def __gen_db_path(self, sball, bucket_path):
dir_name = os.path.basename(sball.export_fasta).split(".")[0]
dir_name += "_db/"
db_name = os.path.basename(sball.export_fasta).split(".")[0]
db_name += "_genome.fasta"
dir_path = bucket_path + dir_name
plumber.force_dir(dir_path)
return dir_path + db_name
def __split_queries(self):
query_recs = []
for rec in SeqIO.parse(self.query_fpath, "fasta"):
query_recs.append(rec)
query_list = []
query_path = self.base_blast_path + args.query_dir
plumber.force_dir(query_path)
for rec in query_recs:
write_out = [rec]
write_path = query_path + self.__format_id(rec.id) + ".fasta"
SeqIO.write(write_out, write_path, "fasta")
query_list.append(write_path)
return query_list
def bucket_blast(self, archive_path):
blast_path = archive_path + self.__gen_archive_subdir()
plumber.force_dir(blast_path)
for db_path in self.bucket:
out_path = blast_path + self.__gen_out_name(db_path)
blast_cmd = [
"blastn", "-db", db_path, "-query", self.query, "-out",
out_path, "-evalue", "1e-20", "-outfmt", "11"
]
try:
blast_ret = subprocess.run(blast_cmd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
check=True,
universal_newlines=True)
except subprocess.CalledProcessError as perror:
raise (RuntimeError("ERROR: BLAST failed!"))
self.archives.append(out_path)
def __build_bucket(self):
bucket_path = self.base_blast_path + args.bucket_dir
plumber.force_dir(bucket_path)
for sball in self.sample_pit:
print("\t\tMaking database for sample", sball.sample_id)
out_path = self.__gen_db_path(sball, bucket_path)
blast_cmd = [
"makeblastdb", "-in", sball.export_fasta, "-dbtype", "nucl",
"-out", out_path
]
try:
blast_ret = subprocess.run(blast_cmd,
stdout=subprocess.PIPE,
stdin=subprocess.PIPE,
check=True,
universal_newlines=True)
except subprocess.CalledProcessError as perror:
raise (RuntimeError("ERROR: makeblastdb failed!"))
self.db_bucket.append(out_path)
def convert_hr_reports(self, reports_path):
convert_path = reports_path + self.__gen_report_subdir()
plumber.force_dir(convert_path)
for arch_path in self.archives:
out_path = convert_path + self.__gen_report_name(arch_path)
bconvert_cmd = [
"blast_formatter", "-archive", arch_path, "-outfmt", "0",
"-out", out_path
]
#print(bconvert_cmd)
try:
convert_ret = subprocess.run(bconvert_cmd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
check=True,
universal_newlines=True)
except subprocess.CalledProcessError as perror:
raise (RuntimeError("ERROR: Report conversion failed!"))
self.reports.append(out_path)
def get_sub_seqs(self, sub_seq_path):
sub_dir = sub_seq_path + self.query_name + "_sseq/"
plumber.force_dir(sub_dir)
temp_name = "sseq_tab_tempBLASTn6"
for arch_path in self.archives:
out_path = sub_seq_path + temp_name
bconvert_cmd = [
"blast_formatter", "-archive", arch_path, "-outfmt",
"6 qseqid sseqid sseq", "-out", out_path
]
try:
convert_ret = subprocess.run(bconvert_cmd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
check=True,
universal_newlines=True)
except subprocess.CalledProcessError as perror:
raise (RuntimeError("ERROR: Tab conversion failed!"))
output_file = sub_dir + self.__gen_sseq_name(arch_path)
with open(sub_seq_path + temp_name) as tf:
dat = tf.readlines()
sseq_count = len(dat)
if sseq_count != 0:
records = []
for line in dat:
qseqid = line.split("\t")[0]
sseqid = line.split("\t")[1]
sseq = line.split("\t")[2]
record_id = qseqid + "." + sseqid
record_seq = Seq(self.seq_format(sseq))
rec = SeqRecord(record_seq, id=record_id, description="")
records.append(rec)
SeqIO.write(records, output_file, "fasta")
os.remove(sub_seq_path + temp_name)
def __extract_sub_seqs(self):
sub_seq_path = self.base_blast_path + args.sub_seq_dir
plumber.force_dir(sub_seq_path)
for query in self.qSeq_list:
print("\t\tBuilding reports for %s ..." % query.query_name)
query.get_sub_seqs(sub_seq_path)
def __gen_hr_reports(self):
reports_path = self.base_blast_path + args.report_dir
plumber.force_dir(reports_path)
for query in self.qSeq_list:
print("\t\tBuilding reports for %s ..." % query.query_name)
query.convert_hr_reports(reports_path)
def __blast_all(self):
archive_path = self.base_blast_path + args.archive_dir
plumber.force_dir(archive_path)
for query in self.qSeq_list:
print("\t\tBlasting %s against bucket ..." % query.query_name)
query.bucket_blast(archive_path)