINGS.pop('pdb_local_divided')
SETTINGS.save()
else:
raise IOError('{0} is not a valid path.'.format(repr(folder)))
def changeDefinitions(**kwargs):
defs = SETTINGS.get(DEFINITIONS_KEY, {})
defs.update(kwargs)
SETTINGS[DEFINITIONS_KEY] = defs
SETTINGS[TIMESTAMP_KEY] = int(time())
SETTINGS.save()
updateDefinitions()
def pathPDBFolder(folder=None, divided=False):
"""Returns or specify local PDB folder for storing PDB files downloaded from
`wwPDB <http://www.wwpdb.org/>`_ servers. Files stored in this folder can
be accessed via :func:`.fetchPDB` from any working directory. To release
the current folder, pass an invalid path, e.g. ``folder=''``.
If *divided* is **True**, the divided folder structure of wwPDB servers
will be assumed when reading from and writing to the local folder. For
example, a structure with identifier **1XYZ** will be present as
:file:`pdblocalfolder/yz/pdb1xyz.pdb.gz`.
If *divided* is **False**, a plain folder structure will be expected and
adopted when saving files. For example, the same structure will be
present as :file:`pdblocalfolder/1xyz.pdb.gz`.
Finally, in either case, lower case letters will be used and compressed
files will be stored."""
if folder is None:
folder = SETTINGS.get('pdb_local_folder')
if folder:
if isdir(folder):
return folder, SETTINGS.get('pdb_local_divided', True)
else:
LOGGER.warn('PDB local folder {0} is not a accessible.'.format(
repr(folder)))
else:
if isdir(folder):
folder = abspath(folder)
LOGGER.info('Local PDB folder is set: {0}'.format(repr(folder)))
if divided:
LOGGER.info('wwPDB divided folder structure will be assumed.')
else:
LOGGER.info('A plain folder structure will be assumed.')
SETTINGS['pdb_local_folder'] = folder
SETTINGS['pdb_local_divided'] = bool(divided)
SETTINGS.save()
else:
current = SETTINGS.pop('pdb_local_folder')
if current:
LOGGER.info('PDB folder {0} is released.'.format(
repr(current)))
SETTINGS.pop('pdb_local_divided')
SETTINGS.save()
else:
raise IOError('{0} is not a valid path.'.format(repr(folder)))
def pathVMD(*path):
"""Return VMD path, or set it to be a user specified *path*."""
if not path:
path = SETTINGS.get('vmd', None)
if isExecutable(path):
return path
else:
LOGGER.warning('VMD path is not set by user, looking for it.')
vmdbin = None
vmddir = None
if PLATFORM == 'Windows':
if PY3K:
import winreg
else:
import _winreg as winreg # PY3K: OK
for vmdversion in ('1.8.7', '1.9', '1.9.1'):
try:
key = winreg.OpenKey(winreg.HKEY_LOCAL_MACHINE,
'Software\\University of Illinois\\VMD\\' +
vmdversion)
vmddir = winreg.QueryValueEx(key, 'VMDDIR')[0]
vmdbin = join(vmddir, 'vmd.exe')
except:
pass
try:
key = winreg.OpenKey(winreg.HKEY_LOCAL_MACHINE,
'Software\\WOW6432node\\University of Illinois\\VMD\\' +
vmdversion)
vmddir = winreg.QueryValueEx(key, 'VMDDIR')[0]
vmdbin = join(vmddir, 'vmd.exe')
except:
pass
else:
vmdbin = which('vmd')
if False:
pipe = os.popen('which vmd')
vmdbin = pipe.next().strip()
vmdfile = open(vmdbin)
for line in vmdfile:
if line.startswith('defaultvmddir='):
vmddir = line.split('=')[1].replace('"', '')
break
vmdfile.close()
if isExecutable(vmdbin):
setVMDpath(vmdbin)
return vmdbin
elif len(path) == 1:
path = path[0]
if isExecutable(path):
SETTINGS['vmd'] = path
SETTINGS.save()
LOGGER.info("VMD path is set to '{0}'.".format(path))
else:
raise OSError('{0} is not executable.'.format(str(path)))
else:
raise ValueError('specify a single path string')
def getVMDpath():
"""Return VMD path set by user or one identified automatically."""
path = SETTINGS.get("vmd", None)
if isExecutable(path):
return path
else:
LOGGER.warning("VMD path is not set by user, looking for it.")
from types import StringType, UnicodeType
vmdbin = None
vmddir = None
if PLATFORM == "Windows":
import _winreg
for vmdversion in ("1.8.7", "1.9", "1.9.1"):
try:
key = _winreg.OpenKey(
_winreg.HKEY_LOCAL_MACHINE, "Software\\University of Illinois\\VMD\\" + vmdversion
)
vmddir = _winreg.QueryValueEx(key, "VMDDIR")[0]
vmdbin = os.path.join(vmddir, "vmd.exe")
except:
pass
try:
key = _winreg.OpenKey(
_winreg.HKEY_LOCAL_MACHINE, "Software\\WOW6432node\\University of Illinois\\VMD\\" + vmdversion
)
vmddir = _winreg.QueryValueEx(key, "VMDDIR")[0]
vmdbin = os.path.join(vmddir, "vmd.exe")
except:
pass
else:
vmdbin = which("vmd")
if False:
pipe = os.popen("which vmd")
vmdbin = pipe.next().strip()
vmdfile = open(vmdbin)
for line in vmdfile:
if line.startswith("defaultvmddir="):
vmddir = line.split("=")[1].replace('"', "")
break
vmdfile.close()
if (
False
and isinstance(vmdbin, (StringType, UnicodeType))
and isinstance(vmddir, (StringType, UnicodeType))
and os.path.isfile(vmdbin)
and os.path.isdir(vmddir)
):
pass # return vmdbin, vmddir
if isExecutable(vmdbin):
setVMDpath(vmdbin)
return vmdbin
def getPDBMirrorPath():
"""Return the path to a local PDB mirror, or **None** if a mirror path is
not set."""
path = SETTINGS.get('pdb_mirror_path')
if path:
if isdir(path):
return path
else:
LOGGER.warning('PDB mirror path {0:s} is not a accessible.'
.format(repr(path)))
def recoverPickle(self, filename=None, folder=None, days=30, **kwargs):
acc = self.uniq_acc
if acc is None:
# assume acc is equal to uniq_acc
acc = self.acc
if folder is None:
folder = SETTINGS.get('rhapsody_local_folder')
if folder is None:
folder = '.'
else:
folder = os.path.join(folder, 'pickles')
if filename is None:
# assume acc is equal to uniq_acc
acc = self.acc
filename = 'UniprotMap-' + acc + '.pkl'
pickle_path = os.path.join(folder, filename)
if not os.path.isfile(pickle_path):
# import unique accession number
acc = pd.queryUniprot(self.acc)['accession 0']
filename = 'UniprotMap-' + acc + '.pkl'
pickle_path = os.path.join(folder, filename)
else:
pickle_path = os.path.join(folder, filename)
# check if pickle exists
if not os.path.isfile(pickle_path):
raise IOError("File '{}' not found".format(filename))
# load pickle
recovered_self = pickle.load(open(pickle_path, "rb"))
if acc not in [recovered_self.acc, recovered_self.uniq_acc]:
raise ValueError('Accession number in recovered pickle (%s) ' %
recovered_self.uniq_acc + 'does not match.')
# check timestamp and ignore pickles that are too old
date_format = "%Y-%m-%d %H:%M:%S.%f"
t_old = datetime.datetime.strptime(recovered_self.timestamp,
date_format)
t_now = datetime.datetime.utcnow()
Delta_t = datetime.timedelta(days=days)
if t_old + Delta_t < t_now:
raise RuntimeError(
'Pickle {} was too old and was ignored.'.format(filename))
self.fullRecord = recovered_self.fullRecord
self.uniq_acc = recovered_self.uniq_acc
self.sequence = recovered_self.sequence
self.PDBrecords = recovered_self.PDBrecords
self.PDBmappings = recovered_self.PDBmappings
self.customPDBmappings = recovered_self.customPDBmappings
self._align_algo_args = recovered_self._align_algo_args
self._align_algo_kwargs = recovered_self._align_algo_kwargs
self.timestamp = recovered_self.timestamp
self.Pfam = recovered_self.Pfam
LOGGER.info("Pickle '{}' recovered.".format(filename))
return
def getNonstdProperties(resname):
"""Return properties of non-standard amino acid *resname*.
>>> getNonstdProperties('PTR')
['acidic', 'aromatic', 'cyclic', 'large', 'polar', 'surface']"""
try:
alist = list(SETTINGS.get(NONSTANDARD_KEY, NONSTANDARD)[resname])
except KeyError:
raise ValueError('{0:s} is not a non-standard residue name'
.format(repr(resname)))
else:
alist.sort()
return alist
def listNonstdAAProps(resname):
"""Returns properties of non-standard amino acid *resname*.
.. ipython:: python
listNonstdAAProps('PTR')"""
try:
alist = list(SETTINGS.get(NONSTANDARD_KEY, NONSTANDARD)[resname])
except KeyError:
raise ValueError('{0} is not a non-standard residue name'
.format(repr(resname)))
else:
alist.sort()
return alist
def savePickle(self, filename=None, folder=None, store_custom_PDBs=False):
if folder is None:
folder = SETTINGS.get('rhapsody_local_folder', '.')
if filename is None:
filename = 'UniprotMap-' + self.uniq_acc + '.pkl'
pickle_path = os.path.join(folder, filename)
cache = self.customPDBmappings
if store_custom_PDBs is not True:
# do not store alignments of custom PDBs
self.customPDBmappings = []
# save pickle
pickle.dump(self, open(pickle_path, "wb"))
self.customPDBmappings = cache
LOGGER.info("Pickle '{}' saved.".format(filename))
return pickle_path
def listNonstdAAProps(resname):
"""Returns properties of non-standard amino acid *resname*.
.. ipython:: python
listNonstdAAProps('PTR')"""
try:
alist = list(SETTINGS.get(NONSTANDARD_KEY, NONSTANDARD)[resname])
except KeyError:
raise ValueError('{0} is not a non-standard residue name'.format(
repr(resname)))
else:
alist.sort()
return alist
def savePickle(self, folder=None, filename=None):
"""Stores a pickle of the current class instance. The pickle will
contain all information and precomputed features, but not GNM and ANM
models. In case a PDBID is missing, the parsed PDB :class:`AtomGroup`
is stored as well.
:arg folder: path of the folder where the pickle will be saved. If not
specified, the local Rhapsody installation folder will be used.
:type folder: str
:arg filename: name of the pickle. By default, the pickle will be
saved as ``'PDBfeatures-[PDBID].pkl'``. If a PDBID is not defined,
the user must provide a filename.
:type filename: str
:return: pickle path
:rtype: str
"""
if folder is None:
# define folder where to look for pickles
folder = SETTINGS.get('rhapsody_local_folder')
if folder is None:
folder = '.'
else:
folder = os.path.join(folder, 'pickles')
if filename is None:
# use the default filename, if possible
if self.PDBID is None:
# when a custom structure is used, there is no
# default filename: the user should provide it
raise ValueError('Please provide a filename.')
filename = 'PDBfeatures-' + self.PDBID + '.pkl'
pickle_path = os.path.join(folder, filename)
# do not store GNM and ANM instances.
# If a valid PDBID is present, do not store parsed PDB
# as well, since it can be easily fetched again
cache = (self._pdb, self._gnm, self._anm)
if self.PDBID is not None:
self._pdb = None
self._gnm = {}
self._anm = {}
for env in ['chain', 'reduced', 'sliced']:
self._gnm[env] = {chID: None for chID in self.chids}
self._anm[env] = {chID: None for chID in self.chids}
# write pickle
pickle.dump(self, open(pickle_path, "wb"))
# restore temporarily cached data
self._pdb, self._gnm, self._anm = cache
LOGGER.info("Pickle '{}' saved.".format(filename))
return pickle_path
def getWWPDBFTPServer():
"""Return a tuple containing name, host, and path of the currently
set `wwPDB <http://www.wwpdb.org/>`_ FTP server."""
server = SETTINGS.get('wwpdb_ftp', None)
if server is None:
LOGGER.warning('A wwPDB FTP server is not set, default FTP server '
'RCSB PDB is used. Use `setWWPDBFTPServer` function '
'to set a server close to your location.')
return _WWPDB_RCSB
else:
if server[2].endswith('data/structures/divided/pdb/'):
return (server[0], server[1],
server[2][:-len('data/structures/divided/pdb/')])
else:
return server
def recoverPickle(self, folder=None, filename=None, days=30, **kwargs):
if folder is None:
# define folder where to look for pickles
folder = SETTINGS.get('rhapsody_local_folder', '.')
if filename is None:
# use the default filename, if possible
if self.PDBID is not None:
filename = 'PDBfeatures-' + self.PDBID + '.pkl'
else:
# when a custom structure is used, there is no
# default filename: the user should provide it
raise ValueError('Please provide a filename.')
pickle_path = os.path.join(folder, filename)
if not os.path.isfile(pickle_path):
raise IOError("File '{}' not found".format(filename))
recovered_self = pickle.load(open(pickle_path, "rb"))
# check consistency of recovered data
if self.PDBID is None:
if self._pdb != recovered_self._pdb:
raise ValueError(
'Incompatible PDB structure in recovered pickle.')
elif self.PDBID != recovered_self.PDBID:
raise ValueError(
'PDBID in recovered pickle ({}) does not match.'.format(
recovered_self.PDBID))
if self.n_modes != recovered_self.n_modes:
raise ValueError(
'Num. of modes in recovered pickle ({}) does not match.'.
format(recovered_self.n_modes))
# check timestamp and ignore pickles that are too old
date_format = "%Y-%m-%d %H:%M:%S.%f"
t_old = datetime.datetime.strptime(recovered_self.timestamp,
date_format)
t_now = datetime.datetime.utcnow()
Delta_t = datetime.timedelta(days=days)
if t_old + Delta_t < t_now:
raise RuntimeError('Pickle was too old and was ignored.')
# import recovered data
self.chids = recovered_self.chids
self.resids = recovered_self.resids
self.feats = recovered_self.feats
self._gnm = recovered_self._gnm
self._anm = recovered_self._anm
self.timestamp = recovered_self.timestamp
LOGGER.info("Pickle '{}' recovered.".format(filename))
return
def delNonstdAminoacid(resname):
"""Delete non-standard amino acid *resname*.
>>> delNonstdAminoacid('PTR')
>>> flagDefinition('nonstdaa') # doctest: +ELLIPSIS
['ASX', 'CSO', 'GLX', ..., 'TPO', 'XAA', 'XLE']
Default set of non-standard amino acids can be restored as follows:
>>> flagDefinition(reset='nonstdaa')"""
nonstd = SETTINGS.get(NONSTANDARD_KEY, NONSTANDARD)
try:
nonstd.pop(resname)
except KeyError:
raise ValueError('{0:s} is not a non-standard residue name'
.format(repr(resname)))
else:
updateNonstandard(nonstd)
def delNonstdAminoacid(resname):
"""Delete non-standard amino acid *resname*.
.. ipython:: python
delNonstdAminoacid('PTR')
flagDefinition('nonstdaa')
Default set of non-standard amino acids can be restored as follows:
.. ipython:: python
flagDefinition(reset='nonstdaa')"""
nonstd = SETTINGS.get(NONSTANDARD_KEY, NONSTANDARD)
try:
nonstd.pop(resname)
except KeyError:
raise ValueError('{0} is not a non-standard residue name'
.format(repr(resname)))
else:
updateNonstandard(nonstd)
def delNonstdAminoacid(resname):
"""Delete non-standard amino acid *resname*.
.. ipython:: python
delNonstdAminoacid('PTR')
flagDefinition('nonstdaa')
Default set of non-standard amino acids can be restored as follows:
.. ipython:: python
flagDefinition(reset='nonstdaa')"""
nonstd = SETTINGS.get(NONSTANDARD_KEY, NONSTANDARD)
try:
nonstd.pop(resname)
except KeyError:
raise ValueError('{0} is not a non-standard residue name'.format(
repr(resname)))
else:
updateNonstandard(nonstd)
def addNonstdAminoacid(resname, *properties):
"""Add non-standard amino acid *resname* with *properties* selected from:
* {props}
.. ipython:: python
addNonstdAminoacid('PTR', 'acidic', 'aromatic', 'cyclic', 'large',
'polar', 'surface')
Default set of non-standard amino acids can be restored as follows:
.. ipython:: python
flagDefinition(reset='nonstdaa')"""
resname = str(resname)
if len(resname) > 4:
LOGGER.warn('Residue name {0} is unusually long.'.format(
repr(resname)))
propset = set(properties)
for cat, val in CATEGORIES.items():
intersection = val.intersection(propset)
if intersection:
if len(intersection) > 1:
raise ValueError('amino acid properties {0} cannot be '
'present together'.format(', '.join(
[repr(prp) for prp in intersection])))
for prop in intersection:
propset.remove(prop)
if propset:
raise ValueError('amino acid property {0} is not valid'.format(
repr(propset.pop())))
nonstd = SETTINGS.get(NONSTANDARD_KEY, NONSTANDARD)
nonstd[resname] = set(properties)
updateNonstandard(nonstd)
def parsePDBStream(stream, **kwargs):
"""Returns an :class:`.AtomGroup` and/or dictionary containing header data
parsed from a stream of PDB lines.
:arg stream: Anything that implements the method ``readlines``
(e.g. :class:`file`, buffer, stdin)"""
model = kwargs.get('model')
header = kwargs.get('header', False)
assert isinstance(header, bool), 'header must be a boolean'
chain = kwargs.get('chain')
subset = kwargs.get('subset')
altloc = kwargs.get('altloc', 'A')
if model is not None:
if isinstance(model, Integral):
if model < 0:
raise ValueError('model must be greater than 0')
else:
raise TypeError('model must be an integer, {0} is invalid'
.format(str(model)))
title_suffix = ''
if subset:
try:
subset = _PDBSubsets[subset.lower()]
except AttributeError:
raise TypeError('subset must be a string')
except KeyError:
raise ValueError('{0} is not a valid subset'
.format(repr(subset)))
title_suffix = '_' + subset
if chain is not None:
if not isinstance(chain, str):
raise TypeError('chain must be a string')
elif len(chain) == 0:
raise ValueError('chain must not be an empty string')
title_suffix = chain + title_suffix
ag = kwargs.pop('ag', None)
if ag is not None:
if not isinstance(ag, AtomGroup):
raise TypeError('ag must be an AtomGroup instance')
n_csets = ag.numCoordsets()
elif model != 0:
ag = AtomGroup(str(kwargs.get('title', 'Unknown')) + title_suffix)
n_csets = 0
biomol = kwargs.get('biomol', False)
auto_secondary = None
secondary = kwargs.get('secondary')
if not secondary:
auto_secondary = SETTINGS.get('auto_secondary')
secondary = auto_secondary
split = 0
hd = None
if model != 0:
LOGGER.timeit()
try:
lines = stream.readlines()
except AttributeError as err:
try:
lines = stream.read().split('\n')
except AttributeError:
raise err
if not len(lines):
raise ValueError('empty PDB file or stream')
if header or biomol or secondary:
hd, split = getHeaderDict(lines)
_parsePDBLines(ag, lines, split, model, chain, subset, altloc)
if ag.numAtoms() > 0:
LOGGER.report('{0} atoms and {1} coordinate set(s) were '
'parsed in %.2fs.'.format(ag.numAtoms(),
ag.numCoordsets() - n_csets))
else:
ag = None
LOGGER.warn('Atomic data could not be parsed, please '
'check the input file.')
elif header:
hd, split = getHeaderDict(stream)
if ag is not None and isinstance(hd, dict):
if secondary:
if auto_secondary:
try:
ag = assignSecstr(hd, ag)
except ValueError:
pass
else:
ag = assignSecstr(hd, ag)
if biomol:
ag = buildBiomolecules(hd, ag)
if isinstance(ag, list):
LOGGER.info('Biomolecular transformations were applied, {0} '
'biomolecule(s) are returned.'.format(len(ag)))
else:
LOGGER.info('Biomolecular transformations were applied to the '
'coordinate data.')
if model != 0:
if header:
return ag, hd
else:
return ag
else:
return hd
def parsePDBStream(stream, **kwargs):
"""Returns an :class:`.AtomGroup` and/or dictionary containing header data
parsed from a stream of PDB lines.
:arg stream: Anything that implements the method ``readlines``
(e.g. :class:`file`, buffer, stdin)"""
model = kwargs.get('model')
header = kwargs.get('header', False)
assert isinstance(header, bool), 'header must be a boolean'
chain = kwargs.get('chain')
subset = kwargs.get('subset')
altloc = kwargs.get('altloc', 'A')
if model is not None:
if isinstance(model, Integral):
if model < 0:
raise ValueError('model must be greater than 0')
else:
raise TypeError('model must be an integer, {0} is invalid'.format(
str(model)))
title_suffix = ''
if subset:
try:
subset = _PDBSubsets[subset.lower()]
except AttributeError:
raise TypeError('subset must be a string')
except KeyError:
raise ValueError('{0} is not a valid subset'.format(repr(subset)))
title_suffix = '_' + subset
if chain is not None:
if not isinstance(chain, str):
raise TypeError('chain must be a string')
elif len(chain) == 0:
raise ValueError('chain must not be an empty string')
title_suffix = chain + title_suffix
ag = None
if 'ag' in kwargs:
ag = kwargs['ag']
if not isinstance(ag, AtomGroup):
raise TypeError('ag must be an AtomGroup instance')
n_csets = ag.numCoordsets()
elif model != 0:
ag = AtomGroup(str(kwargs.get('title', 'Unknown')) + title_suffix)
n_csets = 0
biomol = kwargs.get('biomol', False)
auto_secondary = None
secondary = kwargs.get('secondary')
if not secondary:
auto_secondary = SETTINGS.get('auto_secondary')
secondary = auto_secondary
split = 0
hd = None
if model != 0:
LOGGER.timeit()
try:
lines = stream.readlines()
except AttributeError as err:
try:
lines = stream.read().split('\n')
except AttributeError:
raise err
if not len(lines):
raise ValueError('empty PDB file or stream')
if header or biomol or secondary:
hd, split = getHeaderDict(lines)
_parsePDBLines(ag, lines, split, model, chain, subset, altloc)
if ag.numAtoms() > 0:
LOGGER.report('{0} atoms and {1} coordinate set(s) were '
'parsed in %.2fs.'.format(
ag.numAtoms(),
ag.numCoordsets() - n_csets))
else:
ag = None
LOGGER.warn('Atomic data could not be parsed, please '
'check the input file.')
elif header:
hd, split = getHeaderDict(stream)
if ag is not None and isinstance(hd, dict):
if secondary:
if auto_secondary:
try:
ag = assignSecstr(hd, ag)
except ValueError:
pass
else:
ag = assignSecstr(hd, ag)
if biomol:
ag = buildBiomolecules(hd, ag)
if isinstance(ag, list):
LOGGER.info('Biomolecular transformations were applied, {0} '
'biomolecule(s) are returned.'.format(len(ag)))
else:
LOGGER.info('Biomolecular transformations were applied to the '
'coordinate data.')
if model != 0:
if header:
return ag, hd
else:
return ag
else:
return hd
def calcEVmutPathClasses(EVmut_score):
c = -SETTINGS.get('EVmutation_metrics')['optimal cutoff']
EVmut_class = np.where(EVmut_score < c, 'deleterious', 'neutral')
EVmut_class[np.isnan(EVmut_score)] = '?'
return EVmut_class
def print_sat_mutagen_figure(filename,
rhapsody_obj,
res_interval=None,
PolyPhen2=True,
EVmutation=True,
extra_plot=None,
fig_height=8,
fig_width=None,
dpi=300,
min_interval_size=15,
html=False,
main_clsf='main',
aux_clsf='aux.'):
# check inputs
assert isinstance(filename, str), 'filename must be a string'
assert isinstance(rhapsody_obj, Rhapsody), 'not a Rhapsody object'
assert rhapsody_obj._isColSet('main score'), 'predictions not found'
assert rhapsody_obj._isSaturationMutagenesis(), 'unable to create figure'
if res_interval is not None:
assert isinstance(res_interval, tuple) and len(res_interval) == 2, \
'res_interval must be a tuple of 2 values'
assert res_interval[1] >= res_interval[0], 'invalid res_interval'
if extra_plot is not None:
assert len(extra_plot) == rhapsody_obj.numSAVs, \
'length of additional predictions array is incorrect'
assert isinstance(fig_height, (int, float))
assert isinstance(dpi, int)
matplotlib = _try_import_matplotlib()
if matplotlib is None:
return
# delete extension from filename
filename = os.path.splitext(filename)[0]
# make sure that all variants belong to the same Uniprot sequence
accs = [s.split()[0] for s in rhapsody_obj.data['SAV coords']]
if len(set(accs)) != 1:
m = 'Only variants from a single Uniprot sequence can be accepted'
raise ValueError(m)
# select an appropriate interval, based on available predictions
seq_pos = [int(s.split()[1]) for s in rhapsody_obj.data['SAV coords']]
res_min = np.min(seq_pos)
res_max = np.max(seq_pos)
upper_lim = res_max + min_interval_size
# create empty (20 x num_res) mutagenesis tables
table_best = np.zeros((20, upper_lim), dtype=float)
table_best[:] = 'nan'
table_main = table_best.copy()
if extra_plot is not None:
table_other = table_best.copy()
if PolyPhen2:
table_PP2 = table_best.copy()
if EVmutation:
table_EVmut = table_best.copy()
# import pathogenicity probabilities from Rhapsody object
p_best = rhapsody_obj.getPredictions(classifier='best')['path. prob.']
p_main = rhapsody_obj.data['main path. prob.']
if PolyPhen2:
rhapsody_obj._calcPolyPhen2Predictions()
p_PP2 = rhapsody_obj.data['PolyPhen-2 score']
if EVmutation:
rhapsody_obj._calcEVmutationPredictions()
EVmut_score = np.array(rhapsody_obj.data['EVmutation score'])
EVmut_cutoff = SETTINGS.get('EVmutation_metrics')['optimal cutoff']
p_EVmut = -EVmut_score / EVmut_cutoff * 0.5
# fill tables with predicted probability
# 1: deleterious
# 0: neutral
# 'nan': no prediction/wt
aa_list = 'ACDEFGHIKLMNPQRSTVWY'
aa_map = {aa: i for i, aa in enumerate(aa_list)}
for i, SAV in enumerate(rhapsody_obj.data['SAV coords']):
aa_mut = SAV.split()[3]
index = int(SAV.split()[1]) - 1
table_best[aa_map[aa_mut], index] = p_best[i]
table_main[aa_map[aa_mut], index] = p_main[i]
if extra_plot is not None:
table_other[aa_map[aa_mut], index] = extra_plot[i]
if PolyPhen2:
table_PP2[aa_map[aa_mut], index] = p_PP2[i]
if EVmutation:
table_EVmut[aa_map[aa_mut], index] = p_EVmut[i]
# compute average pathogenicity profiles
# NB: I expect to see RuntimeWarnings in this block
with warnings.catch_warnings():
warnings.simplefilter("ignore", category=RuntimeWarning)
avg_p_best = np.nanmean(table_best, axis=0)
avg_p_main = np.nanmean(table_main, axis=0)
min_p = np.nanmin(table_best, axis=0)
max_p = np.nanmax(table_best, axis=0)
if extra_plot is not None:
avg_p_other = np.nanmean(table_other, axis=0)
if PolyPhen2:
avg_p_PP2 = np.nanmean(table_PP2, axis=0)
if EVmutation:
avg_p_EVmut = np.nanmean(table_EVmut, axis=0)
# use upper strip for showing additional info, such as PDB lengths
upper_strip = np.zeros((1, upper_lim))
upper_strip[:] = 'nan'
PDB_sizes = np.zeros(upper_lim, dtype=int)
PDB_coords = [''] * upper_lim
for s in rhapsody_obj.data:
index = int(s['SAV coords'].split()[1]) - 1
if s['PDB size'] != 0:
PDB_length = int(s['PDB size'])
PDBID_chain = ':'.join(s['PDB SAV coords'][0].split()[:2])
upper_strip[0, index] = PDB_length
PDB_sizes[index] = PDB_length
PDB_coords[index] = PDBID_chain
max_PDB_size = max(PDB_sizes)
if max_PDB_size != 0:
upper_strip[0, :] /= max_PDB_size
# PLOT FIGURE
from matplotlib import pyplot as plt
from matplotlib import gridspec as gridspec
# portion of the sequence to display
if res_interval is None:
res_interval = (res_min, res_max)
# adjust interval
res_i, res_f = _adjust_res_interval(res_interval, upper_lim,
min_interval_size)
nres_shown = res_f - res_i + 1
# figure proportions
if fig_width is None:
fig_width = fig_height / 2 # inches
fig_width *= nres_shown / 20
fig, ax = plt.subplots(3, 2, figsize=(fig_width, fig_height))
wspace = 0.5 # inches
plt.subplots_adjust(wspace=wspace / fig_width, hspace=0.15)
# figure structure
gs = gridspec.GridSpec(3,
2,
width_ratios=[nres_shown, 1],
height_ratios=[1, 20, 10])
ax0 = plt.subplot(gs[0, 0]) # secondary structure strip
ax1 = plt.subplot(gs[1, 0]) # mutagenesis table
axcb = plt.subplot(gs[1, 1]) # colorbar
ax2 = plt.subplot(gs[2, 0]) # average profile
# padding for tick labels
pad = 0.2 / fig_width
# top strip
matplotlib.cm.YlGn.set_bad(color='antiquewhite')
ax0.imshow(upper_strip[0:1, res_i - 1:res_f],
aspect='auto',
cmap='YlGn',
vmin=0,
vmax=1)
ax0.set_ylim((-0.45, .45))
ax0.set_yticks([])
ax0.set_ylabel(f'PDB size \n[0-{max_PDB_size} res] ',
fontsize=14,
ha='right',
va='center',
rotation=0)
ax0.set_xticks(np.arange(5 - res_i % 5, res_f - res_i + 1, 5))
ax0.set_xticklabels([])
# add white grid
ax0.set_xticks(np.arange(-.5, res_f - res_i + 1, 1), minor=True)
ax0.tick_params(axis='both', which='minor', length=0)
ax0.grid(which='minor', color='w', linestyle='-', linewidth=.5)
# mutagenesis table (heatmap)
matplotlib.cm.coolwarm.set_bad(color='#F9E79F')
im = ax1.imshow(table_best[:, res_i - 1:res_f],
aspect='auto',
cmap='coolwarm',
vmin=0,
vmax=1)
axcb.figure.colorbar(im, cax=axcb)
ax1.set_yticks(np.arange(len(aa_list)))
ax1.set_yticklabels(aa_list, ha='center', position=(-pad, 0), fontsize=14)
ax1.set_xticks(np.arange(5 - res_i % 5, res_f - res_i + 1, 5))
ax1.set_xticklabels([])
ax1.set_ylabel('pathog. probability', labelpad=10)
# add white grid
ax1.set_xticks(np.arange(-.5, res_f - res_i + 1, 1), minor=True)
ax1.set_yticks(np.arange(-.5, 20, 1), minor=True)
ax1.tick_params(axis='both', which='minor', length=0)
ax1.grid(which='minor', color='w', linestyle='-', linewidth=.5)
# average pathogenicity profile
x_resids = np.arange(1, upper_lim + 1)
# shading showing range of values
# NB: a bug in pyplot.fill_between() arises when selecting a region with
# set_xlim() in a large plot (e.g. > 1000), causing the shaded area to
# be plotted even though it's outside the selected region. As a workaround,
# here I slice the plot to fit the selected region.
sl = slice(max(0, res_i - 2), min(res_f + 2, upper_lim + 1))
ax2.fill_between(x_resids[sl],
min_p[sl],
max_p[sl],
alpha=0.5,
edgecolor='salmon',
facecolor='salmon')
# plot average profile for other predictions, if available
if extra_plot is not None:
ax2.plot(x_resids, avg_p_other, color='gray', lw=1)
if PolyPhen2:
ax2.plot(x_resids, avg_p_PP2, color='blue', lw=1)
if EVmutation:
ax2.plot(x_resids, avg_p_EVmut, color='green', lw=1)
# solid line for predictions obtained with full classifier
ax2.plot(x_resids, avg_p_main, 'ro-')
# dotted line for predictions obtained with auxiliary classifier
ax2.plot(x_resids, avg_p_best, 'ro-', markerfacecolor='none', ls='dotted')
# cutoff line
ax2.axhline(y=0.5, color='grey', lw=.8, linestyle='dashed')
ax2.set_xlim((res_i - .5, res_f + .5))
ax2.set_xlabel('residue number')
ax2.set_ylim((-0.05, 1.05))
ax2.set_ylabel('average', rotation=90, labelpad=10)
ax2.set_yticklabels([])
ax2r = ax2.twinx()
ax2r.set_ylim((-0.05, 1.05))
ax2r.set_yticks([0, .5, 1])
ax2r.set_yticklabels(['0', '0.5', '1'])
ax2r.tick_params(axis='both', which='major', pad=15)
tight_padding = 0.1
fig.savefig(filename + '.png',
format='png',
bbox_inches='tight',
pad_inches=tight_padding,
dpi=dpi)
plt.close()
plt.rcParams.update(plt.rcParamsDefault)
LOGGER.info(f'Saturation mutagenesis figure saved to {filename}.png')
# write a map in html format, to make figure clickable
if html:
all_axis = {'strip': ax0, 'table': ax1, 'bplot': ax2}
# precompute some useful quantities for html code
html_data = {}
# dpi of printed figure
html_data["dpi"] = dpi
# figure size *before* tight
html_data["fig_size"] = fig.get_size_inches()
# tight bbox as used by fig.savefig()
html_data["tight_bbox"] = fig.get_tightbbox(fig.canvas.get_renderer())
# compute new origin and height, based on tight box and padding
html_data["new_orig"] = html_data["tight_bbox"].min - tight_padding
html_data["new_height"] = (html_data["tight_bbox"].height +
2 * tight_padding)
def get_area_coords(ax, d):
assert ax_type in ("strip", "table", "bplot")
# get bbox coordinates (x0, y0, x1, y1)
bbox = ax.get_position().get_points()
# get bbox coordinates in inches
b_inch = bbox * d["fig_size"]
# adjust bbox coordinates based on tight bbox
b_adj = b_inch - d["new_orig"]
# use html reference system (y = 1 - y)
b_html = b_adj * np.array([1, -1]) + np.array([0, d["new_height"]])
# convert to pixels
b_px = (d["dpi"] * b_html).astype(int)
b_px = np.sort(b_px, axis=0)
# put in html format
coords = '{},{},{},{}'.format(*b_px.flatten())
# output
return coords
# html templates
area_html = Template('<area shape="rect" coords="$coords" '
'id="{{map_id}}_$areaid" {{area_attrs}}> \n')
# write html
with open(filename + '.html', 'w') as f:
f.write('<div>\n')
f.write('<map name="{{map_id}}" id="{{map_id}}" {{map_attrs}}>\n')
for ax_type, ax in all_axis.items():
fields = {'areaid': ax_type}
fields['coords'] = get_area_coords(ax, html_data)
f.write(area_html.substitute(fields))
f.write('</map>\n')
f.write('</div>\n')
# populate info table that will be passed as a javascript variable
best_preds = rhapsody_obj.getPredictions()
best_avg_preds = rhapsody_obj.getResAvgPredictions()
PDB_coords = rhapsody_obj.getPDBcoords()
abbrev = {
'?': '?',
'deleterious': 'del',
'neutral': 'neu',
'prob.delet.': 'p.del',
'prob.neutral': 'p.neu'
}
info = {}
for k in ['strip', 'table', 'bplot']:
n_cols = 20 if k == 'table' else 1
info[k] = [[''] * nres_shown for i in range(n_cols)]
for i, row in enumerate(rhapsody_obj.data):
SAV = row['SAV coords']
acc, resid, aa_wt, aa_mut = SAV.split()
resid = int(resid)
# consider only residues shown in figure
if not (res_i <= resid <= res_f):
continue
# SAV coordinates
SAV_code = f'{aa_wt}{resid}{aa_mut}'
# coordinates on table
t_i = aa_map[aa_mut]
t_j = resid - 1
# coordinates on *shown* table
ts_i = t_i
ts_j = resid - res_i
# compose message for table
bp = best_preds[i]
pprob = bp['path. prob.']
pclass = bp['path. class']
clsf = main_clsf if row['best classifier'] == 'main' else aux_clsf
m = f'{SAV_code}: Rhapsody-{clsf} = {pprob:<3.2f} ({pclass})'
if PolyPhen2:
score = bp['PolyPhen-2 score']
pclass = abbrev[bp['PolyPhen-2 path. class']]
m += f', PolyPhen-2 = {score:<3.2f} ({pclass})'
if EVmutation:
score = bp['EVmutation score']
pclass = abbrev[bp['EVmutation path. class']]
m += f', EVmutation = {score:<3.2f} ({pclass})'
if extra_plot is not None:
score = table_other[t_i, t_j]
m += f', other = {score:<3.2f}'
info['table'][ts_i][ts_j] = m
info['table'][aa_map[aa_wt]][ts_j] = f'{SAV_code[:-1]}: wild-type'
if i % 19 == 0:
# compose message for upper strip
PDBID, ch, resid, aa, size = PDB_coords[i][[
'PDBID', 'chain', 'resid', 'resname', 'PDB size'
]]
if size > 0:
m = f'{PDBID}:{ch}, resid {resid}, aa {aa}, size {size}'
else:
m = 'no PDB found'
info['strip'][0][ts_j] = m
# compose message for bottom plot (residue-averages)
bap = best_avg_preds[int(i / 19)]
pprob = bap['path. prob.']
pcl = bap['path. class']
m = f'{SAV_code[:-1]}: Rhapsody-{clsf} = {pprob:<3.2f} ({pcl})'
if PolyPhen2:
score = bap['PolyPhen-2 score']
pcl = abbrev[bap['PolyPhen-2 path. class']]
m += f', PolyPhen-2 = {score:<3.2f} ({pcl})'
if EVmutation:
score = bap['EVmutation score']
pcl = abbrev[bap['EVmutation path. class']]
m += f', EVmutation = {score:<3.2f} ({pcl})'
if extra_plot is not None:
score = avg_p_other[t_j]
m += f', other = {score:<3.2f}'
info['bplot'][0][ts_j] = m
def create_info_msg(ax_type, d):
text = '[ \n'
for row in d:
text += ' ['
for m in row:
text += f'"{m}",'
text += '], \n'
text += ']'
return text
area_js = Template('{{map_data}}["{{map_id}}_$areaid"] = { \n'
' "img_id": "{{img_id}}", \n'
' "map_id": "{{map_id}}", \n'
' "coords": [$coords], \n'
' "num_rows": $num_rows, \n'
' "num_cols": $num_cols, \n'
' "info_msg": $info_msg, \n'
'}; \n')
# dump info in javascript format
with open(filename + '.js', 'w') as f:
f.write('var {{map_data}} = {}; \n')
for ax_type, d in info.items():
vars = {'areaid': ax_type}
vars['coords'] = get_area_coords(all_axis[ax_type], html_data)
vars['num_rows'] = 20 if ax_type == 'table' else 1
vars['num_cols'] = nres_shown
vars['info_msg'] = create_info_msg(ax_type, d)
f.write(area_js.substitute(vars))
return info
return
def updateDefinitions():
"""Update definitions and set some global variables. This function must be
called at the end of the module."""
global DEFINITIONS, AMINOACIDS, BACKBONE, TIMESTAMP
DEFINITIONS = {}
user = SETTINGS.get('flag_definitions', {})
# nucleics
nucleic = set()
for key in ['nucleobase', 'nucleoside', 'nucleotide']:
aset = set(user.get(key, DEFAULTS[key]))
nucleic.update(aset)
DEFINITIONS[key] = aset
DEFINITIONS['nucleic'] = nucleic
# heteros
for key in [
'water', 'lipid', 'ion', 'sugar', 'heme', 'at', 'cg', 'purine',
'pyrimidine'
]:
DEFINITIONS[key] = set(user.get(key, DEFAULTS[key]))
DEFINITIONS['backbone'] = DEFINITIONS['bb'] = set(
user.get(key, DEFAULTS['bb']))
DEFINITIONS['backbonefull'] = DEFINITIONS['bbfull'] = set(
user.get(key, DEFAULTS['bbfull']))
# element regex
for key in ['hydrogen', 'carbon', 'nitrogen', 'oxygen', 'sulfur']:
DEFINITIONS[key] = recompile(user.get(key, DEFAULTS[key]))
try:
nonstd = SETTINGS[NONSTANDARD_KEY]
except KeyError:
nonstd = NONSTANDARD
DEFINITIONS.update(CATEGORIZED)
else:
for cat in CATEGORIES:
for key in CATEGORIES[cat]:
DEFINITIONS[key] = set(DEFAULTS[key])
DEFINITIONS['charged'] = set(DEFINITIONS['acidic'])
DEFINITIONS['charged'].update(DEFINITIONS['basic'])
for resi, props in nonstd.items():
for prop in props:
DEFINITIONS[prop].add(resi)
DEFINITIONS['stdaa'] = DEFAULTS['stdaa']
DEFINITIONS['nonstdaa'] = set(nonstd)
AMINOACIDS = set(DEFINITIONS['stdaa'])
AMINOACIDS.update(DEFINITIONS['nonstdaa'])
DEFINITIONS['protein'] = DEFINITIONS['aminoacid'] = AMINOACIDS
BACKBONE = DEFINITIONS['bb']
global TIMESTAMP
TIMESTAMP = SETTINGS.get(TIMESTAMP_KEY, 0)
def recoverPickle(self, folder=None, filename=None, days=30, **kwargs):
"""Looks for precomputed pickle for the current PDB structure.
:arg folder: path of folder where pickles are stored. If not specified,
pickles will be searched for in the local Rhapsody installation
folder.
:type folder: str
:arg filename: name of the pickle. If not specified, the default
filename ``'PDBfeatures-[PDBID].pkl'`` will be used. If a PDBID is
not found, user must specify a valid filename.
:type filename: str
:arg days: number of days after which a pickle will be considered too
old and won't be recovered.
:type days: int
"""
if folder is None:
# define folder where to look for pickles
folder = SETTINGS.get('rhapsody_local_folder')
if folder is None:
folder = '.'
else:
folder = os.path.join(folder, 'pickles')
if filename is None:
# use the default filename, if possible
if self.PDBID is not None:
filename = 'PDBfeatures-' + self.PDBID + '.pkl'
else:
# when a custom structure is used, there is no
# default filename: the user should provide it
raise ValueError('Please provide a filename.')
pickle_path = os.path.join(folder, filename)
if not os.path.isfile(pickle_path):
raise IOError("File '{}' not found".format(filename))
recovered_self = pickle.load(open(pickle_path, "rb"))
# check consistency of recovered data
if self.PDBID is None:
if self._pdb != recovered_self._pdb:
raise ValueError('Incompatible PDB structure in recovered pickle.')
elif self.PDBID != recovered_self.PDBID:
raise ValueError('PDBID in recovered pickle ({}) does not match.'
.format(recovered_self.PDBID))
if self.n_modes != recovered_self.n_modes:
raise ValueError('Num. of modes in recovered pickle ({}) does not match.'
.format(recovered_self.n_modes))
# check timestamp and ignore pickles that are too old
date_format = "%Y-%m-%d %H:%M:%S.%f"
t_old = datetime.datetime.strptime(
recovered_self.timestamp, date_format)
t_now = datetime.datetime.utcnow()
Delta_t = datetime.timedelta(days=days)
if t_old + Delta_t < t_now:
raise RuntimeError('Pickle was too old and was ignored.')
# import recovered data
self.chids = recovered_self.chids
self.resids = recovered_self.resids
self.feats = recovered_self.feats
self._gnm = recovered_self._gnm
self._anm = recovered_self._anm
self.timestamp = recovered_self.timestamp
LOGGER.info("Pickle '{}' recovered.".format(filename))
return
def recoverEVmutFeatures(SAVs):
"""Compute EVmutation features by fetching precomputed scores from the
downloaded local folder. If multiple values are found for a given variant,
the average will be taken.
:arg SAVs: list of SAV coordinates, e.g. ``'P17516 135 G E'``.
:type SAVs: list or tuple of strings
:return: an array of EVmutation features for each SAV
:rtype: NumPy structured array
"""
LOGGER.timeit('_EVmut')
LOGGER.info('Recovering EVmutation data...')
# extracts precomputed EVmutation scores for given mutants
# NB:
# negative DeltaE_epist --> deleterious effect
# DeltaE_epist == 0 --> neutral effect (wild-type)
# positive DeltaE_epist --> neutral/benign effect
def find_matching_files(file_list, acc, pos):
match_files = []
for fname in [f for f in file_list if f.startswith(acc)]:
basename = splitext(fname)[0]
res_range = basename.split("_")[-1]
res_i = int(res_range.split("-")[0])
res_f = int(res_range.split("-")[1])
if res_i <= int(pos) <= res_f:
match_files.append(fname)
return match_files
feat_dtype = np.dtype([(f, 'f') for f in EVMUT_FEATS])
features = np.zeros(len(SAVs), dtype=feat_dtype)
features[:] = np.nan
# recover EVmutation data
EVmut_dir = SETTINGS.get('EVmutation_local_folder')
if EVmut_dir is None:
raise RuntimeError('EVmutation folder not set')
file_list = [basename(f) for f in glob(join(EVmut_dir, '*.csv'))]
if not file_list:
raise RuntimeError('EVmutation folder does not contain any .csv files')
for i, SAV in enumerate(SAVs):
acc, pos, wt_aa, mut_aa = SAV.split()
pos = int(pos)
# LOGGER.info('Recovering EVmutation data for {}.'.format(SAV))
# find files containing given SAV coordinates
match_files = find_matching_files(file_list, acc, pos)
# recover data and average them if multiple values are found
mutant = f'{wt_aa}{pos}{mut_aa}'
data = []
for fname in match_files:
with open(join(EVmut_dir, fname), 'r') as f:
for line in f:
if line.startswith(mutant):
ll = line.strip().split(';')[4:8]
data.append(ll)
break
data = np.array(data, dtype=float)
if len(data) == 0:
# LOGGER.warn(f"EVmutation data not found for '{SAV}'")
continue
else:
features[i] = tuple(np.mean(data, axis=0))
LOGGER.report('EVmutation scores recovered in %.1fs.', '_EVmut')
return features
def viewNMDinVMD(filename):
"""Start VMD in the current Python session and load NMD data."""
vmd = SETTINGS.get("vmd")
if vmd:
os.system("{0:s} -e {1:s}".format(vmd, os.path.abspath(filename)))
def fetchPDBLigand(cci, filename=None):
"""Fetch PDB ligand data from PDB_ for chemical component *cci*.
*cci* may be 3-letter chemical component identifier or a valid XML
filename. If *filename* is given, XML file will be saved with that name.
If you query ligand data frequently, you may configure ProDy to save XML
files in your computer. Set ``ligand_xml_save`` option **True**, i.e.
``confProDy(ligand_xml_save=True)``. Compressed XML files will be save
to ProDy package folder, e.g. :file:`/home/user/.prody/pdbligands`. Each
file is around 5Kb when compressed.
This function is compatible with PDBx/PDBML v 4.0.
Ligand data is returned in a dictionary. Ligand coordinate atom data with
*model* and *ideal* coordinate sets are also stored in this dictionary.
Note that this dictionary will contain data that is present in the XML
file and all Ligand Expo XML files do not contain every possible data
field. So, it may be better if you use :meth:`dict.get` instead of
indexing the dictionary, e.g. to retrieve formula weight (or relative
molar mass) of the chemical component use ``data.get('formula_weight')``
instead of ``data['formula_weight']`` to avoid exceptions when this data
field is not found in the XML file. URL and/or path of the XML file are
returned in the dictionary with keys ``url`` and ``path``, respectively.
Following example downloads data for ligand STI (a.k.a. Gleevec and
Imatinib) and calculates RMSD between model (X-ray structure 1IEP) and
ideal (energy minimized) coordinate sets:
.. ipython:: python
from prody import *
ligand_data = fetchPDBLigand('STI')
ligand_data['model_coordinates_db_code']
ligand_model = ligand_data['model']
ligand_ideal = ligand_data['ideal']
transformation = superpose(ligand_ideal.noh, ligand_model.noh)
calcRMSD(ligand_ideal.noh, ligand_model.noh)"""
if not isinstance(cci, str):
raise TypeError('cci must be a string')
if isfile(cci):
inp = openFile(cci)
xml = inp.read()
inp.close()
url = None
path = cci
cci = splitext(splitext(split(cci)[1])[0])[0].upper()
elif len(cci) > 4 or not cci.isalnum():
raise ValueError('cci must be 3-letters long and alphanumeric or '
'a valid filename')
else:
xml = None
cci = cci.upper()
if SETTINGS.get('ligand_xml_save'):
folder = join(getPackagePath(), 'pdbligands')
if not isdir(folder):
makePath(folder)
xmlgz = path = join(folder, cci + '.xml.gz')
if isfile(xmlgz):
with openFile(xmlgz) as inp:
xml = inp.read()
else:
path = None
#url = ('http://ligand-expo.rcsb.org/reports/{0[0]}/{0}/{0}'
# '.xml'.format(cci.upper()))
url = 'http://www.pdb.org/pdb/files/ligand/{0}.xml'.format(cci.upper())
if not xml:
#'http://www.pdb.org/pdb/files/ligand/{0}.xml'
try:
inp = openURL(url)
except IOError:
raise IOError('XML file for ligand {0} is not found online'
.format(cci))
else:
xml = inp.read()
inp.close()
if filename:
out = openFile(filename, mode='w', folder=folder)
out.write(xml)
out.close()
if SETTINGS.get('ligand_xml_save'):
with openFile(xmlgz, 'w') as out:
out.write(xml)
import xml.etree.cElementTree as ET
root = ET.XML(xml)
if (root.get('{http://www.w3.org/2001/XMLSchema-instance}'
'schemaLocation') !=
'http://pdbml.pdb.org/schema/pdbx-v40.xsd pdbx-v40.xsd'):
LOGGER.warn('XML is not in PDBx/PDBML v 4.0 format, resulting '
'dictionary may not contain all data fields')
ns = root.tag[:root.tag.rfind('}')+1]
len_ns = len(ns)
dict_ = {'url': url, 'path': path}
for child in list(root.find(ns + 'chem_compCategory')[0]):
tag = child.tag[len_ns:]
if tag.startswith('pdbx_'):
tag = tag[5:]
dict_[tag] = child.text
dict_['formula_weight'] = float(dict_.get('formula_weight'))
identifiers_and_descriptors = []
results = root.find(ns + 'pdbx_chem_comp_identifierCategory')
if results:
identifiers_and_descriptors.extend(results)
results = root.find(ns + 'pdbx_chem_comp_descriptorCategory')
if results:
identifiers_and_descriptors.extend(results)
for child in identifiers_and_descriptors:
program = child.get('program').replace(' ', '_')
type_ = child.get('type').replace(' ', '_')
dict_[program + '_' + type_] = child[0].text
dict_[program + '_version'] = child.get('program_version')
dict_['audits'] = [(audit.get('action_type'), audit.get('date'))
for audit in
list(root.find(ns + 'pdbx_chem_comp_auditCategory'))]
atoms = list(root.find(ns + 'chem_comp_atomCategory'))
n_atoms = len(atoms)
ideal_coords = np.zeros((n_atoms, 3))
model_coords = np.zeros((n_atoms, 3))
atomnames = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['name'].dtype)
elements = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['element'].dtype)
resnames = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['resname'].dtype)
charges = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['charge'].dtype)
resnums = np.ones(n_atoms, dtype=ATOMIC_FIELDS['charge'].dtype)
alternate_atomnames = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['name'].dtype)
leaving_atom_flags = np.zeros(n_atoms, np.bool)
aromatic_flags = np.zeros(n_atoms, np.bool)
stereo_configs = np.zeros(n_atoms, np.bool)
ordinals = np.zeros(n_atoms, int)
name2index = {}
for i, atom in enumerate(atoms):
data = dict([(child.tag[len_ns:], child.text) for child in list(atom)])
name = data.get('pdbx_component_atom_id', 'X')
name2index[name] = i
atomnames[i] = name
elements[i] = data.get('type_symbol', 'X')
resnames[i] = data.get('pdbx_component_comp_id', 'UNK')
charges[i] = float(data.get('charge', 0))
alternate_atomnames[i] = data.get('alt_atom_id', 'X')
leaving_atom_flags[i] = data.get('pdbx_leaving_atom_flag') == 'Y'
aromatic_flags[i] = data.get('pdbx_atomatic_flag') == 'Y'
stereo_configs[i] = data.get('pdbx_stereo_config') == 'Y'
ordinals[i] = int(data.get('pdbx_ordinal', 0))
model_coords[i, 0] = float(data.get('model_Cartn_x', 0))
model_coords[i, 1] = float(data.get('model_Cartn_y', 0))
model_coords[i, 2] = float(data.get('model_Cartn_z', 0))
ideal_coords[i, 0] = float(data.get('pdbx_model_Cartn_x_ideal', 0))
ideal_coords[i, 1] = float(data.get('pdbx_model_Cartn_y_ideal', 0))
ideal_coords[i, 2] = float(data.get('pdbx_model_Cartn_z_ideal', 0))
pdbid = dict_.get('model_coordinates_db_code')
if pdbid:
model = AtomGroup(cci + ' model ({0})'.format(pdbid))
else:
model = AtomGroup(cci + ' model')
model.setCoords(model_coords)
model.setNames(atomnames)
model.setResnames(resnames)
model.setResnums(resnums)
model.setElements(elements)
model.setCharges(charges)
model.setFlags('leaving_atom_flags', leaving_atom_flags)
model.setFlags('aromatic_flags', aromatic_flags)
model.setFlags('stereo_configs', stereo_configs)
model.setData('ordinals', ordinals)
model.setData('alternate_atomnames', alternate_atomnames)
dict_['model'] = model
ideal = model.copy()
ideal.setTitle(cci + ' ideal')
ideal.setCoords(ideal_coords)
dict_['ideal'] = ideal
bonds = []
warned = set()
for bond in list(root.find(ns + 'chem_comp_bondCategory') or bonds):
name_1 = bond.get('atom_id_1')
name_2 = bond.get('atom_id_2')
try:
bonds.append((name2index[name_1], name2index[name_2]))
except KeyError:
if name_1 not in warned and name_1 not in name2index:
warned.add(name_1)
LOGGER.warn('{0} specified {1} in bond category is not '
'a valid atom name.'.format(repr(name_1), cci))
if name_2 not in warned and name_2 not in name2index:
warned.add(name_2)
LOGGER.warn('{0} specified {1} in bond category is not '
'a valid atom name.'.format(repr(name_2), cci))
if bonds:
bonds = np.array(bonds, int)
model.setBonds(bonds)
ideal.setBonds(bonds)
return dict_
def getResAvgPredictions(self,
resid=None,
classifier='best',
PolyPhen2=True,
EVmutation=True,
refresh=False):
if not self._isSaturationMutagenesis():
return None
# initialize output array
cols = [('sequence index', 'i4'), ('PDB SAV coords', 'U100'),
('PDBID', 'U100'), ('chain', 'U1'), ('resid', 'i4'),
('resname', 'U1'), ('PDB size', 'i4'), ('score', 'f4'),
('path. prob.', 'f4'), ('path. class', 'U12')]
if PolyPhen2:
cols.extend([('PolyPhen-2 score', 'f4'),
('PolyPhen-2 path. class', 'U12')])
if EVmutation:
cols.extend([('EVmutation score', 'f4'),
('EVmutation path. class', 'U12')])
output = np.empty(int(self.numSAVs / 19), dtype=np.dtype(cols))
# fetch unique SAV coords, PDB coords and predictions
uSAVc = self.getUniqueSAVcoords()
PDBc = self.getPDBcoords()
preds = self.getPredictions(classifier=classifier,
PolyPhen2=PolyPhen2,
EVmutation=EVmutation,
refresh=refresh)
# compute residue-averaged quantities
output['sequence index'] = self._calcResAvg(uSAVc['position'])
for field in [
'PDB SAV coords', 'PDBID', 'chain', 'resid', 'resname',
'PDB size'
]:
output[field] = self._calcResAvg(PDBc[field])
# NB: I expect to see RuntimeWarnings in this block
with warnings.catch_warnings():
warnings.simplefilter("ignore", category=RuntimeWarning)
output['score'] = self._calcResAvg(preds['score'])
pp = self._calcResAvg(preds['path. prob.'])
pc = np.where(pp > 0.5, 'deleterious', 'neutral')
pc = np.where(np.isnan(pp), '?', pc)
output['path. prob.'] = pp
output['path. class'] = pc
if PolyPhen2:
ps = self._calcResAvg(preds['PolyPhen-2 score'])
pc = np.where(ps > 0.5, 'deleterious', 'neutral')
pc = np.where(np.isnan(ps), '?', pc)
output['PolyPhen-2 score'] = ps
output['PolyPhen-2 path. class'] = pc
if EVmutation:
ps = self._calcResAvg(preds['EVmutation score'])
cutoff = -SETTINGS.get('EVmutation_metrics')['optimal cutoff']
pc = np.where(ps < cutoff, 'deleterious', 'neutral')
pc = np.where(np.isnan(ps), '?', pc)
output['EVmutation score'] = ps
output['EVmutation path. class'] = pc
if resid is None:
return output
elif isinstance(resid, int):
return output[output['resid'] == resid][0]
else:
raise ValueError('Invalid resid.')
def fetchPDBLigand(cci, filename=None):
"""Fetch PDB ligand data from PDB_ for chemical component *cci*.
*cci* may be 3-letter chemical component identifier or a valid XML
filename. If *filename* is given, XML file will be saved with that name.
If you query ligand data frequently, you may configure ProDy to save XML
files in your computer. Set ``ligand_xml_save`` option **True**, i.e.
``confProDy(ligand_xml_save=True)``. Compressed XML files will be save
to ProDy package folder, e.g. :file:`/home/user/.prody/pdbligands`. Each
file is around 5Kb when compressed.
This function is compatible with PDBx/PDBML v 4.0.
Ligand data is returned in a dictionary. Ligand coordinate atom data with
*model* and *ideal* coordinate sets are also stored in this dictionary.
Note that this dictionary will contain data that is present in the XML
file and all Ligand Expo XML files do not contain every possible data
field. So, it may be better if you use :meth:`dict.get` instead of
indexing the dictionary, e.g. to retrieve formula weight (or relative
molar mass) of the chemical component use ``data.get('formula_weight')``
instead of ``data['formula_weight']`` to avoid exceptions when this data
field is not found in the XML file. URL and/or path of the XML file are
returned in the dictionary with keys ``url`` and ``path``, respectively.
Following example downloads data for ligand STI (a.k.a. Gleevec and
Imatinib) and calculates RMSD between model (X-ray structure 1IEP) and
ideal (energy minimized) coordinate sets:
.. ipython:: python
from prody import *
ligand_data = fetchPDBLigand('STI')
ligand_data['model_coordinates_db_code']
ligand_model = ligand_data['model']
ligand_ideal = ligand_data['ideal']
transformation = superpose(ligand_ideal.noh, ligand_model.noh)
calcRMSD(ligand_ideal.noh, ligand_model.noh)"""
if not isinstance(cci, str):
raise TypeError('cci must be a string')
if isfile(cci):
inp = openFile(cci)
xml = inp.read()
inp.close()
url = None
path = cci
cci = splitext(splitext(split(cci)[1])[0])[0].upper()
elif len(cci) > 4 or not cci.isalnum():
raise ValueError('cci must be 3-letters long and alphanumeric or '
'a valid filename')
else:
xml = None
cci = cci.upper()
if SETTINGS.get('ligand_xml_save'):
folder = join(getPackagePath(), 'pdbligands')
if not isdir(folder):
makePath(folder)
xmlgz = path = join(folder, cci + '.xml.gz')
if isfile(xmlgz):
with openFile(xmlgz) as inp:
xml = inp.read()
else:
path = None
#url = ('http://ligand-expo.rcsb.org/reports/{0[0]}/{0}/{0}'
# '.xml'.format(cci.upper()))
url = 'http://files.rcsb.org/ligands/download/{0}.xml'.format(
cci.upper())
if not xml:
#'http://www.pdb.org/pdb/files/ligand/{0}.xml'
try:
inp = openURL(url)
except IOError:
raise IOError(
'XML file for ligand {0} is not found online'.format(cci))
else:
xml = inp.read()
inp.close()
if filename:
out = openFile(filename, mode='w', folder=folder)
out.write(xml)
out.close()
if SETTINGS.get('ligand_xml_save'):
with openFile(xmlgz, 'w') as out:
out.write(xml)
import xml.etree.cElementTree as ET
root = ET.XML(xml)
if (root.get('{http://www.w3.org/2001/XMLSchema-instance}'
'schemaLocation') !=
'http://pdbml.pdb.org/schema/pdbx-v40.xsd pdbx-v40.xsd'):
LOGGER.warn('XML is not in PDBx/PDBML v 4.0 format, resulting '
'dictionary may not contain all data fields')
ns = root.tag[:root.tag.rfind('}') + 1]
len_ns = len(ns)
dict_ = {'url': url, 'path': path}
for child in list(root.find(ns + 'chem_compCategory')[0]):
tag = child.tag[len_ns:]
if tag.startswith('pdbx_'):
tag = tag[5:]
dict_[tag] = child.text
dict_['formula_weight'] = float(dict_.get('formula_weight'))
identifiers_and_descriptors = []
results = root.find(ns + 'pdbx_chem_comp_identifierCategory')
if results:
identifiers_and_descriptors.extend(results)
results = root.find(ns + 'pdbx_chem_comp_descriptorCategory')
if results:
identifiers_and_descriptors.extend(results)
for child in identifiers_and_descriptors:
program = child.get('program').replace(' ', '_')
type_ = child.get('type').replace(' ', '_')
dict_[program + '_' + type_] = child[0].text
dict_[program + '_version'] = child.get('program_version')
dict_['audits'] = [
(audit.get('action_type'), audit.get('date'))
for audit in list(root.find(ns + 'pdbx_chem_comp_auditCategory'))
]
atoms = list(root.find(ns + 'chem_comp_atomCategory'))
n_atoms = len(atoms)
ideal_coords = np.zeros((n_atoms, 3))
model_coords = np.zeros((n_atoms, 3))
atomnames = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['name'].dtype)
elements = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['element'].dtype)
resnames = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['resname'].dtype)
charges = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['charge'].dtype)
resnums = np.ones(n_atoms, dtype=ATOMIC_FIELDS['charge'].dtype)
alternate_atomnames = np.zeros(n_atoms, dtype=ATOMIC_FIELDS['name'].dtype)
leaving_atom_flags = np.zeros(n_atoms, np.bool)
aromatic_flags = np.zeros(n_atoms, np.bool)
stereo_configs = np.zeros(n_atoms, np.bool)
ordinals = np.zeros(n_atoms, int)
name2index = {}
for i, atom in enumerate(atoms):
data = dict([(child.tag[len_ns:], child.text) for child in list(atom)])
name = data.get('pdbx_component_atom_id', 'X')
name2index[name] = i
atomnames[i] = name
elements[i] = data.get('type_symbol', 'X')
resnames[i] = data.get('pdbx_component_comp_id', 'UNK')
charges[i] = float(data.get('charge', 0))
alternate_atomnames[i] = data.get('alt_atom_id', 'X')
leaving_atom_flags[i] = data.get('pdbx_leaving_atom_flag') == 'Y'
aromatic_flags[i] = data.get('pdbx_atomatic_flag') == 'Y'
stereo_configs[i] = data.get('pdbx_stereo_config') == 'Y'
ordinals[i] = int(data.get('pdbx_ordinal', 0))
model_coords[i, 0] = float(data.get('model_Cartn_x', 0))
model_coords[i, 1] = float(data.get('model_Cartn_y', 0))
model_coords[i, 2] = float(data.get('model_Cartn_z', 0))
ideal_coords[i, 0] = float(data.get('pdbx_model_Cartn_x_ideal', 0))
ideal_coords[i, 1] = float(data.get('pdbx_model_Cartn_y_ideal', 0))
ideal_coords[i, 2] = float(data.get('pdbx_model_Cartn_z_ideal', 0))
pdbid = dict_.get('model_coordinates_db_code')
if pdbid:
model = AtomGroup(cci + ' model ({0})'.format(pdbid))
else:
model = AtomGroup(cci + ' model')
model.setCoords(model_coords)
model.setNames(atomnames)
model.setResnames(resnames)
model.setResnums(resnums)
model.setElements(elements)
model.setCharges(charges)
model.setFlags('leaving_atom_flags', leaving_atom_flags)
model.setFlags('aromatic_flags', aromatic_flags)
model.setFlags('stereo_configs', stereo_configs)
model.setData('ordinals', ordinals)
model.setData('alternate_atomnames', alternate_atomnames)
dict_['model'] = model
ideal = model.copy()
ideal.setTitle(cci + ' ideal')
ideal.setCoords(ideal_coords)
dict_['ideal'] = ideal
bonds = []
warned = set()
for bond in list(root.find(ns + 'chem_comp_bondCategory') or bonds):
name_1 = bond.get('atom_id_1')
name_2 = bond.get('atom_id_2')
try:
bonds.append((name2index[name_1], name2index[name_2]))
except KeyError:
if name_1 not in warned and name_1 not in name2index:
warned.add(name_1)
LOGGER.warn('{0} specified {1} in bond category is not '
'a valid atom name.'.format(repr(name_1), cci))
if name_2 not in warned and name_2 not in name2index:
warned.add(name_2)
LOGGER.warn('{0} specified {1} in bond category is not '
'a valid atom name.'.format(repr(name_2), cci))
if bonds:
bonds = np.array(bonds, int)
model.setBonds(bonds)
ideal.setBonds(bonds)
return dict_
def calcEvolProperties(self,
resid='all',
refresh=False,
folder=None,
max_cols=None,
max_seqs=25000,
**kwargs):
''' Computes Evol properties, i.e. Shannon entropy, Mutual
Information and Direct Information, from Pfam Multiple
Sequence Alignments, for a given residue.
'''
assert type(refresh) is bool
# recover Pfam mapping (if not found already)
self._searchPfam(refresh=refresh)
if resid == 'all':
PF_list = self.Pfam.keys()
else:
# get list of Pfam domains containing resid
PF_list = [
k for k in self.Pfam if any([
resid >= int(segment['start'])
and resid <= int(segment['end'])
for segment in self.Pfam[k]['locations']
])
]
if len(PF_list) == 0:
raise RuntimeError(
'No Pfam domain for resid {}.'.format(resid))
if len(PF_list) > 1:
LOGGER.warn('Residue {} is found in multiple '.format(resid) + \
'({}) Pfam domains.'.format(len(PF_list)))
if folder is None:
folder = SETTINGS.get('rhapsody_local_folder', './')
# iterate over Pfam families
for PF in PF_list:
d = self.Pfam[PF]
# skip if properties are pre-computed
if not refresh and d.get('mapping') is not None:
continue
d['mapping'] = None
d['ref_MSA'] = None
d['entropy'] = np.nan
d['MutInfo'] = np.nan
d['DirInfo'] = np.nan
try:
LOGGER.info('Processing {}...'.format(PF))
# fetch & parse MSA
# fname = PF + '_full.sth'
# fullname = os.path.join(folder, fname)
# if not os.path.isfile(fullname):
# f = fetchPfamMSA(PF)
# os.rename(f, fullname)
# msa = parseMSA(fullname, **kwargs)
# fetch & parse MSA without saving downloaded MSA
f = fetchPfamMSA(PF)
msa = parseMSA(f, **kwargs)
os.remove(f)
# slice MSA to match all segments of the Uniprot sequence
sliced_msa, indexes = self._sliceMSA(msa)
# if max_cols is not None and sliced_msa.numResidues() > max_cols:
# raise Exception('Unable to compute DI: MSA has ' +\
# 'too many columns (max: {}).'.format(max_cols))
# get mapping between Uniprot sequence and Pfam domain
d['mapping'] = self._mapUniprot2Pfam(PF, sliced_msa, indexes)
except Exception as e:
LOGGER.warn('{}: {}'.format(PF, e))
d['mapping'] = str(e)
continue
try:
# refine MSA ('seqid' param. is set as in PolyPhen-2)
rowocc = 0.6
while True:
sliced_msa = refineMSA(sliced_msa, rowocc=rowocc)
rowocc += 0.02
if sliced_msa.numSequences() <= max_seqs or rowocc >= 1:
break
ref_msa = refineMSA(sliced_msa, seqid=0.94, **kwargs)
d['ref_MSA'] = ref_msa
# compute evolutionary properties
d['entropy'] = calcShannonEntropy(ref_msa)
d['MutInfo'] = buildMutinfoMatrix(ref_msa)
# d['DirInfo'] = buildDirectInfoMatrix(ref_msa)
except Exception as e:
LOGGER.warn('{}: {}'.format(PF, e))
return {k: self.Pfam[k] for k in PF_list}