def optimize_tss(args_ops):
if len(os.listdir(args_ops.gffs)) == 0:
print("Error: there is no gff files!!!")
sys.exit()
if len(os.listdir(args_ops.fastas)) == 0:
print("Error: there is no fasta files!!!")
sys.exit()
if len(os.listdir(args_ops.wigs)) == 0:
print("Error: there is no wiggle files!!!")
sys.exit()
Multiparser().parser_wig(args_ops.wigs)
Multiparser().parser_gff(args_ops.gffs, None)
Multiparser().parser_fasta(args_ops.fastas)
gff_path = os.path.join(args_ops.gffs, "tmp")
wig_path = os.path.join(args_ops.wigs, "tmp")
fasta_path = os.path.join(args_ops.fastas, "tmp")
for gff in os.listdir(gff_path):
if args_ops.project_strain in gff:
gff_file = os.path.join(gff_path, gff)
break
for fa in os.listdir(fasta_path):
if args_ops.project_strain in fa:
fasta_file = os.path.join(fasta_path, fa)
break
Helper().check_uni_attributes(gff_file)
optimization(wig_path, fasta_file, gff_file, args_ops)
Helper().remove_all_content(os.path.join(args_ops.output_folder,
"optimized_TSSpredator"), "config", "file")
Helper().remove_all_content(os.path.join(args_ops.output_folder,
"optimized_TSSpredator"), "Master", "dir")
Helper().remove_tmp(args_ops.wigs)
Helper().remove_tmp(args_ops.gffs)
Helper().remove_tmp(args_ops.fastas)
def optimize_tss(args_ops):
if len(os.listdir(args_ops.gffs)) == 0:
print("Error: There is no gff files!!!")
sys.exit()
if len(os.listdir(args_ops.fastas)) == 0:
print("Error: There is no fasta files!!!")
sys.exit()
if len(os.listdir(args_ops.wigs)) == 0:
print("Error: There is no wiggle files!!!")
sys.exit()
Multiparser().parser_wig(args_ops.wigs)
Multiparser().parser_gff(args_ops.gffs, None)
Multiparser().parser_fasta(args_ops.fastas)
gff_path = os.path.join(args_ops.gffs, "tmp")
wig_path = os.path.join(args_ops.wigs, "tmp")
fasta_path = os.path.join(args_ops.fastas, "tmp")
for gff in os.listdir(gff_path):
if args_ops.project_strain in gff:
gff_file = os.path.join(gff_path, gff)
break
for fa in os.listdir(fasta_path):
if args_ops.project_strain in fa:
fasta_file = os.path.join(fasta_path, fa)
break
Helper().check_uni_attributes(gff_file)
optimization(wig_path, fasta_file, gff_file, args_ops)
Helper().remove_all_content(
os.path.join(args_ops.output_folder, "optimized_TSSpredator"),
"config", "file")
Helper().remove_all_content(
os.path.join(args_ops.output_folder, "optimized_TSSpredator"),
"Master", "dir")
Helper().remove_tmp_dir(args_ops.wigs)
Helper().remove_tmp_dir(args_ops.gffs)
Helper().remove_tmp_dir(args_ops.fastas)
def test_optimization(self):
ot.run_TSSpredator_paralle = Mock_func().mock_run_TSSpredator_paralle
ot.convert2gff = Mock_func().mock_convert2gff
if not os.path.exists(os.path.join(self.test_folder, "wigs")):
os.mkdir(os.path.join(self.test_folder, "wigs"))
wig_folder = os.path.join(self.test_folder, "wigs", "tmp")
if not os.path.exists(wig_folder):
os.mkdir(wig_folder)
fasta = os.path.join(self.test_folder, "aaa.fa")
gff = os.path.join(self.test_folder, "aaa.gff")
gen_file(fasta, self.example.fasta)
gen_file(gff, self.example.gff_file)
output_prefix = ["test_1"]
args = self.mock_args.mock()
args.libs = self.example.libs
args.cores = 1
args.cluster = 3
args.program = "TSS"
args.project_strain = "aaa"
args.replicate = "all_1"
args.utr = 200
args.steps = 2
args.gene_length = 2000
args.height = 0.9
args.height_reduction = 0.8
args.factor = 0.9
args.factor_reduction = 0.8
args.base_height = 0.01
args.enrichment = 0.5
args.processing = 0.5
args.length = None
args.replicate_name = "test"
args.tsspredator_path = "test"
args.manual = os.path.join(self.test_folder, "manual.gff")
gen_file(args.manual, self.example.manual_file)
log = open(os.path.join(self.test_folder, "test.log"), "w")
args.output_folder = self.test_folder
os.mkdir(os.path.join(self.test_folder, "optimized_TSSpredator"))
ot.optimization(wig_folder, fasta, gff, args, args.manual, 2000, "aaa",
log)
self.assertTrue(
os.path.exists(
os.path.join(self.test_folder, "optimized_TSSpredator",
"stat_aaa.csv")))
def test_optimization(self):
ot.run_TSSpredator_paralle = Mock_func().mock_run_TSSpredator_paralle
ot.convert2gff = Mock_func().mock_convert2gff
if not os.path.exists(os.path.join(self.test_folder, "wigs")):
os.mkdir(os.path.join(self.test_folder, "wigs"))
wig_folder = os.path.join(self.test_folder, "wigs", "tmp")
if not os.path.exists(wig_folder):
os.mkdir(wig_folder)
fasta = os.path.join(self.test_folder, "aaa.fa")
gff = os.path.join(self.test_folder, "aaa.gff")
gen_file(fasta, self.example.fasta)
gen_file(gff, self.example.gff_file)
output_prefix = ["test_1"]
args = self.mock_args.mock()
args.libs = self.example.libs
args.cores = 1
args.cluster = 3
args.program = "TSS"
args.project_strain = "aaa"
args.replicate = "all_1"
args.utr = 200
args.steps = 2
args.gene_length = 2000
args.height = 0.9
args.height_reduction = 0.8
args.factor = 0.9
args.factor_reduction = 0.8
args.base_height = 0.01
args.enrichment = 0.5
args.processing = 0.5
args.length = None
args.replicate_name = "test"
args.tsspredator_path = "test"
args.manual = os.path.join(self.test_folder, "manual.gff")
gen_file(args.manual, self.example.manual_file)
log = open(os.path.join(self.test_folder, "test.log"), "w")
args.output_folder = self.test_folder
os.mkdir(os.path.join(self.test_folder, "optimized_TSSpredator"))
ot.optimization(wig_folder, fasta, gff, args, args.manual, 2000, "aaa", log)
self.assertTrue(os.path.exists(os.path.join(
self.test_folder, "optimized_TSSpredator", "stat_aaa.csv")))
def optimize_tss(args_ops, log):
if len(os.listdir(args_ops.gffs)) == 0:
print("Error: There is no gff file!")
sys.exit()
if len(os.listdir(args_ops.fastas)) == 0:
print("Error: There is no fasta file!")
sys.exit()
if len(os.listdir(args_ops.wigs)) == 0:
print("Error: There is no wiggle file!")
sys.exit()
Multiparser().parser_wig(args_ops.wigs)
Multiparser().parser_gff(args_ops.gffs, None)
Multiparser().parser_fasta(args_ops.fastas)
Multiparser().parser_gff(args_ops.manuals, None)
gff_path = os.path.join(args_ops.gffs, "tmp")
wig_path = os.path.join(args_ops.wigs, "tmp")
fasta_path = os.path.join(args_ops.fastas, "tmp")
manual_path = os.path.join(args_ops.manuals, "tmp")
if "all" not in args_ops.strain_lengths.keys():
for strain in args_ops.strain_lengths.keys():
detect = False
for man in os.listdir(manual_path):
if strain == man.replace(".gff", ""):
detect = True
log.write("The manually-curated set is found - "
"{0}\n".format(os.path.join(manual_path, man)))
if not detect:
log.write(
"The manually-curated set of {0} is not found.\n".format(
strain))
print("Error: There are genomes in --genome_lengths "
"which is not contained in manually-detected "
"TSS gff files!")
sys.exit()
for man in os.listdir(manual_path):
run = False
prefix = man.replace(".gff", "")
man_file = os.path.join(manual_path, man)
if (prefix in args_ops.strain_lengths.keys()):
length = args_ops.strain_lengths[prefix]
run = True
elif ("all" in args_ops.strain_lengths.keys()):
length = "all"
run = True
log.write("The comparing sequence region of {0} is ".format(
prefix, length))
if run:
for gff in os.listdir(gff_path):
if (gff[:-4] == prefix) and (".gff" in gff):
gff_file = os.path.join(gff_path, gff)
break
for fa in os.listdir(fasta_path):
if (".".join(fa.split(".")[:-1]) == prefix) and (".fa" in fa):
fasta_file = os.path.join(fasta_path, fa)
break
if length == "all":
length = get_length(fasta_file)
log.write(str(length) + "\n")
Helper().check_uni_attributes(gff_file)
log.write("Running optimize_TSSpredator.py for optimization.\n")
optimization(wig_path, fasta_file, gff_file, args_ops, man_file,
length, prefix, log)
Helper().remove_all_content(
os.path.join(args_ops.output_folder, "optimized_TSSpredator"),
"config", "file")
Helper().remove_all_content(
os.path.join(args_ops.output_folder, "optimized_TSSpredator"),
"Master", "dir")
Helper().remove_tmp_dir(args_ops.wigs)
Helper().remove_tmp_dir(args_ops.gffs)
Helper().remove_tmp_dir(args_ops.fastas)
Helper().remove_tmp_dir(args_ops.manuals)