# the directories with the output of breseq
directories = [str(21), str(22), str(43)]
for directory in directories:
if directory == '21':
sample = 'Ancestor_S1'
elif directory == '22':
sample = 'Ancestor_S2'
elif directory == '43':
sample = 'Ancestor_S3'
else:
raise RuntimeError('somehow not in an anticipated directory')
with pushd(directory + '/output/evidence'):
anno = open('annotated.gd')
gd_dicts = []
for line in anno:
try:
gd_dicts.append(lgd.gd_line_to_dict(line))
except ValueError:
pass
dictIDs_to_save = []
for gd_dict in gd_dicts:
if gd_dict['type'] == 'SNP':
dictIDs_to_save.append(gd_dict['parent_id'])
for gd_dict in gd_dicts:
if gd_dict['id'] in dictIDs_to_save:
gene, snp_mutation, snp_evidence = \
lgd.RA_dict_to_SNP_dicts(sample, gd_dict)
lgd.commit_SNP_dicts(gene, snp_mutation, snp_evidence, session)
for directory in directories:
if directory == 'Agg_over_ancestors_against_self':
sample = 'Aggregate_NS001_Ancestors'
elif directory == '21':
sample = 'Ancestor_S1'
elif directory == '22':
sample = 'Ancestor_S2'
elif directory == '43':
sample = 'Ancestor_S3'
else:
with pushd(directory + '/output'):
output_file = open('output.gd')
line_with_samplename = output_file.readlines()[5]
sample = re.findall('[a-zA-Z]+[1-4]t[1-2]',
line_with_samplename)[0] + '_S1'
with pushd(directory + '/output/evidence'):
anno = open('annotated.gd')
for line in anno:
try:
gd_dict = lgd.gd_line_to_dict(line)
if gd_dict['type'] == 'MC':
mc = lgd.gdMC_dict_to_MC_dict(sample, gd_dict)
lgd.add_MC_dict(mc, session)
except ValueError:
pass
try:
session.commit()
except DBAPIError:
print('failed in directory {0}'.format(directory))
session.rollback()