def test_NegativeLookBehind_Regex(self):
"""
Test the NLB regex used in differentiating Tonic vs Asymmetric Tonic, atonic, generalised tonic, tonic-clonic etc in SemioDict.
Note also that the SemioDict is the live one unless specifically specified:
e.g. as an argument to QUERY_SEMIOLOGY(semiology_dict_path=dummy_semiology_dict_path=dummy_semiology_dict_path)
"""
patient = Semiology('Tonic', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
tonic_result, num_query_loc = patient.query_semiology()
self.assertIs(type(tonic_result), pd.DataFrame)
assert not tonic_result.empty
assert (tonic_result['Localising'].sum() == 1 + 10)
assert (tonic_result['Lateralising'].sum() == 1)
# Now test for Atonic
patient_two = Semiology('Atonic', Laterality.NEUTRAL,
Laterality.NEUTRAL)
patient_two.data_frame = self.df
atonic_result, num_query_loc = patient_two.query_semiology()
self.assertIs(type(atonic_result), pd.DataFrame)
assert not atonic_result.empty
assert (atonic_result['Localising'].sum() == 999)
assert (atonic_result['Lateralising'].sum() == 2)
print('\n9 negative lookbehind regex\n')
def test_temporal_and_frontal_hierarchy_reversal(self):
"""
test combined temporal and frontals
"""
patient = Semiology('Aphasia', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
patient.granular = False
inspect_result, _ = patient.query_semiology()
hierarchy_df = Hierarchy(inspect_result)
hierarchy_df.temporal_hierarchy_reversal() # deafult max option
hierarchy_df.frontal_hierarchy_reversal()
inspect_result_reversed = hierarchy_df.new_df
assert hierarchy_df.frontal_hr.equals(hierarchy_df.new_df)
assert inspect_result.all().all() == inspect_result_reversed.all().all(
)
assert not hierarchy_df.temporal_hr.equals(hierarchy_df.new_df)
# ^ because self.new_df.copy() in the Hierarchy.temporal_hierarchy_reversal() method, so order matters.
assert (inspect_result_reversed['TL'].sum() == 3)
assert (inspect_result_reversed['Anterior (temporal pole)'].sum() == 5)
assert (inspect_result_reversed['Lateral Temporal'].sum() == 0)
assert (inspect_result_reversed['ITG'].sum() == 4)
assert (inspect_result_reversed['Mesial Temporal'].sum() == 5)
assert (inspect_result_reversed['FL'].sum() == 0)
assert (inspect_result_reversed['IFG (F3)\n(BA 44,45,47)'].sum() == 1)
print('\n10 combined T & F hierarchy reversals\n')
def test_MappingsCalibration_HierarchyReversal_SemioBrainBeta_TLd(self):
"""
Test the mappings TL for mapping calibration visualisation in the systReview (SemioBrain Database - beta version)
i.e. not dummy data
"""
patient = Semiology('mappings TLd', Laterality.NEUTRAL,
Laterality.NEUTRAL)
patient.data_frame = self.use_SemioBrainBeta_notDummyData()
patient.granular = False
inspect_result, _ = patient.query_semiology()
hierarchy_df = Hierarchy(inspect_result)
hierarchy_df.all_hierarchy_reversal() # deafult max option
inspect_result_reversed = hierarchy_df.new_df
# test postcodes
assert (inspect_result['TL'].sum() == 1)
assert (inspect_result['Lateral Temporal'].sum() == 1)
assert (inspect_result[
'STG (includes Transverse Temporal Gyrus, Both Planum)'].sum() == 1
)
assert 'Planum Temporale' not in inspect_result
# test hierarchy reversal
assert (inspect_result_reversed['TL'].sum() == 0)
assert (inspect_result_reversed['Lateral Temporal'].sum() == 0)
assert (inspect_result_reversed[
'STG (includes Transverse Temporal Gyrus, Both Planum)'].sum() == 1
)
assert 'Planum Temporale' not in inspect_result
assert 'Planum Polare' not in inspect_result
def test_latexceedsloc_3(self):
"""
Test capturing lateralisation value when it exceeds localising value (part1) and combining with lat_no_loc and lat_and_loc (part 2).
Note that the default in Q_L of normalise_lat_to_loc = False and using norm_ratio = lower_value / higher_value
results in capping of lateralisation influence on data visualisation.
In the specific case of latexceedsloc semiology, despite 500 lat cumulative datapoints and 2 localising points,
the GIF results are:
{155: 2.0, 156:1.0}
"""
patient = Semiology('latexceedsloc', Laterality.LEFT, Laterality.LEFT)
patient.data_frame = self.df
# (part 1) test latexceedsloc alone using norm_ratio/oddsratio method of Q_L:
heatmap, _ = patient.get_num_datapoints_dict(method='minmax')
assert heatmap[156] == 200.0
assert heatmap[155] == 300.0
# (part 2) combine above with lat_not_loc and lat_and_loc:
patient = Semiology('lat', Laterality.LEFT, Laterality.LEFT)
patient.data_frame = self.df
lat_allgifs = patient.query_lateralisation(one_map_dummy)
# drop the zero entries - should be only the IL left ones which aren't MTG of TL:
lat_allgifs = lat_allgifs[['Gif Parcellations', 'pt #s']].astype({
'Gif Parcellations':
'int32',
'pt #s':
'int32'
})
lat_allgifs.set_index('Gif Parcellations', inplace=True)
lat_allgifs = lat_allgifs.loc[lat_allgifs['pt #s'] != 0, :]
# assert using shape this not all are 1 so should not be the same:
assert not lat_allgifs['pt #s'].sum() == lat_allgifs.shape[0]
# check the MTG on the left (IL) GIF # 156 is == 3,
# which it isn't due to the norm_ratio method in Q_L - as can be seen from part 1
# so instead we see in part 1 156 was 1, but in the spreadsheet it was 2. so will give 2.
# 3 is better but lost due to norm_ratio method in Q_L
assert (lat_allgifs.loc[156, 'pt #s'] == 201)
# for the right sided GIF, 155, latexceedsloc gives 2 [/],
# lat_not_loc gives 1 (CL) (See test_lat_not_loc_1)[/], and
# lat_and_loc adds none [/]
assert (lat_allgifs.loc[155, 'pt #s'] == 301)
def factor_ql(self, term, normalise_to_localising_values=False):
"""
factor function.
NB normalise_to_localising_values default is False in semiology.py
"""
patient = Semiology(term, Laterality.LEFT, Laterality.LEFT)
patient.data_frame = self.df
patient.normalise_to_localising_values = normalise_to_localising_values
all_combined_gifs = patient.query_lateralisation(one_map_dummy)
return all_combined_gifs
def test_paed_exclusions_query_semio(self):
"""
Test query_semiology then exclude paediatric cases.
"""
patient = Semiology('spasm', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
inspect_result, num_query_loc = patient.query_semiology()
inspect_result = exclude_paediatric_cases(inspect_result)
self.assertIs(type(inspect_result), pd.DataFrame)
assert not inspect_result.empty
assert (inspect_result['Localising'].sum() == 5 + 1)
assert (inspect_result['Lateralising'].sum() == 0)
print('\n8 paed exclusions query_semio()\n')
def test_prelim2_dummy_ILV_control(self):
"""
Preliminary test as per issues #169 on GitHub.
This uses "ILV_4" as semiology term, which is Example 2 on GitHub:
https://github.com/thenineteen/Semiology-Visualisation-Tool/issues/169
"""
patient = Semiology('ILV_4', Laterality.LEFT, Laterality.LEFT)
patient.data_frame = self.df
inspect_result, _ = patient.query_semiology()
self.assertIs(type(inspect_result), pd.DataFrame)
assert not inspect_result.empty
assert (inspect_result['Localising'].sum() == 4)
assert (inspect_result['Lateralising'].sum() == 0)
def test_toplevel_aphasia_parentheses_and_caps(self):
"""
test query_semiology which calls QUERY_SEMIOLOGY
Need to change the call to the semiology_dictionary to
make it the dummy_semio_dict
"""
patient = Semiology('Aphasia', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
inspect_result, num_query_loc = patient.query_semiology()
self.assertIs(type(inspect_result), pd.DataFrame)
assert not inspect_result.empty
# deafult excludes postictals
assert (inspect_result['Localising'].sum() == 14 - 1)
assert (inspect_result['Lateralising'].sum() == 6 - 1)
def test_prelim4_ql_doesnot_lateralise(self):
"""
"""
patient = Semiology('pipeline_notlaat_semioB',
symptoms_side=Laterality.LEFT,
dominant_hemisphere=Laterality.LEFT)
patient.data_frame = self.df
# as no lateralising data, the below will run a manual pipeline
# involving melt_then_pivot_query and pivot_result_to_one_map:
all_combined_gifs = patient.query_lateralisation(
one_map_dummy)
self.assertIs(type(all_combined_gifs), pd.DataFrame)
assert not all_combined_gifs.empty
return all_combined_gifs
def test_prelim5_ql_50_50(self):
"""
Pipeline sequence test of ql: lateralises 50% CL and 50%IL.
"""
patient = Semiology('pipeline_50_50_laateralises_semioC',
symptoms_side=Laterality.LEFT,
dominant_hemisphere=Laterality.LEFT)
patient.data_frame = self.df
# returns the results from Q_L pipeline
all_combined_gifs = patient.query_lateralisation(
one_map_dummy)
self.assertIs(type(all_combined_gifs), pd.DataFrame)
assert not all_combined_gifs.empty
return all_combined_gifs
def test_latnotloc_and_latandloc_2(self):
"""
Test capturing the lateralising but not localising data rather than skipping it.
integrated with lat and loc data.
"""
patient = Semiology('lat_', Laterality.LEFT, Laterality.LEFT)
patient.data_frame = self.df
lat_not_loc_all_combined_gifs = patient.query_lateralisation(
one_map_dummy)
# inspect result
lat_not_loc_result, _ = patient.query_semiology()
self.assertIs(type(lat_not_loc_all_combined_gifs), pd.DataFrame)
assert not lat_not_loc_all_combined_gifs.empty
# drop the zero entries - should be only the IL left ones which aren't MTG of TL:
lat_not_loc_all_combined_gifs = lat_not_loc_all_combined_gifs[[
'Gif Parcellations', 'pt #s'
]].astype({
'Gif Parcellations': 'int32',
'pt #s': 'int32'
})
lat_not_loc_all_combined_gifs.set_index('Gif Parcellations',
inplace=True)
lat_not_loc_gifsclean = lat_not_loc_all_combined_gifs.loc[
lat_not_loc_all_combined_gifs['pt #s'] != 0, :]
gifs_right, gifs_left = gifs_lat_factor()
lat_not_loc_gifsclean_rights = (lat_not_loc_gifsclean.drop(
index=156).index.isin(gifs_right).all())
# inspect result assertions
assert (lat_not_loc_result.Localising.sum() == 1)
assert (lat_not_loc_result['Lateralising'].sum() == 2)
# all_combined_gifs assertions
# all except GIF 156 (L MTG) are in the right GIFs:
assert ((lat_not_loc_gifsclean_rights == True))
assert ((lat_not_loc_gifsclean.index.isin(gifs_left)).any() == True)
# assert using shape as all pt #s are 1:
assert (lat_not_loc_gifsclean['pt #s'].sum() ==
lat_not_loc_gifsclean.shape[0])
# check that latnotloc gives 1 and latandloc adds zero to right MTG GIF #155
heatmap, _ = patient.get_num_datapoints_dict(method='minmax')
assert heatmap[155] == 1 # right
def test_paed_default_query_semio(self):
"""
Test query_semiology for excluding paediatric cases.
This test now shows that default query_semiology() DOES filter paediatric cases.
"""
patient = Semiology('spasm', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
# revert default paed exclusions
patient.include_only_paediatric_cases = True
inspect_result, num_query_loc = patient.query_semiology()
self.assertIs(type(inspect_result), pd.DataFrame)
assert not inspect_result.empty
assert (inspect_result['Localising'].sum() == 10)
assert (inspect_result['Lateralising'].sum() == 0)
print('\n7 paed query_semio()\n')
def test_lat_not_loc_1(self):
"""
Test capturing the lateralising but not localising data rather than skipping it.
As implemented in QUERY_LATERALISATION in branch "Lateralising but no localising value".
"""
patient = Semiology('lat_not_loc', Laterality.LEFT, Laterality.LEFT)
patient.data_frame = self.df
lat_not_loc_all_combined_gifs = patient.query_lateralisation(
one_map_dummy)
# inspect result
lat_not_loc_result, num_query_loc = patient.query_semiology()
self.assertIs(type(lat_not_loc_all_combined_gifs), pd.DataFrame)
assert not lat_not_loc_all_combined_gifs.empty
# drop the zero entries as these are from the CL/IL zeros:
lat_not_loc_all_combined_gifs = lat_not_loc_all_combined_gifs[[
'Gif Parcellations', 'pt #s'
]].astype({
'Gif Parcellations': 'int32',
'pt #s': 'int32'
})
lat_not_loc_all_combined_gifs.set_index('Gif Parcellations',
inplace=True)
lat_not_loc_gifsclean = lat_not_loc_all_combined_gifs.loc[
lat_not_loc_all_combined_gifs['pt #s'] != 0, :]
# now we know only the CL data remains in this dummy data, which is on the RIGHT.
gifs_right, gifs_left = gifs_lat_factor()
lat_not_loc_gifsclean_rights = (
lat_not_loc_gifsclean.index.isin(gifs_right).all())
# inspect result assertions
assert (lat_not_loc_result.Localising.sum() == 0)
assert (lat_not_loc_result['Lateralising'].sum() == 1)
# all_combined_gifs assertions
assert ((lat_not_loc_gifsclean_rights == True))
assert ((lat_not_loc_gifsclean.index.isin(gifs_left)).any() == False)
assert (lat_not_loc_gifsclean['pt #s'].sum() ==
lat_not_loc_gifsclean.shape[0])
# test MTG on right 155 gif # gives 1:
heatmap, _ = patient.get_num_datapoints_dict(method='minmax')
assert 156 not in heatmap # left
assert heatmap[155] == 1 # right
def test_frontal_hierarchy_reversal(self):
"""
continuing testing reversal of postcode system
for new class Hierarchy
"""
patient = Semiology('Aphasia', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
inspect_result, _ = patient.query_semiology()
hierarchy_df = Hierarchy(inspect_result)
hierarchy_df.frontal_hierarchy_reversal() # deafult max option
inspect_result_reversed = hierarchy_df.frontal_hr
# # note default is to exclude post ictals, otherwise add +1 to both below
assert (inspect_result_reversed['FL'].sum() == 0)
assert (inspect_result_reversed['IFG (F3)\n(BA 44,45,47)'].sum() == 1)
print('\n9 frontal dictionary hierarchy reversal\n')
def test_prelim2_qs_doesnot_lateralise(self):
"""
Pipeline sequence testing of dummy data: does not lateralise.
This is a standalone test to ensure qs/QS find dummy semiology.
Test query_semiology which calls QUERY_SEMIOLOGY.
The call to the semiology_dictionary is the dummy_semio_dict.
"""
patient = Semiology('pipeline_notlaat_semioB',
Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
inspect_result, _ = patient.query_semiology()
self.assertIs(type(inspect_result), pd.DataFrame)
assert not inspect_result.empty
assert(inspect_result['Localising'].sum() == 10)
assert(inspect_result['Lateralising'].sum() == 0)
def test_only_postictal_cases(self):
"""
Include only the single postictal aphasia. Note Dominant in dummy data.
cf test_postictal_exclusions
"""
# Low level test
df_postictal = only_postictal_cases(self.df)
query, num_query_lat, num_query_loc = QUERY_SEMIOLOGY(
df_postictal,
semiology_term=['aphasia'],
ignore_case=True,
semiology_dict_path=None,
col1='Reported Semiology',
col2='Semiology Category',
)
assert (query['Localising'].sum() == 1)
assert (query['Lateralising'].sum() == 1)
# High level test
patient = Semiology('aphasia',
Laterality.NEUTRAL,
dominant_hemisphere=Laterality.LEFT)
patient.data_frame = self.df
patient.include_postictals = True
patient.include_only_postictals = True
patient.granular = True
lat_allgifs = patient.query_lateralisation(one_map_dummy)
lat_allgifs = lat_allgifs[['Gif Parcellations', 'pt #s']].astype({
'Gif Parcellations':
'int32',
'pt #s':
'int32'
})
lat_allgifs.set_index('Gif Parcellations', inplace=True)
# note that dominant_hemisphere == Laterality.LEFT as set above. Just to clarify results change if dominace changes. Also 1's become 2's if not granular.
assert (lat_allgifs.loc[164, 'pt #s'] == 1)
assert (lat_allgifs.loc[166, 'pt #s'] == 1)
assert (lat_allgifs.loc[163, 'pt #s'] == 0)
assert (lat_allgifs.loc[165, 'pt #s'] == 0)
lat_allgifs = lat_allgifs.loc[lat_allgifs['pt #s'] != 0, :]
SVT_output, _ = patient.get_num_datapoints_dict(method='minmax')
assert SVT_output == dict((lat_allgifs)['pt #s'])
def test_prelim3_ql_lateralises(self):
"""
Pipeline sequence test of ql: lateralises
The call to the semiology_dictionary is the dummy_semio_dict as passed as an argument to q_l.
Relies on mapping strategy as uses gifs.
"""
patient = Semiology('pipeline_laateralises_semioA',
symptoms_side=Laterality.LEFT,
dominant_hemisphere=Laterality.LEFT)
patient.data_frame = self.df
# returns the results from Q_L pipeline
all_combined_gifs = patient.query_lateralisation(
one_map_dummy)
self.assertIs(type(all_combined_gifs), pd.DataFrame)
assert not all_combined_gifs.empty
return all_combined_gifs
def test_hierarchy(self):
"""
first tested postcode system which duplicates mapping
before testing its reversal.
"""
patient = Semiology('Aphasia', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
# default behaviour of query_semiology changed to use granular hierarchy reversal:
patient.granular = False
inspect_result, _ = patient.query_semiology()
assert (inspect_result['TL'].sum() == 12)
assert (inspect_result['Anterior (temporal pole)'].sum() == 5)
assert (inspect_result['Lateral Temporal'].sum() == 4)
assert (inspect_result['ITG'].sum() == 4)
assert (inspect_result['Mesial Temporal'].sum() == 5)
# # by default postictal are excluded. Otherwise add +1 to both below
assert (inspect_result['FL'].sum() == 1)
assert (inspect_result['IFG (F3)\n(BA 44,45,47)'].sum() == 1)
print('\n7 hierarchy\n')
def test_toplevel_query_lat_mappings(self):
"""
The call to the semiology_dictionary is the dummy_semio_dict as passed as an argument to q_l
Relies on mapping strategy as uses gifs
"""
patient = Semiology('Aphasia', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
all_combined_gifs = patient.query_lateralisation(one_map_dummy)
self.assertIs(type(all_combined_gifs), pd.DataFrame)
assert not all_combined_gifs.empty
labels = ['Gif Parcellations', 'pt #s']
all_combined_gifs = all_combined_gifs.astype({
'Gif Parcellations': 'int32',
'pt #s': 'int32'
})
new_all_combined_gifindexed = all_combined_gifs.loc[:, labels]
new_all_combined_gifindexed.set_index('Gif Parcellations',
inplace=True)
# new_all_combined_gifindexed.to_csv(r'D:\aphasia_fixture.csv')
# load fixture:
fixture = pd.read_excel(
dummy_data_path,
header=0,
usecols='A:B',
sheet_name='fixture_aphasia',
index_col=0,
engine="openpyxl",
)
# fixture.sort_index(inplace=True)
assert ((new_all_combined_gifindexed.shape) == (fixture.shape))
# print('new_all_combined_gifindexed.shape is: ',
# new_all_combined_gifindexed.shape)
# print('fixture.shape.shape is: ', fixture.shape)
assert (new_all_combined_gifindexed.index == fixture.index).all()
assert (new_all_combined_gifindexed.values == fixture.values).all()
def test_temporal_hierarchy_reversal(self):
"""
now test reversal postcode system which duplicates mapping
for new class Hierarchy
"""
patient = Semiology('Aphasia', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
patient.granular = False
inspect_result, _ = patient.query_semiology()
# # the three lines below were integrated into default query_semiology() using.granular = True
hierarchy_df = Hierarchy(inspect_result)
hierarchy_df.temporal_hierarchy_reversal() # deafult max option
inspect_result_reversed = hierarchy_df.temporal_hr
assert (inspect_result_reversed['TL'].sum() == 3)
assert (inspect_result_reversed['Anterior (temporal pole)'].sum() == 5)
assert (inspect_result_reversed['Lateral Temporal'].sum() == 0)
assert (inspect_result_reversed['ITG'].sum() == 4)
assert (inspect_result_reversed['Mesial Temporal'].sum() == 5)
print('\n8 temporal hierarchy reversal\n')
def test_all_hierarchy_reversals(self):
"""
test all.
"""
patient = Semiology('Aphasia', Laterality.NEUTRAL, Laterality.NEUTRAL)
patient.data_frame = self.df
inspect_result, _ = patient.query_semiology()
hierarchy_df = Hierarchy(inspect_result)
hierarchy_df.all_hierarchy_reversal() # deafult max option
inspect_result_reversed = hierarchy_df.new_df
# assert object doesn't have any _hr aatribute:
assert not hasattr(hierarchy_df, 'temporal_hr')
assert (inspect_result_reversed['TL'].sum() == 3)
assert (inspect_result_reversed['Anterior (temporal pole)'].sum() == 5)
assert (inspect_result_reversed['Lateral Temporal'].sum() == 0)
assert (inspect_result_reversed['ITG'].sum() == 4)
assert (inspect_result_reversed['Mesial Temporal'].sum() == 5)
# these two lines should be +1 if not excluding postictals
assert (inspect_result_reversed['FL'].sum() == 0)
assert (inspect_result_reversed['IFG (F3)\n(BA 44,45,47)'].sum() == 1)
print('\n11 all_hierarchy reversals\n')