adding additional queries based on existing examples

sib-swiss · May 22, 2024 · e6daf73 · e6daf73
1 parent 2534096
commit e6daf73
Show file tree

Hide file tree

Showing 11 changed files with 289 additions and 0 deletions.
diff --git a/oma/03-ins-encoded-proteins.ttl b/oma/03-ins-encoded-proteins.ttl
@@ -0,0 +1,21 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:3
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all proteins in OMA that is encoded by the INS gene and their mnemonics and evidence types from Uniprot database (federated query).""" ;
+    sh:select """SELECT DISTINCT ?proteinOMA ?species  ?mnemonic  ?evidenceType ?UniProt_URI
+WHERE {
+    ?proteinOMA a orth:Protein;
+    orth:organism/obo:RO_0002162/up:scientificName ?species;
+    rdfs:label 'INS'.
+    ?proteinOMA lscr:xrefUniprot ?UniProt_URI.
+    #Search the INS gene mnemonics and evidence types from Uniprot database.
+    service <http://sparql.uniprot.org/sparql> { 
+    ?UniProt_URI up:mnemonic ?mnemonic;
+    up:existence/rdfs:label ?evidenceType. }
+}""" .
+
diff --git a/oma/04-orthologs-of-ensembl-gene.ttl b/oma/04-orthologs-of-ensembl-gene.ttl
@@ -0,0 +1,31 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:4
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all genes that are orthologous to ENSLACG00000002497 Ensembl gene (identifier)""" ;
+    sh:select """select ?protein2 ?OMA_LINK 
+where {
+    #The three that contains Orthologs. The leafs are proteins.
+    #This graph pattern defines the relationship protein1 is Orthologs to protein2
+    ?cluster a orth:OrthologsCluster.
+    ?cluster orth:hasHomologousMember ?node1.
+    ?cluster orth:hasHomologousMember ?node2. 
+    ?node2 orth:hasHomologousMember* ?protein2. 
+    ?node1 orth:hasHomologousMember* ?protein1.
+    ########
+     
+    #Specify the protein to look for its orthologs
+    ?protein1 sio:SIO_010079/lscr:xrefEnsemblGene  ensembl:ENSLACG00000002497.
+    ########
+     
+    #The OMA link to the second protein
+    ?protein2 rdfs:seeAlso ?OMA_LINK. 
+    ########
+     
+    filter(?node1 != ?node2) 
+}""" .
+
diff --git a/oma/05-paralogs-of-ensembl-gene.ttl b/oma/05-paralogs-of-ensembl-gene.ttl
@@ -0,0 +1,31 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:5
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all genes that are paralogous to ENSG00000244734 Ensembl gene (identifier).""" ;
+    sh:select """select ?protein2 ?OMA_LINK 
+where {
+    #The three that contains paralogs. The leafs are proteins.
+    #This graph pattern defines the relationship protein1 is paralogous to protein2
+    ?cluster a orth:ParalogsCluster.
+    ?cluster orth:hasHomologousMember ?node1.
+    ?cluster orth:hasHomologousMember ?node2.
+    ?node2 orth:hasHomologousMember* ?protein2.
+    ?node1 orth:hasHomologousMember* ?protein1.
+    ########
+
+    #Specify the protein to look for its paralogs
+    ?protein1 sio:SIO_010079/lscr:xrefEnsemblGene  ensembl:ENSG00000244734.
+    ########
+
+    #The OMA link to the second protein
+    ?protein2 rdfs:seeAlso ?OMA_LINK. 
+    ########
+     
+    filter(?node1 != ?node2) 
+}""" .
+
diff --git a/oma/06-paralogs-with-uniprot-xrefs.ttl b/oma/06-paralogs-with-uniprot-xrefs.ttl
@@ -0,0 +1,23 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:6
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all genes that are paralogous to HUMAN00529 OMA protein (identifier) and their cross-reference links to OMA and Uniprot.""" ;
+    sh:select """select ?protein2 ?Uniprot_link
+where {
+    ?cluster a orth:ParalogsCluster.
+    ?cluster orth:hasHomologousMember ?node1.
+    ?cluster orth:hasHomologousMember ?node2.
+    ?node2 orth:hasHomologousMember* ?protein2.
+    ?node1 orth:hasHomologousMember* ?protein1.
+    ?protein1 a orth:Protein.
+    ?protein1 dc:identifier 'HUMAN00529'.
+    ?protein2 a orth:Protein. 
+    ?protein2  lscr:xrefUniprot ?Uniprot_link. 
+    filter(?node1 != ?node2)
+}""" .
+
diff --git a/oma/07-orthologs-with-uniprot-xrefs.ttl b/oma/07-orthologs-with-uniprot-xrefs.ttl
@@ -0,0 +1,23 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:7
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all genes that are orthologous to HUMAN22169 OMA protein (identifier) and their cross-reference links to OMA and Uniprot.""" ;
+    sh:select """select ?protein2 ?Uniprot_link
+where {
+    ?cluster a orth:OrthologsCluster.
+    ?cluster orth:hasHomologousMember ?node1.
+    ?cluster orth:hasHomologousMember ?node2.
+    ?node2 orth:hasHomologousMember* ?protein2.
+    ?node1 orth:hasHomologousMember* ?protein1.
+    ?protein1 a orth:Protein.
+    ?protein1 dc:identifier 'HUMAN22169'.
+    ?protein2 a orth:Protein. 
+    ?protein2  lscr:xrefUniprot ?Uniprot_link. 
+    filter(?node1 != ?node2)
+}""" .
+
diff --git a/oma/08-rabbit-apoci-orthologs.ttl b/oma/08-rabbit-apoci-orthologs.ttl
@@ -0,0 +1,26 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:8
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all genes per species that are orthologous to Rabbit's APOCI or APOC1 gene and their cross-reference links to OMA and Uniprot including the corresponding Ensembl gene identifier.""" ;
+    sh:select """select ?protein1    ?protein2  ?geneName2  ?species2 ?Prot2_uniprot ?prot2_ensemblGeneId  
+where {
+    ?cluster a orth:OrthologsCluster.
+    ?cluster orth:hasHomologousMember ?node1.
+    ?cluster orth:hasHomologousMember ?node2. 
+    ?node2 orth:hasHomologousMember* ?protein2. 
+    ?node1 orth:hasHomologousMember* ?protein1.
+    ?protein1 a orth:Protein;
+    orth:organism/obo:RO_0002162/up:scientificName 'Oryctolagus cuniculus';
+    rdfs:label 'APOCI'.
+    ?protein2 a orth:Protein; 
+    lscr:xrefUniprot ?Prot2_uniprot;
+    sio:SIO_010079/lscr:xrefEnsemblGene/dc:identifier  ?prot2_ensemblGeneId;
+    rdfs:label ?geneName2;
+    orth:organism/obo:RO_0002162/up:scientificName ?species2.
+    filter(?node1 != ?node2)
+}""" .
diff --git a/oma/09-rabbit-orthologs-of-mouse-homoglobinY.ttl b/oma/09-rabbit-orthologs-of-mouse-homoglobinY.ttl
@@ -0,0 +1,25 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:9
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all Rabbit's proteins encoded by genes that are orthologous to Mouses's hemoglobin Y gene and their cross-reference links to Uniprot.""" ;
+    sh:select """select distinct ?MOUSE_PROTEIN ?RABIT_PROTEIN ?MOUSE_UNIPROT_XREF ?RABIT_UNIPROT_XREF 
+where {
+    ?cluster a orth:OrthologsCluster.
+    ?cluster orth:hasHomologousMember ?node1.
+    ?cluster orth:hasHomologousMember ?node2. 
+    ?node2 orth:hasHomologousMember* ?RABIT_PROTEIN. 
+    ?node1 orth:hasHomologousMember* ?MOUSE_PROTEIN.
+    ?MOUSE_PROTEIN a orth:Protein.
+    ?MOUSE_PROTEIN  orth:organism/obo:RO_0002162/up:scientificName 'Mus musculus';
+    rdfs:label 'HBB-Y';
+    lscr:xrefUniprot ?MOUSE_UNIPROT_XREF.
+    ?RABIT_PROTEIN a orth:Protein.
+    ?RABIT_PROTEIN orth:organism/obo:RO_0002162/up:scientificName 'Oryctolagus cuniculus' .
+    ?RABIT_PROTEIN lscr:xrefUniprot ?RABIT_UNIPROT_XREF.
+    filter(?node1 != ?node2)
+}""" .
diff --git a/oma/10-paralogs-in-human-of-hbb.ttl b/oma/10-paralogs-in-human-of-hbb.ttl
@@ -0,0 +1,26 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:10
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all human proteins that are paralogous to the HBB gene and their UniProt cross-references.""" ;
+    sh:select """select distinct ?PROTEIN_HBB ?IS_PARALOGOUS_TO_PROTEIN ?PARALOG_GENE_LABEL ?HBB_UNIPROT_XREF ?PARALOG_UNIPROT_XREF 
+where {
+    ?cluster a orth:OrthologsCluster.
+    ?cluster orth:hasHomologousMember ?node1.
+    ?cluster orth:hasHomologousMember ?node2. 
+    ?node2 orth:hasHomologousMember* ?PROTEIN_HBB. 
+    ?node1 orth:hasHomologousMember* ?IS_PARALOGOUS_TO_PROTEIN.
+    ?PROTEIN_HBB a orth:Protein ;
+     orth:organism/obo:RO_0002162/up:scientificName 'Homo sapiens';
+     rdfs:label 'HBB';
+     lscr:xrefUniprot ?MOUSE_UNIPROT_XREF .
+    ?IS_PARALOGOUS_TO_PROTEIN a orth:Protein;
+     orth:organism/obo:RO_0002162/up:scientificName 'Homo sapiens' ;
+     lscr:xrefUniprot ?RABIT_UNIPROT_XREF ;
+     rdfs:label ?PARALOG_GENE_LABEL .
+    filter(?node1 != ?node2)
+}""" .
diff --git a/oma/11-percentage-of-proteins-with-paralogs.ttl b/oma/11-percentage-of-proteins-with-paralogs.ttl
@@ -0,0 +1,36 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:11
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """The percentage of proteins in Drosophila melanogaster that has at least one paralogous sequence (protein).""" ;
+    sh:select """select (((xsd:float(?num_paralogy)*100)/xsd:float(?total)) as ?result) 
+  where   { 
+    { 
+       select (count(distinct ?PROTEIN) as ?num_paralogy ) 
+           where { 
+              ?cluster a orth:ParalogsCluster. 
+              ?cluster orth:hasHomologousMember ?node1. 
+              ?cluster orth:hasHomologousMember ?node2.  
+              ?node2 orth:hasHomologousMember* ?PROTEIN.  
+              ?node1 orth:hasHomologousMember* ?IS_PARALOGOUS_TO_PROTEIN. 
+              ?PROTEIN a orth:Protein. 
+              ?PROTEIN orth:organism/obo:RO_0002162/up:scientificName ?species. 
+              ?IS_PARALOGOUS_TO_PROTEIN a orth:Protein. 
+              ?IS_PARALOGOUS_TO_PROTEIN orth:organism/obo:RO_0002162/up:scientificName ?species . 
+              values(?species ){( 'Drosophila melanogaster' )} 
+              filter(?node1 != ?node2) 
+           }
+    } 
+    { 
+        select (count(distinct ?protein_total) as ?total)
+            where {
+               ?protein_total a orth:Protein .
+               ?protein_total orth:organism/obo:RO_0002162/up:scientificName ?species .
+               values(?species ){( 'Drosophila melanogaster' )} 
+            } 
+    }
+}""" .
diff --git a/oma/12-orthologs-between-two-species.ttl b/oma/12-orthologs-between-two-species.ttl
@@ -0,0 +1,22 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:12
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all orthologs between mouse and rabbit, together with their HOG id""" ;
+    sh:select """select distinct ?MOUSE_PROTEIN ?RABIT_PROTEIN ?HOG
+where {
+?HOG a orth:OrthologsCluster ;
+ orth:hasHomologousMember ?node1 ;
+ orth:hasHomologousMember ?node2 . 
+?node2 orth:hasHomologousMember* ?RABIT_PROTEIN. 
+?node1 orth:hasHomologousMember* ?MOUSE_PROTEIN.
+?MOUSE_PROTEIN a orth:Protein ;
+ orth:organism/obo:RO_0002162/up:scientificName 'Mus musculus'.
+?RABIT_PROTEIN a orth:Protein ;
+ orth:organism/obo:RO_0002162/up:scientificName 'Oryctolagus cuniculus' .  
+filter(?node1 != ?node2)
+}""" .
diff --git a/oma/13-hog-members-at-level-from-query-protein.ttl b/oma/13-hog-members-at-level-from-query-protein.ttl
@@ -0,0 +1,25 @@
+prefix ex: <https://sparql.omabrowser.org/.well-known/sparql-examples/>
+prefix sh: <http://www.w3.org/ns/shacl#>
+prefix rdfs:<http://www.w3.org/2000/01/rdf-schema#>
+prefix schema:<https://schema.org/>
+ex:13
+    a sh:SPARQLSelectExecutable, sh:SPARQLExecutable ;
+    sh:prefixes _:sparql_examples_prefixes ;
+    schema:target <https://sparql.omabrowser.org/sparql/> ;
+    rdfs:comment """Retrieve all proteins belongong to the Hierarchical Orthologous Group (HOG) at the level 'Vertebrata' to which humans' CDIN1 gene belong, together with their gene name symbol if available.""" ;
+    sh:select """select distinct ?HOG ?MEMBER ?GENE_LABEL
+where {
+    ?HOG a orth:OrthologsCluster ;
+      orth:hasHomologousMember ?node1 ;
+      orth:hasTaxonomicRange ?taxRange .
+    ?taxRange orth:taxRange 'Vertebrata' .
+    ?node1 orth:hasHomologousMember* ?query ;
+      orth:hasHomologousMember* ?MEMBER .
+    ?MEMBER a orth:Protein .
+    OPTIONAL {
+        ?MEMBER rdfs:label ?GENE_LABEL .
+    }
+    ?query a orth:Protein ;
+      orth:organism/obo:RO_0002162/up:scientificName 'Homo sapiens';
+      rdfs:label 'CDIN1'.
+}""" .