From bffde88061d5df100a0925e5d74cc3a462524766 Mon Sep 17 00:00:00 2001
From: Kevin Schaper <kevinschaper@gmail.com>
Date: Sun, 9 Jun 2024 09:44:06 -0700
Subject: [PATCH] update ingest config & expect yamls

---
 src/monarch_ingest/ingests.yaml   | 45 ++++++++++++++++---------------
 src/monarch_ingest/qc_expect.yaml | 22 +++++++++++----
 2 files changed, 41 insertions(+), 26 deletions(-)

diff --git a/src/monarch_ingest/ingests.yaml b/src/monarch_ingest/ingests.yaml
index 906e8694..44d5b642 100644
--- a/src/monarch_ingest/ingests.yaml
+++ b/src/monarch_ingest/ingests.yaml
@@ -1,3 +1,27 @@
+## Pass-through modular ingests
+alliance_genotype:
+  url:
+    - 'https://github.com/monarch-initiative/alliance-genotype-ingest/releases/latest/download/alliance_genotype_nodes.tsv'
+alliance_phenotype:
+  url:
+    - 'https://github.com/monarch-initiative/alliance-phenotype-association-ingest/releases/latest/download/alliance_phenotype_edges.tsv'
+alliance_disease_association:
+  url:
+    - 'https://github.com/monarch-initiative/alliance-disease-association-ingest/releases/latest/download/alliance_disease_edges.tsv'
+zfin_genotype_to_phenotype:
+  url:
+    - 'https://github.com/monarch-initiative/zfin-genotype-to-phenotype-ingest/releases/latest/download/zfin_genotype_to_phenotype_edges.tsv'
+clingen_variant:
+  url:
+    - 'https://github.com/monarch-initiative/clingen-ingest/releases/latest/download/clingen_variant_nodes.tsv'
+    - 'https://github.com/monarch-initiative/clingen-ingest/releases/latest/download/clingen_variant_edges.tsv'
+clinvar_variant:
+  url:
+    - 'https://github.com/monarch-initiative/clinvar-ingest/releases/latest/download/clinvar_variant_nodes.tsv'
+    - 'https://github.com/monarch-initiative/clinvar-ingest/releases/latest/download/clinvar_variant_edges.tsv'
+
+## Ingests within this repository
+
 alliance_gene:
   config: 'ingests/alliance/gene.yaml'
 alliance_gene_to_expression:
@@ -63,24 +87,3 @@ zfin_gene_to_phenotype:
 # zfin_publication_to_gene:
 #   config: 'ingests/zfin/publication_to_gene.yaml'
 
-## Pass-through modular ingests
-alliance_genotype:
-  url:
-    - 'https://github.com/monarch-initiative/alliance-genotype-ingest/releases/latest/download/alliance_genotype_nodes.tsv'
-alliance_phenotype:
-  url:
-    - 'https://github.com/monarch-initiative/alliance-phenotype-association-ingest/releases/latest/download/alliance_phenotype_edges.tsv'
-alliance_disease_association:
-  url:
-    - 'https://github.com/monarch-initiative/alliance-disease-association-ingest/releases/latest/download/alliance_disease_edges.tsv'
-zfin_genotype_to_phenotype:
-  url:
-    - 'https://github.com/monarch-initiative/zfin-genotype-to-phenotype-ingest/releases/latest/download/zfin_genotype_to_phenotype_edges.tsv'
-clingen_variant:
-  url:
-    - 'https://github.com/monarch-initiative/clingen-ingest/releases/latest/download/clingen_variant_nodes.tsv'
-    - 'https://github.com/monarch-initiative/clingen-ingest/releases/latest/download/clingen_variant_edges.tsv'
-clinvar_variant:
-  url:
-    - 'https://github.com/monarch-initiative/clinvar-ingest/releases/latest/download/clinvar_variant_nodes.tsv'
-    - 'https://github.com/monarch-initiative/clinvar-ingest/releases/latest/download/clinvar_variant_edges.tsv'
diff --git a/src/monarch_ingest/qc_expect.yaml b/src/monarch_ingest/qc_expect.yaml
index c28daf97..3d840a54 100644
--- a/src/monarch_ingest/qc_expect.yaml
+++ b/src/monarch_ingest/qc_expect.yaml
@@ -2,6 +2,8 @@ nodes:
   provided_by:
     alliance_gene_nodes:
       min: 290000
+    alliance_genotype_nodes:
+      min: 130000
     dictybase_gene_nodes:
       min: 14000
     hgnc_gene_nodes:
@@ -14,12 +16,14 @@ nodes:
       min: 5000
     reactome_pathway_nodes:
       min: 21000
+    clinvar_variant_nodes:
+      min: 1280000
+    clingen_variant_nodes:
+      min: 7000
 edges:
   provided_by:
     alliance_gene_to_expression_edges:
       min: 1870000
-    alliance_gene_to_phenotype_edges:
-      min: 300000
     bgee_gene_to_expression_edges:
       min: 430000
     biogrid_edges:
@@ -49,8 +53,16 @@ edges:
     reactome_gene_to_pathway_edges:
       min: 200000
     string_protein_links_edges:
-      min: 1490000
+      min: 1470000
     xenbase_gene_to_phenotype_edges:
       min: 2000
-    zfin_gene_to_phenotype_edges:
-      min: 148000
+    alliance_phenotype_edges:
+      min: 650000
+    alliance_disease_edges:
+      min: 10000
+    zfin_genotype_to_phenotype_edges:
+      min: 125000
+    clinvar_variant_edges:
+      min: 1400000
+    clingen_variant_edges:
+      min: 5000