def fix_empty(taxid=9606):
from collections import Counter
global_settings.verbose = False
glitchy = 0
fine = 0
fixed = 0
path = os.path.join(global_settings.pickle_folder, f'taxid{taxid}')
for pf in os.listdir(path):
p = ProteinGatherer().load(file=os.path.join(path, pf))
if len(p.sequence) == 0:
print('****************************************')
print(f'Attempting to fix {p.gene_name}')
try:
global_settings.verbose = True
p.parse_uniprot()
p.parse_swissmodel()
p.compute_params()
p.parse_gnomAD()
p.get_PTM()
assert len(p.sequence) > 0, 'Darn. Sequence is zero AA long'
p.dump()
fixed += 1
global_settings.verbose = False
except Exception:
traceback.print_exc(file=sys.stdout)
glitchy += 1
else:
fine += 1
print('****************************************')
print(f'Fine: {fine:,}, Fixed {fixed:,}, Glitchy: {glitchy:,}')
def iterate_taxon(taxid=9606):
"""
This is an ad hoc fix to fix humans or similar. For full deployment use ProteomeParser.
:param taxid:
:return:
"""
path = os.path.join(global_settings.pickle_folder, f'taxid{taxid}')
for pf in os.listdir(path):
try:
protein = ProteinGatherer().load(file=os.path.join(path, pf))
protein.gnomAD = []
protein.parse_gnomAD()
protein.get_PTM()
protein.compute_params()
protein.dump()
#michelanglo_protein.get_offsets().parse_gnomAD().compute_params()
#michelanglo_protein.dump()
except:
pass
namedexfile=os.path.join(global_settings.dictionary_folder,
'taxid9606-names2uniprot.json'),
folder=os.path.join(global_settings.temp_folder, 'gnomAD')).split()
if __name__ == '__main__':
global_settings.verbose = True #False
global_settings.error_tolerant = True
global_settings.startup(data_folder='../protein-data')
global_settings.retrieve_references(ask=False, refresh=False)
## Phosphosite
#Phosphosite().split().write()
message('Phosphosite split')
## Uniprot
UniprotMasterReader(first_n_protein=0)
message('Uniprot split')
# gnomAD data needs to be split up after that the dictionaries are made.
# _gnomad()
message('gnomAD split')
taxid = 9606 # that's humans
path = os.path.join(global_settings.pickle_folder, f'taxid{taxid}')
for pf in os.listdir(path):
try:
protein = ProteinGatherer().load(file=os.path.join(path, pf))
protein.gnomAD = []
protein.parse_gnomAD()
protein.get_PTM()
protein.dump()
except:
pass
message('Done.')