rbbt-entities 1.1.0 → 1.1.1

data/lib/rbbt/entity.rb

@@ -2,8 +2,11 @@ require 'rbbt/annotations'
 
 module Entity
   class << self
-    attr_accessor :formats
+    attr_accessor :formats, :entity_property_cache, :entity_list_cache
   end
+
+  self.entity_property_cache = "cache/entity_property"
+  self.entity_list_cache = "cache/entity_list"
   self.formats = {}
   
   def self.extended(base)
@@ -11,15 +14,24 @@ module Entity
 
     Entity.formats[base.to_s] = base
     base.module_eval do
-      class << self
-        attr_accessor :template, :list_template, :action_template, :list_action_template
-        alias prev_entity_extended extended
-      end
+      if not methods.include? "prev_entity_extended"
+        class << self
+          attr_accessor :template, :list_template, :action_template, :list_action_template
+          alias prev_entity_extended extended 
+        end 
+
+        def self.extended(data)
+          prev_entity_extended(data)
+
+          if Array === data and 
+            not AnnotatedArray === data and 
+            not (Annotated === data.compact.first and (data.annotation_types - data.compact.first.annotation_types).any?)
 
-      def self.extended(data)
-        prev_entity_extended(data)
-        data.extend AnnotatedArray if Array === data
-        data
+            data.extend AnnotatedArray
+          end
+
+          data
+        end
       end
 
       def self.format=(formats)
@@ -31,10 +43,16 @@ module Entity
 
       def clean_annotations
         case
-        when Array === self
-          self.annotated_array_clean_collect{|e| e.respond_to?(:clean_annotations)? e.clean_annotations : e}
+        when self.nil?
+          self
         when String === self
           "" << self
+        when self.respond_to?(:annotated_array_clean_collect)
+          self.annotated_array_clean_collect{|e| e.respond_to?(:clean_annotations)? e.clean_annotations : e}
+        when Array === self
+          self.collect{|e| e.respond_to?(:clean_annotations)? e.clean_annotations : e}
+        else
+          raise "Unknown casuistic in clean_annotations for object: #{self.inspect}"
         end
       end
 
@@ -55,7 +73,7 @@ module Entity
           end
         }
       end
-    end
+    end 
   end
 
   def property(name, &block)
@@ -112,13 +130,19 @@ module Entity
   UNPERSISTED_PREFIX = "entity_unpersisted_property_"
   def persist(method_name, type = nil, options = {})
     type = :memory if type.nil?
+    options = Misc.add_defaults options, :dir => Entity.entity_property_cache
 
     self.module_eval do
       orig_name = UNPERSISTED_PREFIX + method_name.to_s
       alias_method orig_name, method_name unless instance_methods.include? orig_name
 
       define_method method_name do |*args|
-        Persist.persist(__method__.to_s, type, options.merge(:other => {:args => args, :id => self.id})) do
+        persist_name = __method__.to_s << ":" << self.id
+
+        persist_options = options
+        persist_options = persist_options.merge(:other => {:args => args}) if args.any?
+
+        Persist.persist(persist_name, type, persist_options) do
           self.send(orig_name, *args)
         end
       end

data/lib/rbbt/entity/gene.rb

@@ -47,17 +47,15 @@ module Gene
   end
   persist :ortholog 
 
-  property :to! => :array2single do |new_format|
+  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
-  persist :to!
 
-  property :to => :array2single do |new_format|
+  property :__to => :array2single do |new_format|
     return self if format == new_format
     to!(new_format).collect!{|v| Array === v ? v.first : v}
   end
-  persist :to 
 
   property :strand => :array2single do 
     Organism.gene_positions(organism).tsv(:fields => ["Strand"], :type => :single, :persist => true).values_at *self
@@ -67,12 +65,10 @@ module Gene
   property :ensembl => :array2single do
     to "Ensembl Gene ID"
   end
-  persist :ensembl
 
   property :entrez => :array2single do
     to "Entrez Gene ID"
   end
-  persist :entrez
 
   property :uniprot => :array2single do
     to "UniProt/SwissProt Accession"
@@ -94,9 +90,9 @@ module Gene
   end
   persist :go_bp_terms
 
-  property :long_name => :single2array do
-    gene = Entrez.get_gene(to("Entrez Gene ID"))
-    gene.nil? ? nil : gene.description.flatten.first
+  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}
   end
   persist :long_name
 
@@ -180,7 +176,7 @@ module Gene
   property :matador_drugs => :array2single do
     @@matador ||= Matador.protein_drug.tsv(:persist => false).tap{|o| o.unnamed = true}
 
-    ensg = self._to("Ensembl Gene ID")
+    ensg = self.to("Ensembl Gene ID")
 
     transcripts = Gene.ensg2enst(organism, ensg)
 
@@ -239,12 +235,70 @@ module Gene
     clean_organism = self.organism.sub(/\/.*/,'') + '/jun2011'
     names.organism = clean_organism
     ranges = names.chromosome.zip(name.range).collect do |chromosome, range|
+      next if range.nil?
       [chromosome, range.begin, range.end] * ":"
     end
     Sequence.job(:somatic_snvs_at_genomic_ranges, File.join("Gene", (names.compact.sort * ", ")[0..80]), :organism => clean_organism, :ranges  => ranges).fork.join.load.values_at *ranges
   end
   persist :somatic_snvs
 
-end
 
+  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
+
+  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
+      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
+    end
+
+    ihop_interactions.collect do |sentence|
+      sentence.css('iHOPatom').each{|atom|
+        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 : interactor.name
+        end
+      }
+      sentence.to_s
+    end
+  end
+end
 

data/lib/rbbt/entity/genomic_mutation.rb

@@ -142,7 +142,7 @@ module GenomicMutation
   property :mutated_isoforms => :array2single do
     res = Sequence.job(:mutated_isoforms_for_genomic_mutations, jobname, :watson => watson, :organism => organism, :mutations => self.clean_annotations).run.values_at *self
     res.each{|list| list.organism = organism unless list.nil?}
-    res[0].annotate res if res[0].respond_to? :annotate
+    res.compact[0].annotate res if res.compact[0].respond_to? :annotate
     res
   end
   persist :mutated_isoforms
@@ -158,22 +158,18 @@ module GenomicMutation
   persist :in_exon_junction?
 
   property :over_gene? => :array2single do |gene|
-    if Gene === gene
-      range = gene.range
-      chromosome = gene.chromosome
-    else
-      range = Gene.setup(gene.dup, "Ensembl Gene ID", organism).range
-      chromosome = Gene.setup(gene.dup, "Ensembl Gene ID", organism).chromosome
-    end
+    gene = Gene.setup(gene.dup, "Ensembl Gene ID", organism) unless Gene === gene
 
-    if range.nil?
+    gene_range = gene.range
+    gene_chromosome = gene.chromosome
+
+    if gene_range.nil?
       [false] * self.length
     else
-      chromosome.zip(position).collect{|chr,pos| chr == chromosome and range.include? pos}
+      chromosome.zip(position).collect{|chr,pos| chr == gene_chromosome and gene_range.include? pos}
     end
-
-    #genes.clean_annotations.collect{|list| list.include? gene}
   end
+  persist :over_gene?
 
   property :affected_exons  => :array2single do
     Sequence.job(:exons_at_genomic_positions, jobname, :organism => organism, :positions => self.clean_annotations).run.values_at *self
@@ -181,9 +177,10 @@ module GenomicMutation
   persist :affected_exons
 
   property :damaging? => :array2single do |*args|
+    damaged_mutated_isoforms = mutated_isoforms.compact.flatten.select{|mi| mi.damaged?(*args)}
     exon_junctions.zip(mutated_isoforms).collect do |exs, mis|
       (Array === exs and exs.any?) or
-      (Array === mis and mis.select{|mi| mi.damaged?(*args)}.any?)
+      (Array === mis and (damaged_mutated_isoforms & mis).any?)
     end
   end
   persist :damaging?

data/lib/rbbt/entity/genotype.rb

@@ -20,6 +20,7 @@ module Genotype
 
   module Cohort
     extend Workflow
+    extend Entity
 
     if self.respond_to? :extended
       class << self
@@ -30,26 +31,6 @@ module Genotype
     def self.extended(cohort)
       prev_genotype_cohort_extended(cohort) if self.respond_to? :prev_genotype_cohort_extended
 
-      class << cohort
-        attr_accessor :metagenotype
-
-        def jobname
-          if @jobname.nil?
-            @jobname ||= "Meta-genotype: " + self.collect{|g| g.jobname} * ", "
-            @jobname[100..-1] = " (etc; #{self.length} genotypes)" if @jobname.length > 100
-          end
-          @jobname
-        end
-
-        def metagenotype
-          if @metagenotype.nil?
-            @metagenotype = GenomicMutation.setup(self.dup.flatten, jobname, self[0].organism, self[0].orig_watson)
-            @metagenotype.extend Genotype unless Genotype === @metagenotype
-          end
-          @metagenotype
-        end
-      end unless cohort.respond_to? :metagenotype
-
       cohort.each do |genotype| genotype.extend Genotype unless Genotype === genotype end
 
       cohort.helper :metagenotype do
@@ -68,6 +49,19 @@ module Genotype
       new.extend Cohort
     end
 
+    def jobname
+      if @jobname.nil?
+        @jobname ||= "Meta-genotype: " + self.collect{|g| g.jobname} * ", "
+        @jobname[100..-1] = " (etc; #{self.length} genotypes)" if @jobname.length > 100
+      end
+      @jobname
+    end
+
+    def metagenotype
+      GenomicMutation.setup(self.dup.flatten, jobname, self[0].organism, self[0].orig_watson).extend Genotype
+    end
+
+
     returns "Ensembl Gene ID"
     task :all_affected_genes => :array do
       set_info :organism, metagenotype.organism
@@ -145,9 +139,11 @@ module Genotype
   task :truncated => :array do
     set_info :organism, genotype.organism
     truncated_isoforms = MutatedIsoform.setup(genotype.mutated_isoforms.flatten.compact, "Hsa/jun2011").select{|isoform_mutation| isoform_mutation.truncated }
+
     proteins = truncated_isoforms.protein
     genes = proteins.gene
-    genes.to("Ensembl Gene ID").uniq.clean_annotations
+    genes = genes.to("Ensembl Gene ID").uniq
+    genes.respond_to?(:clean_annotations)? genes.clean_annotations : genes
   end
 
   returns "Ensembl Gene ID"

data/lib/rbbt/entity/mutated_isoform.rb

@@ -8,6 +8,8 @@ require 'rbbt/sources/uniprot'
 require 'rbbt/entity/gene'
 require 'nokogiri'
 
+Workflow.require_workflow 'structure'
+
 module MutatedIsoform
   extend Entity
   self.annotation :organism
@@ -51,11 +53,13 @@ module MutatedIsoform
 
   property :marked_svg => :single2array do
     svg = Open.read(protein.ensembl_protein_image_url)
+    
     seq_len = protein.sequence_length
     position = self.position
 
 
     doc = Nokogiri::XML(svg)
+    return nil unless doc.css('svg')
     width = doc.css('svg').first.attr('width').to_f
     height = doc.css('svg').first.attr('height').to_f
     start = doc.css('rect.ac').first.attr('x').to_f
@@ -95,7 +99,9 @@ module MutatedIsoform
 
   property :truncated => :array2single do
     begin
-      protein2sequence_length = Misc.process_to_hash(self.protein.flatten){|list| list.sequence_length}
+      proteins = self.protein.compact.flatten
+      protein2sequence_length = Misc.process_to_hash(proteins){|list| proteins.sequence_length}
+
       self.collect do |isoform_mutation|
 
                        next if isoform_mutation.consequence != "FRAMESHIFT" and isoform_mutation.consequence != "NONSENSE"
@@ -196,12 +202,10 @@ module MutatedIsoform
         uniprot_change
       end.compact
 
-      #return TSV.setup({}, :key_field => "Mutated Isoform", :fields => ["Func. Impact"], :type => :list) if list.empty?
       return [nil] * self.length if list.empty?
 
       tsv = MutationAssessor.chunked_predict(list.sort_by{|p| p * "_"})
 
-      #return TSV.setup({}, :key_field => "Mutated Isoform", :fields => ["Func. Impact"], :type => :list) if tsv.nil? or tsv.empty? 
       return [nil] * self.length if tsv.empty?
 
       new = TSV.setup({}, :key_field => "Mutated Isoform", :fields => ["Func. Impact"], :type => :list)
@@ -209,8 +213,13 @@ module MutatedIsoform
       tsv.each do |key, values|
         uniprot, change = key.split(" ")
         uniprot_change = [uniprot.upcase, change.upcase]
-        correspondance[uniprot_change].each do |mutation|
-          new[mutation] = values["Func. Impact"]
+        
+        if correspondance.include? uniprot_change
+          correspondance[uniprot_change].each do |mutation|
+            new[mutation] = values["Func. Impact"]
+          end
+        else
+          Log.medium "Correspondace value missing: #{uniprot_change.inspect}"
         end
       end
 
@@ -230,7 +239,14 @@ module MutatedIsoform
   property :pdbs => :single do
     uniprot = self.transcript.protein.uniprot
     next if uniprot.nil?
-    Uniprot.pdbs_covering_aa_position(uniprot, self.position)
+    UniProt.pdbs_covering_aa_position(uniprot, self.position)
   end
   persist :pdbs
+
+  property :pdbs_and_positions => :single do
+    pdbs.collect do |pdb, info|
+      [pdb, Structure.job(:sequence_position_in_pdb, "Protein: #{ self }", :sequence => protein.sequence, :organism => organism, :position => position, :pdb => pdb).run]
+    end
+  end
 end
+

data/lib/rbbt/entity/pmid.rb

@@ -1,25 +1,40 @@
 require 'rbbt/entity'
+require 'rbbt/entity/document'
 require 'rbbt/sources/pubmed'
 
 module PMID
   extend Entity
+  include Document
 
   self.format = "PMID"
 
   property :article => :array2single do
     PubMed.get_article(self).values_at(*self)
   end
-  persist :article
+
+  property :abstract => :array2single do
+    article.collect{|a| a.nil? ? nil : a.abstract}
+  end
+  persist :abstract
 
   property :title => :array2single do
     article.collect{|a| a.nil? ? nil : a.title}
   end
   persist :title
 
-  property :text => :array2single do
+  property :journal => :array2single do
+    article.collect{|a| a.nil? ? nil : a.journal}
+  end
+  persist :journal
+
+  property :year => :array2single do
+    article.collect{|a| a.nil? ? nil : a.year}
+  end
+  persist :year
+
+  property :_get_text => :array2single do
     article.collect{|a| a.nil? ? nil : a.text}
   end
-  persist :text 
 
   property :pubmed_url => :single2array do
     "<a class='pmid' href='http://www.ncbi.nlm.nih.gov/pubmed/#{self}'>#{ self }</a>"

data/lib/rbbt/entity/protein.rb

@@ -36,17 +36,17 @@ module Protein
   property :uniprot => :array2single do
     to "UniProt/SwissProt Accession"
   end
-  persist :uniprot
 
   property :ensembl => :array2single do
     to "Ensembl Protein ID"
   end
-  persist :ensembl
 
   property :transcript => :array2single do
-    ensembl.collect{|ensp|
+    res = ensembl.collect{|ensp|
       Protein.ensp2enst(organism, ensp)
     }
+    Transcript.setup(res, "Ensembl Transcript ID", self.organism) if defined? Transcript
+    res
   end
   persist :transcript
 
@@ -56,22 +56,20 @@ module Protein
   end
   persist :ensembl_protein_image_url
 
-  property :to! => :array2single do |new_format|
+  property :to => :array2single do |new_format|
     return self if format == new_format
-    Protein.setup(Translation.job(:translate_protein, "", :organism => organism, :proteins => self, :format => new_format).exec, new_format, organism)
+    Protein.setup(Translation.job(:tsv_translate_protein, "", :organism => organism, :proteins => self, :format => new_format).exec.values_at(*self), new_format, organism)
   end
-  persist :to!
 
-  property :to => :array2single do |new_format|
+  property :__to => :array2single do |new_format|
     return self if format == new_format
     to!(new_format).collect!{|v| v.nil? ? nil : v.first}
   end
-  persist :to
 
   property :gene => :array do
     Gene.setup(to("Ensembl Protein ID").clean_annotations, "Ensembl Protein ID", organism)
   end
-  persist :gene
+  persist :gene #, :yaml, :file => '/tmp/testes'
 
   property :pfam => :array2single do
     index = Organism.gene_pfam(organism).tsv :flat, :persist => true
@@ -90,5 +88,6 @@ module Protein
     sequence.collect{|seq| seq.nil? ? nil : seq.length}
   end
   persist :sequence_length
+
 end
 

data/lib/rbbt/entity/transcript.rb

@@ -1,5 +1,4 @@
 require 'rbbt/entity'
-require 'rbbt/entity/gene'
 
 module Transcript
   extend Entity
@@ -17,7 +16,8 @@ module Transcript
           else
             @@enst2ensg[organism][transcript]
           end
-    Gene.setup(res, "Ensembl Gene ID", organism)
+    Gene.setup(res, "Ensembl Gene ID", organism) if defined? Gene
+    res
   end
 
   def self.enst2ensp(organism, transcript)
@@ -34,7 +34,9 @@ module Transcript
 
   property :to! => :array2single do |new_format|
     return self if format == new_format
-    Gene.setup(Translation.job(:tsv_translate_probe, "", :organism => organism, :probes => self, :format => new_format).exec.values_at(*self), new_format, organism)
+    res = Translation.job(:tsv_translate_probe, "", :organism => organism, :probes => self, :format => new_format).exec.values_at(*self)
+    Gene.setup(res, "Ensembl Gene ID", organism) if defined? Gene
+    res
   end
 
   property :to => :array2single do |new_format|

data/test/rbbt/entity/test_protein.rb

@@ -4,6 +4,7 @@ require 'test/unit'
 require 'rbbt/util/tmpfile'
 require 'test/unit'
 require 'rbbt/entity/protein'
+require 'rbbt/entity/mutated_isoform'
 
 class TestProtein < Test::Unit::TestCase
   PROTEIN = Protein.setup("ENSP00000275493", "Hsa/jun2011")
@@ -22,6 +23,11 @@ class TestProtein < Test::Unit::TestCase
     assert_equal "EGFR", PROTEIN.gene.to("Associated Gene Name")
   end
 
+  def test_protein
+    ddd MutatedIsoform.setup(["ENSP00000322422:K168FrameShift"], "Hsa/may2009").info
+    ddd MutatedIsoform.setup(["ENSP00000322422:K168FrameShift"], "Hsa/may2009").protein
+  end
+
 end
 
 

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: rbbt-entities
 version: !ruby/object:Gem::Version 
-  hash: 19
+  hash: 17
   prerelease: 
   segments: 
   - 1
   - 1
-  - 0
-  version: 1.1.0
+  - 1
+  version: 1.1.1
 platform: ruby
 authors: 
 - Miguel Vazquez
@@ -15,8 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2012-01-13 00:00:00 +01:00
-default_executable: 
+date: 2012-01-31 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: rbbt-util
@@ -86,7 +85,6 @@ files:
 - test/rbbt/entity/test_mutated_isoform.rb
 - test/rbbt/entity/test_protein.rb
 - test/rbbt/test_entity.rb
-has_rdoc: true
 homepage: http://github.com/mikisvaz/rbbt-util
 licenses: []
 
@@ -116,7 +114,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: 
-rubygems_version: 1.6.2
+rubygems_version: 1.8.12
 signing_key: 
 specification_version: 3
 summary: Entities for the Ruby Bioinformatics Toolkit (rbbt)