rbbt-entities 1.1.1 → 2.0.0

data/lib/rbbt/entity/mutated_isoform.rb

@@ -5,10 +5,12 @@ require 'rbbt/mutation/mutation_assessor'
 require 'rbbt/mutation/sift'
 require 'rbbt/entity/protein'
 require 'rbbt/sources/uniprot'
+require 'rbbt/sources/InterPro'
 require 'rbbt/entity/gene'
 require 'nokogiri'
 
 Workflow.require_workflow 'structure'
+Workflow.require_workflow 'MutEval'
 
 module MutatedIsoform
   extend Entity
@@ -16,10 +18,12 @@ module MutatedIsoform
 
   self.format = "Mutated Isoform"
 
+  DEFAULT_DAMAGE_PREDICTORS = [:sift, :mutation_assessor]
+
   property :protein => :array2single do
-    Protein.setup(self.collect{|mutation| mutation.split(":").first if mutation =~ /^ENSP/}, "Ensembl Protein ID", organism)
+    proteins = self.collect{|mutation| mutation.split(":").first if mutation[0..3] == "ENSP"}
+    Protein.setup(proteins, "Ensembl Protein ID", organism)
   end
-  persist :protein
 
   property :transcript => :array2single do
     begin
@@ -27,12 +31,10 @@ module MutatedIsoform
       Transcript.setup(protein.transcript.zip(self.collect{|mutation| mutation.split(":").first}).collect{|p| p.compact.first}, "Ensembl Transcript ID", organism)
     end
   end
-  persist :transcript
 
   property :change => :array2single do
     self.collect{|mi| mi.split(":").last}
   end
-  persist :change
 
   property :position => :array2single do
     change.collect{|c|
@@ -43,13 +45,11 @@ module MutatedIsoform
       end
     }
   end
-  persist :position
 
   property :ensembl_protein_image_url => :single2array do
     ensembl_url = if organism == "Hsa" then "www.ensembl.org" else "#{organism.sub(/.*\//,'')}.archive.ensembl.org" end
     "http://#{ensembl_url}/Homo_sapiens/Component/Transcript/Web/TranslationImage?db=core;p=#{protein};_rmd=d2a8;export=svg"
   end
-  persist :ensembl_protein_image_url
 
   property :marked_svg => :single2array do
     svg = Open.read(protein.ensembl_protein_image_url)
@@ -57,7 +57,6 @@ module MutatedIsoform
     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
@@ -71,14 +70,17 @@ module MutatedIsoform
       svg
     end
   end
-  persist :marked_svg
 
   ASTERISK = "*"[0]
   CONSECUENCES = %w(UTR SYNONYMOUS NOSTOP MISS-SENSE INDEL FRAMESHIFT NONSENSE)
   property :consequence => :single2array do
+    return nil if self.nil?
+
     prot, change = self.split(":")
 
     case
+    when change.nil?
+      nil
     when change =~ /UTR/
       "UTR"
     when (change[0] == ASTERISK and not change[0] == change[-1])
@@ -95,37 +97,111 @@ module MutatedIsoform
       "MISS-SENSE"
     end
   end
-  persist :consequence
+
+  property :in_utr => :array2single do
+    consequence.collect{|c|
+      c == "UTR"
+    }
+  end
+
+  property :synonymous => :array2single do
+    consequence.collect{|c|
+      c == "SYNONYMOUS"
+    }
+  end
+
+  property :non_synonymous => :array2single do
+    consequence.collect{|c|
+      not c.nil? and c != "SYNONYMOUS" and c != "UTR"
+    }
+  end
+
+  property :affected_interpro_domains => :single do
+    if protein.nil?
+      []
+    else
+      InterProDomain.setup(Misc.zip_fields(protein.interpro_domain_positions || []).select{|d,s,e|
+        e.to_i > position and s.to_i < position
+      }.collect{|d,s,e| d }, organism)
+    end
+  end
+
+  property :affected_interpro_domain_positions => :single do
+    if protein.nil?
+      []
+    else
+      Misc.zip_fields(protein.interpro_domain_positions || []).select{|d,s,e|
+        e.to_i > position and s.to_i < position
+      }.collect{|d,s,e| [d, position - s.to_i, s.to_i, e.to_i]}
+    end
+  end
+
+  property :affected_domain_positions => :single do
+    affected_interpro_domain_positions
+  end
+
+  property :affected_domains => :single do
+    affected_interpro_domains
+  end
+
+  property :ablated_interpro_domains => :single do
+    if protein.nil?
+      []
+    else
+      InterProDomain.setup(Misc.zip_fields(protein.interpro_domain_positions || []).select{|d,s,e|
+        e.to_i > position
+      }.collect{|d,s,e| d }, organism)
+    end
+  end
+
+  property :ablated_interpro_domain_positions => :single do
+    if protein.nil?
+      []
+    else
+      Misc.zip_fields(protein.interpro_domain_positions || []).select{|d,s,e|
+        e.to_i > position
+      }.collect{|d,s,e| [d, s.to_i, e.to_i]}
+    end
+  end
+
+  property :ablated_domain_positions => :single do
+    ablated_interpro_domain_positions
+  end
+
+  property :ablated_domains => :single do
+    ablated_interpro_domains
+  end
 
   property :truncated => :array2single do
     begin
       proteins = self.protein.compact.flatten
-      protein2sequence_length = Misc.process_to_hash(proteins){|list| proteins.sequence_length}
+      protein2sequence_length = Misc.process_to_hash(proteins){|list| proteins.any? ? proteins.sequence_length : []}
 
       self.collect do |isoform_mutation|
 
-                       next if isoform_mutation.consequence != "FRAMESHIFT" and isoform_mutation.consequence != "NONSENSE"
-                       protein  = isoform_mutation.protein
-                       position = isoform_mutation.position
-                       sequence_length = protein2sequence_length[protein]
-
-                       case
-                       when (sequence_length.nil? or position.nil?)
-                         nil
-                       when position < sequence_length.to_f * 0.7
-                         true
-                       else
-                         false
-                       end
-                     end
-                   end
+        next if isoform_mutation.consequence != "FRAMESHIFT" and isoform_mutation.consequence != "NONSENSE"
+        protein  = isoform_mutation.protein
+        position = isoform_mutation.position
+        sequence_length = protein2sequence_length[protein]
+
+        case
+        when (sequence_length.nil? or position.nil?)
+          nil
+        when position < sequence_length.to_f * 0.7
+          true
+        when (isoform_mutation.ablated_domains.any?)
+          true
+        else
+          false
+        end
+      end
+    end
   end
-  persist :truncated
 
   property :damage_scores => :array2single do |*args|
     begin
       methods = args.first
-      methods = [:sift, :mutation_assessor] if methods.nil?
+      methods = MutatedIsoform::DEFAULT_DAMAGE_PREDICTORS if methods.nil?
       methods = [methods] unless Array === methods
       values = methods.collect{|method|
         case method.to_sym
@@ -133,6 +209,12 @@ module MutatedIsoform
           sift_scores
         when :mutation_assessor
           mutation_assessor_scores
+        when :polyphen
+          polyphen_scores
+        when :snps_and_go
+          snps_and_go_scores
+        when :transFIC
+          transFIC_scores(:mutation_assessor)
         else
           raise "Unknown predictive method: #{ method }"
         end
@@ -154,94 +236,100 @@ module MutatedIsoform
       end
     end
   end
-  persist :damage_scores
 
   property :damaged? => :array2single do |*args|
     begin
       methods, threshold = args
+      threshold, methods = methods, nil if threshold.nil? and not Array === methods
       threshold     = 0.8 if threshold.nil?
       damage_scores = self.damage_scores(methods)
       truncated     = self.truncated
+
       damage_scores.zip(truncated).collect{|damage, truncated| truncated or (not damage.nil? and damage > threshold) }
     end
   end
-  persist :damaged?
 
-  property :sift_scores => :array2single do
+  property :snps_and_go_scores => :array2single do
     begin
-      missense = self.select{|iso_mut| iso_mut.consequence == "MISS-SENSE"}
-
-      values = SIFT.chunked_predict(missense).values_at(*self).collect{|v|
-        v.nil? ? nil : 1.0 - v["Score 1"].to_f
+      missense = self.select{|mutation| mutation.consequence == "MISS-SENSE"}
+      res = MutEval.job(:snps_and_go, "MutatedIsoforms (#{self.length})", :mutations => missense.sort, :organism => organism).run
+      res.values_at(*self).collect{|v| (v.nil? or v["SNPSandGO Score"].nil? or v["SNPSandGO Score"].empty?) ? 
+        nil : 
+        (v["SNPSandGO Prediction"] == "Disease" ? 1.0 - (10.0 - v["SNPSandGO Score"].to_f) / 20 : 0 + (10.0 - v["SNPSandGO Score"].to_f) / 20)
       }
-
-      values
-
-      #range = {nil => nil,
-      #  ""  => nil,
-      #  "TOLERATED" => 0,
-      #  "*DAMAGING" => 1,
-      #  "DAMAGING" => 1}
-
-      #range.values_at *values
+    rescue
+      Log.warn $!.message
+      [nil] * self.length
     end
   end
-  persist :sift_scores
 
-  property :mutation_assessor_scores => :array2single do
+  property :polyphen_scores => :array2single do
     begin
       missense = self.select{|mutation| mutation.consequence == "MISS-SENSE"}
+      res = MutEval.job(:polyphen, "MutatedIsoforms (#{self.length})", :mutations => missense.sort, :organism => organism).run
+      res.values_at(*self).collect{|v| (v.nil? or v["Polyphen Score"].nil? or v["Polyphen Score"].empty?) ? nil : v["Polyphen Score"].to_f / 10}
+    rescue
+      Log.warn $!.message
+      [nil] * self.length
+    end
+  end
 
-      correspondance = {}
-      list = missense.zip(missense.protein.to "UniProt/SwissProt ID").collect do |mutation, uniprot|
-        prot, change = mutation.split(":")
-        next if uniprot.nil?
-        uniprot_change = [uniprot.upcase, change.upcase]
-        correspondance[uniprot_change] ||= []
-        correspondance[uniprot_change] << mutation
-        uniprot_change
-      end.compact
-
-      return [nil] * self.length if list.empty?
-
-      tsv = MutationAssessor.chunked_predict(list.sort_by{|p| p * "_"})
-
-      return [nil] * self.length if tsv.empty?
+  property :sift_scores => :array2single do
+    begin
+      missense = self.select{|mutation| mutation.consequence == "MISS-SENSE"}
+      res = MutEval.job(:sift, "MutatedIsoforms (#{self.length})", :mutations => missense.sort, :organism => organism).run
+      res.values_at(*self).collect{|v| (v.nil? or v["SIFT Score"].nil? or v["SIFT Score"].empty?) ? nil : 1.0 - v["SIFT Score"].to_f}
+    rescue
+      Log.warn $!.message
+      [nil] * self.length
+    end
+  end
 
-      new = TSV.setup({}, :key_field => "Mutated Isoform", :fields => ["Func. Impact"], :type => :list)
+  property :transFIC_scores => :array2single do |*args|
+    method = args.first || :mutation_assessor
+    range = {nil => nil,
+      ""  => nil,
+      "low_impact" => 0,
+      "medium_impact" => 0.7,
+      "high_impact" => 1.0}
+    field_names = {
+    }
+    begin
+      missense = self.select{|mutation| mutation.consequence == "MISS-SENSE"}
 
-      tsv.each do |key, values|
-        uniprot, change = key.split(" ")
-        uniprot_change = [uniprot.upcase, change.upcase]
-        
-        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
+      field_name = {
+        :mutation_assessor => "maTransfic",
+      }[method.to_sym]
 
+      MutEval.job(:transFIC, "MutatedIsoforms (#{self.length})", :mutations => missense.sort, :organism => organism).run.values_at(*self).collect{|v| (v.nil? or v[field_name].nil? or v[field_name].empty?) ? nil : v[field_name].to_f}
+    rescue
+      Log.warn $!.message
+      [nil] * self.length
+    end
+  end
 
-      range = {nil => nil,
-        ""  => nil,
-        "neutral" => 0,
-        "low" => 0.5,
-        "medium" => 0.7,
-        "high" => 1.0}
+  property :mutation_assessor_scores => :array2single do
+    range = {nil => nil,
+      ""  => nil,
+      "neutral" => 0,
+      "low" => 0.5,
+      "medium" => 0.7,
+      "high" => 1.0}
 
-      range.values_at *new.values_at(*self)
+    begin
+      missense = self.select{|mutation| mutation.consequence == "MISS-SENSE"}
+      MutEval.job(:mutation_assessor, "MutatedIsoforms (#{self.length})", :mutations => missense.sort, :organism => organism).run.values_at(*self).collect{|v| (v.nil? or v["Mutation Assessor Prediction"].nil? or v["Mutation Assessor Prediction"].empty?) ? nil : range[v["Mutation Assessor Prediction"]]}
+    rescue
+      Log.warn $!.message
+      [nil] * self.length
     end
   end
-  persist :mutation_assessor_scores
 
   property :pdbs => :single do
     uniprot = self.transcript.protein.uniprot
     next if uniprot.nil?
     UniProt.pdbs_covering_aa_position(uniprot, self.position)
   end
-  persist :pdbs
 
   property :pdbs_and_positions => :single do
     pdbs.collect do |pdb, info|

data/lib/rbbt/entity/pmid.rb

@@ -1,13 +1,21 @@
 require 'rbbt/entity'
 require 'rbbt/entity/document'
 require 'rbbt/sources/pubmed'
+require 'rbbt/sources/gscholar'
 
 module PMID
   extend Entity
   include Document
 
+  self.annotation :default_type
+
   self.format = "PMID"
 
+  property :docid => :single do |*args|
+    type = args.first || default_type
+    ["PMID", self, type].compact * ":"
+  end
+
   property :article => :array2single do
     PubMed.get_article(self).values_at(*self)
   end
@@ -15,25 +23,44 @@ module PMID
   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 :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}
+  property :cites => :single2array do
+    if title
+      begin
+        GoogleScholar.number_cites(title)
+      rescue
+        nil
+      end
+    else
+      nil
+    end
+  end
+
+  property :_get_text => :array2single do |*args|
+    type = args.first || default_type
+
+    case type.to_s
+    when "full_text", 'fulltext'
+      article.collect{|a| a.nil? ? nil : a.full_text}
+    when "abstract"
+      article.collect{|a| a.nil? ? nil : a.abstract }
+    when "best"
+      article.collect{|a| a.nil? ? nil : (a.full_text || a.text) }
+    else
+      article.collect{|a| a.nil? ? nil : a.text}
+    end
   end
 
   property :pubmed_url => :single2array do

data/lib/rbbt/entity/protein.rb

@@ -19,7 +19,7 @@ module Protein
 
   def self.ensp2sequence(organism, protein)
     @@ensp2sequence ||= {}
-    @@ensp2sequence[organism] ||= Organism.protein_sequence(organism).tsv :persist => true
+    @@ensp2sequence[organism] ||= Organism.protein_sequence(organism).tsv :persist => true, :unnamed => true
     if Array === protein
       @@ensp2sequence[organism].values_at *protein
     else
@@ -29,7 +29,7 @@ module Protein
 
   def self.ensp2enst(organism, protein)
     @@ensp2enst ||= {}
-    @@ensp2enst[organism] ||= Organism.transcripts(organism).tsv(:type => :single, :key_field => "Ensembl Protein ID", :fields => ["Ensembl Transcript ID"], :persist => true)
+    @@ensp2enst[organism] ||= Organism.transcripts(organism).tsv(:type => :single, :key_field => "Ensembl Protein ID", :fields => ["Ensembl Transcript ID"], :persist => true, :unnamed => true)
     @@ensp2enst[organism][protein]
   end
 
@@ -51,6 +51,7 @@ module Protein
   persist :transcript
 
   property :ensembl_protein_image_url => :single2array do
+    organism = self.organism || "Hsa"
     ensembl_url = if organism == "Hsa" then "www.ensembl.org" else "#{organism.sub(/.*\//,'')}.archive.ensembl.org" end
     "http://#{ensembl_url}/Homo_sapiens/Component/Transcript/Web/TranslationImage?db=core;p=#{ensembl};_rmd=d2a8;export=svg"
   end
@@ -67,13 +68,12 @@ module Protein
   end
 
   property :gene => :array do
-    Gene.setup(to("Ensembl Protein ID").clean_annotations, "Ensembl Protein ID", organism)
+    Gene.setup(to("Ensembl Protein ID").clean_annotations, "Ensembl Protein ID", organism).ensembl
   end
   persist :gene #, :yaml, :file => '/tmp/testes'
 
   property :pfam => :array2single do
-    index = Organism.gene_pfam(organism).tsv :flat, :persist => true
-    index.unnamed = true
+    index = Organism.gene_pfam(organism).tsv :flat, :persist => true, :unnamed => true
     pfam = index.values_at(*self).flatten
     Pfam.setup pfam
   end
@@ -82,12 +82,41 @@ module Protein
   property :sequence => :array2single do
     Protein.ensp2sequence(organism, self.ensembl)
   end
-  persist :sequence
+  persist :_ary_sequence
 
   property :sequence_length => :array2single do
     sequence.collect{|seq| seq.nil? ? nil : seq.length}
   end
-  persist :sequence_length
+  persist :_ary_sequence_length
+
+  property :marked_svg => :single2array do |*args|
+    positions = args.first
+    svg = Open.read(ensembl_protein_image_url)
+    
+    seq_len = sequence_length
+
+    doc = Nokogiri::XML(svg)
+    return nil unless doc.css('svg').any?
+    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
+
+    positions.each do |position|
+      if width and height and start and seq_len and position
+        offset = (width - start)/seq_len * position + start + rand * 10
+        svg = svg.sub(/<\/svg>/,"<rect x='#{offset}' y='1' width='1' height='#{height}' style='fill:rgb(255,0,0);opacity:0.5;stroke:none;'></svg>")
+      end
+    end
+
+    svg
+  end
+
+  property :pdbs => :single do
+    next if uniprot.nil?
+    UniProt.pdbs(uniprot)
+  end
+  persist :pdbs
+
 
 end
 

data/lib/rbbt/entity/transcript.rb

@@ -10,7 +10,7 @@ module Transcript
 
   def self.enst2ensg(organism, transcript)
     @@enst2ensg ||= {}
-    @@enst2ensg[organism] ||= Organism.gene_transcripts(organism).tsv(:type => :single, :key_field => "Ensembl Transcript ID", :fields => ["Ensembl Gene ID"], :persist => true).tap{|o| o.unnamed = true}
+    @@enst2ensg[organism] ||= Organism.gene_transcripts(organism).tsv(:type => :single, :key_field => "Ensembl Transcript ID", :fields => ["Ensembl Gene ID"], :persist => true, :unnamed => true)
     res = if Array === transcript
             @@enst2ensg[organism].values_at *transcript
           else
@@ -22,7 +22,7 @@ module Transcript
 
   def self.enst2ensp(organism, transcript)
     @@enst2ensp ||= {}
-    @@enst2ensp[organism] ||= Organism.transcripts(organism).tsv(:type => :single, :key_field => "Ensembl Transcript ID", :fields => ["Ensembl Protein ID"], :persist => true)
+    @@enst2ensp[organism] ||= Organism.transcripts(organism).tsv(:type => :single, :key_field => "Ensembl Transcript ID", :fields => ["Ensembl Protein ID"], :persist => true, :unnamed => true)
     res = if Array === transcript
             @@enst2ensp[organism].values_at *transcript
           else
@@ -31,6 +31,18 @@ module Transcript
     Protein.setup(res, "Ensembl Protein ID", organism)
   end
 
+  def self.enst2ense(organism, transcript)
+    @@enst2ense ||= {}
+    @@enst2ense[organism] ||= Organism.transcript_exons(organism).tsv(:persist => true, :fields => "Ensembl Exon ID", :unnamed => true)
+    res = if Array === transcript
+            @@enst2ense[organism].chunked_values_at transcript
+          else
+            @@enst2ense[organism][transcript]
+          end
+    res
+  end
+
+
 
   property :to! => :array2single do |new_format|
     return self if format == new_format
@@ -44,13 +56,17 @@ module Transcript
     to!(new_format).collect!{|v| v.nil? ? nil : v.first}
   end
 
+  property :exons => :array2single do 
+    Transcript.enst2ense(organism, self)
+  end
+  persist :_ary_exons
+
   property :ensembl => :array2single do
     to "Ensembl Transcript ID"
   end
 
   property :sequence => :array2single do
-    transcript_sequence = Organism.transcript_sequence(organism).tsv :persist => true
-    transcript_sequence.unnamed = true
+    transcript_sequence = Organism.transcript_sequence(organism).tsv :persist => true, :unnamed => true
     transcript_sequence.values_at *self.ensembl
   end