def mapping():
out_f, hit_f, map_f, taxid_d = None, None, None, {}
hit_f = megablast_output #hit file: BLAST similarity search result (TSV format)
map_f = "/home/nancy/assembly_app/blobtools/blobtools-master/taxon_n" #mapping file (TSV format), in which one column lists a sequence ID (of a subject) and another the NCBI TaxID
map_col_sseqid = "0" #column of mapping file containing sequence IDs (of the subject)
map_col_taxid = "2" #column of mapping file containing the TaxID of the subject
hit_col_qseqid = "0" #column of the hit file containing query ID
hit_col_sseqid = "1" #column of the hit file containing subject ID
hit_col_score = "11" #column of the hit file containing (bit)score
try:
hit_col_qseqid = int(hit_col_qseqid)
hit_col_sseqid = int(hit_col_sseqid)
hit_col_score = int(hit_col_score)
except ValueError:
BtLog.error('41' % (
"--hit_column_qseqid, --hit_column_sseqid and --hit_column_score"
))
if map_f:
if map_col_sseqid and map_col_taxid:
try:
map_col_sseqid = int(map_col_sseqid)
map_col_taxid = int(map_col_taxid)
except ValueError:
BtLog.error('44')
print BtLog.status_d['1'] % ("Mapping file", map_f)
taxid_d = BtIO.parseDict(map_f, map_col_sseqid, map_col_taxid)
out_f = BtIO.getOutFile("taxified", hit_f, "out")
else:
BtLog.error('44')
else:
BtLog.error('41')
output = []
print BtLog.status_d['1'] % ("similarity search result", hit_f)
with open(hit_f) as fh:
for idx, line in enumerate(fh):
col = line.rstrip("\n").split()
qseqid = col[hit_col_qseqid]
sseqid = col[hit_col_sseqid]
score = col[hit_col_score]
tax_id = None
if sseqid not in taxid_d:
BtLog.warn_d['12'] % (sseqid, map_f)
tax_id = taxid_d.get(sseqid, "N/A")
output.append("%s\t%s\t%s\t%s" %
(qseqid, tax_id, score, sseqid))
if output:
with open(out_f, "w") as fh:
print BtLog.status_d['24'] % out_f
fh.write("\n".join(output) + "\n")
def main():
args = docopt(__doc__)
out_f, hit_f, map_f, taxid_d = None, None, None, {}
hit_f = args['--hit_file']
hit_col_qseqid = args['--hit_column_qseqid']
hit_col_sseqid = args['--hit_column_sseqid']
hit_col_score = args['--hit_column_score']
map_f = args['--taxid_mapping_file']
map_col_sseqid = args['--map_col_sseqid']
map_col_taxid = args['--map_col_taxid']
custom_f = args['--custom']
custom_taxid = args['--custom_taxid']
custom_score = args['--custom_score']
prefix = args['--out']
try:
hit_col_qseqid = int(hit_col_qseqid)
hit_col_sseqid = int(hit_col_sseqid)
hit_col_score = int(hit_col_score)
except ValueError:
BtLog.error('41' % (
"--hit_column_qseqid, --hit_column_sseqid and --hit_column_score"))
if custom_taxid:
try:
custom_taxid = int(custom_taxid)
except TypeError:
BtLog.error('26')
out_f = BtIO.getOutFile(hit_f, prefix, "taxID_%s.out" % custom_taxid)
taxid_d = defaultdict(lambda: custom_taxid)
elif map_f:
if map_col_sseqid and map_col_taxid:
try:
map_col_sseqid = int(map_col_sseqid)
map_col_taxid = int(map_col_taxid)
except ValueError:
BtLog.error('44')
print BtLog.status_d['1'] % ("Mapping file", map_f)
taxid_d = BtIO.parseDict(map_f, map_col_sseqid, map_col_taxid)
out_f = BtIO.getOutFile(hit_f, prefix, "taxified.out")
else:
BtLog.error('44')
else:
BtLog.error('41')
output = []
print BtLog.status_d['1'] % ("similarity search result", hit_f)
with open(hit_f) as fh:
for idx, line in enumerate(fh):
col = line.rstrip("\n").split()
qseqid = col[hit_col_qseqid]
sseqid = col[hit_col_sseqid]
score = col[hit_col_score]
tax_id = None
if custom_taxid:
tax_id = taxid_d[sseqid]
else:
if sseqid not in taxid_d:
BtLog.warn_d['12'] % (sseqid, map_f)
tax_id = taxid_d.get(sseqid, "N/A")
output.append("%s\t%s\t%s\t%s" % (qseqid, tax_id, score, sseqid))
if output:
with open(out_f, "w") as fh:
print BtLog.status_d['24'] % out_f
fh.write("\n".join(output) + "\n")