def test_strip_non_alphabetic(self):
test_string = "abcdefghi"
result = _utils.strip_non_alphabetic(test_string)
self.assertEqual(test_string, result)
test_string = "abcd890"
result = _utils.strip_non_alphabetic(test_string)
self.assertEqual("abcd",result)
test_string = "123456789"
result = _utils.strip_non_alphabetic(test_string)
self.assertEqual("", result)
def genome_name(cls, contig):
"""
Returns a string of the name of a non-complete genome from its contig
name.
Args:
contig(str): contig's accession name
"""
if len(contig) == 12:
return strip_non_alphabetic(contig) + "00000000"
else:
return contig
def from_ftp(self, seqdata):
"""Obtains the FASTA sequence via the NCBI FTP server in the WGS genome
pipeline and labels the sequence as being from the WGS piepline.
Args:
seqdata: a SequenceMetadata instance storing sequence-related data
that would otherwise be a data clump
"""
seq_id = strip_non_alphabetic(str(seqdata.accession))
self.download_file(seq_id, 'fsa_nt.gz')
with open(generate_path('tmp/loading.fasta'), 'rb') as handle:
self.read_fasta(handle, seqdata)
seqdata.dict["is_from"] = "WGS"
def get_seqdata(self, contigswrapper):
"""
Args:
contigswrapper: a ContigsWrapper instance that holds contig metadata for a genome
Returns: a BLAST record for self.load_contigs to use
"""
Entrez.email = "*****@*****.**"
handle = None
i = 0
while i < 3:
try:
print "Getting data from Entrez..."
handle = Entrez.efetch(
db="nuccore", id=contigswrapper.genome,
rettype="fasta", retmode="text"
)
for record in SeqIO.parse(handle, 'fasta'):
if "complete" in record.description.lower():
contigswrapper.dict["is_from"] = "CORE"
print "Getting data from Entrez..."
handle = Entrez.efetch(
db="nuccore", id=contigswrapper.genome,
rettype="fasta", retmode="text"
)
self.load_contigs(handle, contigswrapper)
break
else:
print "Downloading data from WGS"
self.download_file(
strip_non_alphabetic(str(contigswrapper.genome)),
'fsa_nt.gz'
)
with open(
generate_path('tmp/loading.fasta'),
'rb'
) as handle:
contigswrapper.dict["is_from"] = "WGS"
self.load_contigs(handle, contigswrapper)
except HTTPError:
i += 1
continue
break
try:
handle is None
except NameError:
raise TypeError("Could not retrieve file for analysis")