def setup(self):
"""Setup main parameters"""
if base.check_snap() == True:
print ('running in a snap')
cmd = base.set_netmhcpan_cmd()
print ('netmhcpan cmd set to:', cmd)
pd.set_option('display.width', 120)
#override base.defaults entries if provided in conf
set_defaults(self.__dict__)
self.sequences = None
self.peptides = None
if self.sequence_file is not '':
self.sequences = get_sequences(self.sequence_file, header_sep=self.fasta_header_sep)
print ('input is %s protein sequences' %len(self.sequences))
elif self.peptide_file is not '':
self.peptides = read_names(self.peptide_file)
print ('input is %s peptides' %len(self.peptides))
if self.sequences is None and self.peptides is None:
print ('no valid inputs found')
return False
self.mhc1_alleles = get_alleles(self.mhc1_alleles)
self.mhc2_alleles = get_alleles(self.mhc2_alleles)
if len(self.mhc1_alleles)==0 and len(self.mhc2_alleles)==0:
return False
self.predictors = self.predictors.split(',')
for p in self.predictors:
if p not in base.predictors:
print ('unknown predictor in config file. Use:')
show_predictors()
return False
if self.mhc1_alleles[0] in base.mhc1_presets:
self.mhc1_alleles = base.get_preset_alleles(self.mhc1_alleles[0])
if self.mhc2_alleles[0] in base.mhc2_presets:
self.mhc2_alleles = base.get_preset_alleles(self.mhc2_alleles[0])
if type(self.cutoffs) is int:
self.cutoffs = [self.cutoffs]
else:
self.cutoffs = [float(i) for i in self.cutoffs.split(',')]
if os.path.exists(self.names):
self.names = read_names(self.names)
elif self.names == '':
self.names=None
else:
self.names = self.names.split(',')
if self.names is not None:
print ('selected sequences:', self.names)
if not os.path.exists(self.path) and self.path != '':
os.mkdir(self.path)
#copy input seqs to path
if self.sequences is not None:
self.sequences.to_csv(os.path.join(self.path, 'input.csv'),index=False)
return True
def check_ensembl(release='75'):
"""Check pyensembl ref genome cached. Needed for running in snap"""
#check if running inside a snap package so we can download
#the genome files for pyensembl
cache_dir = None
if base.check_snap() is True:
#home = os.path.join('/home', os.environ['USER'])
home = os.environ['SNAP_USER_COMMON']
cache_dir = os.path.join(home, '.cache')
os.environ['PYENSEMBL_CACHE_DIR'] = cache_dir
print('checking for ref human genome')
fetch_ensembl_release(cache_dir, release)
return