def _main(self):
genomes = load_dir(self.abort_event, self.genomedir, 'gb', '.*\.gb$')
if not genomes: return 1
pseqs = {}
for g in genomes:
if g.id not in self.promoters: continue
if g.id in pseqs: continue
pseqs[g.id] = self.promoters[g.id].extract(g)
print g.id
print pseqs[g.id]
print
finder = SchemaFinder()
repo = finder.find(pseqs.values())
with open('FC-promoter.mfs', 'w') as out:
writer = csv.writer(out)
writer.writerow(('motif', 'count'))
for pi in repo.get_all():
writer.writerow((pi, repo.count(pi)))
from BioUtils.Tools.Output import user_message
from BioUtils.SeqUtils import load_files, load_dir
_pid = os.getpid()
#setup signal handler
signal.signal(signal.SIGINT, sig_handler)
signal.signal(signal.SIGTERM, sig_handler)
signal.signal(signal.SIGQUIT, sig_handler)
abort_event = Event()
with user_message('Loading genomes...', '\n'):
genomes_dir = u'/home/allis/Documents/INMI/Aerobic-CODH/genomes/'
genome_names = ['Thermococcus_barophilus_Ch5-complete.gb',
'Thermococcus_onnurineus_NA1-complete-genome.gb',
'Thermococcus_sp._ES1.gb',
'Thermococcus-DS1-preliminary.gb']
genomes = load_dir(abort_event, genomes_dir, 'gb', r'.*\.gb')
if not genomes: sys.exit(1)
# load_files(abort_event, [os.path.join(genomes_dir, f) for f in genome_names], 'gb')
hmm = u'/home/allis/Documents/INMI/Aerobic-CODH/COX-EC/COX-EC_1.2.99.2_CoxL.hmm'
hmmer = Hmmer(abort_event)
for g in genomes:
results = hmmer.hmmsearch_genome(hmm, g, table=11, decorate=True)
if results:
SeqIO.write(g, '%s.gb' % g.name, 'gb')
print '='*80
print g.name, g.description
for fi in results:
print results[fi][0]
