Merge pull request #244 from emilazy/push-qonlzoyrmkms

Migrate tests to pytest

Author: timodonnell

docs/intro.rst

@@ -53,8 +53,8 @@ From a checkout you can run the unit tests with:
 
 .. code-block:: shell
 
-    $ pip install nose
-    $ nosetests .
+    $ pip install pytest
+    $ pytest
 
 
 Using conda

pylintrc

@@ -2,4 +2,3 @@
 # Without ignoring this, we get errors like:
 # E:249,20: Module 'numpy' has no 'nan' member (no-member)
 ignored-modules = numpy
-ignored-classes = nose.tools

test/expensive_verify_pretrain_optimizable.py

@@ -1,4 +1,4 @@
-# Expensive test - not run by nose.
+# Expensive test - not run by pytest.
 
 from mhcflurry import train_pan_allele_models_command
 from mhcflurry.downloads import get_path

test/test_amino_acid.py

@@ -2,7 +2,6 @@
 initialize()
 
 from mhcflurry import amino_acid
-from nose.tools import eq_
 from numpy.testing import assert_equal
 import pandas
 
@@ -33,7 +32,7 @@ def test_index_and_one_hot_encoding():
     one_hot = amino_acid.fixed_vectors_encoding(
         index_encoding,
         letter_to_vector_df)
-    eq_(one_hot.shape, (2, 4, 3))
+    assert one_hot.shape == (2, 4, 3)
     assert_equal(
         one_hot[0],
         [

test/test_calibrate_percentile_ranks_command.py

@@ -41,7 +41,7 @@ def run_and_check(n_jobs=0, delete=True, additional_args=[]):
     print("Saved predictor to", dest_models_dir)
 
     new_predictor = Class1AffinityPredictor.load(dest_models_dir)
-    assert_equal(len(new_predictor.allele_to_percent_rank_transform), 0)
+    assert len(new_predictor.allele_to_percent_rank_transform) == 0
 
     args = [
         "mhcflurry-calibrate-percentile-ranks",
@@ -58,7 +58,7 @@ def run_and_check(n_jobs=0, delete=True, additional_args=[]):
     subprocess.check_call(args)
 
     new_predictor = Class1AffinityPredictor.load(dest_models_dir)
-    assert_equal(len(new_predictor.allele_to_percent_rank_transform), 2)
+    assert len(new_predictor.allele_to_percent_rank_transform) == 2
 
     if delete:
         print("Deleting: %s" % dest_models_dir)

test/test_class1_affinity_predictor.py

@@ -13,7 +13,6 @@
 
 from mhcflurry import Class1AffinityPredictor
 
-from nose.tools import eq_, assert_raises
 from numpy import testing
 
 from mhcflurry.downloads import get_path
@@ -167,7 +166,7 @@ def test_class1_affinity_predictor_a0205_memorize_training_data():
     predictor.calibrate_percentile_ranks(num_peptides_per_length=1000)
     ic50_pred = predictor.predict(df.peptide.values, allele=allele)
     ic50_true = df.measurement_value.values
-    eq_(len(ic50_pred), len(ic50_true))
+    assert len(ic50_pred) == len(ic50_true)
     testing.assert_allclose(
         numpy.log(ic50_pred), numpy.log(ic50_true), rtol=0.2, atol=0.2
     )
@@ -184,17 +183,17 @@ def test_class1_affinity_predictor_a0205_memorize_training_data():
 
     # Test an unknown allele
     print("Starting unknown allele check")
-    eq_(predictor.supported_alleles, [allele])
+    assert predictor.supported_alleles == [allele]
     ic50_pred = predictor.predict(df.peptide.values, allele="HLA-A*02:01", throw=False)
     assert numpy.isnan(ic50_pred).all()
 
-    assert_raises(
+    testing.assert_raises(
         ValueError, predictor.predict, df.peptide.values, allele="HLA-A*02:01"
     )
 
-    eq_(predictor.supported_alleles, [allele])
-    assert_raises(ValueError, predictor.predict, ["AAAAA"], allele=allele)  # too short
-    assert_raises(
+    assert predictor.supported_alleles == [allele]
+    testing.assert_raises(ValueError, predictor.predict, ["AAAAA"], allele=allele)  # too short
+    testing.assert_raises(
         ValueError,
         predictor.predict,
         ["AAAAAAAAAAAAAAAAAAAA"],  # too long

test/test_class1_neural_network.py

@@ -6,8 +6,6 @@
 
 import pytest
 
-from nose.tools import eq_, assert_less, assert_greater, assert_almost_equal
-
 import pandas
 
 from mhcflurry.class1_neural_network import Class1NeuralNetwork
@@ -53,7 +51,7 @@ def test_class1_neural_network_a0205_training_accuracy(setup_module):
     predictor.fit(df.peptide.values, df.measurement_value.values)
     ic50_pred = predictor.predict(df.peptide.values)
     ic50_true = df.measurement_value.values
-    eq_(len(ic50_pred), len(ic50_true))
+    assert len(ic50_pred) == len(ic50_true)
     testing.assert_allclose(
         numpy.log(ic50_pred), numpy.log(ic50_true), rtol=0.2, atol=0.2
     )
@@ -75,7 +73,7 @@ def test_class1_neural_network_a0205_training_accuracy(setup_module):
     )
     predictor2 = Class1NeuralNetwork(**hyperparameters2)
     predictor2.fit(df.peptide.values, df.measurement_value.values, verbose=0)
-    eq_(predictor.network().to_json(), predictor2.network().to_json())
+    assert predictor.network().to_json() == predictor2.network().to_json()
 
 
 def test_inequalities(setup_module):
@@ -146,8 +144,8 @@ def test_inequalities(setup_module):
 
     # Binders should be stronger
     for pred in ["prediction1", "prediction2"]:
-        assert_less(df.loc[df.value < 1000, pred].mean(), 500)
-        assert_greater(df.loc[df.value >= 1000, pred].mean(), 500)
+        assert df.loc[df.value < 1000, pred].mean() < 500
+        assert df.loc[df.value >= 1000, pred].mean() > 500
 
     # For the binders, the (=) on the weak-binding measurement (100) in
     # inequality1 should make the prediction weaker, whereas for inequality2

test/test_class1_presentation_predictor.py

@@ -6,7 +6,6 @@
 import pickle
 
 from numpy.testing import assert_, assert_equal, assert_allclose, assert_array_equal
-from nose.tools import assert_greater, assert_less
 import pytest
 import numpy
 
@@ -108,17 +107,17 @@ def add_prediction_cols(test_df, predictor):
 
     print("AUC", score1, score2)
 
-    assert_greater(score1, 0.8)
-    assert_greater(score2, 0.8)
+    assert score1 > 0.8
+    assert score2 > 0.8
 
     score1 = roc_auc_score(
         test_df.hit.values, -test_df.prediction1_percentile.values)
     score2 = roc_auc_score(
         test_df.hit.values, -test_df.prediction2_percentile.values)
     print("AUC (using percentiles)", score1, score2)
 
-    assert_greater(score1, 0.8)
-    assert_greater(score2, 0.8)
+    assert score1 > 0.8
+    assert score2 > 0.8
 
     # Test saving, loading, pickling
     models_dir = tempfile.mkdtemp("_models")
@@ -161,7 +160,7 @@ def test_downloaded_predictor_small(predictors):
         peptide_lengths=[9],
         result="best")
     print(scan_results)
-    assert_equal(len(scan_results), 6)
+    assert len(scan_results) == 6
 
     scan_results = presentation_predictor.predict_sequences(
         sequences=[
@@ -176,7 +175,7 @@ def test_downloaded_predictor_small(predictors):
         peptide_lengths=[8, 9],
         result="best")
     print(scan_results)
-    assert_equal(len(scan_results), 6)
+    assert len(scan_results) == 6
 
     scan_results = presentation_predictor.predict_sequences(
         sequences=[
@@ -191,7 +190,7 @@ def test_downloaded_predictor_small(predictors):
         peptide_lengths=[9],
         result="all")
     print(scan_results)
-    assert_equal(len(scan_results), 6)
+    assert len(scan_results) == 6
 
     scan_results = presentation_predictor.predict_sequences(
         sequences=[
@@ -206,7 +205,7 @@ def test_downloaded_predictor_small(predictors):
         peptide_lengths=[8, 9],
         result="all")
     print(scan_results)
-    assert_equal(len(scan_results), 18)
+    assert len(scan_results) == 18
 
     scan_results = presentation_predictor.predict_sequences(
         sequences=[
@@ -221,7 +220,7 @@ def test_downloaded_predictor_small(predictors):
         peptide_lengths=[10],
         result="all")
     print(scan_results)
-    assert_equal(len(scan_results), 0)
+    assert len(scan_results) == 0
 
 
 def test_downloaded_predictor(predictors):
@@ -244,7 +243,7 @@ def test_downloaded_predictor(predictors):
         ])
     print(scan_results1)
 
-    assert_equal(len(scan_results1), 3), str(scan_results1)
+    assert len(scan_results1) == 3, str(scan_results1)
     assert (scan_results1.affinity < 200).all(), str(scan_results1)
     assert (scan_results1.presentation_score > 0.7).all(), str(scan_results1)
 
@@ -311,7 +310,7 @@ def test_downloaded_predictor(predictors):
     print(scan_results4)
 
     assert len(scan_results4) > 200, len(scan_results4)
-    assert_less(scan_results4.iloc[0].affinity, 100)
+    assert scan_results4.iloc[0].affinity < 100
 
     sequences = {
         "seq1":
@@ -345,7 +344,7 @@ def test_downloaded_predictor(predictors):
             ],
         })
     print(scan_results5)
-    assert_equal(len(scan_results5), len(scan_results4) * 2)
+    assert len(scan_results5) == len(scan_results4) * 2
 
     # Test case-insensitive.
     scan_results6 = presentation_predictor.predict_sequences(

test/test_class1_processing_neural_network.py

@@ -7,8 +7,6 @@
 
 from sklearn.metrics import roc_auc_score
 
-from nose.tools import eq_, assert_less, assert_greater, assert_almost_equal
-
 import pandas
 import pytest
 
@@ -233,8 +231,8 @@ def is_hit(n_flank, c_flank, peptide):
     print("Test auc", test_auc)
 
     if do_assertions:
-        assert_greater(train_auc, 0.9)
-        assert_greater(test_auc, 0.85)
+        assert train_auc > 0.9
+        assert test_auc > 0.85
 
     return network
 

test/test_custom_loss.py

@@ -1,8 +1,6 @@
 from . import initialize
 initialize()
 
-from nose.tools import eq_, assert_less, assert_greater, assert_almost_equal
-
 import numpy
 import tensorflow as tf
 import pytest
@@ -40,39 +38,39 @@ def test_mse_with_inequalities(loss_obj=CUSTOM_LOSSES['mse_with_inequalities']):
     print(adjusted_y)
     loss0 = evaluate_loss(loss_obj.loss, adjusted_y, y_values)
     print(loss0)
-    eq_(loss0, 0.0)
+    assert loss0 == 0.0
 
     adjusted_y = loss_obj.encode_y(y_values, [">", ">", ">", ">"])
     loss0 = evaluate_loss(loss_obj.loss, adjusted_y, y_values)
-    eq_(loss0, 0.0)
+    assert loss0 == 0.0
 
     adjusted_y = loss_obj.encode_y(y_values, ["<", "<", "<", "<"])
     loss0 = evaluate_loss(loss_obj.loss, adjusted_y, y_values)
-    eq_(loss0, 0.0)
+    assert loss0 == 0.0
 
     adjusted_y = loss_obj.encode_y(y_values, ["=", "<", "=", ">"])
     loss0 = evaluate_loss(loss_obj.loss, adjusted_y, y_values)
-    eq_(loss0, 0.0)
+    assert loss0 == 0.0
 
     adjusted_y = loss_obj.encode_y(y_values, ["=", "<", "=", ">"])
     loss0 = evaluate_loss(loss_obj.loss, adjusted_y, [0.0, 0.4, 0.8, 1.0])
-    eq_(loss0, 0.0)
+    assert loss0 == 0.0
 
     adjusted_y = loss_obj.encode_y(y_values, [">", "<", ">", ">"])
     loss0 = evaluate_loss(loss_obj.loss, adjusted_y, [0.1, 0.4, 0.9, 1.0])
-    eq_(loss0, 0.0)
+    assert loss0 == 0.0
 
     adjusted_y = loss_obj.encode_y(y_values, [">", "<", ">", ">"])
     loss0 = evaluate_loss(loss_obj.loss, adjusted_y, [0.1, 0.6, 0.9, 1.0])
-    assert_greater(loss0, 0.0)
+    assert loss0 > 0.0
 
     adjusted_y = loss_obj.encode_y(y_values, ["=", "<", ">", ">"])
     loss0 = evaluate_loss(loss_obj.loss, adjusted_y, [0.1, 0.6, 0.9, 1.0])
-    assert_almost_equal(loss0, 0.02 / 4)
+    assert loss0 == pytest.approx(0.02 / 4, abs=1e-7)
 
     adjusted_y = loss_obj.encode_y(y_values, ["=", "<", "=", ">"])
     loss0 = evaluate_loss(loss_obj.loss, adjusted_y, [0.1, 0.6, 0.9, 1.0])
-    assert_almost_equal(loss0, 0.03 / 4)
+    assert loss0 == pytest.approx(0.03 / 4, abs=1e-7)
 
 
 def test_mse_with_inequalities_and_multiple_outputs():
@@ -91,7 +89,7 @@ def test_mse_with_inequalities_and_multiple_outputs():
             [3000, 0.8],
             [4000, 1.0],
         ])
-    assert_almost_equal(loss0, 0.0)
+    assert loss0 == pytest.approx(0.0, abs=1e-7)
 
     y_values = [0.0, 0.5, 0.8, 1.0]
     adjusted_y = loss_obj.encode_y(
@@ -105,7 +103,7 @@ def test_mse_with_inequalities_and_multiple_outputs():
             [3000, 0.8],
             [1.0, 4000],
         ])
-    assert_almost_equal(loss0, 0.02 / 4)
+    assert loss0 == pytest.approx(0.02 / 4, abs=1e-7)
 
     y_values = [0.0, 0.5, 0.8, 1.0]
     adjusted_y = loss_obj.encode_y(
@@ -119,7 +117,7 @@ def test_mse_with_inequalities_and_multiple_outputs():
             [3000, 0.8],
             [1.0, 4000],
         ])
-    assert_almost_equal(loss0, 0.01 / 4)
+    assert loss0 == pytest.approx(0.01 / 4, abs=1e-7)
 
     y_values = [0.0, 0.5, 0.8, 1.0]
     adjusted_y = loss_obj.encode_y(
@@ -133,7 +131,7 @@ def test_mse_with_inequalities_and_multiple_outputs():
             [3000, 0.8],
             [1.0, 4000],
         ])
-    assert_almost_equal(loss0, 0.02 / 4)
+    assert loss0 == pytest.approx(0.02 / 4, abs=1e-7)
 
 
 def test_multiallelic_mass_spec_loss():

test/test_hyperparameters.py

@@ -14,7 +14,7 @@ def test_all_combinations_of_hyperparameters():
         Class1NeuralNetwork
         .hyperparameter_defaults
         .models_grid(**combinations_dict))
-    assert_equal(len(results), 4)
+    assert len(results) == 4
 
 if __name__ == "__main__":
     test_all_combinations_of_hyperparameters()

test/test_predict_command.py

@@ -50,7 +50,7 @@ def test_csv():
         for delete in deletes:
             os.unlink(delete)
 
-    assert_equal(result.shape, (3, 8))
+    assert result.shape == (3, 8)
 
 
 def test_no_csv():
@@ -75,7 +75,7 @@ def test_no_csv():
             os.unlink(delete)
 
     print(result)
-    assert_equal(len(result), 6)
+    assert len(result) == 6
     sub_result1 = result.loc[result.peptide == "SIINFEKL"].set_index("allele")
     print(sub_result1)
     assert (