fix: bug where Phenotypes.subset(inplace=True) would raise an Attri…

…buteError (#226)

.readthedocs.yaml

@@ -6,12 +6,16 @@
 
 version: 2
 
+build:
+  os: "ubuntu-22.04"
+  tools:
+    python: "3.7"
+
 sphinx:
   configuration: docs/conf.py
   fail_on_warning: true
 
 python:
-  version: 3.7
   install:
     - method: pip
       path: .

haptools/data/genotypes.py

@@ -599,8 +599,8 @@ def check_maf(
                 self.variants = np.delete(self.variants, idx)
                 maf = np.delete(maf, idx)
                 self.log.info(
-                    "Ignoring missing genotypes from "
-                    f"{original_num_variants - len(self.variants)} samples"
+                    f"Ignoring {original_num_variants - len(self.variants)} variants "
+                    f"with MAF < {threshold}"
                 )
                 self._var_idx = None
             else:

haptools/data/phenotypes.py

@@ -402,7 +402,7 @@ def subset(
                 )
             name_idx = tuple(self._name_idx[name] for name in pts.names)
             if inplace:
-                self._pts.names = None
+                self._name_idx = None
             pts.data = pts.data[:, name_idx]
         if not inplace:
             return pts

tests/test_data.py

@@ -989,6 +989,18 @@ def test_subset_phenotypes(self):
         np.testing.assert_allclose(pts_sub.data, expected_data)
         assert np.array_equal(pts_sub.names, expected_names)
 
+        # also try in-place
+        expected_data = pts.data[[3, 4], [1]]
+        expected_data = expected_data[:, np.newaxis]
+        assert len(expected_data.shape) == 2
+        expected_names = (pts.names[1],)
+        samples = ("HG00100", "HG00101")
+        names = ("bmi",)
+        pts.subset(samples=samples, names=names, inplace=True)
+        assert pts_sub.samples == samples
+        np.testing.assert_allclose(pts_sub.data, expected_data)
+        assert np.array_equal(pts_sub.names, expected_names)
+
 
 class TestCovariates:
     def _get_expected_covariates(self):