def test_add_tag_zeroInputsIncluded(self):
rec1 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "4"
},
"SB": {
"X": "0"
}
})
rec2 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "0"
},
"SB": {
"X": "8"
}
})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
tag.add_tag_values(rec1)
sampleA_tag_values = rec1.sample_tag_values["SA"]
self.assertEquals("0.3015", sampleA_tag_values["ZScoreX"])
sampleB_tag_values = rec1.sample_tag_values["SB"]
self.assertEquals("-0.9045", sampleB_tag_values["ZScoreX"])
def test_add_tag_nullInputsProduceNullZScores(self):
rec1 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "4"
},
"SB": {
"X": "."
}
})
rec2 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "."
},
"SB": {
"X": "8"
}
})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
tag.add_tag_values(rec1)
sampleA_tag_values = rec1.sample_tag_values["SA"]
self.assertEquals("-1.0", sampleA_tag_values["ZScoreX"])
sampleB_tag_values = rec1.sample_tag_values["SB"]
self.assertEquals(".", sampleB_tag_values["ZScoreX"])
def test_add_tag_doesNothingIfNoStdev(self):
rec1 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "4"
},
"SB": {
"X": "4"
}
})
rec2 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "4"
},
"SB": {
"X": "4"
}
})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
tag.add_tag_values(rec1)
self.assertEquals(0, tag._stdev)
self.assertEqual(["X"], sorted(rec1.sample_tag_values["SA"].keys()))
self.assertEqual(["X"], sorted(rec1.sample_tag_values["SB"].keys()))
def test_add_tag(self):
rec1 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "4"
},
"SB": {
"X": "7"
}
})
rec2 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "13"
},
"SB": {
"X": "16"
}
})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
tag.add_tag_values(rec1)
sampleA_tag_values = rec1.sample_tag_values["SA"]
self.assertEquals("-1.2649", sampleA_tag_values["ZScoreX"])
sampleB_tag_values = rec1.sample_tag_values["SB"]
self.assertEquals("-0.6325", sampleB_tag_values["ZScoreX"])
def test_init_setsPopulationStatistics(self):
rec1 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "4"
},
"SB": {
"X": "7"
}
})
rec2 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "13"
},
"SB": {
"X": "16"
}
})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
values = [4, 7, 13, 16]
self.assertAlmostEquals(mean(values), tag._mean,
_ZScoreTag._MAX_PRECISION)
self.assertAlmostEquals(stdev(values), tag._stdev,
_ZScoreTag._MAX_PRECISION)
def test_init_setsPopulationStatisticsAssignsStddevCorrectlyWhenOneValue(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"."}})
reader = MockVcfReader(records=[rec1])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
values = [4]
self.assertAlmostEquals(mean(values), tag._mean, _ZScoreTag._MAX_PRECISION)
self.assertAlmostEquals(stdev(values), tag._stdev, _ZScoreTag._MAX_PRECISION)
def test_init_setsPopulationStatisticsAssignsStddevCorrectlyWhenNoValues(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"."}, "SB":{"X":"."}})
reader = MockVcfReader(records=[rec1])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
expected_mean = None
expected_stdev = None
self.assertEquals(expected_mean, tag._mean)
self.assertEquals(expected_stdev, tag._stdev)
def test_init_setsPopulationStatisticsRoundsTo13digits(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"1"},
"SB":{"X":"1"},
"SC":{"X" : "0"}})
reader = MockVcfReader(records=[rec1])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
self.assertEquals(round(2/3, 13), tag._mean, repr(tag._mean))
self.assertEquals(round(stdev([1,1,0]), 13), tag._stdev)
def test_init_setsPopulationStatisticsUsingMaxRangeForMultiValuedInput(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"7"}})
rec2 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"13,0"}, "SB":{"X":"0,16"}})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
values = [4, 7, 13, 16]
self.assertAlmostEquals(mean(values), tag._mean, _ZScoreTag._MAX_PRECISION)
self.assertAlmostEquals(stdev(values), tag._stdev, _ZScoreTag._MAX_PRECISION)
def test_init_setsPopulationStatisticsParsesFloats(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"2"}, "SB":{"X":"3.5"}})
rec2 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"6.5"}, "SB":{"X":"8"}})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
values = [2, 3.5, 6.5, 8]
self.assertAlmostEquals(mean(values), tag._mean, _ZScoreTag._MAX_PRECISION)
self.assertAlmostEquals(stdev(values), tag._stdev, _ZScoreTag._MAX_PRECISION)
def test_add_tag(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"7"}})
rec2 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"13"}, "SB":{"X":"16"}})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
tag.add_tag_values(rec1)
sampleA_tag_values = rec1.sample_tag_values["SA"]
self.assertEquals("-1.2649", sampleA_tag_values["ZScoreX"])
sampleB_tag_values = rec1.sample_tag_values["SB"]
self.assertEquals("-0.6325", sampleB_tag_values["ZScoreX"])
def test_add_tag_nullInputsProduceNullZScores(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"."}})
rec2 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"."}, "SB":{"X":"8"}})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
tag.add_tag_values(rec1)
sampleA_tag_values = rec1.sample_tag_values["SA"]
self.assertEquals("-1.0", sampleA_tag_values["ZScoreX"])
sampleB_tag_values = rec1.sample_tag_values["SB"]
self.assertEquals(".", sampleB_tag_values["ZScoreX"])
def test_add_tag_zeroInputsIncluded(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"0"}})
rec2 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"0"}, "SB":{"X":"8"}})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
tag.add_tag_values(rec1)
sampleA_tag_values = rec1.sample_tag_values["SA"]
self.assertEquals("0.3015", sampleA_tag_values["ZScoreX"])
sampleB_tag_values = rec1.sample_tag_values["SB"]
self.assertEquals("-0.9045", sampleB_tag_values["ZScoreX"])
def test_add_tag_doesNothingIfNoStdev(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"4"}})
rec2 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"4"}})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
tag.add_tag_values(rec1)
self.assertEquals(0, tag._stdev)
self.assertEqual(["X"], sorted(rec1.sample_tag_values["SA"].keys()))
self.assertEqual(["X"], sorted(rec1.sample_tag_values["SB"].keys()))
def test_init_setsPopulationStatisticsConsidersZeros(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"0"}})
rec2 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"7"}, "SB":{"X":"13"}})
rec3 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"16"}, "SB":{"X":"0"}})
reader = MockVcfReader(records=[rec1, rec2, rec3])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
values = [4, 0, 7, 13, 16, 0]
self.assertAlmostEquals(mean(values), float(tag._mean), _ZScoreTag._MAX_PRECISION)
self.assertAlmostEquals(stdev(values), float(tag._stdev), _ZScoreTag._MAX_PRECISION)
def test_init_setsPopulationStatisticsIgnoresUnparsableValues(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"7"}})
rec2 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"A1"}, "SB":{"X":"2A"}})
rec3 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"13"}, "SB":{"X":"16"}})
reader = MockVcfReader(records=[rec1, rec2, rec3])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
values = [4, 7, 13, 16]
self.assertAlmostEquals(mean(values), tag._mean, _ZScoreTag._MAX_PRECISION)
self.assertAlmostEquals(stdev(values), tag._stdev, _ZScoreTag._MAX_PRECISION)
def test_init_metaheaders(self):
rec1 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"4"}, "SB":{"X":"7"}})
rec2 = vcf.VcfRecord("1", "42", "A", "C",
sample_tag_values={"SA":{"X":"13"}, "SB":{"X":"16"}})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag("ZScoreX",
'##FORMAT=<ID=ZScoreX,Number=1,Type=Float,Description="ZScore for X">',
"X",
reader)
self.assertEquals(3, len(tag.metaheaders))
it = iter(tag.metaheaders)
self.assertEquals(next(it), '##jacquard.summarize.ZScoreX.X_mean=' + repr(tag._mean))
self.assertEquals(next(it), '##jacquard.summarize.ZScoreX.X_stdev=' + repr(tag._stdev))
self.assertRegexpMatches(next(it), '##FORMAT=<ID=ZScoreX,Number=1,Type=Float,Description="ZScore for X">')
def test_init_metaheaders(self):
rec1 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "4"
},
"SB": {
"X": "7"
}
})
rec2 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "13"
},
"SB": {
"X": "16"
}
})
reader = MockVcfReader(records=[rec1, rec2])
tag = zscore_caller._ZScoreTag(
"ZScoreX",
'##FORMAT=<ID=ZScoreX,Number=1,Type=Float,Description="ZScore for X">',
"X", reader)
self.assertEquals(3, len(tag.metaheaders))
it = iter(tag.metaheaders)
self.assertEquals(
next(it), '##jacquard.summarize.ZScoreX.X_mean=' + repr(tag._mean))
self.assertEquals(
next(it),
'##jacquard.summarize.ZScoreX.X_stdev=' + repr(tag._stdev))
self.assertRegexpMatches(
next(it),
'##FORMAT=<ID=ZScoreX,Number=1,Type=Float,Description="ZScore for X">'
)
def test_init_setsPopulationStatisticsAssignsStddevCorrectlyWhenNoValues(
self):
rec1 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "."
},
"SB": {
"X": "."
}
})
reader = MockVcfReader(records=[rec1])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
expected_mean = None
expected_stdev = None
self.assertEquals(expected_mean, tag._mean)
self.assertEquals(expected_stdev, tag._stdev)
def test_init_setsPopulationStatisticsRoundsTo13digits(self):
rec1 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "1"
},
"SB": {
"X": "1"
},
"SC": {
"X": "0"
}
})
reader = MockVcfReader(records=[rec1])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
self.assertEquals(round(2 / 3, 13), tag._mean, repr(tag._mean))
self.assertEquals(round(stdev([1, 1, 0]), 13), tag._stdev)
def test_init_setsPopulationStatisticsAssignsStddevCorrectlyWhenOneValue(
self):
rec1 = vcf.VcfRecord("1",
"42",
"A",
"C",
sample_tag_values={
"SA": {
"X": "4"
},
"SB": {
"X": "."
}
})
reader = MockVcfReader(records=[rec1])
tag = zscore_caller._ZScoreTag("ZScoreX", "ZScore for X", "X", reader)
values = [4]
self.assertAlmostEquals(mean(values), tag._mean,
_ZScoreTag._MAX_PRECISION)
self.assertAlmostEquals(stdev(values), tag._stdev,
_ZScoreTag._MAX_PRECISION)
