fix res dict building

bioml-tools · Nov 15, 2024 · 13e8203 · 13e8203
1 parent 994366b
commit 13e8203
Showing 1 changed file with 4 additions and 1 deletion.
diff --git a/src/bio_datasets/structure/residue.py b/src/bio_datasets/structure/residue.py
@@ -312,7 +312,7 @@ def from_ccd(
         chem_component_3to1 = get_component_3to1()
         chem_component_categories = get_component_categories(get_component_types())
         frequencies = get_residue_frequencies()
-        res_names = np.unique(ccd_data["chem_comp_atom"]["comp_id"].as_array(str))
+        res_names = list(np.unique(ccd_data["chem_comp_atom"]["comp_id"].as_array(str)))
 
         def keep_res(res_name):
             res_filter = frequencies.get(res_name, 0) >= minimum_pdb_entries
@@ -326,6 +326,9 @@ def keep_res(res_name):
             res_filter = res_filter and (
                 keep_hydrogens or res_name not in ["H", "D", "D8U"]
             )
+            res_filter = res_filter and (
+                res_name in chem_component_3to1 and res_name in chem_component_categories
+            )
             return res_filter
 
         res_names = [res for res in res_names if keep_res(res)]