Merge pull request #291 from biocore/fix-matching

Fixing sparse matching

gneiss/tests/test_util.py

@@ -379,6 +379,31 @@ def test_biom_match_no_common_ids(self):
         with self.assertRaises(ValueError):
             match(table, md)
 
+    def test_biom_match_intersect(self):
+        table = Table(
+            np.array([[0, 0, 1, 1],
+                      [2, 3, 4, 4],
+                      [5, 5, 3, 3]]).T,
+            ['a', 'b', 'c', 'd'],
+            ['s1', 's2', 'y4'])
+        md = pd.DataFrame([[0, 1], [1, 0], [1, 1]],
+                          index=['s2', 's1', 's3'],
+                          columns=['x1', 'x2'])
+        exp_table = Table(
+            np.array([[0, 0, 1, 1],
+                      [2, 3, 4, 4]]).T,
+            ['a', 'b', 'c', 'd'],
+            ['s1', 's2'])
+        exp_md = pd.DataFrame([[1, 0], [0, 1]],
+                              columns=['x1', 'x2'],
+                              index=['s1', 's2'])
+        res_table, res_md = match(table, md)
+        pdt.assert_frame_equal(res_md, exp_md)
+        exp_df = pd.DataFrame(exp_table.to_dataframe())
+        res_df = pd.DataFrame(res_table.to_dataframe())
+        pdt.assert_frame_equal(res_df, exp_df)
+
+
     def test_biom_match_tips_intersect_tips(self):
         # there are less tree tips than table columns
         table = Table(

gneiss/util.py

@@ -131,7 +131,10 @@ def _dense_match(table, metadata):
     if len(submetadataids) != len(metadata.index):
         raise ValueError("`metadata` has duplicate sample ids.")
 
-    idx = subtableids & submetadataids
+    idx = list(subtableids & submetadataids)
+    # make sure that the list is always the same to remove
+    # unwanted random behavior
+    idx.sort()
     if len(idx) == 0:
         raise ValueError(("No more samples left.  Check to make sure that "
                           "the sample names between `metadata` and `table` "
@@ -148,7 +151,10 @@ def _sparse_match(table, metadata):
     submetadataids = set(metadata.index)
     if len(submetadataids) != len(metadata.index):
         raise ValueError("`metadata` has duplicate sample ids.")
-    idx = subtableids & submetadataids
+    idx = list(subtableids & submetadataids)
+    # make sure that the list is always the same to remove
+    # unwanted random behavior
+    idx.sort()
     if len(idx) == 0:
         raise ValueError(("No more samples left.  Check to make sure that "
                           "the sample names between `metadata` and `table` "
@@ -165,6 +171,7 @@ def sort_f(xs):
         return [xs[out_metadata.index.get_loc(x)] for x in xs]
 
     out_table = out_table.sort(sort_f=sort_f, axis='sample')
+    out_metadata = out_metadata.loc[out_table.ids()]
     return out_table, out_metadata