LifeSciencesQueries/SPARQL

From W3C Wiki

SPARQL queries

Transations of queries from http://www.isb-sib.ch/~ejain/expasy4j-webng/query.html#examples into SPARQL. 


# ==== Ex 1 : Names of proteins with a gene with a specific name or synonym
# 
# SELECT 
#   ?protein, ?name 
# WHERE 
#   [?protein rdf:type :Protein] AND 
#   [?protein :name ?name] AND 
#   [?protein :encodedBy ?gene] AND 
#   [?gene :name "CRB"] 
# USING 
#   "urn:lsid:uniprot.org:ontology:",
#   "http://www.w3.org/1999/02/22-rdf-syntax-ns#" AS rdf

PREFIX  rdf:   <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX  :      <urn:lsid:uniprot.org:ontology:>
SELECT  ?protein ?name
WHERE
  { ?protein  rdf:type    :Protein ;
              :name       ?name ;
              :encodedBy  ?gene .
    ?gene     :name       "CRB" .
  }



# ==== Ex 2 :Ranges of transmembrane regions
# 
# SELECT 
#   ?protein, ?begin, ?end 
# WHERE 
#   [?protein rdf:type :Protein] AND 
#   [?protein :annotation ?annotation] AND
#   [?annotation rdf:type :Transmembrane_Annotation] AND 
#   [?annotation  :range ?range] AND 
#   [?range :begin ?begin] AND  -- Note: Should use MAYBE...
#   [?range :end ?end] 
# USING 
#   "urn:lsid:uniprot.org:ontology:",
#   "http://www.w3.org/1999/02/22-rdf-syntax-ns#" AS rdf

PREFIX  rdf:   <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX  :      <urn:lsid:uniprot.org:ontology:>
SELECT  ?protein ?begin ?end
WHERE
  { ?protein  rdf:type     :Protein ;
              :annotation  ?annotation .
    ?annotation  rdf:type  :Transmembrane_Annotation .

    # As given
    ?annotation  :range ?range .
    ?range :begin ?begin .
    ?range :end ?end .
    # Optional
##     OPTIONAL
##       { ?annotation  :range  ?range .
##         ?range    :begin  ?begin ;
##                   :end    ?end .
##       }
  }



# ==== Ex 3 : Proteins with publications by authors with matching names
# 
# SELECT 
#   ?protein, ?author, ?title
# WHERE 
#   [?protein rdf:type :Protein] AND 
#   [?protein :modified ?modified] AND 
#   [?protein :citation ?citation] AND 
#   [?citation :author ?author] AND 
#   ?author ~ "bairoch %" AND 
#   [?citation :title ?title] 
# USING 
#   "urn:lsid:uniprot.org:ontology:",
#   "http://www.w3.org/1999/02/22-rdf-syntax-ns#" AS rdf

PREFIX :     <urn:lsid:uniprot.org:ontology:>
PREFIX rdf:  <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

SELECT 
  ?protein ?author ?title
{
  ?protein rdf:type :Protein .
  ?protein :modified ?modified .
  ?protein :citation ?citation .
  ?citation :author ?author .
  FILTER ( regex(?author, "^bairoch ", "i") ) .
  ?citation :title ?title
}



# ==== Ex 4 :Number of times a publication by a specific author is cited
# 
# COUNT 
#   ?protein
# WHERE 
#   [?protein rdf:type :Protein] AND 
#   [?protein :citation ?citation] AND 
#   [?citation :author "Bairoch A."] 
# USING 
#   "urn:lsid:uniprot.org:ontology:",
#   "http://www.w3.org/1999/02/22-rdf-syntax-ns#" AS rdf

# No COUNT in SPARQL

PREFIX :     <urn:lsid:uniprot.org:ontology:>
PREFIX rdf:  <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

SELECT ?protein 
WHERE 
{
   ?protein rdf:type :Protein ;
           :citation [ :author "Bairoch A." ] .
}



# ==== Ex 5 :Resources that are related to proteins annotated with a specific keyword
# 
# SELECT 
#   ?related
# WHERE 
#   [?protein rdf:type :Protein] AND 
#   [?protein :classifiedWith <urn:lsid:uniprot.org:keywords:48>] AND 
#   [?protein rdfs:seeAlso ?related] 
# USING 
#   "urn:lsid:uniprot.org:ontology:",
#   "http://www.w3.org/1999/02/22-rdf-syntax-ns#" AS rdf, 
#   "http://www.w3.org/2000/01/rdf-schema#" AS rdfs

PREFIX  rdfs:  <http://www.w3.org/2000/01/rdf-schema#>
PREFIX  rdf:   <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX  :      <urn:lsid:uniprot.org:ontology:>
SELECT  ?related
WHERE
  { ?protein  rdf:type         :Protein ;
              :classifiedWith  <urn:lsid:uniprot.org:keywords:48> ;
              rdfs:seeAlso     ?related .
  }



# ==== Ex 6 : Genes associated with human diseases
# 
# SELECT 
#   ?gene AS "Gene", ?name AS "Name", ?text AS "Disease"
# WHERE 
#   [?protein rdf:type :Protein] AND
#   [?protein :gene ?gene] AND
#   [?gene :name ?name] AND 
#   [?protein :organism taxon:9606] AND 
#   [?protein :annotation ?annotation] AND 
#   [?annotation rdf:type :Disease_Annotation] AND 
#   [?annotation rdfs:comment ?text] 
# USING 
#   "urn:lsid:uniprot.org:ontology:",
#   "http://www.w3.org/1999/02/22-rdf-syntax-ns#" AS rdf, 
#   "http://www.w3.org/2000/01/rdf-schema#" AS rdfs,
#   "urn:lsid:uniprot.org:taxonomy:" AS taxon

PREFIX :      <urn:lsid:uniprot.org:ontology:>
PREFIX rdf:   <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs:  <http://www.w3.org/2000/01/rdf-schema#>
PREFIX taxon: <urn:lsid:uniprot.org:taxonomy:>

# No variable aliases
SELECT ?gene ?name ?text
{
  ?protein rdf:type :Protein .
  ?protein :gene ?gene .
  ?gene :name ?name .
  ?protein :organism <urn:lsid:uniprot.org:taxonomy:9606> .
  ?protein :annotation ?annotation .
  ?annotation rdf:type :Disease_Annotation .
  ?annotation rdfs:comment ?text .
}



# ==== Ex 7 : Sequences of bacterial proteins
# 
# SELECT 
#   ?protein, ?aa
# WHERE 
#   [?protein rdf:type :Protein] AND 
#   [?protein :organism taxon:2] AND 
#   [?protein :sequence ?s] AND 
#   [?s rdf:value ?aa]
# USING 
#   "urn:lsid:uniprot.org:ontology:",
#   "http://www.w3.org/1999/02/22-rdf-syntax-ns#" AS rdf, 
#   "http://www.w3.org/2000/01/rdf-schema#" AS rdfs,
#   "urn:lsid:uniprot.org:taxonomy:" AS taxon

PREFIX :      <urn:lsid:uniprot.org:ontology:>
PREFIX rdf:   <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs:  <http://www.w3.org/2000/01/rdf-schema#>
PREFIX taxon: <urn:lsid:uniprot.org:taxonomy:>

SELECT  ?protein ?aa
{
  ?protein rdf:type :Protein .
  ?protein :organism <urn:lsid:uniprot.org:taxonomy:2> .
  ?protein :sequence ?s .
  ?s rdf:value ?aa
}



# ==== Ex 8 : Recently modified entries
# 
# SELECT 
#   ?protein, ?date
# WHERE 
#   [?protein rdf:type :Protein] AND 
#   [?protein :modified ?date] AND 
#   ?date > "2004-08"
# USING 
#   "urn:lsid:uniprot.org:ontology:",
#   "http://www.w3.org/1999/02/22-rdf-syntax-ns#" AS rdf

PREFIX  rdfs:  <http://www.w3.org/2000/01/rdf-schema#>
PREFIX  rdf:   <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX  :      <urn:lsid:uniprot.org:ontology:>
SELECT  ?protein ?date
WHERE
  { ?protein  rdf:type   :Protein ;
              :modified  ?date .
    # Not a date!
    FILTER ( ?date > "2004-08" )
  }



# ==== Ex 9 : Interactions between two specific enzymes
# 
# SELECT 
#   ?interaction, ?p1, ?p2
# WHERE 
#   [?p1 rdf:type :Protein] AND 
#   [?p1 :enzyme ec:2.7.7.-] AND 
#   [?p2 rdf:type :Protein] AND 
#   [?p2 :enzyme ec:3.1.3.16] AND 
#   [?interaction rdf:type :Interaction] AND 
#   [?interaction :participant ?p1] AND 
#   [?interaction :participant ?p2] 
# USING 
#   "urn:lsid:uniprot.org:ontology:",
#   "http://www.w3.org/1999/02/22-rdf-syntax-ns#" AS rdf, 
#   "http://www.w3.org/2000/01/rdf-schema#" AS rdfs,
#   "urn:lsid:uniprot.org:enzymes:" AS ec

PREFIX  rdfs:  <http://www.w3.org/2000/01/rdf-schema#>
PREFIX  rdf:   <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX  :      <urn:lsid:uniprot.org:ontology:>
SELECT  ?interaction ?p1 ?p2
WHERE
  { ?p1       rdf:type      :Protein ;
              :enzyme       <urn:lsid:uniprot.org:enzymes:2.7.7.-> .
    ?p2       rdf:type      :Protein ;
              :enzyme       <urn:lsid:uniprot.org:enzymes:3.1.3.16> .
    ?interaction  rdf:type  :Interaction ;
              :participant  ?p1 ;
              :participant  ?p2 .
  }
Retrieved from "https://www.w3.org/wiki/index.php?title=LifeSciencesQueries/SPARQL&oldid=13028"