add ROBOKOP

00_fetch_data/robokop/Makefile

@@ -0,0 +1,96 @@
+
+all: binding_nodes.jsonl.gz binding_edges.jsonl.gz cam_nodes.jsonl.gz cam_edges.jsonl.gz ctd_nodes.jsonl.gz ctd_edges.jsonl.gz drugcentral_nodes.jsonl.gz drugcentral_edges.jsonl.gz alliance_nodes.jsonl.gz alliance_edges.jsonl.gz gtex_nodes.jsonl.gz gtex_edges.jsonl.gz gtopdb_nodes.jsonl.gz gtopdb_edges.jsonl.gz hetionet_nodes.jsonl.gz hetionet_edges.jsonl.gz hgnc_nodes.jsonl.gz hgnc_edges.jsonl.gz hmdb_nodes.jsonl.gz hmdb_edges.jsonl.gz hgoa_nodes.jsonl.gz hgoa_edges.jsonl.gz icees_nodes.jsonl.gz icees_edges.jsonl.gz intact_nodes.jsonl.gz intact_edges.jsonl.gz panther_nodes.jsonl.gz panther_edges.jsonl.gz pharos_nodes.jsonl.gz pharos_edges.jsonl.gz string_nodes.jsonl.gz string_edges.jsonl.gz textmining_nodes.jsonl.gz textmining_edges.jsonl.gz viralproteome_nodes.jsonl.gz viralproteome_edges.jsonl.gz
+
+binding_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/BINDING_Automat/9164067231dc9b8c/nodes.jsonl.gz > $@
+binding_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/BINDING_Automat/9164067231dc9b8c/edges.jsonl.gz > $@
+
+cam_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/CAMKP_Automat/e92bd7b217535f2d/nodes.jsonl.gz > $@
+cam_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/CAMKP_Automat/e92bd7b217535f2d/edges.jsonl.gz > $@
+
+ctd_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/CTD_Automat/f92c663160ec5e36/nodes.jsonl.gz > $@
+ctd_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/CTD_Automat/f92c663160ec5e36/edges.jsonl.gz  > $@
+
+drugcentral_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/DrugCentral_Automat/dec0617490b49c7a/nodes.jsonl.gz > $@
+drugcentral_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/DrugCentral_Automat/dec0617490b49c7a/edges.jsonl.gz > $@
+
+alliance_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/GenomeAllianceOrthologs_Automat/06e2a3c1eeea5251/nodes.jsonl.gz > $@
+alliance_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/GenomeAllianceOrthologs_Automat/06e2a3c1eeea5251/edges.jsonl.gz > $@
+
+gtex_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/GTEx_Automat/a6448b9092bb81a1/nodes.jsonl.gz > $@
+gtex_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/GTEx_Automat/a6448b9092bb81a1/edges.jsonl.gz > $@
+
+gtopdb_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/GtoPdb_Automat/0ea6074c824c2236/nodes.jsonl.gz > $@
+gtopdb_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/GtoPdb_Automat/0ea6074c824c2236/edges.jsonl.gz  > $@
+
+hetionet_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/Hetio_Automat/85a5f53e63150e1e/nodes.jsonl.gz > $@
+hetionet_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/Hetio_Automat/85a5f53e63150e1e/edges.jsonl.gz > $@
+
+hgnc_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/HGNC_Automat/dee31cfce74e5944/nodes.jsonl.gz > $@
+hgnc_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/HGNC_Automat/dee31cfce74e5944/edges.jsonl.gz > $@
+
+hmdb_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/HMDB_Automat/6715124699b6dbf0/nodes.jsonl.gz > $@
+hmdb_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/HMDB_Automat/6715124699b6dbf0/edges.jsonl.gz > $@
+
+hgoa_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/HumanGOA_Automat/06f107a4e9e8e547/nodes.jsonl.gz > $@
+hgoa_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/HumanGOA_Automat/06f107a4e9e8e547/edges.jsonl.gz > $@
+
+icees_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/icees-kg/1.5.0/nodes.jsonl.gz > $@
+icees_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/icees-kg/1.5.0/edges.jsonl.gz > $@
+
+intact_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/IntAct_Automat/e5b936f966a02c2c/nodes.jsonl.gz > $@
+intact_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/IntAct_Automat/e5b936f966a02c2c/edges.jsonl.gz > $@
+
+panther_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/PANTHER_Automat/c0189f14ba41da6c/nodes.jsonl.gz > $@
+panther_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/PANTHER_Automat/c0189f14ba41da6c/edges.jsonl.gz > $@
+
+pharos_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/PHAROS_Automat/d3068b509bf17ff3/nodes.jsonl.gz > $@
+pharos_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/PHAROS_Automat/d3068b509bf17ff3/edges.jsonl.gz > $@
+
+string_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/STRING-DB_Automat/4ca5a0ce557e2c18/nodes.jsonl.gz > $@
+string_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/STRING-DB_Automat/4ca5a0ce557e2c18/edges.jsonl.gz > $@
+
+textmining_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/TMKP_Automat/2ae17047aa1b5f3c/nodes.jsonl.gz > $@
+textmining_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/TMKP_Automat/2ae17047aa1b5f3c/edges.jsonl.gz > $@
+
+viralproteome_nodes.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/ViralProteome_Automat/8ca116ae0bba6945/nodes.jsonl.gz > $@
+viralproteome_edges.jsonl.gz:
+	curl -L https://stars.renci.org/var/plater/bl-4.2.1/ViralProteome_Automat/8ca116ae0bba6945/edges.jsonl.gz > $@
+
+.PHONY: all
+
+

README.md

@@ -10,10 +10,11 @@ EBI Codon HPC pipeline for building integrated knowledge graphs from [EMBL-EBI r
 * [Metabolights](https://www.ebi.ac.uk/metabolights)
 * [ChEMBL](https://www.ebi.ac.uk/chembl/)
 
-GrEBI also imports complementary datasets, so far:
+In addition to loading from EBI databases GrEBI also imports other knowledge graphs, so far:
 
 * [MONARCH Initiative KG](https://monarch-initiative.github.io/monarch-ingest/Sources/)
 * [Ubergraph](https://github.com/INCATools/ubergraph)
+* [ROBOKOP](https://robokop.renci.org/)
 * [Human Reference Atlas KG](https://humanatlas.io/)
 * [AOPWiki](https://aopwiki.org/) (via [AOPWikiRDF](https://github.com/marvinm2/AOPWikiRDF))
 

configs/datasource_configs/robokop_alliance.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.Alliance",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/alliance_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/alliance_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_binding.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.BINDING",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/binding_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/binding_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_cam.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.CAM",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/cam_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/cam_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_ctd.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.CTD",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/ctd_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/ctd_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_drugcentral.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.DrugCentral",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/drugcentral_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/drugcentral_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_gtex.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.GTEx",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/gtex_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/gtex_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_gtopdb.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.GtoPdb",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/gtopdb_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/gtopdb_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_hetionet.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.Hetionet",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/hetionet_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/hetionet_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_hgoa.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.HumanGOA",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/hgoa_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/hgoa_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_hmdb.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.HMDB",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/hmdb_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/hmdb_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_icees.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.ICEES",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/icees_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/icees_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}

configs/datasource_configs/robokop_intact.json

@@ -0,0 +1,21 @@
+{
+    "name": "Robokop.IntAct",
+    "enabled": true,
+    "ingests": [
+        {
+            "ingest_files": ["./00_fetch_data/robokop/intact_edges.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_kgx_edges",
+            "ingest_args": [
+                { "name": "--kgx-inject-key-prefix", "value": "robokop:" }
+            ]
+        },
+        {
+            "ingest_files": ["./00_fetch_data/robokop/intact_nodes.jsonl.gz"],
+            "ingest_script": "./target/release/grebi_ingest_json",
+            "ingest_args": [
+                { "name": "--json-rename-field", "value": "category:grebi:type" },
+                { "name": "--json-inject-key-prefix", "value": "robokop:" }
+             ]
+        }
+    ]
+}