rbbt-entities 1.1.1 → 2.0.0

data/bin/icgc2rbbt.rb

@@ -0,0 +1,23 @@
+#!/usr/bin/env ruby
+require 'rbbt/util/open'
+
+file = ARGV.shift
+directory = ARGV.shift
+
+genotypes = {}
+Open.read(file).split("\n").each do |line|
+  next if line =~ /^Cancer Type/
+
+  chr, pos, ref, mut, sample = line.split(/\t/).values_at 2, 3, 6, 10, 35 
+
+  chr.sub!(/chr/,'')
+  mut = '-' * (mut.length - 1) if mut =~/^-[ACGT]/
+
+  genotypes[sample] ||= []
+  genotypes[sample] << [chr, pos, mut] * ":"
+end
+
+genotypes.each do |sample, mutations|
+  mutations.uniq!
+  Open.write(File.join(directory, sample), mutations.uniq * "\n")
+end

data/bin/vcf2rbbt.rb

@@ -0,0 +1,15 @@
+#!/usr/bin/env ruby
+require 'rbbt/util/open'
+
+file = ARGV.shift
+
+Open.read(file).split("\n").each do |line|
+  next if line =~ /^#/
+
+  chr, pos, id, ref, mut, score = line.split(/\t/)
+
+  chr.sub!(/chr/,'')
+  mut = mut + '-' * (ref.length - mut.length) if ref.length > mut.length
+
+  puts [chr, pos, mut, score] * ":"
+end

data/lib/rbbt/entity/chromosome_range.rb

@@ -0,0 +1,73 @@
+require 'rbbt/entity'
+require 'rbbt/workflow'
+require 'rbbt/sources/organism'
+require 'rbbt/entity/gene'
+
+Workflow.require_workflow "Sequence"
+
+module ChromosomeRange
+  extend Entity
+
+  self.annotation :organism
+ 
+  self.format = "Chromosome Range"
+
+
+  def self.text_to_unit(text)
+    text = text.sub('^', '+')
+    base = text.to_f
+    case
+    when text =~ /KB?$/
+      base * 1000
+    when text =~ /MB?$/
+      base * 1000_000
+    when text =~ /^\d+(\.\d+)?(e\+\d+)?$/
+      base
+    else 
+      raise "Text format not understood: #{ text }"
+    end.to_i
+  end
+
+  property :unit => :array2single do
+    self.collect{|range|
+      chr, start, eend = range.split(":")
+      [chr, ChromosomeRange.text_to_unit(start), ChromosomeRange.text_to_unit(eend)] * ":"
+    }
+  end
+  persist :unit
+
+  property :genes => :array2single do
+    Sequence.job(:genes_at_genomic_ranges, "ChromosomeRange", :organism => organism, :ranges => self.unit).run.tap{|t| t.namespace = organism}.values_at *self.unit
+  end
+
+  property :ensembl_browser => :single2array do
+    "http://#{Misc.ensembl_server(self.organism)}/Homo_sapiens/Location/View?db=core&r=#{chromosome}:#{start}-#{eend}"
+  end
+ 
+  property :chromosome => :array2single do
+    self.clean_annotations.collect{|r| r.split(":")[0]}
+  end
+  persist :_ary_chromosome
+
+  property :start => :array2single do
+    self.clean_annotations.collect{|r| r.split(":")[1].to_i}
+  end
+  persist :_ary_start
+
+  property :eend => :array2single do
+    self.clean_annotations.collect{|r| r.split(":")[2].to_i}
+  end
+  persist :_ary_eend
+
+  property :end => :array2single do
+    self.eend
+  end
+  persist :_ary_end
+
+  property :range => :array2single do
+    start.zip(self.end).collect{|s,e| (s..e)}
+  end
+  persist :_ary_range
+
+
+end

data/lib/rbbt/entity/cnv.rb

@@ -26,12 +26,30 @@ module CNV
 
   property :genes => :array2single do
     @genes ||= begin
-                 genes = Sequence.job(:genes_at_genomic_ranges, jobname, :organism => organism, :ranges => self).run
-                 genes.unnamed = true
+                 genes = Sequence.job(:genes_at_genomic_ranges, jobname, :organism => organism, :ranges => self, :unnamed => true).run
                  genes = genes.values_at *self
                  Gene.setup(genes, "Ensembl Gene ID", organism)
                end
   end
 
+  property :chromosome => :array2single do
+    self.clean_annotations.collect{|mut| mut.split(":")[0]}
+  end
+  persist :_ary_chromosome
+
+  property :start => :array2single do
+    self.clean_annotations.collect{|mut| mut.split(":")[1].to_i}
+  end
+  persist :_ary_start
+
+  property :end => :array2single do
+    self.clean_annotations.collect{|mut| mut.split(":")[2].to_i}
+  end
+  persist :_ary_end
+
+  property :range => :array2single do
+    start.zip(self.end).collect{|s,e| (s..e)}
+  end
+
 end
  

data/lib/rbbt/entity/gene.rb

@@ -7,6 +7,7 @@ require 'rbbt/sources/cancer'
 require 'rbbt/entity/protein'
 require 'rbbt/entity/pmid'
 require 'rbbt/entity/transcript'
+require 'rbbt/bow/bow'
 
 Workflow.require_workflow "Translation"
 
@@ -15,7 +16,7 @@ module Gene
 
   def self.ensg2enst(organism, gene)
     @@ensg2enst ||= {}
-    @@ensg2enst[organism] ||= Organism.gene_transcripts(organism).tsv(:type => :flat, :key_field => "Ensembl Gene ID", :fields => ["Ensembl Transcript ID"], :persist => true).tap{|o| o.unnamed = true}
+    @@ensg2enst[organism] ||= Organism.gene_transcripts(organism).tsv(:type => :flat, :key_field => "Ensembl Gene ID", :fields => ["Ensembl Transcript ID"], :persist => true, :unnamed => true)
 
     if Array === gene
       @@ensg2enst[organism].values_at *gene
@@ -24,7 +25,6 @@ module Gene
     end
   end
 
-
   def self.filter(query, field = nil, options = nil, entity = nil)
     return true if query == entity
 
@@ -33,39 +33,86 @@ module Gene
     false
   end
 
+  def self.gene_list_bases(genes)
+    genes = genes.ensembl
+    chromosome_genes = {}
+    Misc.process_to_hash(genes){|genes| genes.chromosome}.each{|gene, chr| chromosome_genes[chr] ||= []; chromosome_genes[chr] << gene}
+    total = 0
+    chromosome_genes.each do |chr,gs|
+      next if chr.nil?
+      total += Misc.total_length(genes.annotate(gs).chr_range.compact)
+    end
+    
+    total
+  end
+
+  def self.gene_list_exon_bases(genes)
+    genes = genes.ensembl
+    chromosome_genes = {}
+    Misc.process_to_hash(genes){|genes| genes.chromosome}.each{|gene, chr| chromosome_genes[chr] ||= []; chromosome_genes[chr] << gene}
+
+    @@exon_range_tsv ||= {}
+    organism = genes.organism
+    @@exon_range_tsv[organism] ||= Organism.exons(organism).tsv :persist => true, :fields => ["Exon Chr Start", "Exon Chr End"], :type => :list, :cast => :to_i, :unnamed => true
+    total = 0
+
+    chromosome_genes.each do |chr,gs|
+      next if chr.nil?
+      exons = genes.annotate(gs).transcripts.compact.flatten.exons.compact.flatten.uniq
+
+      exon_ranges = exons.collect{|exon|
+        Log.low "Exon #{ exon } does not have range" unless @@exon_range_tsv[organism].include? exon
+        next unless @@exon_range_tsv[organism].include? exon
+        pos = @@exon_range_tsv[organism][exon]
+        (pos.first..pos.last)
+      }.compact
+      total += Misc.total_length(exon_ranges)
+    end
+    
+    total
+  end
+
+
+
   self.annotation :format
   self.annotation :organism
 
-  self.format = Organism::Hsa.identifiers.all_fields - ["Ensembl Protein ID", "Ensembl Transcript ID"]
+  self.format = Organism.identifiers("Hsa").all_fields - ["Ensembl Protein ID", "Ensembl Transcript ID"]
 
   property :ortholog => :array2single do |other|
     return self if organism =~ /^#{ other }(?!\w)/
     new_organism = organism.split(":")
     new_organism[0] = other
     new_organism = new_organism * "/"
-    Gene.setup(Organism[organism]["ortholog_#{other}"].tsv(:persist => true).values_at(*self.ensembl).collect{|l| l.first}, "Ensembl Gene ID", new_organism)
+    Gene.setup(Organism[organism]["ortholog_#{other}"].tsv(:persist => true, :unnamed => true).values_at(*self.ensembl).collect{|l| l.first}, "Ensembl Gene ID", new_organism)
   end
   persist :ortholog 
 
   property :to => :array2single do |new_format|
     return self if format == new_format
-    Gene.setup(Translation.job(:tsv_translate, "", :organism => organism, :genes => self, :format => new_format).exec.values_at(*self), new_format, organism)
-  end
-
-  property :__to => :array2single do |new_format|
-    return self if format == new_format
-    to!(new_format).collect!{|v| Array === v ? v.first : v}
+    genes = Translation.job(:tsv_translate, "", :organism => organism, :genes => self, :format => new_format).exec.values_at(*self)
+    Gene.setup(genes, new_format, organism)
+    genes
   end
 
   property :strand => :array2single do 
-    Organism.gene_positions(organism).tsv(:fields => ["Strand"], :type => :single, :persist => true).values_at *self
+    @@strand_tsv ||= {}
+    @@strand_tsv[organism] ||= Organism.gene_positions(organism).tsv(:fields => ["Strand"], :type => :single, :persist => true, :unnamed => true)
+    to("Ensembl Gene ID").collect do |gene|
+      @@strand_tsv[organism][gene]
+    end
   end
-  persist :strand
+  persist :_ary_strand
 
   property :ensembl => :array2single do
     to "Ensembl Gene ID"
   end
 
+  property :biotype => :array2single do
+    Organism.gene_biotype(organism).tsv(:persist => true, :type => :single, :unnamed => true).values_at *self.ensembl
+  end
+  persist :biotype
+
   property :entrez => :array2single do
     to "Entrez Gene ID"
   end
@@ -73,26 +120,25 @@ module Gene
   property :uniprot => :array2single do
     to "UniProt/SwissProt Accession"
   end
-  persist :uniprot
 
   property :name => :array2single do
+    return self if self.format == "Associated Gene Name"
     to "Associated Gene Name"
   end
-  persist :name
 
   property :chr_start => :array2single do
-    Organism.gene_positions(organism).tsv(:persist => true, :type => :single, :cast => :to_i, :fields => ["Gene Start"]).values_at *self
+    Organism.gene_positions(organism).tsv(:persist => true, :type => :single, :cast => :to_i, :fields => ["Gene Start"], :unnamed => true).values_at *self
   end
   persist :chr_start
 
   property :go_bp_terms => :array2single do
-    Organism.gene_go_bp(organism).tsv(:persist => true, :key_field => "Ensembl Gene ID", :fields => ["GO ID"], :type => :flat).values_at *self.ensembl
+    Organism.gene_go_bp(organism).tsv(:persist => true, :key_field => "Ensembl Gene ID", :fields => ["GO ID"], :type => :flat, :unnamed => true).values_at *self.ensembl
   end
   persist :go_bp_terms
 
   property :long_name => :array2single do
     entre = self.entrez
-    gene = Entrez.get_gene(entrez).values_at(*entrez).collect{|gene| gene.nil? ? nil : gene.description.flatten.first}
+    gene = Entrez.get_gene(entrez).values_at(*entrez).collect{|gene| gene.nil? ? nil : (gene.description || []).flatten.first}
   end
   persist :long_name
 
@@ -140,26 +186,26 @@ module Gene
   persist :max_protein_length
 
   property :chromosome => :array2single do
-    chr = Organism.gene_positions(organism).tsv :fields => ["Chromosome Name"], :type => :single, :persist => true
-    chr.unnamed = true
+    @@chromosome_tsv ||= {}
+    @@chromosome_tsv[organism] ||= Organism.gene_positions(organism).tsv :fields => ["Chromosome Name"], :type => :single, :persist => true, :unnamed => true
     if Array === self
       to("Ensembl Gene ID").collect do |gene|
-        chr[gene]
+        @@chromosome_tsv[organism][gene]
       end
     else
-      chr[to("Ensembl Gene ID")]
+      @@chromosome_tsv[organism][to("Ensembl Gene ID")]
     end
   end
   persist :chromosome
 
-  property :range => :array2single do
-    pos = Organism.gene_positions(organism).tsv :fields => ["Gene Start", "Gene End"], :type => :list, :persist => true, :cast => :to_i
+  property :chr_range => :array2single do
+    chr_range_index ||= Organism.gene_positions(organism).tsv :fields => ["Gene Start", "Gene End"], :type => :list, :persist => true, :cast => :to_i, :unnamed => true
     to("Ensembl Gene ID").collect do |gene|
-      next if not pos.include? gene
-      Range.new *pos[gene]
+      next if not chr_range_index.include? gene
+      Range.new *chr_range_index[gene]
     end
   end
-  persist :range
+  persist :chr_range
 
   property :articles => :array2single do
     PMID.setup(Organism.gene_pmids(organism).tsv(:persist => true, :fields => ["PMID"], :type => :flat, :unnamed => true).values_at *self.entrez)
@@ -167,14 +213,14 @@ module Gene
   persist :articles
 
   property :sequence => :array2single do
-    Organism.gene_sequence(organism).tsv :persist => true
-    @gene_sequence.unnamed = true
-    @gene_sequence.values_at *self.ensembl
+    @@sequence_tsv ||= {}
+    @@sequence_tsv[organism] ||= Organism.gene_sequence(organism).tsv :persist => true, :unnamed => true
+    @@sequence_tsv[organism].values_at *self.ensembl
   end
   persist :sequence
 
   property :matador_drugs => :array2single do
-    @@matador ||= Matador.protein_drug.tsv(:persist => false).tap{|o| o.unnamed = true}
+    @@matador ||= Matador.protein_drug.tsv(:persist => false, :unnamed => true)
 
     ensg = self.to("Ensembl Gene ID")
 
@@ -225,7 +271,7 @@ module Gene
   persist :pathway_drugs
 
   property :related_cancers => :array2single do
-    Cancer["cancer_genes.tsv"].tsv(:persist => true, :type => :list).values_at(*self.name).collect{|v| v.nil? ? nil : v["Tumour Types (Somatic Mutations)"].split(", ") + v["Tumour Types (Germline Mutations)"].split(", ")}
+    Cancer["cancer_genes.tsv"].tsv(:persist => true, :type => :list).values_at(*self.name).collect{|v| v.nil? ? nil : (v["Tumour Types (Somatic Mutations)"].split(", ") + v["Tumour Types (Germline Mutations)"].split(", ")).uniq}
   end
   persist :related_cancers
 
@@ -234,7 +280,7 @@ module Gene
     raise "No organism defined" if self.organism.nil?
     clean_organism = self.organism.sub(/\/.*/,'') + '/jun2011'
     names.organism = clean_organism
-    ranges = names.chromosome.zip(name.range).collect do |chromosome, range|
+    ranges = names.chromosome.zip(name.chr_range).collect do |chromosome, range|
       next if range.nil?
       [chromosome, range.begin, range.end] * ":"
     end
@@ -243,62 +289,89 @@ module Gene
   persist :somatic_snvs
 
 
+  property :literature_score do |terms|
+    terms = terms.collect{|t| t.stem}
+    articles = self.articles
+    if articles.nil? or articles.empty?
+      0
+    else
+      articles.inject(0){|acc,article| acc += article.text.words.select{|word| terms.include? word}.length }.to_f / articles.length
+    end
+  end
+  persist :literature_score
+
+
   property :ihop_interactions => :single do
     uniprot = self.uniprot
-    url = "http://ws.bioinfo.cnio.es/iHOP/cgi-bin/getSymbolInteractions?ncbiTaxId=9606&reference=#{uniprot}&namespace=UNIPROT__AC" 
-    doc = Nokogiri::XML(Open.read(url))
-    sentences = doc.css("iHOPsentence")
-    sentences
-  end
+    if uniprot.nil?
+      nil
+    else
+      sentences = []
 
-  property :tagged_ihop_interactions => :single do
-    interactors = []
-    ihop_interactions.each do |sentence|
-      sentence.css('iHOPatom').collect{|atom|
-        atom.css('evidence');
-      }.compact.flatten.each do |evidence|
-        symbol =  evidence.attr('symbol')
-        taxid  =  evidence.attr('ncbiTaxId')
-
-        if Organism.entrez_taxids(self.organism).list.include? taxid
-          interactors << symbol
-        end
+      begin
+        url = "http://ws.bioinfo.cnio.es/iHOP/cgi-bin/getSymbolInteractions?ncbiTaxId=9606&reference=#{uniprot}&namespace=UNIPROT__AC" 
+        doc = Nokogiri::XML(Open.read(url))
+        sentences = doc.css("iHOPsentence")
+      rescue
       end
-    end
-
-    Gene.setup(interactors, "Associated Gene Name", self.organism).organism
-    
-    interactors_ensembl = interactors.ensembl
 
-    interactors2ensembl = {}
-    interactors.collect{|i| i}.zip(interactors_ensembl.collect{|i| i}).each do |o,e|
-      interactors2ensembl[o] = e
+      sentences
     end
+  end
 
-    ihop_interactions.collect do |sentence|
-      sentence.css('iHOPatom').each{|atom|
-        evidences = atom.css('evidence')
-        symbol = evidences.collect do |evidence|
+  property :tagged_ihop_interactions => :single do
+    interactors = []
+    ihop_interactions = self.ihop_interactions
+    if ihop_interactions.nil?
+      nil
+    else
+      ihop_interactions.each do |sentence|
+        sentence.css('iHOPatom').collect{|atom|
+          atom.css('evidence');
+        }.compact.flatten.each do |evidence|
           symbol =  evidence.attr('symbol')
           taxid  =  evidence.attr('ncbiTaxId')
 
           if Organism.entrez_taxids(self.organism).list.include? taxid
-            symbol
-          else
-            nil
+            interactors << symbol
           end
-        end.compact.first
+        end
+      end
 
-        evidences.remove
+      Gene.setup(interactors, "Associated Gene Name", self.organism).organism
+      
+      interactors_ensembl = interactors.ensembl
 
-        if interactors2ensembl.include? symbol and not interactors2ensembl[symbol].nil?
-          atom.children.remove
-          interactor = interactors2ensembl[symbol]
-          atom.replace interactor.respond_to?(:link)? interactor.link : interactor.name
-        end
-      }
-      sentence.to_s
+      interactors2ensembl = {}
+      interactors.collect{|i| i}.zip(interactors_ensembl.collect{|i| i}).each do |o,e|
+        interactors2ensembl[o] = e
+      end
+
+      ihop_interactions.collect do |sentence|
+        sentence.css('iHOPatom').each{|atom|
+          literal = atom.content()
+          evidences = atom.css('evidence')
+          symbol = evidences.collect do |evidence|
+            symbol =  evidence.attr('symbol')
+            taxid  =  evidence.attr('ncbiTaxId')
+
+            if Organism.entrez_taxids(self.organism).list.include? taxid
+              symbol
+            else
+              nil
+            end
+          end.compact.first
+
+          evidences.remove
+
+          if interactors2ensembl.include? symbol and not interactors2ensembl[symbol].nil?
+            atom.children.remove
+            interactor = interactors2ensembl[symbol]
+            atom.replace interactor.respond_to?(:link)? interactor.link(nil, nil, :html_link_extra_attrs => "title='#{literal}'") : interactor.name
+          end
+        }
+        sentence.to_s
+      end
     end
   end
 end
-