def merge(args):
filename = args.sample_set_file
sep = args.sep
name = args.name
desc = args.desc
sample_set_names = args.sample_set_names
# check arguments
if name is None:
name = 'merge'
if desc is None:
desc = name
if not sample_set_names:
logging.error('No sample set names specified')
return 1
sample_sets = dict((ss.name,ss) for ss in _parse_sample_sets(filename, sep))
new_value_dict = collections.defaultdict(lambda: 0)
for ss_name in sample_set_names:
if ss_name not in sample_sets:
logging.error('Sample set name "%s" not found.. Exiting.' % (ss_name))
return 1
ss = sample_sets[ss_name]
for k,v in ss.value_dict.iteritems():
if k in new_value_dict:
continue
new_value_dict[k] = v
ss = SampleSet(name, desc, new_value_dict.items())
print ss.to_json()
def newcohort(args):
sample_set_file = args.sample_set_file
cohort_file = args.cohort_file
sep = args.sep
cohort_samples = set(line.strip() for line in open(cohort_file))
for ss in _parse_sample_sets(sample_set_file, sep):
new_value_dict = {}
hits = 0
for k,v in ss.value_dict.iteritems():
if k in cohort_samples:
if v == 1:
hits += 1
new_value_dict[k] = v
if hits > 0:
ss = SampleSet(ss.name, ss.desc, new_value_dict.items())
print ss.to_json()
else:
logging.warning('Sample set %s has no hits' % (ss.name))
def subset(args):
filename = args.sample_set_file
sep = args.sep
name = args.name
desc = args.desc
hit_set_names = args.hit_sets
miss_set_names = args.miss_sets
# check arguments
if name is None:
name = 'subset'
if desc is None:
desc = name
if not hit_set_names or not miss_set_names:
logging.error('Sample sets to be considered "hits" or "misses" '
'should be specified using --hit and --miss')
return 1
sample_sets = dict((ss.name,ss) for ss in _parse_sample_sets(filename, sep))
new_value_dict = {}
for ss_name in hit_set_names:
if ss_name not in sample_sets:
logging.error('Sample set name "%s" not found.. Exiting.' % (ss_name))
return 1
ss = sample_sets[ss_name]
for k,v in ss.value_dict.iteritems():
if v == 1:
new_value_dict[k] = 1
for ss_name in miss_set_names:
if ss_name not in sample_sets:
logging.error('Sample set name "%s" not found.. Exiting.' % (ss_name))
return 1
ss = sample_sets[ss_name]
for k,v in ss.value_dict.iteritems():
if v == 1:
new_value_dict[k] = 0
ss = SampleSet(name, desc, new_value_dict.items())
print ss.to_json()
