def test_varicella_big(self):
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.hhv3.fasta')
assembly.order_and_orient(os.path.join(inDir, 'contigs.hhv3.fasta'),
os.path.join(inDir, 'ref.hhv3.fasta'),
outFasta)
self.assertEqual(str(Bio.SeqIO.read(outFasta, 'fasta').seq),
str(Bio.SeqIO.read(expected, 'fasta').seq))
def test_multi_overlap(self):
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.ebov.small.fasta')
assembly.order_and_orient(os.path.join(inDir, 'contigs.ebov.fasta'),
os.path.join(inDir, 'ref.ebov.small.fasta'),
outFasta)
self.assertEqual(str(Bio.SeqIO.read(outFasta, 'fasta').seq),
str(Bio.SeqIO.read(expected, 'fasta').seq))
def setUp(self):
super(TestOrderOrientAndImputeFromReference, self).setUp()
self.inDir = util.file.get_test_input_path(self)
self.refFasta = os.path.join(self.inDir, 'ref.influenza_partial.fasta')
self.outOrientFasta = util.file.mkstempfname('.fasta')
assembly.order_and_orient(
os.path.join(self.inDir, 'contigs.influenza.fasta'), self.refFasta,
self.outOrientFasta)
def test_influenza_multisegment(self):
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.influenza.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.influenza.fasta'),
os.path.join(inDir, 'ref.influenza.fasta'), outFasta)
self.assertEqualContents(outFasta, expected)
os.unlink(outFasta)
def test_ebov_palindrome_refsel(self):
# this tests a scenario where show-aligns has more alignments than show-tiling
with util.file.tempfnames(('.out.fasta', '.stats.tsv')) as (outFasta, outStats):
contigs, refs, expected, expectedStats = self.inputs('contigs.ebov.doublehit.fasta',
'refs.ebov.fasta',
'expected.ebov.doublehit.fasta',
'expected.refsel.ebov.stats.tsv')
assembly.order_and_orient(contigs, refs, outFasta, n_genome_segments=1, outStats=outStats)
self.assertEqualFastaSeqs(outFasta, expected)
self.assertEqualContents(outStats, expectedStats)
def test_hiv_wraparound(self):
# this tests a misassembly from Trinity and checks that we still use some of the contig
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.hiv.wrapped.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.hiv.wrapped.fasta'),
os.path.join(inDir, 'ref.hiv.fasta'), outFasta)
self.assertEqual(str(Bio.SeqIO.read(outFasta, 'fasta').seq),
str(Bio.SeqIO.read(expected, 'fasta').seq))
def test_lassa_protein(self):
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.lasv.promer.fasta')
assembly.order_and_orient(os.path.join(inDir, 'contigs.lasv.fasta'),
os.path.join(inDir, 'ref.lasv.fasta'),
outFasta,
aligner='promer')
self.assertEqualContents(outFasta, expected)
os.unlink(outFasta)
def test_ebov_palindrome_refsel(self):
# this tests a scenario where show-aligns has more alignments than show-tiling
with util.file.tempfnames(('.out.fasta', '.stats.tsv')) as (outFasta, outStats):
contigs, refs, expected, expectedStats = self.inputs('contigs.ebov.doublehit.fasta',
'refs.ebov.fasta',
'expected.ebov.doublehit.fasta',
'expected.refsel.ebov.stats.tsv')
assembly.order_and_orient(contigs, refs, outFasta, n_genome_segments=1, outStats=outStats)
self.assertEqualFastaSeqs(outFasta, expected)
self.assertEqualContents(outStats, expectedStats)
def setUp(self):
super(TestOrderOrientAndImputeFromReference, self).setUp()
self.inDir = util.file.get_test_input_path(self)
self.refFasta = os.path.join(self.inDir, 'ref.influenza_partial.fasta')
self.outOrientFasta = util.file.mkstempfname('.fasta')
assembly.order_and_orient(
os.path.join(self.inDir, 'contigs.influenza.fasta'),
self.refFasta,
self.outOrientFasta)
def test_influenza_multisegment(self):
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.influenza.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.influenza.fasta'),
os.path.join(inDir, 'ref.influenza.fasta'),
outFasta)
self.assertEqualContents(outFasta, expected)
os.unlink(outFasta)
def test_ebov_palindrome(self):
# this tests a scenario where show-aligns has more alignments than show-tiling
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.ebov.doublehit.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.ebov.doublehit.fasta'),
os.path.join(util.file.get_test_input_path(), 'ebov-makona.fasta'),
outFasta)
self.assertEqual(str(Bio.SeqIO.read(outFasta, 'fasta').seq),
str(Bio.SeqIO.read(expected, 'fasta').seq))
def test_varicella_big(self):
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.hhv3.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.hhv3.fasta'),
os.path.join(inDir, 'ref.hhv3.fasta'),
outFasta)
self.assertEqual(
str(Bio.SeqIO.read(outFasta, 'fasta').seq),
str(Bio.SeqIO.read(expected, 'fasta').seq))
def test_lassa_protein(self):
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.lasv.promer.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.lasv.fasta'),
os.path.join(inDir, 'ref.lasv.fasta'),
outFasta,
aligner='promer')
self.assertEqualContents(outFasta, expected)
os.unlink(outFasta)
def test_multi_overlap(self):
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.ebov.small.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.ebov.fasta'),
os.path.join(inDir, 'ref.ebov.small.fasta'),
outFasta)
self.assertEqual(
str(Bio.SeqIO.read(outFasta, 'fasta').seq),
str(Bio.SeqIO.read(expected, 'fasta').seq))
def test_hiv_wraparound(self):
# this tests a misassembly from Trinity and checks that we still use some of the contig
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.hiv.wrapped.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.hiv.wrapped.fasta'),
os.path.join(inDir, 'ref.hiv.fasta'),
outFasta)
self.assertEqual(
str(Bio.SeqIO.read(outFasta, 'fasta').seq),
str(Bio.SeqIO.read(expected, 'fasta').seq))
def test_ebov_palindrome(self):
# this tests a scenario where show-aligns has more alignments than show-tiling
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.ebov.doublehit.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.ebov.doublehit.fasta'),
os.path.join(util.file.get_test_input_path(), 'ebov-makona.fasta'),
outFasta)
self.assertEqual(
str(Bio.SeqIO.read(outFasta, 'fasta').seq),
str(Bio.SeqIO.read(expected, 'fasta').seq))
def test_lassa_multisegment_refsel(self):
with util.file.tempfnames(('.out.fasta', '.out_ref.fasta', '.stats.tsv')) \
as (outFasta, outReference, outStats):
contigs, expected, expectedStats = self.inputs('contigs.lasv.fasta',
'expected.lasv.fasta',
'expected.refsel.lasv.stats.tsv')
refs = [self.input('ref.lasv.{}.fasta'.format(strain))
for strain in ('josiah', 'pinneo', 'KGH_G502', 'BNI_Nig08_A19', 'nomatch')]
assembly.order_and_orient(contigs, refs, outFasta,
outReference=outReference, outStats=outStats)
self.assertEqualContents(outFasta, expected)
self.assertEqualFasta(outReference, refs[0])
self.assertEqualContents(outStats, expectedStats)
def test_lassa_multisegment_refsel(self):
with util.file.tempfnames(('.out.fasta', '.out_ref.fasta', '.stats.tsv')) \
as (outFasta, outReference, outStats):
contigs, expected, expectedStats = self.inputs('contigs.lasv.fasta',
'expected.lasv.fasta',
'expected.refsel.lasv.stats.tsv')
refs = [self.input('ref.lasv.{}.fasta'.format(strain))
for strain in ('josiah', 'pinneo', 'KGH_G502', 'BNI_Nig08_A19', 'nomatch')]
assembly.order_and_orient(contigs, refs, outFasta,
outReference=outReference, outStats=outStats)
self.assertEqualContents(outFasta, expected)
self.assertEqualFasta(outReference, refs[0])
self.assertEqualContents(outStats, expectedStats)
def test_alternate_contigs(self):
# this tests that --outAlternateContigs works as expected
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
altFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.hiv.big_indel.fasta')
expectedAlt = os.path.join(inDir, 'expected.hiv.big_indel.alternates.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.hiv.big_indel.fasta'),
os.path.join(inDir, 'ref.hiv.fasta'),
outFasta,
outAlternateContigs=altFasta)
self.assertEqualContents(outFasta, expected)
self.assertEqualContents(altFasta, expectedAlt)
def test_alternate_contigs(self):
# this tests that --outAlternateContigs works as expected
inDir = util.file.get_test_input_path(self)
outFasta = util.file.mkstempfname('.fasta')
altFasta = util.file.mkstempfname('.fasta')
expected = os.path.join(inDir, 'expected.hiv.big_indel.fasta')
expectedAlt = os.path.join(inDir, 'expected.hiv.big_indel.alternates.fasta')
assembly.order_and_orient(
os.path.join(inDir, 'contigs.hiv.big_indel.fasta'),
os.path.join(inDir, 'ref.hiv.fasta'),
outFasta,
outAlternateContigs=altFasta)
self.assertEqualContents(outFasta, expected)
self.assertEqualContents(altFasta, expectedAlt)
def test_ambig_align(self):
inDir = util.file.get_test_input_path(self)
contigs_gz = os.path.join(inDir, 'contigs.lasv.ambig.fasta.gz')
contigs = util.file.mkstempfname('.fasta')
with util.file.open_or_gzopen(contigs_gz, 'rb') as f_in:
with open(contigs, 'wb') as f_out:
shutil.copyfileobj(f_in, f_out)
expected = os.path.join(inDir, 'expected.lasv.ambig.fasta')
outFasta = util.file.mkstempfname('.fasta')
assembly.order_and_orient(
contigs, os.path.join(inDir, 'ref.lasv.ISTH2376.fasta'), outFasta)
def get_seqs(fasta):
return [str(s.seq) for s in Bio.SeqIO.parse(fasta, 'fasta')]
self.assertEqual(get_seqs(outFasta), get_seqs(expected))
def test_ambig_align_ebov(self):
inDir = util.file.get_test_input_path(self)
contigs_gz = os.path.join(inDir, 'contigs.ebov.ambig.fasta.gz')
contigs = util.file.mkstempfname('.fasta')
with util.file.open_or_gzopen(contigs_gz, 'rb') as f_in:
with open(contigs, 'wb') as f_out:
shutil.copyfileobj(f_in, f_out)
expected = os.path.join(inDir, 'expected.ebov.ambig.fasta')
outFasta = util.file.mkstempfname('.fasta')
assembly.order_and_orient(
contigs,
os.path.join(inDir, 'ref.ebov.makona_C15.fasta'),
outFasta)
def get_seqs(fasta):
return [str(s.seq) for s in Bio.SeqIO.parse(fasta, 'fasta')]
self.assertEqual(get_seqs(outFasta), get_seqs(expected))