Fix edgeR rpy2 tests (#692)

* fix broken rpy2 edger tests

* updated edger tests

pertpy/tools/_differential_gene_expression/_edger.py

@@ -60,8 +60,10 @@ def fit(self, **kwargs):  # adata, design, mask, layer
         logger.info("Calculating NormFactors")
         dge = edger.calcNormFactors(dge)
 
-        with localconverter(get_conversion() + pandas2ri.converter):
-            design_r = ro.conversion.py2rpy(pd.DataFrame(self.design))
+        with localconverter(get_conversion() + numpy2ri.converter):
+            # dt = np.dtype([(name, 'float64') for name in self.design.columns])
+            # design_array = np.array(self.design.values, dtype=dt)
+            design_r = ro.conversion.py2rpy(self.design.values)
 
         logger.info("Estimating Dispersions")
         dge = edger.estimateDisp(dge, design=design_r)

tests/tools/_differential_gene_expression/test_edger.py

@@ -1,4 +1,5 @@
-from pertpy.tools._differential_gene_expression import EdgeR
+from pertpy.tools._differential_gene_expression import EdgeR, PyDESeq2
+import numpy.testing as npt
 
 
 def test_edger_simple(test_adata):
@@ -13,6 +14,17 @@ def test_edger_simple(test_adata):
     res_df = method.test_contrasts(method.contrast("condition", "A", "B"))
 
     assert len(res_df) == test_adata.n_vars
+    # Compare against snapshot
+    npt.assert_almost_equal(
+        res_df.p_value.values,
+        [8.0000e-05, 1.8000e-04, 5.3000e-04, 1.1800e-03, 3.3800e-02, 3.3820e-02, 7.7980e-02, 1.3715e-01, 2.5052e-01, 9.2485e-01],
+        decimal=4,
+    )
+    npt.assert_almost_equal(
+        res_df.log_fc.values,
+        [ 0.61208, -0.39374,  0.57944,  0.7343 , -0.58675,  0.42575, -0.23951, -0.20761,  0.17489,  0.0247],
+        decimal=4,
+    )
 
 
 def test_edger_complex(test_adata):
@@ -28,6 +40,12 @@ def test_edger_complex(test_adata):
     # Check that the index of the result matches the var_names of the AnnData object
     assert set(test_adata.var_names) == set(res_df["variable"])
 
+    # Compare ranking of genes from a different method (without design matrix handling)
+    down_gene = res_df.set_index("variable").loc['gene3', 'log_fc'] 
+    up_gene = res_df.set_index("variable").loc['gene1', 'log_fc'] 
+    assert down_gene < up_gene
 
-# TODO: there should be a test checking if, for a concrete example, the output p-values and effect sizes are what
-# we expect (-> frozen snapshot, that way we also get a heads-up if something changes upstream)
+    method = PyDESeq2(adata=test_adata, design="~condition1+group")
+    method.fit()
+    deseq_res_df = method.test_contrasts(method.contrast("condition1", "A", "B"))
+    assert all(res_df.sort_values('log_fc')['variable'].values == deseq_res_df.sort_values('log_fc')['variable'].values)