def _extract_clusters(self, tag, qual='ugene_name'):
tagre = re.compile(tag)
clusters = {}
for record in load_files(self._abort_event, self.genomes_files, 'gb'):
for f in record.features:
if qual in f.qualifiers:
q = ' '.join(f.qualifiers[qual])
if not tagre.match(q): continue
c = f.extract(record)
c.id = c.name = q
c.description = record.description
if c.seq.alphabet is not NucleotideAlphabet \
or c.seq.alphabet is not ProteinAlphabet:
c.seq.alphabet = IUPAC.IUPACAmbiguousDNA()
self._process_features(c)
clusters[c.id] = c
return clusters
def _main(self):
accessions_file = '/home/allis/Documents/INMI/Geobacillus-COX/CoxL-markers.txt'
with open(accessions_file) as inp:
accessions = set(line for line in (l.strip() for l in inp) if line)
seqs = load_files(self.abort_event,
['/home/allis/Documents/INMI/Geobacillus-COX/CoxL-analysis.files/CoxL-analysis.gb'])
targets = filter(lambda s: s.id in accessions, seqs)
to_fetch = set()
for t in targets:
for f in t.features:
coded = f.qualifiers.get('coded_by')
if not coded: continue
m = coded_re.match(coded[0])
if m: to_fetch.add(m.group(2))
entrez = BatchEntrez(self.abort_event, '*****@*****.**')
recs = entrez.get_records_for_terms(list(to_fetch), 'nucleotide')
if recs:
for r in recs:
fname = filename_for_record(r)
print 'Saving: %s' % fname
safe_write(r, '/home/allis/Documents/INMI/Geobacillus-COX/genomes/'+fname)
else: print 'No records were fetched'
print 'Done'
if True:
# from DegenPrimer import MultiprocessingBase
# MultiprocessingBase.cpu_count = 1
abort_event = Event()
lb = BlastCLI(abort_event)
with user_message('Loading genomes...', '\n'):
genomes_dir = u'/home/allis/Dropbox/Science/Микра/Thermococcus/sequence/GenBank/Thermococcus'
genome_names = [
'Thermococcus_barophilus_Ch5-complete.gb',
'Thermococcus_onnurineus_NA1-complete-genome.gb',
'Thermococcus_sp._ES1.gb', 'Thermococcus-DS1-preliminary.gb'
]
genomes = load_files(
abort_event,
[os.path.join(genomes_dir, f) for f in genome_names], 'gb')
ref = genomes[0]
subj = genomes[1:]
@shelf_result
def g2g2shelf():
return lb.g2g_blastp(ref,
subj,
11,
features_of_interest=[{
'ugene_name': 'FC-full'
}, {
'ugene_name': 'COC-full'
}])
signal.signal(signal.SIGTERM, sig_handler)
signal.signal(signal.SIGQUIT, sig_handler)
if True:
# from DegenPrimer import MultiprocessingBase
# MultiprocessingBase.cpu_count = 1
abort_event = Event()
lb = BlastCLI(abort_event)
with user_message('Loading genomes...', '\n'):
genomes_dir = u'/home/allis/Dropbox/Science/Микра/Thermococcus/sequence/GenBank/Thermococcus'
genome_names = ['Thermococcus_barophilus_Ch5-complete.gb',
'Thermococcus_onnurineus_NA1-complete-genome.gb',
'Thermococcus_sp._ES1.gb',
'Thermococcus-DS1-preliminary.gb']
genomes = load_files(abort_event, [os.path.join(genomes_dir, f) for f in genome_names], 'gb')
ref = genomes[0]
subj = genomes[1:]
@shelf_result
def g2g2shelf():
return lb.g2g_blastp(ref, subj, 11, features_of_interest=[{'ugene_name': 'FC-full'}, {'ugene_name': 'COC-full'}])
g2g_res = '/tmp/DP-PCR-N_KAEs'
if not os.path.isfile(g2g_res):
g2g_res = g2g2shelf()
print g2g_res
if g2g_res:
with roDict(g2g_res) as db:
results = db['result']
