def getTaxIdMetaData(ncbi_id):
etal_bool = False
if len(ncbi_id) > 1:
ncbi_id = ncbi_id[0]
etal_bool = True
record = etools.eFetch(ncbi_id, logger=logger, db="taxonomy")[0]
metadata = [record['Rank'], record['ScientificName']]
if etal_bool:
metadata = [e + ' et al.' for e in metadata]
return metadata[0], metadata[1]
def _download(self, seqids):
"""Download records from GenBank given sequence ids"""
records = []
i = 1
while i <= self.nseqs and len(seqids) > 0:
if len(seqids) > 100:
n = 100 # Download in chunks of 100
else:
n = len(seqids)
seqs = []
for _ in range(n):
randi = random.randint(0, len(seqids)-1)
seqs.append(seqids.pop(randi))
for record in etools.eFetch(seqs, logger=self.logger):
record = self._parse(record)
if record:
records.append(record)
i += 1
return records
def findParent(parentid):
return etools.eFetch(parentid, logger=logger,
db="taxonomy")[0]['ParentTaxId']