From 46e369f42f16ec025b0752c35466a5bc592c2fff Mon Sep 17 00:00:00 2001
From: naglepuff <mike.nagler@kitware.com>
Date: Thu, 19 Dec 2024 12:36:14 -0500
Subject: [PATCH 1/2] Fix metap ingest

---
 nmdc_server/data_object_filters.py | 2 +-
 nmdc_server/ingest/all.py          | 2 +-
 nmdc_server/ingest/pipeline.py     | 4 +---
 3 files changed, 3 insertions(+), 5 deletions(-)

diff --git a/nmdc_server/data_object_filters.py b/nmdc_server/data_object_filters.py
index d6b77ed2..0cd8c6d6 100644
--- a/nmdc_server/data_object_filters.py
+++ b/nmdc_server/data_object_filters.py
@@ -34,7 +34,7 @@ class WorkflowActivityTypeEnum(Enum):
     metabolomics_analysis = "nmdc:MetabolomicsAnalysis"
     metagenome_assembly = "nmdc:MetagenomeAssembly"
     metagenome_annotation = "nmdc:MetagenomeAnnotation"
-    metaproteomic_analysis = "nmdc:MetaproteomicAnalysis"
+    metaproteomic_analysis = "nmdc:MetaproteomicsAnalysis"
     metatranscriptome = "nmdc:MetatranscriptomeAnalysis"
     metatranscriptome_assembly = "nmdc:MetatranscriptomeAssembly"
     metatranscriptome_annotation = "nmdc:MetatranscriptomeAnnotation"
diff --git a/nmdc_server/ingest/all.py b/nmdc_server/ingest/all.py
index f3c79058..fa300799 100644
--- a/nmdc_server/ingest/all.py
+++ b/nmdc_server/ingest/all.py
@@ -220,7 +220,7 @@ def load(db: Session, function_limit=None, skip_annotation=False):
         pipeline.load(
             db,
             mongodb[workflow_set].find(
-                {"type": "nmdc:MetaproteomicsAnalysis"},
+                {"type": WorkflowActivityTypeEnum.metaproteomic_analysis.value},
                 no_cursor_timeout=True,
             ),
             pipeline.load_mp_analysis,
diff --git a/nmdc_server/ingest/pipeline.py b/nmdc_server/ingest/pipeline.py
index c7039675..56182508 100644
--- a/nmdc_server/ingest/pipeline.py
+++ b/nmdc_server/ingest/pipeline.py
@@ -92,7 +92,6 @@ def load_mp_analysis(db: Session, obj: Dict[str, Any], **kwargs) -> LoadObjectRe
             "gene_function_id": {
                 "$regex": gene_regex,
             },
-            "best_protein": True,
         },
         no_cursor_timeout=True,
         projection={
@@ -100,7 +99,6 @@ def load_mp_analysis(db: Session, obj: Dict[str, Any], **kwargs) -> LoadObjectRe
             "was_generated_by": True,
             "count": True,
             "gene_function_id": True,
-            "best_protein": True,
         },
     )
     if kwargs.get("function_limit"):
@@ -116,7 +114,7 @@ def load_mp_analysis(db: Session, obj: Dict[str, Any], **kwargs) -> LoadObjectRe
                 metaproteomic_analysis_id=pipeline.id,
                 gene_function_id=function_id,
                 count=annotation["count"],
-                best_protein=annotation["best_protein"],
+                best_protein=True,
             )
         )
     if metap_gene_function_aggregations:

From 204bcd649f9cf4a4e5f741302abbe37a2d7eb759 Mon Sep 17 00:00:00 2001
From: naglepuff <mike.nagler@kitware.com>
Date: Thu, 19 Dec 2024 14:37:17 -0500
Subject: [PATCH 2/2] Use gene query logic for all gene table values

---
 nmdc_server/query.py | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/nmdc_server/query.py b/nmdc_server/query.py
index 4e19b946..9cdfe17d 100644
--- a/nmdc_server/query.py
+++ b/nmdc_server/query.py
@@ -384,7 +384,13 @@ def query(self, db) -> Query:
 
             # Gene function queries are treated differently because they join
             # in three different places (metaT, metaG and metaP).
-            if table == Table.gene_function:
+            if table in [
+                Table.gene_function,
+                Table.kegg_function,
+                Table.go_function,
+                Table.pfam_function,
+                Table.cog_function,
+            ]:
                 metag_matches = filter.matches(db, self.table)
                 metap_conditions = [
                     SimpleConditionSchema(