fix ECs in bacdive, cell_shape in bacdive and madin_etal

kg_microbe/transform_utils/bacdive/bacdive.py

@@ -67,6 +67,7 @@
     DOMAIN,
     DSM_NUMBER,
     DSM_NUMBER_COLUMN,
+    EC_CATEGORY,
     EC_KEY,
     EC_PREFIX,
     ENZYME_TO_ASSAY_EDGE,
@@ -484,7 +485,7 @@ def run(self, data_file: Union[Optional[Path], Optional[str]] = None, show_statu
                     assay_nodes_to_write.append(
                         [
                             ec_id,
-                            PHENOTYPIC_CATEGORY,
+                            EC_CATEGORY,
                             assay[BACDIVE_MAPPING_ENZYME_LABEL],
                         ]
                         + [None] * (len(self.node_header) - 3)

kg_microbe/transform_utils/bactotraits/bactotraits.py

@@ -245,6 +245,15 @@ def run(
             combo_curie_map = {
                 key: value for key, value in custom_curie_map.items() if COMBO_KEY in value
             }
+            unique_combo_node_data = [
+                (
+                    inner_curie_map[CURIE_COLUMN],
+                    inner_curie_map[CATEGORY_COLUMN],
+                    inner_curie_map[NAME_COLUMN],
+                )
+                for _, v in combo_curie_map.items()
+                for inner_curie_map in v[COMBO_KEY]
+            ]
             unique_combo_edge_data = [
                 (
                     v[CURIE_COLUMN],
@@ -257,6 +266,7 @@ def run(
                 for inner_curie_map in v[COMBO_KEY]
             ]
             combo_edge_data = [list(edge) for edge in unique_combo_edge_data]
+            combo_node_data = [list(edge) for edge in unique_combo_node_data]
 
             progress_class = tqdm if show_status else DummyTqdm
             with progress_class() as progress:
@@ -333,6 +343,7 @@ def run(
                     progress.set_description(f"Processing line #{i}")
                     # After each iteration, call the update method to advance the progress bar.
                     progress.update(1)
+                node_writer.writerows(combo_node_data)
                 edge_writer.writerows(combo_edge_data)
         drop_duplicates(self.output_node_file)
         drop_duplicates(self.output_edge_file)

kg_microbe/transform_utils/constants.py

@@ -210,7 +210,6 @@
 MEDIUM_TYPE_CATEGORY = "biolink:ChemicalMixture"
 SOLUTION_CATEGORY = "biolink:ChemicalEntity"
 INGREDIENT_CATEGORY = "biolink:ChemicalEntity"
-SHAPE_CATEGORY = "biolink:AbstractEntity"
 METABOLISM_CATEGORY = "biolink:ActivityAndBehavior"
 PATHWAY_CATEGORY = "biolink:BiologicalProcess"
 CARBON_SUBSTRATE_CATEGORY = "biolink:ChemicalEntity"

kg_microbe/transform_utils/custom_curies.yaml

@@ -150,7 +150,7 @@ trophic_type:
     curie: "trophic_type:heterotrophy"
     name: "heterotroph"
     <<: *biological_process_block
-  tt_heterotrph: *heterotroph_block
+  tt_heterotroph: *heterotroph_block
   chemoautolithotroph:
     curie: "trophic_type:chemoautolithotrophy"
     name: "chemoautolithotroph"
@@ -363,7 +363,7 @@ cell_shape:
     name: "ellipsoidal shaped cell"
     <<: *phenotypic_quality_block
 
-  curved_spiral_shaped:
+  s_curved_spiral:
     curie: "cell_shape:curved_spiral"
     name: "curved spiral shaped cell"
     <<: *phenotypic_quality_block

kg_microbe/transform_utils/madin_etal/madin_etal.py

@@ -49,9 +49,9 @@
     PATHWAY_CATEGORY,
     PATHWAY_PREFIX,
     PATHWAYS_COLUMN,
+    PHENOTYPIC_CATEGORY,
     PREFERRED_TERM_KEY,
     ROLE_CATEGORY,
-    SHAPE_CATEGORY,
     SHAPE_PREFIX,
     SUBJECT_LABEL_COLUMN,
     TAX_ID_COLUMN,
@@ -363,7 +363,11 @@ def run(self, data_file: Union[Optional[Path], Optional[str]] = None, show_statu
                         else filtered_row[CELL_SHAPE_COLUMN]
                     )
                     if cell_shape:
-                        cell_shape_node = [SHAPE_PREFIX + cell_shape, SHAPE_CATEGORY, cell_shape]
+                        cell_shape_node = [
+                            SHAPE_PREFIX + cell_shape,
+                            PHENOTYPIC_CATEGORY,
+                            cell_shape,
+                        ]
                         tax_to_cell_shape_edge = [
                             tax_id,
                             NCBI_TO_SHAPE_EDGE,