def test_descs(self):
geo_id='GSM32100'
f=Factory()
sample=f.newGEO(geo_id)
self.assertIsInstance(sample, Sample)
self.assertEqual(sample.geo_id, geo_id)
descs=sample.descriptions()
self.assertIsInstance(descs, dict)
def test_new_geo(self):
f=Factory()
pairs={'GSE10072':Series,
'GSM15718':Sample,
'GDS994':Dataset,
'GPL96':Platform
}
for geo_id, geo_class in pairs.items():
geo=f.newGEO(geo_id)
self.assertIsInstance(geo, geo_class)
self.assertEqual(geo.geo_id, geo_id)
def test_new_geo(self):
f = Factory()
pairs = {
'GSE10072': Series,
'GSM15718': Sample,
'GDS994': Dataset,
'GPL96': Platform
}
for geo_id, geo_class in pairs.items():
geo = f.newGEO(geo_id)
self.assertIsInstance(geo, geo_class)
self.assertEqual(geo.geo_id, geo_id)
def test_get_field_words(self):
geo_id='GSE10072'
geo=Factory().newGEO(geo_id)
words=Word2Geo.get_field_words(geo)
self.assertEqual(len(words['title']), 42)
self.assertEqual(len(words['description']), 0)
self.assertEqual(len(words['summary']), 738) # not quite sure why this isn't 741
def main():
options=get_options()
geo_ids=get_geo_ids(options)
f=Factory()
warn("insert_geo_words starting: %s" % (datetime.datetime.now().__str__()))
fuse=options.fuse
for geo_id in geo_ids:
geo=f.newGEO(geo_id)
warn("inserting %s" % (geo.geo_id))
stats=insert_series(geo)
warn("%s: %s" % (geo_id, stats))
fuse-=1
if (fuse==0): break
warn("insert_geo_words done: %s" % (datetime.datetime.now().__str__()))
return 0
def _geolist2pmidlist(geo_ids):
''' converts a list of mixed pmids and geo_ids to all pmids by doing the lookups on the geo objects '''
pmidlist = []
for id in geo_ids:
if re.match('^\d+$', id):
pmidlist.append(id)
else:
try:
geo = Factory().newGEO(id)
pmids = geo.pubmed_id # might be single value or list, so:
except Exception as e:
warn("caught %s" % (e))
continue # id not a geo id, or geo didn't have any pubmed_id
try:
pmidlist.append(pmids)
except:
pmidlist.extend(pmids)
return pmidlist