def taxonomy_ref(in_filename, ref_taxonomy_filename, out_filename):
"""
"""
logger = logging.getLogger('otu.taxonomy_ref')
# load the taxonomy file
ref_taxonomy_handler = open(ref_taxonomy_filename, 'r')
ref_taxonomy_reader = csv.reader(ref_taxonomy_handler, delimiter='\t')
ref_taxonomy = dict()
for row in ref_taxonomy_reader:
ref_taxonomy[row[0]] = utils.parse_taxonomy(row[1])
ref_taxonomy_handler.close()
# write the taxonomy file
#logger.info("write the taxonomy file %s" % os.path.basename(out_filename)
out_handle = open(out_filename, 'w')
out_writer = csv.writer(out_handle, delimiter='\t', lineterminator='\n')
for record in SeqIO.parse(in_filename, "fasta"):
out_writer.writerow([record.id, ref_taxonomy[record.id]])
out_handle.close()
def taxonomy_blast(in_filename, ref_filename, ref_taxonomy_filename,
out_filename, task='blastn', num_threads=1, evalue=10e-30,
perc_identity=90):
"""Assign taxonomy.
1) makeblastdb -in REF_FILENAME -parse_seqids -dbtype nucl -out DB_FILENAME
2) blastn -task blastn -db DB_FILENAME -query IN_FILENAME -out BLAST_OUT_FILENAME[.xml]
-evalue EVALUE -perc_identity PERC_IDENTITY -num_threads NUM_THREADS -outfmt 5
3) load reference taxonomy file (REF_TAXONOMY_FILENAME)
4) write the taxonomy file (OUT_FILENAME)
"""
logger = logging.getLogger('otu.taxonomy_blast')
basepath = os.path.splitext(out_filename)[0]
# makeblastdb
db_prefix = basepath + "_BLAST_DB_TMP"
devnull = open(os.devnull, "w")
cmd = ["makeblastdb", "-in", ref_filename, "-parse_seqids", "-dbtype",
"nucl", "-out", db_prefix]
logger.info(' '.join(cmd))
proc = subprocess.Popen(cmd, stdout=devnull, stderr=subprocess.PIPE)
_, out_stderr = proc.communicate()
devnull.close()
if proc.returncode:
logger.error(out_stderr)
raise Exception(out_stderr)
# blastn
blast_out_filename = basepath + "_BLAST_OUT_TMP.xml"
devnull = open(os.devnull, "w")
cmd = ["blastn", "-task", task, "-db", db_prefix, "-query", in_filename,
"-out", blast_out_filename, "-evalue", str(evalue), "-perc_identity",
str(perc_identity), "-num_threads", str(num_threads), "-outfmt", "5"]
logger.info(' '.join(cmd))
proc = subprocess.Popen(cmd, stdout=devnull, stderr=subprocess.PIPE)
_, out_stderr = proc.communicate()
devnull.close()
if proc.returncode:
logger.error(out_stderr)
raise Exception(out_stderr)
logger.info("write the taxonomy file %s" % os.path.basename(out_filename))
# load reference taxonomy file
ref_taxonomy_handler = open(ref_taxonomy_filename, 'r')
ref_taxonomy_reader = csv.reader(ref_taxonomy_handler, delimiter='\t')
ref_taxonomy = dict()
for row in ref_taxonomy_reader:
ref_taxonomy[row[0]] = utils.parse_taxonomy(row[1])
ref_taxonomy_handler.close()
# write the output file
blast_out_handle = open(blast_out_filename, 'r')
blast_out_records = NCBIXML.parse(blast_out_handle)
out_handle = open(out_filename, 'w')
out_writer = csv.writer(out_handle, delimiter='\t', lineterminator='\n')
for record in blast_out_records:
if len(record.alignments) > 0:
best_hit_id = record.alignments[0].hit_id
ta = ref_taxonomy[best_hit_id]
else:
ta = "Unknown"
out_writer.writerow([record.query, ta])
out_handle.close()
blast_out_handle.close()
# delete tmp files
for filename in glob.glob(db_prefix + '*'):
os.remove(filename)
os.remove(blast_out_filename)
