publisci 0.1.2

data/resources/primer.ttl

@@ -0,0 +1,38 @@
+@prefix prov: <http://www.w3.org/ns/prov#> .
+@prefix rdf:  <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
+@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
+@prefix foaf: <http://xmlns.com/foaf/0.1/> .
+
+<http://example.org/entity/chart1> a prov:Entity ;
+	prov:wasAttributedTo <http://example.org/agent/derek> ;
+	rdfs:label "chart1" .
+
+<http://example.org/entity/composition1> a prov:Entity ;
+	rdfs:label "composition1" .
+
+<http://example.org/entity/regionList1> a prov:Entity ;
+	rdfs:label "regionList1" .
+
+<http://example.org/entity/dataSet1> a prov:Entity ;
+	rdfs:label "dataSet1" .
+
+<http://example.org/agent/derek> a prov:Agent ;
+	foaf:name "Derek" ;
+	prov:actedOnBehalfOf <http://example.org/agent/chartgen> ;
+	rdfs:label "derek" .
+
+<http://example.org/agent/chartgen> a prov:Organization ;
+	rdfs:label "chartgen" .
+
+<http://example.org/activity/illustrate> a prov:Activity ;
+	prov:generated <http://example.org/entity/chart1> ;
+	prov:used <http://example.org/entity/composition1>;
+	prov:wasAssociatedWith <http://example.org/agent/derek> ;
+	rdfs:label "illustrate" .
+
+<http://example.org/activity/compose1> a prov:Activity ;
+	prov:generated <http://example.org/entity/composition1> ;
+	prov:used <http://example.org/entity/regionList1>, <http://example.org/entity/dataSet1>;
+	prov:wasAssociatedWith <http://example.org/agent/derek> ;
+	rdfs:label "compose1" .
+

data/resources/queries/code_resources.rq

@@ -0,0 +1,10 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#> 
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> 
+PREFIX skos:  <http://www.w3.org/2004/02/skos/core#> 
+
+SELECT DISTINCT ?dimension ?dimLabel ?codeList ?class WHERE{
+  ?dimension a qb:DimensionProperty;
+    rdfs:label ?dimLabel;
+    qb:codeList ?codeList;
+    rdfs:range ?class.
+}

data/resources/queries/codes.rq

@@ -0,0 +1,18 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX skos:  <http://www.w3.org/2004/02/skos/core#>
+
+
+SELECT DISTINCT ?dimension ?code ?codeLabel WHERE {
+	?dim a qb:DimensionProperty;
+		rdfs:label ?dimension;
+		qb:codeList ?codes.
+
+  ?component qb:dimension ?dim.
+  ?dsd qb:component ?component.
+
+  ?dataSet qb:structure ?dsd.
+
+	?codes skos:hasTopConcept ?code.
+	?code skos:prefLabel ?codeLabel.
+}

data/resources/queries/dataset.rq

@@ -0,0 +1,7 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#> 
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> 
+
+SELECT DISTINCT ?dataset ?label WHERE{
+	?dataset a qb:DataSet;
+		rdfs:label ?label.		
+}

data/resources/queries/dimension_ranges.rq

@@ -0,0 +1,8 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#> 
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> 
+
+SELECT DISTINCT ?dimension ?range ?label WHERE {
+	?dimension a qb:DimensionProperty;
+		rdfs:label ?label;
+		rdfs:range ?range.
+}

data/resources/queries/dimensions.rq

@@ -0,0 +1,12 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+
+SELECT DISTINCT ?dimension ?label WHERE{
+  ?dimension a qb:DimensionProperty;
+		rdfs:label ?label.
+
+  ?component qb:dimension ?dimension.
+  ?dsd qb:component ?component.
+
+  ?dataSet qb:structure ?dsd.
+}

data/resources/queries/gene.rq

@@ -0,0 +1,16 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+
+SELECT DISTINCT ?gene WHERE {
+  [] a qb:ComponentSpecification;
+    rdfs:label "Tumor_Sample_Barcode";
+    qb:measure ?barcode_measure.
+
+  ?patient_obs a qb:Observation;
+    ?barcode_measure ?barcode;
+    <http://onto.strinz.me/properties/patient_id> "%{patient}";
+    <http://onto.strinz.me/properties/Hugo_Symbol> ?gene.
+
+
+  #  ?node <http://semanticscience.org/resource/SIO_000300> ?gene ;  
+}

data/resources/queries/hugo_to_ensembl.rq

@@ -0,0 +1,7 @@
+SELECT distinct ?hgnc ?ensembl where {
+ {?hgnc <http://bio2rdf.org/hgnc_vocabulary:approved_symbol> "%{hugo_symbol}"}
+ UNION
+ {?hgnc <http://bio2rdf.org/hgnc_vocabulary:synonym> "%{hugo_symbol}"}
+ 
+ ?hgnc <http://bio2rdf.org/hgnc_vocabulary:x-ensembl> ?ensembl
+}

data/resources/queries/maf_column.rq

@@ -0,0 +1,26 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX sio: <http://semanticscience.org/resource/>
+PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
+
+SELECT DISTINCT ?column WHERE {
+  [] a qb:ComponentSpecification;
+    rdfs:label "%{column}";
+    qb:measure ?column_measure.
+
+  [] a qb:ComponentSpecification;
+    rdfs:label "patient_id";
+    qb:measure ?patient_id.
+
+  ?obs a qb:Observation;
+    ?patient_id "%{patient}" ;
+    ?column_measure ?column.
+ # {
+ #   ?obs ?column_measure ?column.
+ #   FILTER isLiteral(?column)
+ # }
+ # UNION
+ # {?obs ?column_measure [ sio:SIO_000300 ?column ]}
+ # UNION
+ # {?obs ?column_measure [ sio:SIO_000008 [sio:SIO_000300 ?column] ]}
+}

data/resources/queries/measures.rq

@@ -0,0 +1,12 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+
+SELECT DISTINCT ?measure ?label WHERE{
+	?measure a qb:MeasureProperty;
+		rdfs:label ?label.
+
+  ?component qb:measure ?measure.
+  ?dsd qb:component ?component.
+
+  ?dataSet qb:structure ?dsd.
+}

data/resources/queries/observation_labels.rq

@@ -0,0 +1,8 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#> 
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> 
+PREFIX prop:  <http://www.rqtl.org/dc/properties/> 
+
+SELECT DISTINCT ?observation ?label WHERE{
+  ?observation a qb:Observation;
+    rdfs:label ?label.
+}

data/resources/queries/observations.rq

@@ -0,0 +1,13 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX prop:  <http://www.rqtl.org/dc/properties/>
+
+SELECT DISTINCT ?observation ?property ?value WHERE{
+	?observation a qb:Observation;
+		?property ?value ;
+    qb:dataSet ?dataSet .
+
+	{ ?property a qb:DimensionProperty. }
+	UNION
+	{ ?property a qb:MeasureProperty. }
+}

data/resources/queries/patient.rq

@@ -0,0 +1,11 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+
+SELECT (COUNT(?barcode) as ?barcodes) WHERE {
+  [] a qb:ComponentSpecification;
+    rdfs:label "patient_id";
+    qb:measure ?patient_id.
+
+  [] a qb:Observation;
+    ?patient_id ?barcode.
+}

data/resources/queries/patient_list.rq

@@ -0,0 +1,11 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+
+SELECT DISTINCT ?id WHERE {
+  [] a qb:ComponentSpecification;
+    rdfs:label "patient_id";
+    qb:measure ?patient_id.
+
+  [] a qb:Observation;
+    ?patient_id ?id.
+}

data/resources/queries/patients_with_mutation.rq

@@ -0,0 +1,18 @@
+PREFIX qb:   <http://purl.org/linked-data/cube#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX sio: <http://semanticscience.org/resource/>
+PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
+
+SELECT DISTINCT ?patient_id WHERE {
+  [] a qb:ComponentSpecification;
+    rdfs:label "Hugo_Symbol";
+    qb:measure ?column_measure.
+
+  [] a qb:ComponentSpecification;
+    rdfs:label "patient_id";
+    qb:measure ?patient_id_property.
+
+  ?obs a qb:Observation;
+    ?column_measure <http://identifiers.org/hgnc.symbol/%{hugo_symbol}> ;
+    ?patient_id_property ?patient_id .
+}

data/resources/queries/properties.rq

@@ -0,0 +1,8 @@
+SELECT DISTINCT ?comp ?label
+WHERE {
+  {?comp a qb:MeasureProperty;
+    rdfs:label ?label.}
+  UNION
+  {?comp a qb:DimensionProperty;
+    rdfs:label ?label.}
+}LIMIT 5

data/resources/queries/test.rq

@@ -0,0 +1,3 @@
+SELECT * WHERE{
+	?s ?p ?o
+}LIMIT 10

data/resources/weather.numeric.arff

@@ -0,0 +1,28 @@
+% 1. Title: Weather Database
+%
+% 2. Sources:
+%    (a) Generated from RDF source spec/turtle/weather
+%
+@RELATION weather
+
+@ATTRIBUTE humidity integer
+@ATTRIBUTE outlook {sunny, overcast, rainy}
+@ATTRIBUTE play {no, yes}
+@ATTRIBUTE temperature integer
+@ATTRIBUTE windy {FALSE, TRUE}
+
+@DATA
+85,sunny,no,85,FALSE
+90,sunny,no,80,TRUE
+86,overcast,yes,83,FALSE
+96,rainy,yes,70,FALSE
+80,rainy,yes,68,FALSE
+70,rainy,no,65,TRUE
+65,overcast,yes,64,TRUE
+95,sunny,no,72,FALSE
+70,sunny,yes,69,FALSE
+80,rainy,yes,75,FALSE
+70,sunny,yes,75,TRUE
+90,overcast,yes,72,TRUE
+75,overcast,yes,81,FALSE
+91,rainy,no,71,TRUE

data/scripts/get_gene_lengths.rb

@@ -0,0 +1,50 @@
+# require_relative '../lib/bio-publisci.rb'
+require 'bio-publisci'
+def gene_lengths(repo)
+
+  gene_query = <<-EOF
+  PREFIX qb:   <http://purl.org/linked-data/cube#>
+
+  SELECT DISTINCT ?gene WHERE {
+    ?obs a qb:Observation;
+      <http://onto.strinz.me/properties/Hugo_Symbol> ?gene.
+  }
+  EOF
+
+  genes = PubliSci::QueryHelper.execute(gene_query,repo).map(&:gene)
+  puts "retrieving information for #{genes.size} genes"
+
+  all_genes = []
+  genes.each_slice(250){|slice|
+    sadi_in = <<-EOF
+    @prefix sio: <http://semanticscience.org/resource/>.
+    @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>.
+    @prefix serviceowl: <http://biordf.org:8080/sadi_service_ontologies/HUGO_2_ENSEMBLGENEINFO.owl/>.
+
+    EOF
+
+    slice.each{|gene|
+      sadi_in += <<-EOF
+      <#{gene}> a serviceowl:HUGO_2_ENSEMBLGENEINFO_Input;
+         sio:SIO_000300 "#{gene}".
+      EOF
+    }
+    response = PubliSci::SADI_request.fetch_async('http://biordf.org:8080/cgi-bin/services/HUGO_2_ENSEMBLGENEINFO_async.pl',sadi_in)
+    puts "chunk done #{ all_genes.size}"
+    all_genes << response.join("\n\n")
+  }
+  all_genes.join("\n\n")
+end
+
+repo_uri = ARGV[0] || 'http://localhost:8080'
+
+repo = RDF::FourStore::Repository.new(repo_uri)
+
+lengths = gene_lengths(repo)
+puts lengths
+if ARGV[1] == "save"
+  repo << RDF::Turtle::Reader.new(lengths)
+else
+  open('gene_lengths.ttl','w'){|f| f.write lengths }
+end
+

data/scripts/islet_mlratio.rb

@@ -0,0 +1,6 @@
+load File.dirname(__FILE__) + '/../lib/bio-publisci.rb'
+
+gen = PubliSci::Readers::RMatrix.new
+con = Rserve::Connection.new
+con.eval("load('#{ARGV[0] || './.RData'}')")
+gen.generate_n3(con, "islet.mlratio", "pheno", {measures: ["probe","individual","pheno"], no_labels: true})

data/scripts/scan_islet.rb

@@ -0,0 +1,6 @@
+load File.dirname(__FILE__) + '/../lib/bio-publisci.rb'
+
+gen = PubliSci::Readers::RMatrix.new
+con = Rserve::Connection.new
+con.eval("load('#{ARGV[0] || './.RData'}')")
+gen.generate_n3(con, "scan.islet", "scan", {measures: ["probe","marker","lod"], no_labels: true})

data/scripts/update_reference.rb

@@ -0,0 +1,25 @@
+puts "Really rewrite reference files? This could mess up the tests... [y/N]"
+exit unless gets.chomp == 'y'
+
+puts "overwriting #{File.absolute_path(File.dirname(__FILE__) + '/../spec/turtle/bacon')}"
+load File.dirname(__FILE__) + '/../lib/bio-publisci.rb'
+
+gen = PubliSci::Readers::CSV.new
+turtle_string = gen.generate_n3(File.dirname(__FILE__) + '/../spec/csv/bacon.csv','bacon',{dimensions:["producer","pricerange"], label_column:0})
+open(File.dirname(__FILE__) + '/../spec/turtle/bacon', 'w'){|f| f.write turtle_string}
+
+rcon = Rserve::Connection.new
+gen = PubliSci::Readers::Dataframe.new
+rcon.void_eval <<-EOF
+library(qtl)
+data(listeria)
+mr = scanone(listeria,method="mr")
+EOF
+rexp = rcon.eval 'mr'
+turtle_string = gen.generate_n3(rexp,'mr')
+open(File.dirname(__FILE__) + '/../spec/turtle/reference', 'w'){|f| f.write turtle_string}
+puts "overwriting #{File.absolute_path(File.dirname(__FILE__) + '/../spec/turtle/reference')}"
+
+turtle_string = PubliSci::Dataset.for('resources/weather.numeric.arff',false)
+open(File.dirname(__FILE__) + '/../spec/turtle/weather', 'w'){|f| f.write turtle_string}
+puts "overwriting #{File.absolute_path(File.dirname(__FILE__) + '/../spec/turtle/weather')}"

data/server/helpers.rb

@@ -0,0 +1,215 @@
+class PubliSciServer < Sinatra::Base
+
+  helpers do
+
+    def h(str)
+      CGI::escapeHTML(str.to_s)
+    end
+
+    def format_ttl(str)
+      #TODO make this actually work right
+      str.gsub("\n","\n<br>").gsub('<br>     ','<br>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; ').gsub('<br>   ','<br>&nbsp; &nbsp; &nbsp; ')
+    end
+
+    def self.configure_server(opts)
+      set :port, opts[:port].to_i if opts[:port]
+      set :bind, opts[:bind] if opts[:bind]
+      if opts[:type]
+        if opts[:type] == "fourstore"
+          uri = opts[:uri] || 'http://localhost:80802'
+          set :repository, RDF::FourStore::Repository.new(uri.dup)
+        else
+          set :repository, RDF::Repository.new
+        end
+      else
+        set :repository, RDF::Repository.new
+      end
+    end
+
+    def print_usage
+      str = puts <<-EOF
+Bio-PubliSci Server Interface
+
+Description: Start the server and point your browser to localhost:4567
+
+Usage: bio-publisci-server [options]
+
+      EOF
+    end
+
+    CONTENT_TYPES={'xml' => 'text/xml','json' => 'application/json','ttl' => 'text/n3'}
+
+    def content_for(data,fallback=:to_html,format=(params[:format] || request.accept))
+      if params[:format]
+        format = CONTENT_TYPES[format]
+      else
+        format = format.first
+      end
+      if CONTENT_TYPES.values.include? format
+        content_type format #, :charset => 'utf-8'
+      end
+
+      case format.to_s
+      when 'text/xml'
+        data.to_xml
+      when 'application/json'
+        data.to_json
+      else
+        if data.respond_to? fallback
+          data.send(fallback)
+        else
+          data.to_s
+        end
+      end
+    end
+
+    def rdf_content_for(data,fallback=:to_html,format=(params[:format] || request.accept))
+      if params[:format]
+        format = CONTENT_TYPES[format]
+      else
+        format = format.first
+      end
+      if CONTENT_TYPES.values.include? format
+        content_type format #, :charset => 'utf-8'
+      end
+
+      case format.to_s
+      when 'text/xml'
+        data.to_rdfxml
+      when 'application/rdf+xml'
+        data.to_rdfxml
+      when 'application/json'
+        data.dump(:jsonld, :standard_prefixes => true)
+      when 'text/n3'
+        data.to_ttl
+      else
+        if data.respond_to? fallback
+          data.send(fallback)
+        else
+          data.to_s
+        end
+      end
+    end
+
+    def content_response(html_resp,content=:no_content)
+      if !CONTENT_TYPES.keys.include?(params[:format].to_s) and request.accept? 'text/html'
+        if html_resp.is_a? Symbol
+          haml html_resp
+        else
+          html_resp
+        end
+      else
+        content
+      end
+    end
+
+    def create_repository(type,uri)
+      if type == "in_memory"
+        flash[:notice] = "#{type} repository created!"
+        RDF::Repository.new
+      elsif type == "4store"
+        unless RDF::URI(uri).valid?
+          flash[:notice] = "Need a valid URI for a FourStore repository"
+          nil
+        end
+
+        flash[:notice] = "#{type} repository created!"
+        RDF::FourStore::Repository.new(uri)
+      else
+        raise "UnkownTypeForSomeReason?: #{type}"
+      end
+    end
+
+    def clear_repository
+      repo = settings.repository
+      if repo.is_a? RDF::FourStore::Repository
+      passwd = settings.sudo_pass
+      raise "need sudo password set to clear 4store" unless passwd
+      `echo #{passwd} | sudo -S killall 4s-backend`
+      `echo #{passwd} | sudo -S killall 4s-httpd`
+      `echo #{passwd} | sudo -S 4s-backend-setup test`
+      `echo #{passwd} | sudo -S 4s-backend test`
+      `echo #{passwd} | sudo -S 4s-httpd -U test`
+
+      else
+        repo.clear
+      end
+    end
+
+    def default_prefixes
+      <<-EOF
+PREFIX qb:    <http://purl.org/linked-data/cube#>
+PREFIX rdf:   <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
+PREFIX rdfs:  <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX owl:   <http://www.w3.org/2002/07/owl#>
+PREFIX skos:  <http://www.w3.org/2004/02/skos/core#>
+
+      EOF
+    end
+
+    def example_query
+      default_prefixes +
+      <<-EOF
+SELECT * WHERE {
+  ?s ?p ?o
+} LIMIT 10
+      EOF
+    end
+
+    def example_dsl
+      <<-EOF
+data do
+  object "https://raw.github.com/wstrinz/bioruby-publisci/master/spec/csv/bacon.csv"
+
+  object "https://raw.github.com/wstrinz/bioruby-publisci/master/resources/weather.numeric.arff"
+end
+      EOF
+    end
+
+    def load_dsl(script)
+      ev = PubliSci::DSL::Instance.new
+      begin
+        ev.instance_eval(script)
+      rescue Exception => e
+        raise "Caught error in eval #{e} #{e.backtrace}"
+      end
+
+      import_rdf(ev.instance_eval("generate_n3"),:ttl)
+    end
+
+    def import_rdf(input,type)
+      if input.is_a?(File) || input.is_a?(Tempfile)
+        f = Tempfile.new(['',".#{type}"])
+        begin
+          f.write(input.read)
+          f.close
+          oldsize = settings.repository.size
+          settings.repository.load(f.path, format: type)
+        ensure
+          f.unlink
+        end
+        "#{settings.repository.size - oldsize} triples imported"
+      else
+        oldsize = settings.repository.size
+        read = RDF::Reader.for(type.to_sym)
+        read = read.new(input)
+        settings.repository << read
+        "#{settings.repository.size - oldsize} triples imported"
+      end
+    end
+
+    def query_repository(query,format_result = true)
+      repo = settings.repository
+      if repo.is_a? RDF::FourStore::Repository
+        sols = SPARQL::Client.new("#{settings.repository.uri}/sparql/").query(query)
+      elsif repo.is_a? RDF::Repository
+        sols = SPARQL::Client.new(settings.repository).query(query)
+      else
+        raise "Unrecognized Repository: #{settings.repository.class}"
+      end
+
+     sols
+
+    end
+  end
+end