def seqScanning(Uniprot_coord):
'''Returns a list of SAVs. If the string 'Uniprot_coord' is just a Uniprot ID,
the list will contain all possible amino acid substitutions at all positions
in the sequence. If 'Uniprot_coord' also includes a specific position, the list
will only contain all possible amino acid variants at that position.
'''
assert isinstance(Uniprot_coord, str), "Must be a string."
coord = Uniprot_coord.strip().split()
assert len(coord) < 3, "Invalid format. Examples: 'Q9BW27' or 'Q9BW27 10'."
Uniprot_record = pd.queryUniprot(coord[0])
sequence = Uniprot_record['sequence 0'].replace("\n", "")
if len(coord) == 1:
positions = range(len(sequence))
else:
positions = [int(coord[1]) - 1]
SAV_list = []
acc = coord[0]
for i in positions:
wt_aa = sequence[i]
for aa in 'ACDEFGHIKLMNPQRSTVWY':
if aa == wt_aa:
continue
s = ' '.join([acc, str(i + 1), wt_aa, aa])
SAV_list.append(s)
return SAV_list
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 _print_fasta_file(Uniprot_accs, filename='custom_sequences.fasta'):
date = datetime.date.today().strftime('%Y%m%d')
new_accs = {}
with open(filename, 'w', 1) as f:
for acc in Uniprot_accs:
new_acc = f"{acc}-{date}"
f.write(f">{new_acc}")
record = queryUniprot(acc)
sequence = record['sequence 0']
f.write(sequence)
# store new temporary accession numbers
new_accs[acc] = new_acc
return filename, new_accs
def queryUniprot(*args, n_attempts=3, dt=1, **kwargs):
"""
Redefine prody function to check for no internet connection
"""
attempt = 0
while attempt < n_attempts:
try:
_ = openURL('http://www.uniprot.org/')
break
except:
LOGGER.info(f'Attempt {attempt} to contact www.uniprot.org failed')
attempt += 1
time.sleep((attempt + 1) * dt)
else:
_ = openURL('http://www.uniprot.org/')
return pd.queryUniprot(*args, **kwargs)
def refresh(self):
"""Refresh imported Uniprot records and mappings, and
delete precomputed alignments.
"""
# import Uniprot record and official accession number
self.fullRecord = queryUniprot(self.acc)
self.uniq_acc = self.fullRecord['accession 0']
# import main sequence and PDB records
rec = self.fullRecord
self.sequence = rec['sequence 0'].replace("\n", "")
self.PDBrecords = [
rec[key] for key in rec.keys()
if key.startswith('dbRef') and 'PDB' in rec[key]
]
# parse PDB records into PDB mappings, easier to access
self._initiatePDBmappings()
# set remaining attributes
self.customPDBmappings = []
self._align_algo_args = ['localxs', -0.5, -0.1]
self._align_algo_kwargs = {'one_alignment_only': True}
self._timestamp = str(datetime.datetime.utcnow())
self.Pfam = None
return