def test_gds_info(self):
gds_info = GDSInfo()
self.assertIsNotNone(gds_info)
self.assertGreater(len(gds_info.keys()), 0)
self.assertGreater(len(gds_info.items()), 0)
self.assertGreater(len(gds_info.values()), 0)
self.assertIsNotNone(gds_info[self.test_sample])
self.assertEqual(gds_info[self.test_sample]['genes'], 9561)
self.assertEqual(int(gds_info[self.test_sample]['sample_count']), 4)
self.assertEqual(len(gds_info[self.test_sample]['subsets']), 2)
def valid(info, n=40):
"""Return a set of subset types containing more than n samples in every subset"""
invalid = set()
subsets = set([sinfo["type"] for sinfo in info["subsets"]])
for sampleinfo in info["subsets"]:
if len(sampleinfo["sample_id"]) < n:
invalid.add(sampleinfo["type"])
return subsets.difference(invalid)
def report(stypes, info):
"""Pretty-print GDS and valid susbset types"""
for id, sts in stypes:
print(id)
for st in sts:
gds = info[id]
print(" %s:" % st + ", ".join([
"%s/%d" % (sinfo["description"], len(sinfo["sample_id"]))
for sinfo in gds["subsets"] if sinfo["type"] == st
]))
gdsinfo = GDSInfo()
valid_subset_types = [(id, valid(info)) for id, info in sorted(gdsinfo.items())
if valid(info)]
report(valid_subset_types, gdsinfo)
print('datasets = ' + str(len(valid_subset_types)))
print('type subsets = ' + str(sum(len(b) for _, b in valid_subset_types)))