def annotate_hmm_matches(hits_file, hits_annot_file, args):
hits_annot_header = map(
str.strip, '''#query_name, hit, level, evalue,
sum_score, query_length, hmmfrom, hmmto, seqfrom, seqto, query_coverage,
members_in_og, og_description, og_COG_categories'''.
split(','))
annota.connect()
print colorify("Functional annotation of hits starts now", 'green')
start_time = time.time()
if pexists(hits_file):
OUT = open(hits_annot_file, "w")
if not args.no_file_comments:
print >> OUT, get_call_info()
print >> OUT, '\t'.join(hits_annot_header)
qn = 0
t1 = time.time()
for line in open(hits_file):
if not line.strip() or line.startswith('#'):
continue
qn += 1
if qn and (qn % 10000 == 0):
total_time = time.time() - start_time
print >>sys.stderr, qn, total_time, "%0.2f q/s (refinement)" %\
((float(qn) / total_time))
sys.stderr.flush()
(query, hit, evalue, sum_score, query_length, hmmfrom, hmmto,
seqfrom, seqto, q_coverage) = map(str.strip, line.split('\t'))
if hit not in ['ERROR', '-']:
hitname = cleanup_og_name(hit)
level, nm, desc, cats = annota.get_og_annotations(hitname)
print >> OUT, '\t'.join(
map(str, [
query, hitname, level, evalue, sum_score, query_length,
hmmfrom, hmmto, seqfrom, seqto, q_coverage, nm, desc,
cats
]))
else:
print >> OUT, '\t'.join([query] + [hit] *
(len(hits_annot_header) - 1))
elapsed_time = time.time() - t1
if not args.no_file_comments:
print >> OUT, '# %d queries scanned' % (qn)
print >> OUT, '# Total time (seconds):', elapsed_time
print >> OUT, '# Rate:', "%0.2f q/s" % ((float(qn) / elapsed_time))
OUT.close()
print colorify(" Processed queries:%s total_time:%s rate:%s" %\
(qn, elapsed_time, "%0.2f q/s" % ((float(qn) / elapsed_time))), 'lblue')
def annotate_hmm_matches(hits_file, hits_annot_file, args):
hits_annot_header = map(str.strip, '''#query_name, hit, level, evalue,
sum_score, query_length, hmmfrom, hmmto, seqfrom, seqto, query_coverage,
members_in_og, og_description, og_COG_categories'''.split(','))
annota.connect()
print colorify("Functional annotation of hits starts now", 'green')
start_time = time.time()
if pexists(hits_file):
OUT = open(hits_annot_file, "w")
if not args.no_file_comments:
print >>OUT, get_call_info()
print >>OUT, '\t'.join(hits_annot_header)
qn = 0
t1 = time.time()
for line in open(hits_file):
if not line.strip() or line.startswith('#'):
continue
if qn and (qn % 10000 == 0):
total_time = time.time() - start_time
print >>sys.stderr, qn+1, total_time, "%0.2f q/s (refinement)" %\
((float(qn + 1) / total_time))
sys.stderr.flush()
qn += 1
(query, hit, evalue, sum_score, query_length, hmmfrom, hmmto,
seqfrom, seqto, q_coverage) = map(str.strip, line.split('\t'))
if hit not in ['ERROR', '-']:
hitname = cleanup_og_name(hit)
level, nm, desc, cats = annota.get_og_annotations(hitname)
print >>OUT, '\t'.join(map( str, [query, hitname, level, evalue,
sum_score, query_length,
hmmfrom, hmmto, seqfrom,
seqto, q_coverage, nm, desc,
cats]))
else:
print >>OUT, '\t'.join(
[query] + [hit] * (len(hits_annot_header) - 1))
elapsed_time = time.time() - t1
if not args.no_file_comments:
print >>OUT, '# %d queries scanned' % (qn + 1)
print >>OUT, '# Total time (seconds):', elapsed_time
print >>OUT, '# Rate:', "%0.2f q/s" % ((float(qn + 1) / elapsed_time))
OUT.close()
print colorify(" Processed queries:%s total_time:%s rate:%s" %\
(qn+1, elapsed_time, "%0.2f q/s" % ((float(qn+1) / elapsed_time))), 'lblue')