From ad3f1176642a499229554d490037b533ef7da418 Mon Sep 17 00:00:00 2001
From: Pjotr Prins
Date: Wed, 23 Oct 2024 20:46:23 -0500
Subject: wikidata: build first query

---
 gn/db/sources/wikidata.scm | 29 +++++++++++++++++++++++++++++
 gn/db/sparql.scm           |  3 ++-
 2 files changed, 31 insertions(+), 1 deletion(-)
 create mode 100644 gn/db/sources/wikidata.scm

(limited to 'gn')

diff --git a/gn/db/sources/wikidata.scm b/gn/db/sources/wikidata.scm
new file mode 100644
index 0000000..7397426
--- /dev/null
+++ b/gn/db/sources/wikidata.scm
@@ -0,0 +1,29 @@
+#!
+
+Wikidata queries
+
+!#
+
+(define-module (gn db sources wikidata)
+)
+
+(define ps-encoded-by "ps:P702")
+(define wdt-instance-of "wdt:P31")
+(define wdt-in-taxon "wdt:P703")
+(define wd-human "wd:Q15978631")
+(define wd-mouse "wd:Q83310")
+(define wd-rat "wd:Q184224")
+(define wd-gene "wd:Q7187")
+
+(define (wikidata_query_geneids gene_name)
+  "Return the wikidata identifiers pointing to genes of listed species"
+  (string-append
+     "SELECT DISTINCT ?wikidata_id
+            WHERE {
+              ?wikidata_id " wdt-instance-of " " wd-gene ";
+                           " wdt-in-taxon " ?species .
+              VALUES (?species) { (" wd-human " ) ( " wd-mouse" ) ( " wd-rat" ) } .
+              ?wikidata_id rdfs:label \"" gene_name "\"@en .
+        }
+"
+              ))
diff --git a/gn/db/sparql.scm b/gn/db/sparql.scm
index b7d94f3..f03389b 100644
--- a/gn/db/sparql.scm
+++ b/gn/db/sparql.scm
@@ -2,7 +2,7 @@
 
 Module for handling SPARQL primitives.
 
-Note that GN queries should go into gn/data - this is currently not
+Note that GN queries should go into gn/db/sources - this is currently not
 the case.
 
 !#
@@ -18,6 +18,7 @@ the case.
   #:use-module (web request)
   #:use-module (web uri)
   #:use-module (gn cache memoize)
+  #:use-module (gn db sources wikidata)
   #:use-module (web gn-uri)
 
   #:export (memo-sparql-species
-- 
cgit v1.2.3