def make_lists_of_accessions_of_set_size(self):
"""make_lists_of_expanded_slices_of_set_size: should return a
list of lists"""
expected_list = ['HM780503 HM780506 HM780660 HM780780']
observed = make_lists_of_accessions_of_set_size(
['HM780503', 'HM780506', 'HM780660', 'HM780780'], size_limit=3)
self.assertEqual(observed, expected_list)
def bulk_efetch(query_ids):
"""Wraps EUtils call"""
# pre-bin calls because we're potentially obtaining millions of records
# and EUtils would normally store all the records in memory
bins = make_lists_of_accessions_of_set_size(list(query_ids))
handle = EUtils(db='nucleotide',rettype='gb')
for queries in bins:
data = handle[queries].read()
yield data
def __getitem__(self, query):
"""Gets an query from NCBI. Assumes lists are lists of accessions.
Returns a handle to the result (either in memory or file on disk).
WARNING: result is not guaranteed to contain any data.
"""
#check if it's a slice
if isinstance(query, slice):
#query = expand_slice(query)
queries = make_lists_of_expanded_slices_of_set_size(query)
return self.grab_data(queries)
#check if it's a list -- if so, delimit with ' '
if isinstance(query, list) or isinstance(query,tuple):
#query = ' '.join(map(str, query))
queries = make_lists_of_accessions_of_set_size(query)
return self.grab_data(queries)
# most likey a general set of search terms
#e.g. '9606[taxid] OR 28901[taxid]' . So just return.
return self.grab_data([query])
def make_lists_of_accessions_of_set_size(self):
"""make_lists_of_expanded_slices_of_set_size: should return a
list of lists"""
expected_list = ['HM780503 HM780506 HM780660 HM780780']
observed = make_lists_of_accessions_of_set_size(['HM780503','HM780506', 'HM780660', 'HM780780'],size_limit=3)
self.assertEqual(observed,expected_list)
