def test_hmmsearch_22(self):
"""Test parsing hmmsearch 2.2 file (text_22_hmmsearch_001.out)"""
res = read(path.join("Hmmer", "text_22_hmmsearch_001.out"), self.fmt)
# first query
self.assertEqual('Peptidase_C1', res.id)
self.assertEqual('Papain family cysteine protease', res.description)
self.assertEqual('hmmsearch', res.program)
self.assertEqual('2.2g', res.version)
self.assertEqual('cysprot1b.fa', res.target)
self.assertEqual(4, len(res))
# first hit
hit = res[0]
self.assertEqual('CATL_RAT', hit.id)
self.assertEqual('<unknown description>', hit.description)
self.assertAlmostEqual(449.4, hit.bitscore)
self.assertAlmostEqual(2e-135, hit.evalue)
self.assertEqual(1, hit.domain_obs_num)
self.assertEqual(1, len(hit))
# first hit, first hsp
hsp = hit[0]
self.assertEqual('Peptidase_C1', hsp.query_id)
self.assertEqual('Papain family cysteine protease', hsp.query_description)
self.assertEqual('CATL_RAT', hit.id)
self.assertEqual('<unknown description>', hit.description)
self.assertEqual(1, hsp.domain_index)
self.assertAlmostEqual(449.4, hsp.bitscore)
self.assertAlmostEqual(2e-135, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(337, hsp.query_end)
self.assertEqual('[]', hsp.query_endtype)
self.assertEqual('lPesfDWReWkggaVtpVKdQGiqCGSCWAFSavgalEgr', str(hsp.query.seq)[:40])
self.assertEqual('IVKNSWGtdWGEnGYfriaRgknksgkneCGIaseasypi', str(hsp.query.seq)[-40:])
self.assertEqual(337, len(hsp.query.seq))
self.assertEqual('+P+++DWRe kg VtpVK+QG qCGSCWAFSa g lEg+',
str(hsp.aln_annotation['similarity'])[:40])
self.assertEqual('+VKNSWG++WG++GY++ia+++n n+CG+a+ asypi',
str(hsp.aln_annotation['similarity'])[-40:])
self.assertEqual(113, hsp.hit_start)
self.assertEqual(332, hsp.hit_end)
self.assertEqual('..', hsp.hit_endtype)
self.assertEqual('IPKTVDWRE-KG-CVTPVKNQG-QCGSCWAFSASGCLEGQ', str(hsp.hit.seq)[:40])
self.assertEqual('LVKNSWGKEWGMDGYIKIAKDRN----NHCGLATAASYPI', str(hsp.hit.seq)[-40:])
self.assertEqual(337, len(hsp.hit.seq))
# last hit
hit = res[-1]
self.assertEqual('PAPA_CARPA', hit.id)
self.assertEqual('<unknown description>', hit.description)
self.assertAlmostEqual(337.7, hit.bitscore)
self.assertAlmostEqual(9e-102, hit.evalue)
self.assertEqual(1, hit.domain_obs_num)
self.assertEqual(1, len(hit))
# last hit, last hsp
hsp = hit[-1]
self.assertEqual('Peptidase_C1', hsp.query_id)
self.assertEqual('Papain family cysteine protease', hsp.query_description)
self.assertEqual('PAPA_CARPA', hit.id)
self.assertEqual('<unknown description>', hit.description)
self.assertEqual(1, hsp.domain_index)
self.assertAlmostEqual(337.7, hsp.bitscore)
self.assertAlmostEqual(9e-102, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(337, hsp.query_end)
self.assertEqual('[]', hsp.query_endtype)
self.assertEqual('lPesfDWReWkggaVtpVKdQGiqCGSCWAFSavgalEgr', str(hsp.query.seq)[:40])
self.assertEqual('IVKNSWGtdWGEnGYfriaRgknksgkneCGIaseasypi', str(hsp.query.seq)[-40:])
self.assertEqual(337, len(hsp.query.seq))
self.assertEqual('+Pe +DWR+ kg aVtpVK+QG +CGSCWAFSav ++Eg+',
str(hsp.aln_annotation['similarity'])[:40])
self.assertEqual('++KNSWGt WGEnGY+ri+Rg+++s ++ CG+ ++ yp+',
str(hsp.aln_annotation['similarity'])[-40:])
self.assertEqual(133, hsp.hit_start)
self.assertEqual(343, hsp.hit_end)
self.assertEqual('..', hsp.hit_endtype)
self.assertEqual('IPEYVDWRQ-KG-AVTPVKNQG-SCGSCWAFSAVVTIEGI', str(hsp.hit.seq)[:40])
self.assertEqual('LIKNSWGTGWGENGYIRIKRGTGNS-YGVCGLYTSSFYPV', str(hsp.hit.seq)[-40:])
self.assertEqual(337, len(hsp.hit.seq))
def test_hmmsearch_22(self):
"""Test parsing hmmsearch 2.2 file (text_22_hmmsearch_001.out)."""
res = read(path.join("Hmmer", "text_22_hmmsearch_001.out"), self.fmt)
# first query
self.assertEqual("Peptidase_C1", res.id)
self.assertEqual("Papain family cysteine protease", res.description)
self.assertEqual("hmmsearch", res.program)
self.assertEqual("2.2g", res.version)
self.assertEqual("cysprot1b.fa", res.target)
self.assertEqual(4, len(res))
# first hit
hit = res[0]
self.assertEqual("CATL_RAT", hit.id)
self.assertEqual("<unknown description>", hit.description)
self.assertAlmostEqual(449.4, hit.bitscore)
self.assertAlmostEqual(2e-135, hit.evalue)
self.assertEqual(1, hit.domain_obs_num)
self.assertEqual(1, len(hit))
# first hit, first hsp
hsp = hit[0]
self.assertEqual("Peptidase_C1", hsp.query_id)
self.assertEqual("Papain family cysteine protease",
hsp.query_description)
self.assertEqual("CATL_RAT", hit.id)
self.assertEqual("<unknown description>", hit.description)
self.assertEqual(1, hsp.domain_index)
self.assertAlmostEqual(449.4, hsp.bitscore)
self.assertAlmostEqual(2e-135, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(337, hsp.query_end)
self.assertEqual("[]", hsp.query_endtype)
self.assertEqual("lPesfDWReWkggaVtpVKdQGiqCGSCWAFSavgalEgr",
str(hsp.query.seq)[:40])
self.assertEqual("IVKNSWGtdWGEnGYfriaRgknksgkneCGIaseasypi",
str(hsp.query.seq)[-40:])
self.assertEqual(337, len(hsp.query.seq))
self.assertEqual(
"+P+++DWRe kg VtpVK+QG qCGSCWAFSa g lEg+",
str(hsp.aln_annotation["similarity"])[:40],
)
self.assertEqual(
"+VKNSWG++WG++GY++ia+++n n+CG+a+ asypi",
str(hsp.aln_annotation["similarity"])[-40:],
)
self.assertEqual(113, hsp.hit_start)
self.assertEqual(332, hsp.hit_end)
self.assertEqual("..", hsp.hit_endtype)
self.assertEqual("IPKTVDWRE-KG-CVTPVKNQG-QCGSCWAFSASGCLEGQ",
str(hsp.hit.seq)[:40])
self.assertEqual("LVKNSWGKEWGMDGYIKIAKDRN----NHCGLATAASYPI",
str(hsp.hit.seq)[-40:])
self.assertEqual(337, len(hsp.hit.seq))
# last hit
hit = res[-1]
self.assertEqual("PAPA_CARPA", hit.id)
self.assertEqual("<unknown description>", hit.description)
self.assertAlmostEqual(337.7, hit.bitscore)
self.assertAlmostEqual(9e-102, hit.evalue)
self.assertEqual(1, hit.domain_obs_num)
self.assertEqual(1, len(hit))
# last hit, last hsp
hsp = hit[-1]
self.assertEqual("Peptidase_C1", hsp.query_id)
self.assertEqual("Papain family cysteine protease",
hsp.query_description)
self.assertEqual("PAPA_CARPA", hit.id)
self.assertEqual("<unknown description>", hit.description)
self.assertEqual(1, hsp.domain_index)
self.assertAlmostEqual(337.7, hsp.bitscore)
self.assertAlmostEqual(9e-102, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(337, hsp.query_end)
self.assertEqual("[]", hsp.query_endtype)
self.assertEqual("lPesfDWReWkggaVtpVKdQGiqCGSCWAFSavgalEgr",
str(hsp.query.seq)[:40])
self.assertEqual("IVKNSWGtdWGEnGYfriaRgknksgkneCGIaseasypi",
str(hsp.query.seq)[-40:])
self.assertEqual(337, len(hsp.query.seq))
self.assertEqual(
"+Pe +DWR+ kg aVtpVK+QG +CGSCWAFSav ++Eg+",
str(hsp.aln_annotation["similarity"])[:40],
)
self.assertEqual(
"++KNSWGt WGEnGY+ri+Rg+++s ++ CG+ ++ yp+",
str(hsp.aln_annotation["similarity"])[-40:],
)
self.assertEqual(133, hsp.hit_start)
self.assertEqual(343, hsp.hit_end)
self.assertEqual("..", hsp.hit_endtype)
self.assertEqual("IPEYVDWRQ-KG-AVTPVKNQG-SCGSCWAFSAVVTIEGI",
str(hsp.hit.seq)[:40])
self.assertEqual("LIKNSWGTGWGENGYIRIKRGTGNS-YGVCGLYTSSFYPV",
str(hsp.hit.seq)[-40:])
self.assertEqual(337, len(hsp.hit.seq))
def test_hmmsearch_20(self):
"""Test parsing hmmsearch 2.0 file (text_20_hmmsearch_001.out)"""
res = read(path.join("Hmmer", "text_20_hmmsearch_001.out"), self.fmt)
# first query
self.assertEqual('SEED', res.id)
self.assertEqual('<unknown description>', res.description)
self.assertEqual('hmmsearch', res.program)
self.assertEqual('2.0', res.version)
self.assertEqual('HMM.dbtemp.29591', res.target)
self.assertEqual(751, len(res))
# first hit
hit = res[0]
self.assertEqual('PAB2_ARATH', hit.id)
self.assertEqual('P42731 POLYADENYLATE-BINDING PROTEIN 2 (PO', hit.description)
self.assertAlmostEqual(393.8, hit.bitscore)
self.assertAlmostEqual(6.1e-114, hit.evalue)
self.assertEqual(4, hit.domain_obs_num)
self.assertEqual(4, len(hit))
# first hit, first hsp
hsp = hit[0]
self.assertEqual('SEED', hsp.query_id)
self.assertEqual('<unknown description>', hsp.query_description)
self.assertEqual('PAB2_ARATH', hsp.hit_id)
self.assertEqual('P42731 POLYADENYLATE-BINDING PROTEIN 2 (PO', hsp.hit_description)
self.assertEqual(3, hsp.domain_index)
self.assertAlmostEqual(109.1, hsp.bitscore)
self.assertAlmostEqual(3e-28, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(77, hsp.query_end)
self.assertEqual('[]', hsp.query_endtype)
self.assertEqual(216, hsp.hit_start)
self.assertEqual(287, hsp.hit_end)
self.assertEqual('..', hsp.hit_endtype)
# first hit, last hsp
hsp = hit[-1]
self.assertEqual('SEED', hsp.query_id)
self.assertEqual('<unknown description>', hsp.query_description)
self.assertEqual('PAB2_ARATH', hsp.hit_id)
self.assertEqual('P42731 POLYADENYLATE-BINDING PROTEIN 2 (PO', hsp.hit_description)
self.assertEqual(1, hsp.domain_index)
self.assertAlmostEqual(92.1, hsp.bitscore)
self.assertAlmostEqual(3.9e-23, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(77, hsp.query_end)
self.assertEqual('[]', hsp.query_endtype)
self.assertEqual(37, hsp.hit_start)
self.assertEqual(109, hsp.hit_end)
self.assertEqual('..', hsp.hit_endtype)
# last hit
hit = res[-1]
self.assertEqual('O00369', hit.id)
self.assertEqual('O00369 L1 ELEMENT L1.20 P40 AND PUTATIVE P', hit.description)
self.assertAlmostEqual(-23.8, hit.bitscore)
self.assertAlmostEqual(9.9e+02, hit.evalue)
self.assertEqual(1, hit.domain_obs_num)
self.assertEqual(1, len(hit))
# last hit, first hsp
hsp = hit[0]
self.assertEqual('SEED', hsp.query_id)
self.assertEqual('<unknown description>', hsp.query_description)
self.assertEqual('O00369', hsp.hit_id)
self.assertEqual('O00369 L1 ELEMENT L1.20 P40 AND PUTATIVE P', hsp.hit_description)
self.assertEqual(1, hsp.domain_index)
self.assertAlmostEqual(-23.8, hsp.bitscore)
self.assertAlmostEqual(9.9e+02, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(77, hsp.query_end)
self.assertEqual('[]', hsp.query_endtype)
self.assertEqual(180, hsp.hit_start)
self.assertEqual(249, hsp.hit_end)
self.assertEqual('..', hsp.hit_endtype)
def test_hmmsearch_20(self):
"""Test parsing hmmsearch 2.0 file (text_20_hmmsearch_001.out)."""
res = read(path.join("Hmmer", "text_20_hmmsearch_001.out"), self.fmt)
# first query
self.assertEqual("SEED", res.id)
self.assertEqual("<unknown description>", res.description)
self.assertEqual("hmmsearch", res.program)
self.assertEqual("2.0", res.version)
self.assertEqual("HMM.dbtemp.29591", res.target)
self.assertEqual(751, len(res))
# first hit
hit = res[0]
self.assertEqual("PAB2_ARATH", hit.id)
self.assertEqual("P42731 POLYADENYLATE-BINDING PROTEIN 2 (PO",
hit.description)
self.assertAlmostEqual(393.8, hit.bitscore)
self.assertAlmostEqual(6.1e-114, hit.evalue)
self.assertEqual(4, hit.domain_obs_num)
self.assertEqual(4, len(hit))
# first hit, first hsp
hsp = hit[0]
self.assertEqual("SEED", hsp.query_id)
self.assertEqual("<unknown description>", hsp.query_description)
self.assertEqual("PAB2_ARATH", hsp.hit_id)
self.assertEqual("P42731 POLYADENYLATE-BINDING PROTEIN 2 (PO",
hsp.hit_description)
self.assertEqual(3, hsp.domain_index)
self.assertAlmostEqual(109.1, hsp.bitscore)
self.assertAlmostEqual(3e-28, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(77, hsp.query_end)
self.assertEqual("[]", hsp.query_endtype)
self.assertEqual(216, hsp.hit_start)
self.assertEqual(287, hsp.hit_end)
self.assertEqual("..", hsp.hit_endtype)
# first hit, last hsp
hsp = hit[-1]
self.assertEqual("SEED", hsp.query_id)
self.assertEqual("<unknown description>", hsp.query_description)
self.assertEqual("PAB2_ARATH", hsp.hit_id)
self.assertEqual("P42731 POLYADENYLATE-BINDING PROTEIN 2 (PO",
hsp.hit_description)
self.assertEqual(1, hsp.domain_index)
self.assertAlmostEqual(92.1, hsp.bitscore)
self.assertAlmostEqual(3.9e-23, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(77, hsp.query_end)
self.assertEqual("[]", hsp.query_endtype)
self.assertEqual(37, hsp.hit_start)
self.assertEqual(109, hsp.hit_end)
self.assertEqual("..", hsp.hit_endtype)
# last hit
hit = res[-1]
self.assertEqual("O00369", hit.id)
self.assertEqual("O00369 L1 ELEMENT L1.20 P40 AND PUTATIVE P",
hit.description)
self.assertAlmostEqual(-23.8, hit.bitscore)
self.assertAlmostEqual(9.9e02, hit.evalue)
self.assertEqual(1, hit.domain_obs_num)
self.assertEqual(1, len(hit))
# last hit, first hsp
hsp = hit[0]
self.assertEqual("SEED", hsp.query_id)
self.assertEqual("<unknown description>", hsp.query_description)
self.assertEqual("O00369", hsp.hit_id)
self.assertEqual("O00369 L1 ELEMENT L1.20 P40 AND PUTATIVE P",
hsp.hit_description)
self.assertEqual(1, hsp.domain_index)
self.assertAlmostEqual(-23.8, hsp.bitscore)
self.assertAlmostEqual(9.9e02, hsp.evalue)
self.assertEqual(0, hsp.query_start)
self.assertEqual(77, hsp.query_end)
self.assertEqual("[]", hsp.query_endtype)
self.assertEqual(180, hsp.hit_start)
self.assertEqual(249, hsp.hit_end)
self.assertEqual("..", hsp.hit_endtype)