def test_empty_input_succeed(self):
novoalign = tools.novoalign.NovoalignTool()
novoalign.install()
# make the input fasta empty
inFasta = util.file.mkstempfname('.input.fasta')
util.file.touch(inFasta)
novoalign.index_fasta(inFasta)
inBam = os.path.join(util.file.get_test_input_path(), 'empty.bam')
outFasta = util.file.mkstempfname('.refined.fasta')
# run refine_assembly
args = [
inFasta, inBam, outFasta, "--chr_names", 'G5012.3',
"--min_coverage", '3', "--novo_params",
"-r Random -l 30 -g 40 -x 20 -t 502 -c {}".format(_CPUS)
]
args = assembly.parser_refine_assembly(
argparse.ArgumentParser()).parse_args(args)
print(args)
args.func_main(args)
# the expected output is an empty fasta file
self.assertTrue(os.path.isfile(outFasta))
self.assertTrue(os.path.getsize(outFasta) == 0)
def test_ref_assisted_assembly(self):
novoalign = tools.novoalign.NovoalignTool()
novoalign.install()
# prep inputs
orig_ref = os.path.join(util.file.get_test_input_path(), 'ebov-makona.fasta')
refGenome = util.file.mkstempfname('.ref.fasta')
shutil.copyfile(orig_ref, refGenome)
novoalign.index_fasta(refGenome)
inBam = os.path.join(util.file.get_test_input_path(), 'G5012.3.testreads.bam')
outFasta = util.file.mkstempfname('.refined.fasta')
# run refine_assembly
args = [refGenome, inBam, outFasta, "--chr_names", 'G5012.3', "--min_coverage", '3', "--novo_params",
"-r Random -l 30 -g 40 -x 20 -t 502"]
args = assembly.parser_refine_assembly().parse_args(args)
args.func_main(args)
self.assertTrue(os.path.isfile(outFasta))
self.assertTrue(os.path.getsize(outFasta) > 1000)
# check assembly quality
with open(outFasta, 'rt') as inf:
seq = Bio.SeqIO.read(inf, 'fasta')
self.assertGreater(len(seq), 17000)
self.assertGreater(assembly.unambig_count(seq.seq), len(seq) * 0.95)
def test_aligned_empty_input_bam_assembly(self):
mm2 = tools.minimap2.Minimap2()
samtools = tools.samtools.SamtoolsTool()
with util.file.tempfname('.ref.fasta') as inFasta:
with util.file.tempfname('.bam') as inBam:
with util.file.tempfname('.refined.fasta') as outFasta:
with util.file.tempfname('.vcf.gz') as outVcf:
shutil.copyfile(
os.path.join(util.file.get_test_input_path(),
'ebov-makona.fasta'), inFasta)
mm2.align_bam(os.path.join(
util.file.get_test_input_path(), 'empty.bam'),
inFasta,
inBam,
options=['-x', 'sr'])
samtools.index(inBam)
# run refine_assembly
args = [
inFasta, inBam, outFasta,
"--already_realigned_bam", inBam, '--outVcf',
outVcf
]
args = assembly.parser_refine_assembly(
argparse.ArgumentParser()).parse_args(args)
args.func_main(args)
# the expected output is an empty fasta file
self.assertTrue(os.path.isfile(outFasta))
self.assertTrue(os.path.getsize(outFasta) == 0)
self.assertTrue(os.path.isfile(outVcf))
self.assertTrue(os.path.getsize(outVcf) > 0)
def test_empty_input_bam_assembly(self):
novoalign = tools.novoalign.NovoalignTool()
novoalign.install()
# prep inputs
orig_ref = os.path.join(util.file.get_test_input_path(),
'ebov-makona.fasta')
inFasta = util.file.mkstempfname('.ref.fasta')
shutil.copyfile(orig_ref, inFasta)
novoalign.index_fasta(inFasta)
inBam = os.path.join(util.file.get_test_input_path(), 'empty.bam')
outFasta = util.file.mkstempfname('.refined.fasta')
# run refine_assembly
args = [
inFasta, inBam, outFasta, "--chr_names", 'G5012.3',
"--min_coverage", '3', "--novo_params",
"-r Random -l 30 -g 40 -x 20 -t 502"
]
args = assembly.parser_refine_assembly(
argparse.ArgumentParser()).parse_args(args)
args.func_main(args)
# the expected output is an empty fasta file
self.assertTrue(os.path.isfile(outFasta))
self.assertTrue(os.path.getsize(outFasta) == 0)
def test_ref_assisted_assembly(self):
novoalign = tools.novoalign.NovoalignTool()
novoalign.install()
# prep inputs
orig_ref = os.path.join(util.file.get_test_input_path(), 'ebov-makona.fasta')
refGenome = util.file.mkstempfname('.ref.fasta')
shutil.copyfile(orig_ref, refGenome)
novoalign.index_fasta(refGenome)
inBam = os.path.join(util.file.get_test_input_path(), 'G5012.3.testreads.bam')
outFasta = util.file.mkstempfname('.refined.fasta')
# run refine_assembly
args = [refGenome, inBam, outFasta, "--chr_names", 'G5012.3', "--min_coverage", '3', "--novo_params",
"-r Random -l 30 -g 40 -x 20 -t 502"]
args = assembly.parser_refine_assembly().parse_args(args)
args.func_main(args)
self.assertTrue(os.path.isfile(outFasta))
self.assertTrue(os.path.getsize(outFasta) > 1000)
# check assembly quality
with open(outFasta, 'rt') as inf:
seq = Bio.SeqIO.read(inf, 'fasta')
self.assertGreater(len(seq), 17000)
self.assertGreater(assembly.unambig_count(seq.seq), len(seq) * 0.95)
def test_empty_input_fasta_assembly(self):
novoalign = tools.novoalign.NovoalignTool()
novoalign.install()
# make the input fasta empty
inFasta = util.file.mkstempfname('.input.fasta')
util.file.touch(inFasta)
novoalign.index_fasta(inFasta)
inBam = os.path.join(util.file.get_test_input_path(), 'G5012.3.testreads.bam')
outFasta = util.file.mkstempfname('.refined.fasta')
# run refine_assembly
args = [inFasta, inBam, outFasta, "--chr_names", 'G5012.3', "--min_coverage", '3', "--novo_params",
"-r Random -l 30 -g 40 -x 20 -t 502 -c {}".format(_CPUS)]
args = assembly.parser_refine_assembly(argparse.ArgumentParser()).parse_args(args)
args.func_main(args)
# the expected output is an empty fasta file
self.assertTrue(os.path.isfile(outFasta))
self.assertTrue(os.path.getsize(outFasta) == 0)
def test_empty_input_bam_assembly(self):
novoalign = tools.novoalign.NovoalignTool()
novoalign.install()
# prep inputs
orig_ref = os.path.join(util.file.get_test_input_path(), 'ebov-makona.fasta')
inFasta = util.file.mkstempfname('.ref.fasta')
shutil.copyfile(orig_ref, inFasta)
novoalign.index_fasta(inFasta)
inBam = os.path.join(util.file.get_test_input_path(), 'empty.bam')
outFasta = util.file.mkstempfname('.refined.fasta')
# run refine_assembly
args = [inFasta, inBam, outFasta, "--chr_names", 'G5012.3', "--min_coverage", '3', "--novo_params",
"-r Random -l 30 -g 40 -x 20 -t 502"]
args = assembly.parser_refine_assembly(argparse.ArgumentParser()).parse_args(args)
args.func_main(args)
# the expected output is an empty fasta file
self.assertTrue(os.path.isfile(outFasta))
self.assertTrue(os.path.getsize(outFasta) == 0)