def test_split_references(self):
test_file_1 = ('/pbi/dept/secondary/siv/testdata/SA3-RS/lambda/'
'2372215/0007_tiny/Alignment_Results/m150404_1016'
'26_42267_c100807920800000001823174110291514_s1_p'
'0.1.aligned.bam')
test_file_2 = ('/pbi/dept/secondary/siv/testdata/SA3-Sequel/ecoli/'
'315/3150204/r54049_20160508_152025/1_A01/Alignment'
'_Results/m54049_160508_155917.alignmentset.xml')
test_file_3 = ('/pbi/dept/secondary/siv/testdata/SA3-RS/ecoli/'
'tiny-multimovie/Alignment_Results/'
'combined.alignmentset.xml')
NREC1 = len(AlignmentSet(test_file_1))
NREC2 = len(AlignmentSet(test_file_2))
NREC3 = len(AlignmentSet(test_file_3))
NREC = NREC1 + NREC2 + NREC3
self.assertNotEqual(NREC1, 0)
self.assertNotEqual(NREC2, 0)
self.assertNotEqual(NREC3, 0)
self.assertNotEqual(NREC, 0)
ds1 = AlignmentSet(test_file_1, test_file_2, test_file_3)
# used to get total:
#self.assertEqual(sum(1 for _ in ds1), N_RECORDS)
self.assertEqual(len(ds1), NREC)
dss = ds1.split_references(1)
self.assertEqual(len(dss), 1)
self.assertEqual(sum([len(ds_) for ds_ in dss]), NREC)
self.assertEqual(len(ds1), NREC)
self.assertFalse(ds1.filters)
dss = ds1.split_references(12)
self.assertEqual(len(dss), 2)
self.assertEqual(sum([len(ds_) for ds_ in dss]),
NREC)
self.assertEqual(len(set(dss[0].index.tId)), 1)
self.assertEqual(len(set(dss[-1].index.tId)), 1)
self.assertEqual(
dss[0].tid2rname[list(set(dss[0].index.tId))[0]],
'ecoliK12_pbi_March2013')
self.assertEqual(len(dss[0]), NREC2 + NREC3)
self.assertEqual(
dss[-1].tid2rname[list(set(dss[-1].index.tId))[0]],
'lambda_NEB3011')
self.assertEqual(len(dss[-1]), NREC1)
def test_split_references(self):
test_file_1 = ('/pbi/dept/secondary/siv/testdata/SA3-RS/lambda/'
'2372215/0007_tiny/Alignment_Results/m150404_1016'
'26_42267_c100807920800000001823174110291514_s1_p'
'0.1.aligned.bam')
test_file_2 = ('/pbi/dept/secondary/siv/testdata/SA3-Sequel/ecoli/'
'315/3150204/r54049_20160508_152025/1_A01/Alignment'
'_Results/m54049_160508_155917.alignmentset.xml')
test_file_3 = ('/pbi/dept/secondary/siv/testdata/SA3-RS/ecoli/'
'tiny-multimovie/Alignment_Results/'
'combined.alignmentset.xml')
NREC1 = len(AlignmentSet(test_file_1))
NREC2 = len(AlignmentSet(test_file_2))
NREC3 = len(AlignmentSet(test_file_3))
NREC = NREC1 + NREC2 + NREC3
self.assertNotEqual(NREC1, 0)
self.assertNotEqual(NREC2, 0)
self.assertNotEqual(NREC3, 0)
self.assertNotEqual(NREC, 0)
ds1 = AlignmentSet(test_file_1, test_file_2, test_file_3)
# used to get total:
#self.assertEqual(sum(1 for _ in ds1), N_RECORDS)
self.assertEqual(len(ds1), NREC)
dss = ds1.split_references(1)
self.assertEqual(len(dss), 1)
self.assertEqual(sum([len(ds_) for ds_ in dss]), NREC)
self.assertEqual(len(ds1), NREC)
self.assertFalse(ds1.filters)
dss = ds1.split_references(12)
self.assertEqual(len(dss), 2)
self.assertEqual(sum([len(ds_) for ds_ in dss]), NREC)
self.assertEqual(len(set(dss[0].index.tId)), 1)
self.assertEqual(len(set(dss[-1].index.tId)), 1)
self.assertEqual(dss[0].tid2rname[list(set(dss[0].index.tId))[0]],
'ecoliK12_pbi_March2013')
self.assertEqual(len(dss[0]), NREC2 + NREC3)
self.assertEqual(dss[-1].tid2rname[list(set(dss[-1].index.tId))[0]],
'lambda_NEB3011')
self.assertEqual(len(dss[-1]), NREC1)
def test_split_references(self):
test_file_1 = ('/pbi/dept/secondary/siv/testdata/SA3-RS/lambda/'
'2372215/0007_tiny/Alignment_Results/m150404_1016'
'26_42267_c100807920800000001823174110291514_s1_p'
'0.1.aligned.bam')
test_file_2 = ('/pbi/dept/secondary/siv/testdata/SA3-Sequel/ecoli/'
'315/3150204/r54049_20160508_152025/1_A01/Alignment'
'_Results/m54049_160508_155917.alignmentset.xml')
test_file_3 = ('/pbi/dept/secondary/siv/testdata/SA3-RS/ecoli/'
'tiny-multimovie/Alignment_Results/'
'combined.alignmentset.xml')
NREC1 = len(AlignmentSet(test_file_1))
NREC2 = len(AlignmentSet(test_file_2))
NREC3 = len(AlignmentSet(test_file_3))
NREC = NREC1 + NREC2 + NREC3
assert not NREC1 == 0
assert not NREC2 == 0
assert not NREC3 == 0
assert not NREC == 0
ds1 = AlignmentSet(test_file_1, test_file_2, test_file_3)
# used to get total:
#assert sum(1 for _ in ds1) == N_RECORDS
assert len(ds1) == NREC
dss = list(ds1.split_references(1))
assert len(dss) == 1
assert sum([len(ds_) for ds_ in dss]) == NREC
assert len(ds1) == NREC
assert not ds1.filters
dss = list(ds1.split_references(12))
assert len(dss) == 2
assert sum([len(ds_) for ds_ in dss]) == NREC
assert len(set(dss[0].index.tId)) == 1
assert len(set(dss[-1].index.tId)) == 1
assert dss[0].tid2rname[list(set(
dss[0].index.tId))[0]] == 'ecoliK12_pbi_March2013'
assert len(dss[0]) == NREC2 + NREC3
assert dss[-1].tid2rname[list(set(
dss[-1].index.tId))[0]] == 'lambda_NEB3011'
assert len(dss[-1]) == NREC1
