Country trouble shooting

4 files changed, 34 insertions, 18 deletions

diff --git a/doc/INSTALL.md b/doc/INSTALL.md
index 45aca0f..367b452 100644
--- a/doc/INSTALL.md
+++ b/doc/INSTALL.md
@@ -77,6 +77,7 @@ Note: see above on GUIX_PACKAGE_PATH.
 ## Run country semantic enrichment script
 
     cd bh20-seq-resource/scripts/db_enrichment
+    edit input_location.csv
     guix environment guix --ad-hoc git python nss-certs python-rdflib -- python3 country_enrichment.py
 
 ## Run the tests
diff --git a/scripts/db_enrichment/country_enrichment.py b/scripts/db_enrichment/country_enrichment.py
index 1f99d42..f62a64e 100644
--- a/scripts/db_enrichment/country_enrichment.py
+++ b/scripts/db_enrichment/country_enrichment.py
@@ -39,14 +39,36 @@ def callSPARQL(query):
 
 g = Graph()
 
+test_query="""
+# Use with https://query.wikidata.org/
+SELECT DISTINCT ?a ?label ?country ?continent ?coor WHERE {
+    BIND (XXX as ?a) .
+    OPTIONAL {
+        ?a wdt:P625 ?coor.
+    }
+    ?a rdfs:label ?label .
+    ?a wdt:P17 ?country.
+    ?country rdfs:label ?country_label .
+    OPTIONAL {
+        ?country wdt:P30 ?continent.
+        ?continent rdfs:label ?continent_label
+        FILTER (lang(?continent_label)='en')
+    }
+    FILTER (lang(?country_label)='en')
+    FILTER (lang(?label)='en')
+}
+"""
+
+# wdt:P625 are GEO coordinates
+
 query = """
 construct {
     ?a wdt:P625 ?c.
     ?a rdfs:label ?label .
     ?a wdt:P17 ?country.
     ?country rdfs:label ?country_label .
-    ?country wdt:P30 ?continent.
-    ?continent rdfs:label ?continent_label
+    ?country wdt:P30 ?continent .
+    ?continent rdfs:label ?continent_label .
 } WHERE
 {
     BIND (XXX as ?a) .
@@ -59,7 +81,6 @@ construct {
     FILTER (lang(?continent_label)='en')
     FILTER (lang(?country_label)='en')
     FILTER (lang(?label)='en')
-
 }
 """""
 
@@ -72,6 +93,8 @@ with open(outputFile, 'r') as csvfile:
         counter=counter+1
 
         try:
+            testq = test_query.replace("XXX", "<"+row[0]+">")
+            print(testq)
             tmpquery=query.replace("XXX", "<"+row[0]+">")
             print(tmpquery)
 
diff --git a/scripts/db_enrichment/input_location.csv b/scripts/db_enrichment/input_location.csv
index a4246cd..8c3308f 100644
--- a/scripts/db_enrichment/input_location.csv
+++ b/scripts/db_enrichment/input_location.csv
@@ -1,16 +1,6 @@
 http://www.wikidata.org/entity/Q7960498
 http://www.wikidata.org/entity/Q692895
-http://www.wikidata.org/entity/Q928
 http://www.wikidata.org/entity/Q2722074
 http://www.wikidata.org/entity/Q25622187
 http://www.wikidata.org/entity/Q27684996
 http://www.wikidata.org/entity/Q2757125
-http://www.wikidata.org/entity/Q1922283
-http://www.wikidata.org/entity/Q490
-http://www.wikidata.org/entity/Q677037
-http://www.wikidata.org/entity/Q3037
-http://www.wikidata.org/entity/Q843
-http://www.wikidata.org/entity/Q183
-http://www.wikidata.org/entity/Q29
-http://www.wikidata.org/entity/Q17
-http://www.wikidata.org/entity/Q810
diff --git a/scripts/db_enrichment/readme.md b/scripts/db_enrichment/readme.md
index 88e8be5..7539104 100644
--- a/scripts/db_enrichment/readme.md
+++ b/scripts/db_enrichment/readme.md
@@ -11,11 +11,13 @@ File containing information about the countries in our database. Additional info
 This SPARQL query (http://sparql.genenetwork.org/sparql/) retrieves all countries (ids) from our database that do not have a label yet:
 
 
->SELECT DISTINCT ?geoLocation  WHERE
->{
->?fasta ?x [ <<http://purl.obolibrary.org/obo/GAZ_00000448>> ?geoLocation] .
->FILTER NOT EXISTS {?geoLocation <<http://www.w3.org/2000/01/rdf-schema#label>> ?geoLocation_tmp_label}
->}
+```sparql
+SELECT DISTINCT ?geoLocation  WHERE
+{
+    ?fasta ?x [ <http://purl.obolibrary.org/obo/GAZ_00000448> ?geoLocation] .
+    FILTER NOT EXISTS {?geoLocation <http://www.w3.org/2000/01/rdf-schema#label> ?geoLocation_tmp_label}
+}
+```
 
 [Run query](http://sparql.genenetwork.org/sparql/?default-graph-uri=&query=%0D%0ASELECT+DISTINCT+%3FgeoLocation++WHERE%0D%0A%7B%0D%0A++%3Ffasta+%3Fx+%5B+%3Chttp%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FGAZ_00000448%3E+%3FgeoLocation%5D+.%0D%0A++FILTER+NOT+EXISTS+%7B%3FgeoLocation+%3Chttp%3A%2F%2Fwww.w3.org%2F2000%2F01%2Frdf-schema%23label%3E+%3FgeoLocation_tmp_label%7D%0D%0A%7D&format=text%2Fhtml&timeout=0&debug=on&run=+Run+Query+)