def mrnaExt(ID):
"""Extract sequence record for the given sequence ID"""
recData = Entrez.efetch(db="nucleotide",
id=ID,
rettype="gb",
warning=False)
record = SeqIO.read(recData, 'genbank')
return record
def cdsExt(ID, geneName):
"""
returns sequence record object for the input gene refseq ID
"""
retdata = Entrez.efetch(db="nucleotide",
id=ID,
rettype='gb',
retmode='text').read()
with open("Align/" + geneName.split('.')[0] + ".log", "a") as fp:
if 'LOW QUALITY PROTEIN' in retdata:
fp.write('%s CDS is of low quality\n' % ID)
data = retdata.split('\n')
for obj in data:
if ' CDS ' in obj:
try:
cdsRange = [
int(obj.lstrip(' CDS ').split('..')[0]),
int(obj.lstrip(' CDS ').split('..')[1])
]
except ValueError:
try:
cdsRange = [
int(
obj.lstrip(' CDS ').split('..')[0].
lstrip('<').lstrip('>').rstrip('<').rstrip('>')),
int(
obj.lstrip(' CDS ').split('..')[1].
lstrip('<').lstrip('>').rstrip('<').rstrip('>'))
]
except ValueError:
try:
cdsRange = [
int(
obj.lstrip(' CDS ').split('..')
[0]),
int(
obj.lstrip(' CDS ').split('..')
[1].lstrip('>'))
]
except ValueError:
print(
"Problem found while extracting cds from %s. Please report this issue to ambuj (at) ufl (dot) edu"
% obj)
continue
recData = Entrez.efetch(db="nucleotide",
id=ID,
rettype="gb",
warning=False)
record = SeqIO.read(recData, 'genbank')
record.seq = record.seq[cdsRange[0] - 1:cdsRange[1]]
return record