def test_parse_classic_otu_table_floats_in_table(self):
"""parse_classic_otu_table functions using an OTU table containing floats
but cast as int....this will automatically cast into floats"""
data = self.otu_table1_floats
data_f = data.split("\n")
obs = parse_classic_otu_table(data_f)
exp = (
["Fing", "Key", "NA"],
["0", "1", "2", "3", "4"],
array(
[
[19111.0, 44536.0, 42.0],
[1216.0, 3500.0, 6.0],
[1803.0, 1184.0, 2.0],
[1722.1, 4903.2, 17.0],
[589.6, 2074.4, 34.5],
]
),
self.expected_lineages1,
)
# because float comps in arrays always errors
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
self.assertEqual(obs[3], exp[3])
self.assertTrue(all((obs[2] == exp[2]).tolist()))
def test_parse_classic_otu_table_consensus_lineage(self):
"""parse_classic_otu_table should accept 'consensusLineage'"""
data = """#Full OTU Counts
#OTU ID Fing Key NA consensusLineage
0 19111 44536 42 Bacteria; Actinobacteria; Actinobacteridae; Propionibacterineae; Propionibacterium
1 1216 3500 6 Bacteria; Firmicutes; Alicyclobacillaceae; Bacilli; Lactobacillales; Lactobacillales; Streptococcaceae; Streptococcus
2 1803 1184 2 Bacteria; Actinobacteria; Actinobacteridae; Gordoniaceae; Corynebacteriaceae
3 1722 4903 17 Bacteria; Firmicutes; Alicyclobacillaceae; Bacilli; Staphylococcaceae
4 589 2074 34 Bacteria; Cyanobacteria; Chloroplasts; vectors"""
data_f = StringIO(data)
obs = parse_classic_otu_table(data_f)
exp = (['Fing','Key','NA'],
['0','1','2','3','4'],
array([[19111,44536,42],[1216,3500,6],[1803,1184,2],\
[1722,4903,17], [589,2074,34]]),
[['Bacteria','Actinobacteria','Actinobacteridae','Propionibacterineae','Propionibacterium'],
['Bacteria','Firmicutes','Alicyclobacillaceae','Bacilli','Lactobacillales','Lactobacillales','Streptococcaceae','Streptococcus'],
['Bacteria','Actinobacteria','Actinobacteridae','Gordoniaceae','Corynebacteriaceae'],
['Bacteria','Firmicutes','Alicyclobacillaceae','Bacilli','Staphylococcaceae'],
['Bacteria','Cyanobacteria','Chloroplasts','vectors']])
# because float comps in arrays always errors
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
self.assertEqual(obs[3], exp[3])
self.assertTrue(all((obs[2] == exp[2]).tolist()))
def test_parse_classic_otu_table_float_counts(self):
"""parse_classic_otu_table should return correct result from small table"""
data = """#Full OTU Counts
#OTU ID Fing Key NA Consensus Lineage
0 19111 44536 42 Bacteria; Actinobacteria; Actinobacteridae; Propionibacterineae; Propionibacterium
1 1216 3500 6 Bacteria; Firmicutes; Alicyclobacillaceae; Bacilli; Lactobacillales; Lactobacillales; Streptococcaceae; Streptococcus
2 1803 1184 2 Bacteria; Actinobacteria; Actinobacteridae; Gordoniaceae; Corynebacteriaceae
3 1722 4903 17 Bacteria; Firmicutes; Alicyclobacillaceae; Bacilli; Staphylococcaceae
4 589 2074 34 Bacteria; Cyanobacteria; Chloroplasts; vectors"""
data_f = (data.split('\n'))
obs = parse_classic_otu_table(data_f, count_map_f=float)
exp = (['Fing','Key','NA'],
['0','1','2','3','4'],
array([[19111.,44536.,42.],[1216.,3500.,6.],[1803.,1184.,2.],\
[1722.,4903.,17.], [589,2074.,34.]]),
[['Bacteria','Actinobacteria','Actinobacteridae','Propionibacterineae','Propionibacterium'],
['Bacteria','Firmicutes','Alicyclobacillaceae','Bacilli','Lactobacillales','Lactobacillales','Streptococcaceae','Streptococcus'],
['Bacteria','Actinobacteria','Actinobacteridae','Gordoniaceae','Corynebacteriaceae'],
['Bacteria','Firmicutes','Alicyclobacillaceae','Bacilli','Staphylococcaceae'],
['Bacteria','Cyanobacteria','Chloroplasts','vectors']])
# because float comps in arrays always errors
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
self.assertEqual(obs[3], exp[3])
self.assertTrue(all((obs[2] == exp[2]).tolist()))
def test_parse_classic_otu_table(self):
"""parse_classic_otu_table functions as expected with new-style OTU table
"""
data = self.otu_table1
data_f = data.split("\n")
obs = parse_classic_otu_table(data_f)
exp = (
["Fing", "Key", "NA"],
["0", "1", "2", "3", "4"],
array([[19111, 44536, 42], [1216, 3500, 6], [1803, 1184, 2], [1722, 4903, 17], [589, 2074, 34]]),
self.expected_lineages1,
)
# divide the comparisons into their four elements
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
assert_almost_equal(obs[2], exp[2])
self.assertEqual(obs[3], exp[3])
# test that the modified parse_classic performs correctly on OTU tables
# without leading comments
data = self.otu_table_without_leading_comment
data_f = data.split("\n")
obs = parse_classic_otu_table(data_f)
sams = ["let-7i", "miR-7", "miR-17n", "miR-18a", "miR-19a", "miR-22", "miR-25", "miR-26a"]
otus = ["A2M", "AAAS", "AACS", "AADACL1"]
vals = array(
[
[-0.2, 0.03680505, 0.205, 0.23, 0.66, 0.08, -0.373, 0.26],
[-0.09, -0.25, 0.274, 0.15, 0.12, 0.29, 0.029, -0.1148452],
[0.33, 0.19, 0.27, 0.28, 0.19, 0.25, 0.089, 0.14],
[0.49, -0.92, -0.723, -0.23, 0.08, 0.49, -0.386, -0.64],
]
)
exp = (sams, otus, vals, []) # no lineages
# because float comps in arrays always errors
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
self.assertEqual(obs[3], exp[3])
self.assertTrue(all((obs[2] == exp[2]).tolist()))
def test_parse_classic_otu_table(self):
"""parse_classic_otu_table functions as expected with new-style OTU table
"""
data = self.otu_table1
data_f = (data.split('\n'))
obs = parse_classic_otu_table(data_f)
exp = (['Fing', 'Key', 'NA'], ['0', '1', '2', '3', '4'],
array([[19111, 44536, 42], [1216, 3500, 6], [1803, 1184, 2],
[1722, 4903, 17], [589, 2074,
34]]), self.expected_lineages1)
# divide the comparisons into their four elements
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
assert_almost_equal(obs[2], exp[2])
self.assertEqual(obs[3], exp[3])
# test that the modified parse_classic performs correctly on OTU tables
# without leading comments
data = self.otu_table_without_leading_comment
data_f = (data.split('\n'))
obs = parse_classic_otu_table(data_f)
sams = [
'let-7i', 'miR-7', 'miR-17n', 'miR-18a', 'miR-19a', 'miR-22',
'miR-25', 'miR-26a'
]
otus = ['A2M', 'AAAS', 'AACS', 'AADACL1']
vals = array([\
[-0.2, 0.03680505, 0.205, 0.23, 0.66, 0.08, -0.373, 0.26],
[-0.09, -0.25, 0.274, 0.15, 0.12, 0.29, 0.029, -0.1148452],
[0.33, 0.19, 0.27, 0.28, 0.19, 0.25, 0.089, 0.14],
[0.49, -0.92, -0.723, -0.23, 0.08, 0.49, -0.386, -0.64]])
exp = (sams, otus, vals, []) # no lineages
# because float comps in arrays always errors
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
self.assertEqual(obs[3], exp[3])
self.assertTrue(all((obs[2] == exp[2]).tolist()))
def test_parse_classic_otu_table_floats_in_table(self):
"""parse_classic_otu_table functions using an OTU table containing floats
but cast as int....this will automatically cast into floats"""
data = self.otu_table1_floats
data_f = (data.split('\n'))
obs = parse_classic_otu_table(data_f)
exp = (['Fing', 'Key', 'NA'], ['0', '1', '2', '3', '4'],
array([[19111.0, 44536.0, 42.0], [1216.0, 3500.0, 6.0],
[1803.0, 1184.0, 2.0], [1722.1, 4903.2, 17.0],
[589.6, 2074.4, 34.5]]), self.expected_lineages1)
# because float comps in arrays always errors
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
self.assertEqual(obs[3], exp[3])
self.assertTrue(all((obs[2] == exp[2]).tolist()))
def test_parse_classic_otu_table_legacy(self):
"""parse_classic_otu_table functions as expected with legacy OTU table
"""
data = self.legacy_otu_table1
data_f = (data.split('\n'))
obs = parse_classic_otu_table(data_f)
exp = (['Fing', 'Key', 'NA'], ['0', '1', '2', '3', '4'],
array([[19111, 44536, 42], [1216, 3500, 6], [1803, 1184, 2],
[1722, 4903, 17], [589, 2074,
34]]), self.expected_lineages1)
# divide the comparisons into their four elements
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
assert_almost_equal(obs[2], exp[2])
self.assertEqual(obs[3], exp[3])
def test_parse_classic_otu_table_float_counts(self):
"""parse_classic_otu_table should return correct result from small table"""
data = """#Full OTU Counts
#OTU ID Fing Key NA Consensus Lineage
0 19111 44536 42 Bacteria; Actinobacteria; Actinobacteridae; Propionibacterineae; Propionibacterium
1 1216 3500 6 Bacteria; Firmicutes; Alicyclobacillaceae; Bacilli; Lactobacillales; Lactobacillales; Streptococcaceae; Streptococcus
2 1803 1184 2 Bacteria; Actinobacteria; Actinobacteridae; Gordoniaceae; Corynebacteriaceae
3 1722 4903 17 Bacteria; Firmicutes; Alicyclobacillaceae; Bacilli; Staphylococcaceae
4 589 2074 34 Bacteria; Cyanobacteria; Chloroplasts; vectors"""
data_f = data.split("\n")
obs = parse_classic_otu_table(data_f, count_map_f=float)
exp = (
["Fing", "Key", "NA"],
["0", "1", "2", "3", "4"],
array(
[
[19111.0, 44536.0, 42.0],
[1216.0, 3500.0, 6.0],
[1803.0, 1184.0, 2.0],
[1722.0, 4903.0, 17.0],
[589, 2074.0, 34.0],
]
),
[
["Bacteria", "Actinobacteria", "Actinobacteridae", "Propionibacterineae", "Propionibacterium"],
[
"Bacteria",
"Firmicutes",
"Alicyclobacillaceae",
"Bacilli",
"Lactobacillales",
"Lactobacillales",
"Streptococcaceae",
"Streptococcus",
],
["Bacteria", "Actinobacteria", "Actinobacteridae", "Gordoniaceae", "Corynebacteriaceae"],
["Bacteria", "Firmicutes", "Alicyclobacillaceae", "Bacilli", "Staphylococcaceae"],
["Bacteria", "Cyanobacteria", "Chloroplasts", "vectors"],
],
)
# because float comps in arrays always errors
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
self.assertEqual(obs[3], exp[3])
self.assertTrue(all((obs[2] == exp[2]).tolist()))
def test_parse_classic_otu_table_legacy(self):
"""parse_classic_otu_table functions as expected with legacy OTU table
"""
data = self.legacy_otu_table1
data_f = data.split("\n")
obs = parse_classic_otu_table(data_f)
exp = (
["Fing", "Key", "NA"],
["0", "1", "2", "3", "4"],
array([[19111, 44536, 42], [1216, 3500, 6], [1803, 1184, 2], [1722, 4903, 17], [589, 2074, 34]]),
self.expected_lineages1,
)
# divide the comparisons into their four elements
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
assert_almost_equal(obs[2], exp[2])
self.assertEqual(obs[3], exp[3])
def test_parse_classic_otu_table_consensus_lineage(self):
"""parse_classic_otu_table should accept 'consensusLineage'"""
data = """#Full OTU Counts
#OTU ID Fing Key NA consensusLineage
0 19111 44536 42 Bacteria; Actinobacteria; Actinobacteridae; Propionibacterineae; Propionibacterium
1 1216 3500 6 Bacteria; Firmicutes; Alicyclobacillaceae; Bacilli; Lactobacillales; Lactobacillales; Streptococcaceae; Streptococcus
2 1803 1184 2 Bacteria; Actinobacteria; Actinobacteridae; Gordoniaceae; Corynebacteriaceae
3 1722 4903 17 Bacteria; Firmicutes; Alicyclobacillaceae; Bacilli; Staphylococcaceae
4 589 2074 34 Bacteria; Cyanobacteria; Chloroplasts; vectors"""
data_f = StringIO(data)
obs = parse_classic_otu_table(data_f)
exp = (
["Fing", "Key", "NA"],
["0", "1", "2", "3", "4"],
array([[19111, 44536, 42], [1216, 3500, 6], [1803, 1184, 2], [1722, 4903, 17], [589, 2074, 34]]),
[
["Bacteria", "Actinobacteria", "Actinobacteridae", "Propionibacterineae", "Propionibacterium"],
[
"Bacteria",
"Firmicutes",
"Alicyclobacillaceae",
"Bacilli",
"Lactobacillales",
"Lactobacillales",
"Streptococcaceae",
"Streptococcus",
],
["Bacteria", "Actinobacteria", "Actinobacteridae", "Gordoniaceae", "Corynebacteriaceae"],
["Bacteria", "Firmicutes", "Alicyclobacillaceae", "Bacilli", "Staphylococcaceae"],
["Bacteria", "Cyanobacteria", "Chloroplasts", "vectors"],
],
)
# because float comps in arrays always errors
self.assertEqual(obs[0], exp[0])
self.assertEqual(obs[1], exp[1])
self.assertEqual(obs[3], exp[3])
self.assertTrue(all((obs[2] == exp[2]).tolist()))
