Merge pull request #23 from TheJacksonLaboratory/G3-457-view-similar-…

…genesets-button-broken

Reverting change to get_genesets_by_hom_id

deploy/k8s/base/ingress.yaml

@@ -0,0 +1,11 @@
+apiVersion: networking.k8s.io/v1
+kind: Ingress
+metadata:
+  name:  geneweaver-legacy-ingress
+  annotations:
+    # Set the read and send timeouts for long connections
+    nginx.ingress.kubernetes.io/proxy-read-timeout: "3600"  # 1 hour, adjust as needed
+    nginx.ingress.kubernetes.io/proxy-send-timeout: "3600"  # 1 hour, adjust as needed
+    # Optional: Prevent connection closing during idle time
+    nginx.ingress.kubernetes.io/proxy-buffering: "off"      # Ensure SSE is not buffered
+    nginx.ingress.kubernetes.io/keep-alive: "75s"           # Adjust keep-alive duration

deploy/k8s/base/kustomization.yaml

@@ -8,3 +8,4 @@ resources:
   - service.yaml
   - external-secrets.yaml
   - persistent-volume-claim.yaml
+  - ingress.yaml

deploy/k8s/overlays/jax-cluster-dev-10--dev/kustomization.yaml

@@ -3,7 +3,7 @@ kind: Kustomization
 namespace: dev
 resources:
   - ../../base
-  - ingress.yaml
 patchesStrategicMerge:
   - persistent-volume-claim.yaml
   - service.yaml
+  - ingress.yaml

deploy/k8s/overlays/jax-cluster-dev-10--sqa/kustomization.yaml

@@ -3,6 +3,6 @@ kind: Kustomization
 namespace: sqa
 resources:
   - ../../base
-  - ingress.yaml
 patchesStrategicMerge:
   - persistent-volume-claim.yaml
+  - ingress.yaml

deploy/k8s/overlays/jax-cluster-prod-10--prod/kustomization.yaml

@@ -3,7 +3,7 @@ kind: Kustomization
 namespace: prod
 resources:
   - ../../base
-  - ingress.yaml
 patchesStrategicMerge:
   - persistent-volume-claim.yaml
   - configmap.yaml
+  - ingress.yaml

deploy/k8s/overlays/jax-cluster-prod-10--stage/kustomization.yaml

@@ -3,7 +3,7 @@ kind: Kustomization
 namespace: stage
 resources:
   - ../../base
-  - ingress.yaml
 patchesStrategicMerge:
   - persistent-volume-claim.yaml
   - configmap.yaml
+  - ingress.yaml

migration/114-create-view-geneset2hom.sql

@@ -0,0 +1,14 @@
+CREATE MATERIALIZED VIEW extsrc.geneset2hom AS
+SELECT g.gs_id, array_agg(h.hom_id) as hom_id_array
+FROM extsrc.homology h
+         INNER JOIN extsrc.geneset_value gv ON
+    h.ode_gene_id = gv.ode_gene_id
+         INNER JOIN production.geneset g ON gv.gs_id = g.gs_id
+WHERE g.gs_status NOT LIKE 'de%'
+  AND hom_id IN
+      (SELECT DISTINCT hom_id
+       FROM extsrc.homology)
+GROUP BY g.gs_id;
+
+CREATE INDEX geneset2hom_gs_id_index
+    ON extsrc.geneset2hom (gs_id);

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "geneweaver-legacy"
-version = "1.4.5"
+version = "1.4.6"
 description = ""
 authors = ["Alexander Berger <[email protected]>"]
 readme = "README.md"

src/application.py

@@ -3224,9 +3224,9 @@ def calculate_jaccard(gs_id, genesets):
     """
     jaccards = {}
     gs1 = geneweaverdb.get_geneset_hom_ids(gs_id)
+    gs2s = geneweaverdb.get_genesets_hom_ids(genesets)
     for g in genesets:
-        gs2 = geneweaverdb.get_geneset_hom_ids(g)
-        jaccards[g] = jaccard(gs1, gs2)
+        jaccards[g] = jaccard(gs1, gs2s[g])
     return jaccards
 
 

src/geneweaverdb.py

@@ -3620,18 +3620,20 @@ def get_similar_genesets(geneset_id, user_id, grp_by):
             '''SELECT a.* FROM (
                    (SELECT geneset.*,jac_value,gic_value 
                        FROM geneset, geneset_jaccard gj
-                       WHERE gs_id=gs_id_right AND gs_id_left=408167 
+                       WHERE gs_id=gs_id_right AND gs_id_left=%(geneset_id)s 
+                         AND geneset_is_readable(%(user_id)s, gs_id) 
                          AND gs_status NOT LIKE 'de%%' 
                            ORDER BY gj.jac_value DESC LIMIT 500
                      ) 
                      UNION
                    (SELECT geneset.*,jac_value,gic_value 
                        FROM geneset, geneset_jaccard gj 
-                       WHERE gs_id=gs_id_left AND gs_id_right=408167 
+                       WHERE gs_id=gs_id_left AND gs_id_right=%(geneset_id)s 
+                         AND geneset_is_readable(%(user_id)s, gs_id) 
                          AND gs_status NOT LIKE 'de%%' 
                            ORDER BY gj.jac_value DESC LIMIT 500
                      ) 
-                 ) AS a ORDER BY a.jac_value DESC LIMIT 1000''',
+                 ) AS a ORDER BY ''' + order_by + ''' a.jac_value DESC LIMIT 1000''',
             {
                 'geneset_id': geneset_id,
                 'user_id': user_id,
@@ -3921,21 +3923,16 @@ def get_geneset_hom_ids(gs_id):
 
 def get_genesets_by_hom_id(hom_ids):
     """
-    Find which genesets are associated with which arrays. postgres does not have an any to
-    any array comparision so I have to loop :(
+    Find which genesets are associated with which arrays.
     :param hom_ids:
     :return: list of genesets
     """
-    if not isinstance(hom_ids, list):
-        hom_ids = [hom_ids]
-
+    geneset_list = []
     with PooledCursor() as cursor:
-        geneset_list = []
-        for h in hom_ids:
-            cursor.execute('SELECT geneset_array FROM extsrc.hom2geneset WHERE hom_id=ANY(%s)', (h,))
-            geneset_list.append(cursor.fetchone()[0])
-        list(set(geneset_list[0]))
-    return geneset_list
+        cursor.execute("SELECT geneset_array FROM extsrc.hom2geneset WHERE hom_id=ANY(%s)", (hom_ids,))
+        for result in cursor.fetchall():
+            geneset_list.extend(result[0])
+    return list(set(geneset_list))
 
 
 def get_genesets_hom_ids(gs_ids):
@@ -3945,18 +3942,7 @@ def get_genesets_hom_ids(gs_ids):
     """
     with PooledCursor() as cursor:
         cursor.execute("""
-        SELECT
-            g.gs_id,
-            ARRAY_AGG(DISTINCT h.hom_id) AS hom_ids
-        FROM
-            extsrc.homology h
-                INNER JOIN extsrc.geneset_value gsv ON h.ode_gene_id = gsv.ode_gene_id
-                INNER JOIN production.geneset g ON gsv.gs_id = g.gs_id
-        WHERE
-            g.gs_status NOT LIKE 'de%%'
-          AND g.gs_id = ANY(%(gs_ids)s)
-        GROUP BY
-            g.gs_id;
+        SELECT gs_id, hom_id_array FROM extsrc.geneset2hom WHERE gs_id = ANY(%(gs_ids)s)
         """, {"gs_ids": list(gs_ids)})
         if cursor.rowcount == 0:
             return {}