tests for generate_transforms (closes #162)

tests/transform_tests.py

@@ -5,12 +5,15 @@
 import numpy as np
 import warnings
 
-from flowkit import Sample, transforms
+from flowkit import Sample, transforms, generate_transforms
 
 data1_fcs_path = 'data/gate_ref/data1.fcs'
 data1_sample = Sample(data1_fcs_path)
 data1_raw_events = data1_sample.get_events(source='raw')
 
+# Sample with null channel
+null_chan_sample = Sample(data1_fcs_path, null_channel_list=['FL1-H'])
+
 test_data_range1 = np.linspace(0.0, 10.0, 101)
 
 
@@ -190,3 +193,136 @@ def test_inverse_wsp_biex_transform():
 
         np.testing.assert_array_almost_equal(test_data_range1, x, decimal=10)
 
+    def test_generate_transforms_defaults(self):
+        xform_lut = generate_transforms(data1_sample)
+
+        self.assertEqual(len(xform_lut), len(data1_sample.pnn_labels))
+
+        # pick a fluoro channel label and check Transform type
+        fluoro_label = data1_sample.pnn_labels[data1_sample.fluoro_indices[0]]
+        self.assertIsInstance(xform_lut[fluoro_label], transforms.LogicleTransform)
+
+        # verify time use max time
+        time_max = data1_sample.get_channel_events(data1_sample.time_index, source='raw').max()
+        self.assertEqual(xform_lut['Time'].param_t, time_max)
+
+    def test_generate_transforms_default_asinh(self):
+        xform_lut = generate_transforms(
+            data1_sample,
+            fluoro_xform_class=transforms.AsinhTransform
+        )
+
+        self.assertEqual(len(xform_lut), len(data1_sample.pnn_labels))
+
+        # pick a fluoro channel label and check Transform type
+        fluoro_label = data1_sample.pnn_labels[data1_sample.fluoro_indices[0]]
+        self.assertIsInstance(xform_lut[fluoro_label], transforms.AsinhTransform)
+
+    def test_generate_transforms_default_hyperlog(self):
+        xform_lut = generate_transforms(
+            data1_sample,
+            fluoro_xform_class=transforms.HyperlogTransform
+        )
+
+        self.assertEqual(len(xform_lut), len(data1_sample.pnn_labels))
+
+        # pick a fluoro channel label and check Transform type
+        fluoro_label = data1_sample.pnn_labels[data1_sample.fluoro_indices[0]]
+        self.assertIsInstance(xform_lut[fluoro_label], transforms.HyperlogTransform)
+
+    def test_generate_transforms_default_log(self):
+        xform_lut = generate_transforms(
+            data1_sample,
+            fluoro_xform_class=transforms.LogTransform
+        )
+
+        self.assertEqual(len(xform_lut), len(data1_sample.pnn_labels))
+
+        # pick a fluoro channel label and check Transform type
+        fluoro_label = data1_sample.pnn_labels[data1_sample.fluoro_indices[0]]
+        self.assertIsInstance(xform_lut[fluoro_label], transforms.LogTransform)
+
+    def test_generate_transforms_default_wsp_biex(self):
+        xform_lut = generate_transforms(
+            data1_sample,
+            fluoro_xform_class=transforms.WSPBiexTransform
+        )
+
+        self.assertEqual(len(xform_lut), len(data1_sample.pnn_labels))
+
+        # pick a fluoro channel label and check Transform type
+        fluoro_label = data1_sample.pnn_labels[data1_sample.fluoro_indices[0]]
+        self.assertIsInstance(xform_lut[fluoro_label], transforms.WSPBiexTransform)
+
+    def test_generate_transforms_default_null_channel(self):
+        # null channel sample has 'FL1-H' nullified
+        xform_lut = generate_transforms(null_chan_sample)
+
+        self.assertEqual(len(xform_lut), len(data1_sample.pnn_labels) - 1)
+
+        # ensure null channel label is missing from xform LUT
+        null_label = 'FL1-H'
+        self.assertNotIn(null_label, xform_lut)
+
+    def test_generate_transforms_transform_not_supported(self):
+        self.assertRaises(
+            NotImplementedError,
+            generate_transforms,
+            data1_sample,
+            fluoro_xform_class=transforms.RatioTransform
+        )
+
+    def test_generate_transforms_transform_instance_scatter(self):
+        # specify a Transform instance for scatter channels
+        scatter_xform = transforms.LogTransform(param_t=262144, param_m=4.1)
+
+        xform_lut = generate_transforms(
+            data1_sample,
+            scatter_xform_class=scatter_xform
+        )
+
+        self.assertEqual(len(xform_lut), len(data1_sample.pnn_labels))
+
+        # pick a scatter channel label and check Transform type
+        scatter_label = data1_sample.pnn_labels[data1_sample.scatter_indices[0]]
+        scatter_xform = xform_lut[scatter_label]
+        self.assertIsInstance(scatter_xform, transforms.LogTransform)
+
+        self.assertEqual(scatter_xform.param_m, 4.1)
+
+    def test_generate_transforms_transform_instance_fluoro(self):
+        # specify a Transform instance for fluoro channels
+        fluoro_xform = transforms.AsinhTransform(param_t=262144, param_m=4.1, param_a=0.0)
+
+        xform_lut = generate_transforms(
+            data1_sample,
+            fluoro_xform_class=fluoro_xform
+        )
+
+        self.assertEqual(len(xform_lut), len(data1_sample.pnn_labels))
+
+        # pick a fluoro channel label and check Transform type
+        fluoro_label = data1_sample.pnn_labels[data1_sample.fluoro_indices[0]]
+        fluoro_xform = xform_lut[fluoro_label]
+        self.assertIsInstance(fluoro_xform, transforms.AsinhTransform)
+
+        self.assertEqual(fluoro_xform.param_m, 4.1)
+
+    def test_generate_transforms_transform_instance_time(self):
+        # specify a Transform instance for time channel
+        # make up a max time to check the custom instance is returned
+        time_max = 123
+        time_xform = transforms.LinearTransform(param_t=time_max, param_a=0.0)
+
+        xform_lut = generate_transforms(
+            data1_sample,
+            time_xform_class=time_xform
+        )
+
+        self.assertEqual(len(xform_lut), len(data1_sample.pnn_labels))
+
+        # check Transform type & params
+        time_xform = xform_lut['Time']
+        self.assertIsInstance(time_xform, transforms.LinearTransform)
+
+        self.assertEqual(time_xform.param_t, time_max)