def test_fuzzy_filter(self):
pixelset = PixelSetParser(self.pixelset_path)
pixelset.parse()
pixels, na, fuzzy = pixelset.filter(fuzzy_filter=False)
self.assertEqual(len(pixels), 1862)
self.assertEqual(pixels.index[0], 'CAGL0F02695g')
self.assertEqual(pixels.index[-1], 'CAGL0L01925g')
def test_na_filter(self):
pixelset = PixelSetParser(self.pixelset_path)
pixelset.parse()
pixels, na, fuzzy = pixelset.filter(na_filter=False)
self.assertEqual(len(pixels), 1911)
self.assertEqual(pixels.index[0], 'CAGL0F02695g')
self.assertEqual(pixels.index[-1], 'CAGL0M12749g')
def test_parse(self):
pixelset = PixelSetParser(self.pixelset_path)
pixelset.parse()
self.assertIsNotNone(pixelset.pixels)
self.assertEqual(len(pixelset.pixels), 1936)
pixel = pixelset.pixels.loc['CAGL0F02695g', ]
self.assertAlmostEqual(pixel.Value, 4.25954345565357)
self.assertAlmostEqual(pixel.Quality_score, 5.0188163324757298e-5)
def test__to_pixels_with_existing_pixel(self):
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
omics_unit_type=self.omics_unit_type,
strain=self.strain,
)
parser.parse()
parser._set_pixel_set()
self._load_cgd_entries()
existing_db_pixel = self._create_pixel_from_set(parser.pixelset)
parser._to_pixels()
self.assertEqual(len(parser.db_pixels['new']), 1836)
self.assertEqual(len(parser.db_pixels['update']), 1)
# We should have updated existing pixel data
identifier = existing_db_pixel.omics_unit.reference.identifier
pixel = parser.pixels.ix[identifier]
updated_db_pixel = parser.db_pixels['update'][0]
self.assertEqual(updated_db_pixel.value, pixel.Value)
self.assertEqual(updated_db_pixel.quality_score, pixel.Quality_score)
def test__to_pixels(self):
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
omics_unit_type=self.omics_unit_type,
strain=self.strain,
)
# With no pixel file parsed, _to_pixels has no effect
parser._to_pixels()
self.assertEqual(parser.db_pixels, {
'new': [],
'update': [],
})
parser.parse()
self._load_cgd_entries()
parser._to_pixels()
self.assertEqual(len(parser.db_pixels['new']), 1837)
self.assertEqual(len(parser.db_pixels['update']), 0)
db_pixel = parser.db_pixels['new'][0]
pixel = parser.pixels.ix[db_pixel.omics_unit.reference.identifier]
self.assertEqual(db_pixel.value, pixel.Value)
self.assertEqual(db_pixel.quality_score, pixel.Quality_score)
def test_parse_with_force(self):
pixelset = PixelSetParser(self.pixelset_path)
self.assertIsNone(pixelset.pixels)
pixelset.parse()
self.assertIsNotNone(pixelset.pixels)
old_pixels = pixelset.pixels
pixelset.parse()
self.assertEqual(id(pixelset.pixels), id(old_pixels))
pixelset.parse(force=True)
self.assertNotEqual(id(pixelset.pixels), id(old_pixels))
def test_save_with_same_reference_omics_units(self):
# Create two OmicsUnits with the same identifier but for different
# strains/types
identifier = 'CAGL0B00990g'
OmicsUnitFactory(reference=EntryFactory(identifier=identifier),
strain=StrainFactory(),
type=OmicsUnitTypeFactory())
OmicsUnitFactory(reference=EntryFactory(identifier=identifier),
strain=self.strain,
type=self.omics_unit_type)
parser = PixelSetParser(
Path('apps/submission/fixtures/dataset-0002/data.txt'),
description=self.description,
analysis=self.analysis,
omics_unit_type=self.omics_unit_type,
strain=self.strain,
)
parser.parse()
self._load_cgd_entries()
self.assertEqual(Pixel.objects.count(), 0)
parser.save()
# Even if we've created two OmicsUnits with the same reference
# identifier, only the pixel corresponding to the OmicsUnit with the
# relevant strain & OmicsUnitType will be imported. Hence, we expect
# to have created 7 pixels and not 8, see:
# - https://github.com/Candihub/pixel/pull/292
# - https://github.com/Candihub/pixel/pull/296
self.assertEqual(Pixel.objects.count(), 7)
def test_save_with_existing_pixel(self):
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
omics_unit_type=self.omics_unit_type,
strain=self.strain,
)
parser.parse()
parser._set_pixel_set()
self._load_cgd_entries()
self._create_pixel_from_set(parser.pixelset)
self.assertEqual(Pixel.objects.count(), 1)
parser.save()
self.assertEqual(Pixel.objects.count(), 1837)
def test__set_pixel_set(self):
parser = PixelSetParser(self.pixelset_path)
with self.assertRaises(PixelSetParserError):
parser._set_pixel_set()
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
)
parser._set_pixel_set()
pixelset = parser.pixelset
self.assertIsNotNone(pixelset)
self.assertEqual(
# pixelset.pixels_file.name is a PosixPath
pixelset.pixels_file.name.name,
self.pixelset_path.name)
self.assertEqual(pixelset.description, self.description)
self.assertEqual(pixelset.analysis, self.analysis)
def test_save(self):
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
omics_unit_type=self.omics_unit_type,
strain=self.strain,
)
parser.parse()
self._load_cgd_entries()
self.assertEqual(Pixel.objects.count(), 0)
parser.save()
self.assertEqual(Pixel.objects.count(), 1837)
def test_save_with_update(self):
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
omics_unit_type=self.omics_unit_type,
strain=self.strain,
)
parser.parse()
parser._set_pixel_set()
self._load_cgd_entries()
pixel = self._create_pixel_from_set(parser.pixelset)
self.assertEqual(Pixel.objects.count(), 1)
self.assertEqual(pixel.value, 4.2)
self.assertEqual(pixel.quality_score, 0.8)
parser.save(update=True)
self.assertEqual(Pixel.objects.count(), 1837)
pixel = Pixel.objects.get(pk=pixel.pk)
self.assertEqual(pixel.value, 2.703695974165)
self.assertEqual(pixel.quality_score, 0.00268822352590468)
def test_save_populates_cached_fields(self):
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
omics_unit_type=self.omics_unit_type,
strain=self.strain,
)
parser.parse()
self._load_cgd_entries()
parser.save()
self.assertEqual(parser.pixelset.cached_species,
[self.strain.species.name])
self.assertEqual(parser.pixelset.cached_omics_areas, [])
self.assertEqual(parser.pixelset.cached_omics_unit_types,
[self.omics_unit_type.name])
def test_filter(self):
pixelset = PixelSetParser(self.pixelset_path)
self.assertEqual(pixelset.filter(), tuple([None] * 3))
pixelset.parse()
pixels, na, fuzzy = pixelset.filter()
self.assertEqual(len(pixels), 1837)
self.assertEqual(pixels.index[0], 'CAGL0F02695g')
self.assertEqual(pixels.index[-1], 'CAGL0L01925g')
self.assertEqual(len(na), 74)
self.assertEqual(na.index[0], 'CAGL0C02475g')
self.assertEqual(na.index[-1], 'CAGL0M12749g')
self.assertEqual(len(fuzzy), 25)
self.assertEqual(fuzzy.index[0], 'CAGL0A02211g;CAGL0D02640g')
self.assertEqual(fuzzy.index[-1], 'CAGL0G08173g;CAGL0D05082g')
def test_init(self):
pixelset = PixelSetParser(self.pixelset_path)
self.assertEqual(pixelset.pixelset_path, self.pixelset_path)
self.assertIsNone(pixelset.pixels)
def test__get_omics_units(self):
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
)
with self.assertRaises(PixelSetParserSaveError):
parser._get_omics_units(None)
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
strain=self.strain,
)
with self.assertRaises(PixelSetParserSaveError):
parser._get_omics_units(None)
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
omics_unit_type=self.omics_unit_type,
)
with self.assertRaises(PixelSetParserSaveError):
parser._get_omics_units(None)
parser = PixelSetParser(
self.pixelset_path,
description=self.description,
analysis=self.analysis,
omics_unit_type=self.omics_unit_type,
strain=self.strain,
)
parser.parse()
pixels, _, _ = parser.filter()
# We need to load Entries first!
self.assertEqual(OmicsUnit.objects.count(), 0)
with self.assertRaises(PixelSetParserSaveError):
parser._get_omics_units(pixels)
self.assertEqual(OmicsUnit.objects.count(), 0)
# So… load entries
self._load_cgd_entries()
# Then create omics units
omics_units = parser._get_omics_units(pixels, verbose=True)
self.assertEqual(omics_units.count(), 1837)
self.assertEqual(OmicsUnit.objects.count(), 1837)