split up neo4j dataload queries

EBISPOT · Nov 7, 2024 · a998c7e · a998c7e
1 parent d948bf9
commit a998c7e
Show file tree

Hide file tree

Showing 6 changed files with 27 additions and 25 deletions.
diff --git a/dataload/07_create_db/neo4j/create_indexes.cypher b/dataload/07_create_db/neo4j/create_indexes.cypher
diff --git a/dataload/07_create_db/neo4j/cypher/create_indexes.cypher b/dataload/07_create_db/neo4j/cypher/create_indexes.cypher
@@ -0,0 +1,11 @@
+
+CREATE INDEX node_id FOR (n:GraphNode) ON n.`grebi:nodeId`
+;
+CREATE INDEX subgraph FOR (n:GraphNode) ON n.`grebi:subgraph`
+;
+CREATE INDEX id_id FOR (n:Id) ON n.`id`
+;
+CREATE INDEX ic FOR (n:GraphNode) ON (n.ic)
+;
+CALL db.awaitIndexes(10800)
+;
diff --git a/dataload/07_create_db/neo4j/cypher/ic_scores_1.cypher b/dataload/07_create_db/neo4j/cypher/ic_scores_1.cypher
@@ -0,0 +1,4 @@
+MATCH (ancestor:`ols:Class`)<-[:`biolink:broad_match`*1..]-(subclass:`ols:Class`)
+WITH ancestor, count(DISTINCT subclass) AS num_desc
+SET ancestor.num_desc = num_desc
+;
diff --git a/dataload/07_create_db/neo4j/cypher/ic_scores_2.cypher b/dataload/07_create_db/neo4j/cypher/ic_scores_2.cypher
@@ -0,0 +1,4 @@
+WITH COLLECT { MATCH (cl:`ols:Class`) RETURN max(cl.num_desc) }[0] AS max_num_desc
+MATCH (cl2:`ols:Class`)
+SET cl2.ic = 1.0 - (cl2.num_desc/max_num_desc)
+;
diff --git a/dataload/07_create_db/neo4j/neo4j_import.dockersh b/dataload/07_create_db/neo4j/neo4j_import.dockersh
@@ -35,10 +35,14 @@ sleep 20
 
 echo Creating neo4j indexes...
 
-cypher-shell -a neo4j://127.0.0.1:7687 --non-interactive -f /create_indexes.cypher
+cypher-shell -a neo4j://127.0.0.1:7687 --non-interactive -f /cypher/ic_scores_1.cypher
+sleep 20
+cypher-shell -a neo4j://127.0.0.1:7687 --non-interactive -f /cypher/ic_scores_2.cypher
+sleep 20
+cypher-shell -a neo4j://127.0.0.1:7687 --non-interactive -f /cypher/create_indexes.cypher
 
 echo Creating neo4j indexes done
 
-sleep 5
+sleep 20
 neo4j stop
 
diff --git a/dataload/07_create_db/neo4j/neo4j_import.slurm.py b/dataload/07_create_db/neo4j/neo4j_import.slurm.py
@@ -35,7 +35,7 @@ def main():
             '--bind ' + shlex.quote(neo_data_path) + ':/data',
             '--bind ' + shlex.quote(neo_logs_path) + ':/logs',
             '--bind ' + os.path.abspath(os.path.join(os.environ['GREBI_HOME'], '07_create_db/neo4j/neo4j_import.dockersh')) + ':/import.sh',
-            '--bind ' + os.path.abspath(os.path.join(os.environ['GREBI_HOME'], '07_create_db/neo4j/create_indexes.cypher')) + ':/create_indexes.cypher',
+            '--bind ' + os.path.abspath(os.path.join(os.environ['GREBI_HOME'], '07_create_db/neo4j/cypher')) + ':/cypher',
             '--writable-tmpfs',
             '--network=none',
             '--env NEO4J_AUTH=none',
@@ -50,7 +50,7 @@ def main():
             '-v ' + shlex.quote(neo_data_path) + ':/data',
             '-v ' + shlex.quote(neo_logs_path) + ':/logs',
             '-v ' + os.path.abspath(os.path.join(os.environ['GREBI_HOME'], '07_create_db/neo4j/neo4j_import.dockersh')) + ':/import.sh',
-            '-v ' + os.path.abspath(os.path.join(os.environ['GREBI_HOME'], '07_create_db/neo4j/create_indexes.cypher')) + ':/create_indexes.cypher',
+            '-v ' + os.path.abspath(os.path.join(os.environ['GREBI_HOME'], '07_create_db/neo4j/cypher')) + ':/cypher',
             '-e NEO4J_AUTH=none',
             'neo4j:5.18.0',
             'bash /import.sh'