def test_extract_blast(self):
esi.read_gff = Mock_func().mock_read_gff
nr_blast = os.path.join(self.test_folder, "nr_table")
gen_file(nr_blast, self.example.blast_nr_all)
srna_blast = os.path.join(self.test_folder, "srna_table")
gen_file(srna_blast, self.example.blast_srna_all)
output_file = os.path.join(self.test_folder, "out.gff")
output_table = os.path.join(self.test_folder, "out.csv")
esi.extract_blast(nr_blast, "test.srna", output_file, output_table,
"nr", None, None)
datas, attributes = extract_info(output_file, "file")
refs, ref_attributes = extract_info(self.example.out_nr_gff, "string")
self.assertEqual(set(datas), set(refs[1:]))
self.assertEqual(set(attributes[0]), set(attributes[0]))
self.assertEqual(set(attributes[1]), set(attributes[1]))
datas = import_data(output_table)
esi.extract_blast(srna_blast, "test.srna", output_file, output_table,
"sRNA", None, None)
datas, attributes = extract_info(output_file, "file")
refs, ref_attributes = extract_info(self.example.out_srna_gff,
"string")
self.assertEqual(set(datas), set(refs[1:]))
self.assertEqual(set(attributes[0]), set(attributes[0]))
self.assertEqual(set(attributes[1]), set(attributes[1]))
datas = import_data(output_table)
self.assertEqual(set(datas),
set(self.example.out_srna_csv.split("\n")))
def _blast(self, database, database_format, data_type, args_srna, prefixs,
program, database_type, e, filters):
if (database is None):
print("Error: No database was assigned!")
else:
if database_format:
database = self._formatdb(database, data_type,
args_srna.out_folder,
args_srna.blastdb, database_type)
for prefix in prefixs:
blast_file = os.path.join(
args_srna.out_folder, "blast_results_and_misc",
"_".join([database_type, "blast", prefix + ".txt"]))
if os.path.exists(blast_file):
os.remove(blast_file)
srna_file = "_".join([self.prefixs["basic"], prefix])
out_file = os.path.join(
args_srna.out_folder,
"_".join(["tmp", database_type, prefix]))
print("Running Blast of {0} in {1}".format(prefix, database))
seq_file = os.path.join(args_srna.out_folder,
"_".join(["sRNA_seq", prefix]))
if (seq_file not in os.listdir(
args_srna.out_folder)) or ((database_type == "nr") and
("sec_str" in filters)):
self.helper.get_seq(
srna_file, os.path.join(self.fasta_path,
prefix + ".fa"), seq_file)
if database_type == "nr":
tmp_plus, tmp_minus = self._get_strand_fasta(
seq_file, args_srna.out_folder)
tmp_blast = os.path.join(args_srna.out_folder,
"blast_results_and_misc",
"tmp_blast.txt")
if os.path.exists(tmp_blast):
os.remove(tmp_blast)
self._run_para_blast(program, database, e, tmp_plus,
tmp_blast, "plus",
args_srna.para_blast)
self._run_para_blast(program, database, e, tmp_minus,
blast_file, "minus",
args_srna.para_blast)
self.helper.merge_file(tmp_blast, blast_file)
os.remove(tmp_plus)
os.remove(tmp_minus)
else:
self._run_para_blast(program, database, e, seq_file,
blast_file, "both",
args_srna.para_blast)
extract_blast(blast_file, srna_file, out_file,
out_file + ".csv", database_type)
shutil.move(out_file, srna_file)
def _blast(self, database, database_format, data_type, args_srna,
prefixs, program, database_type, e):
if (database is None):
print("Error: No database assigned!")
else:
if database_format:
self._formatdb(database, data_type, args_srna.out_folder,
args_srna.blast_path, database_type)
for prefix in prefixs:
blast_file = os.path.join(
args_srna.out_folder, "blast_result_and_misc",
"_".join([database_type, "blast", prefix + ".txt"]))
srna_file = "_".join([self.prefixs["basic"], prefix])
out_file = os.path.join(
args_srna.out_folder,
"_".join(["tmp", database_type, prefix]))
print("Running Blast of {0}".format(prefix))
seq_file = os.path.join(
args_srna.out_folder, "_".join(["sRNA_seq", prefix]))
if seq_file not in os.listdir(args_srna.out_folder):
self.helper.get_seq(
srna_file,
os.path.join(self.fasta_path, prefix + ".fa"),
seq_file)
if database_type == "nr":
tmp_plus, tmp_minus = self._get_strand_fasta(
seq_file, args_srna.out_folder)
tmp_blast = os.path.join("tmp_blast.txt")
self._run_blast(args_srna.blast_path, program, database, e,
tmp_plus, tmp_blast, "plus")
self._run_blast(args_srna.blast_path, program, database, e,
tmp_minus, blast_file, "minus")
self.helper.merge_file(tmp_blast, blast_file)
os.remove(tmp_blast)
os.remove(tmp_plus)
os.remove(tmp_minus)
else:
self._run_blast(args_srna.blast_path, program, database, e,
seq_file, blast_file, "both")
extract_blast(blast_file, srna_file, out_file,
out_file + ".csv", database_type)
shutil.move(out_file, srna_file)
def test_extract_blast(self):
esi.read_gff = Mock_func().mock_read_gff
nr_blast = os.path.join(self.test_folder, "nr_table")
gen_file(nr_blast, self.example.blast_nr_all)
srna_blast = os.path.join(self.test_folder, "srna_table")
gen_file(srna_blast, self.example.blast_srna_all)
output_file = os.path.join(self.test_folder, "out.gff")
output_table = os.path.join(self.test_folder, "out.csv")
esi.extract_blast(nr_blast, "test.srna", output_file, output_table, "nr")
datas, attributes = extract_info(output_file, "file")
refs, ref_attributes = extract_info(self.example.out_nr_gff, "string")
self.assertEqual(set(datas), set(refs[1:]))
self.assertEqual(set(attributes[0]), set(attributes[0]))
self.assertEqual(set(attributes[1]), set(attributes[1]))
datas = import_data(output_table)
# self.assertEqual(set(datas), set(self.example.out_nr_csv.split("\n")))
esi.extract_blast(srna_blast, "test.srna", output_file, output_table, "sRNA")
datas, attributes = extract_info(output_file, "file")
refs, ref_attributes = extract_info(self.example.out_srna_gff, "string")
self.assertEqual(set(datas), set(refs[1:]))
self.assertEqual(set(attributes[0]), set(attributes[0]))
self.assertEqual(set(attributes[1]), set(attributes[1]))
datas = import_data(output_table)
self.assertEqual(set(datas), set(self.example.out_srna_csv.split("\n")))