bacterial-annotator 0.3.7 → 0.3.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 0f8c89d68e4afea5b49e88e68159f573b8fe44c1
-  data.tar.gz: a0742812f0e42a3346b63c79f590df774fb5e16c
+  metadata.gz: b46c0c48f45550ea7deb1580acb72927435df345
+  data.tar.gz: b5511795149b832c27fc8867dc3904c9c96d9561
 SHA512:
-  metadata.gz: 0f8d59637356bc752bea3271d9f9d733eee83699928312becf7fbbf319ffc482e976c5dc7927f5ba300aa1b263b8909c63b8d17afb2e0c3d4a0b9497c5c49f80
-  data.tar.gz: 55857b474d7e88373d295f3bf8ea7d9d99d431b8192e8f24d63dec012a511725bad7f1e43edc6e7b1259688c9c84cc868d08c3add2b2e082efca085cf2636489
+  metadata.gz: ac790dd5249a74cf559912dc88ebaad614d3f934a43c297848e1bda2db146c682b11862489170ddc21683fed070509f505cdc6759530ab0e1d2f8c80d8951ac7
+  data.tar.gz: 0de96df50d2e3a1ad4274134969007b8efef2f1a63b24e7063c45f102c9020ba3e6c337de2214bc1a2f26ae1d33057d9ec7c6ef0ea182e23c77633b3cd8d5e0a

data/bin/ba_blat

@@ -1,7 +1,7 @@
 #!/usr/bin/env ruby
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.01

data/bin/ba_mafft

@@ -1,7 +1,7 @@
 #!/usr/bin/env ruby
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.01

data/bin/ba_prodigal

@@ -1,7 +1,7 @@
 #!/usr/bin/env ruby
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.01

data/bin/ba_raxml

@@ -1,7 +1,7 @@
 #!/usr/bin/env ruby
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.01

data/bin/bacterial-annotator

@@ -1,7 +1,7 @@
 #!/usr/bin/env ruby
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.01
@@ -64,7 +64,6 @@ annotate [OPTIONS]
     --minlength		Minimum contig length for annotation [default=500]
 
     --meta		Better for metagenome and plasmid annotations because of disparate codon usage [default=off]
-    --gff			Will also generate gff annotation files [off by default]
 
 OEM
 
@@ -92,8 +91,6 @@ def parseOptions_annotate
       options[:outdir] = ARGV.shift
     when "--force", "-f"
       options[:force] = 1
-    when "--gff"
-      options[:gff] = 1
     when "--minlength"
       options[:minlength] = ARGV.shift
     when "--pidentity"
@@ -192,7 +189,10 @@ def parseOptions_compare
 end
 
 
-# Main
+########
+# MAIN #
+########
+
 if ARGV.size > 1
 
   ROOT = File.dirname(__FILE__)

data/lib/bacterial-annotator/fasta-manip.rb

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.0.1

data/lib/bacterial-annotator/genbank-manip.rb

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.0.1
@@ -10,7 +10,7 @@
 
 class GenbankManip
 
-  attr_accessor :gbk, :coding_seq, :cds_file
+  attr_accessor :gbk, :coding_seq, :cds_file, :rna_file
 
   # Initialize then genbank file
   def initialize gbk_file, outdir
@@ -58,13 +58,6 @@ class GenbankManip
         protId = ftH["protein_id"][0] if !ftH["protein_id"].nil?
         locustag = ftH["locus_tag"][0] if !ftH["locus_tag"].nil?
 
-        # if ftH.has_key? "translation"
-        #   pep = ftH["translation"][0] if !ftH["translation"].nil?
-        # else
-        #   dna = get_DNA(ft,@bioseq)
-        #   pep = dna.translate
-        # end
-
         dna = get_DNA(ft,@bioseq)
         pep = dna.translate
         pepBioSeq = Bio::Sequence.auto(pep)
@@ -89,6 +82,49 @@ class GenbankManip
 
   end
 
+  # Prepare rRNA tRNA
+  def get_rna
+
+    if @rna_seq == nil
+
+      @rna_seq = {}
+      @gbk.features do |ft|
+
+        next if ! ft.feature.to_s.include? "RNA"
+
+        ftH = ft.to_hash
+        loc = ft.locations
+        # seqBeg = loc[0].from.to_s
+        # seqEnd = loc[0].to.to_s
+        # strand = loc[0].strand.to_s
+        if ftH.has_key? "pseudo"
+          next
+        end
+        # gene = ftH["gene"] if !ftH["gene"].nil?
+        # protId = ftH["protein_id"][0] if !ftH["protein_id"].nil?
+        product = ""
+        product = ftH["product"][0] if !ftH["product"].nil?
+        locustag = ftH["locus_tag"][0] if !ftH["locus_tag"].nil?
+
+        # puts "#{@accession}\t#{seqBeg}\t#{seqEnd}\t#{strand}\t#{protId}\t#{locustag}\t#{gene[0]}\t#{product[0]}"
+        dna = get_DNA(ft,@bioseq)
+        dnaBioSeq = Bio::Sequence.auto(dna)
+
+        @rna_seq[locustag] = {type: ft.feature.to_s,
+                              location: loc,
+                              locustag: locustag,
+                              product: product,
+                              bioseq_gene: dnaBioSeq}
+
+      end
+
+    end
+
+    @rna_seq
+
+  end
+
+
 
   # Print CDS to files
   # RETURN : cds_file path
@@ -116,79 +152,136 @@ class GenbankManip
 
   end
 
+  # Print RNA to files
+  # RETURN : rna_file path
+  def write_rna_to_file outdir
+
+    rna_file = "#{@gbk.accession}.rna"
+
+    if @rna_seq == nil
+      get_rna
+    end
+
+    File.open("#{outdir}/#{rna_file}", "w") do |fwrite|
+      @rna_seq.each_key do |k|
+        seqout_dna = @rna_seq[k][:bioseq_gene].output_fasta("#{k}|#{@rna_seq[k][:type]}|#{@rna_seq[k][:product]}",60)
+        fwrite.write(seqout_dna)
+      end
+    end
+
+    @rna_file = "#{outdir}/" + rna_file
+
+  end
+
 
   # add annotation to a genbank file produced by prodigal
-  def add_annotation annotations, outdir, mode, reference_locus
+  def add_annotations annotations, mode, reference_locus=nil
 
-    nb_of_added_ft = 0
+    # nb_of_added_ft = 0
     i = 0
 
     contig = @gbk.definition
 
-    # iterate through
-    @gbk.features.each_with_index do |cds, ft_index|
+    if mode == "inplace"
+
+      # iterate through
+      @gbk.features.each_with_index do |cds, ft_index|
 
-      next if cds.feature != "CDS"
+        next if cds.feature != "CDS"
 
-      if mode == 0
         ftArray = []
         cds.qualifiers = []
-      else
-        ftArray = cds.qualifiers
-      end
 
-      i += 1
-      prot_id = contig+"_"+i.to_s
-      hit = nil
-      hit = annotations[prot_id] if annotations.has_key? prot_id
-
-      if hit != nil
-        locus, gene, product, note = nil
-        locus = hit[:locustag]
-        gene = hit[:gene]
-        product = hit[:product]
-        note = hit[:note]
-        pId = hit[:pId]
-
-        if gene != nil
-          qGene = Bio::Feature::Qualifier.new('gene', gene)
-          ftArray.push(qGene)
-        end
+        i += 1
+        prot_id = contig+"_"+i.to_s
+        hit = nil
+        hit = annotations[prot_id] if annotations.has_key? prot_id
+
+        if hit != nil
+          locus, gene, product, note = nil
+          locus = hit[:locustag]
+          gene = hit[:gene]
+          product = hit[:product]
+          note = hit[:note]
+          pId = hit[:pId]
+
+          if gene != nil
+            qGene = Bio::Feature::Qualifier.new('gene', gene)
+            ftArray.push(qGene)
+          end
+
+          if product != nil
+            qProd = Bio::Feature::Qualifier.new('product', product)
+            ftArray.push(qProd)
+          end
+
+          # check if there is a reference genome.. reference_locus shouldn't be nil in that case
+          if locus != nil
+            qNote = Bio::Feature::Qualifier.new('note', "corresponds to #{locus} locus (#{pId}% identity) from #{reference_locus.entry_id}")
+            ftArray.push(qNote)
+          end
+
+          if note != nil
+            qNote = Bio::Feature::Qualifier.new('note', note)
+            ftArray.push(qNote)
+          end
 
-        if product != nil
-          qProd = Bio::Feature::Qualifier.new('product', product)
-          ftArray.push(qProd)
         end
+        cds.qualifiers = ftArray
 
-        # check if there is a reference genome.. reference_locus shouldn't be nil in that case
-        if locus != nil
-          qNote = Bio::Feature::Qualifier.new('note', "corresponds to #{locus} locus (#{pId}% identity) from #{reference_locus.entry_id}")
-          ftArray.push(qNote)
-        end
+      end
+
+
+    elsif mode == "new"
+
+      sorted_annotations = annotations.sort_by { |k, v| v[:query_location][0][0] }
+
+      new_features = {}
+      annotations_done = {}
+
+      @gbk.features.each_with_index do |ft, ft_index|
+
+        sorted_annotations.each do |k,v|
+
+          next if annotations_done.has_key? k
+
+          if v[:query_location][0][0] < ft.locations[0].from
+
+            if v[:subject_location][0][0] > v[:subject_location][0][1]
+              location = "complement(#{v[:query_location][0][0]}..#{v[:query_location][0][1]})"
+            else
+              location = "#{v[:query_location][0][0]}..#{v[:query_location][0][1]}"
+            end
+
+            feature = Bio::Feature.new(v[:feature][0],location)
+            feature.qualifiers.push(Bio::Feature::Qualifier.new('product',v[:product][0])) if ! v[:product][0].nil? or v[:product][0] != ""
+            new_features[ft_index] = feature
+            annotations_done[k] = 1
+            break
+
+          end
 
-        if note != nil
-          qNote = Bio::Feature::Qualifier.new('note', note)
-          ftArray.push(qNote)
         end
 
+      end
 
+      new_features.each do |k,v|
+        @gbk.features.insert(k,v)
       end
-      cds.qualifiers = ftArray
 
     end
 
-    File.open("#{outdir}/#{contig}.gbk", "w") do |f| 
+  end
+
+
+  def save_genbank_to_file outdir
+
+    File.open("#{outdir}/#{@gbk.definition}.gbk", "w") do |f|
       f.write(@gbk.to_biosequence.output(:genbank))
     end
 
-    # Bioruby doesn't support gff at this point
-    # File.open("#{outdir}/#{contig}.gff", "w") do |f| 
-    #   f.write(@gbk.to_biosequence.output(:gff))
-    # end
-
   end
 
-
   ###################
   # Private Methods #
   ###################

data/lib/bacterial-annotator/remote-ncbi.rb

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.0.1

data/lib/bacterial-annotator/synteny-manip.rb

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.0.1
@@ -12,34 +12,42 @@ class SyntenyManip
 
   attr_reader :query_file, :subject_file, :aln_hits
 
-  def initialize query_file, subject_file, name, pidentity
+  def initialize query_file, subject_file, name, pidentity, type
     @query_file = query_file
     @subject_file = subject_file
     @name = name
     @pidentity = pidentity
     @aln_file = nil
+    @type = type
   end                           # end of initialize
 
   # run blat on proteins
   def run_blat root, outdir
-    system("#{root}/blat.linux -out=blast8 -minIdentity=#{@pidentity} -prot #{@subject_file} #{@query_file} #{outdir}/#{@name}.blat8.tsv")
+    base_cmd = "#{root}/blat.linux -out=blast8 -minIdentity=#{@pidentity}"
+    system("#{base_cmd} #{@subject_file} #{@query_file} #{outdir}/#{@name}.blat8.tsv")
+    if @type == "prot"
+      system("#{base_cmd} -prot #{@subject_file} #{@query_file} #{outdir}/#{@name}.blat8.tsv")
+    end
     @aln_file = "#{outdir}/#{@name}.blat8.tsv"
     # extract_hits
   end                           # end of method
 
   # Extract Hit from blast8 file and save it in hash
   # contig-0_1      ABJ71957.1      96.92   65      2       0       1       65      1       65      9.2e-31 131.0
-  def extract_hits mode
+  def extract_hits_prodigal mode, ref_cds=nil
 
     @aln_hits = {}
+    feature = ""
     File.open(@aln_file,"r") do |fread|
       while l = fread.gets
         lA = l.chomp!.split("\t")
         key = lA[0]
         if mode == :refgenome
           hit = lA[1]
+          feature = "cds"
         elsif mode == :externaldb
-          hit = lA[1].chomp.split("|")[1]
+          hit = lA[1].chomp.split("|")[3]
+          feature = "cds"
         end
         if ! @aln_hits.has_key? key
           next if lA[2].to_f < @pidentity
@@ -50,7 +58,8 @@ class SyntenyManip
             hits: [hit],
             length: [lA[3].to_i],
             query_location: [[lA[6].to_i,lA[7].to_i]],
-            subject_location: [[lA[8].to_i,lA[9].to_i]]
+            subject_location: [[lA[8].to_i,lA[9].to_i]],
+            feature: feature
           }
         elsif lA[11].to_f > @aln_hits[key][:score]
           @aln_hits[key] = {
@@ -60,7 +69,8 @@ class SyntenyManip
             hits: [hit],
             length: [lA[3].to_i],
             query_location: [[lA[6].to_i,lA[7].to_i]],
-            subject_location: [[lA[8].to_i,lA[9].to_i]]
+            subject_location: [[lA[8].to_i,lA[9].to_i]],
+            feature: feature
           }
         elsif lA[11].to_f == @aln_hits[key][:score]
           @aln_hits[key][:hits] << hit
@@ -73,49 +83,116 @@ class SyntenyManip
 
   end                           # end of method
 
+  # Extract Hit from blast8 file and save it in hash
+  # prpa    PA0668.4|rRNA|23S       99.97   2891    1       0       705042  707932  1       2891    0.0e+00 5671.0
+  def extract_hits_dna mode
 
+    @aln_hits = {}
+    feature = ""
+    File.open(@aln_file,"r") do |fread|
+      while l = fread.gets
+        lA = l.chomp!.split("\t")
+        key = lA[0]+"_"+lA[6]+"_"+lA[7]
+        if mode == :rna
+          hit_split = lA[1].chomp.split("|")
+          hit = hit_split[0]
+          feature = hit_split[1]
+          product = hit_split[2]
+        end
+        if ! @aln_hits.has_key? key
+          next if lA[2].to_f < @pidentity
+          @aln_hits[key] = {
+            pId: lA[2].to_f.round(2),
+            evalue: lA[10],
+            score: lA[11].to_f,
+            hits: [hit],
+            product: [product],
+            length: [lA[3].to_i],
+            query_location: [[lA[6].to_i,lA[7].to_i]],
+            subject_location: [[lA[8].to_i,lA[9].to_i]],
+            feature: [feature]
+          }
+        elsif lA[11].to_f > @aln_hits[key][:score]
+          @aln_hits[key] = {
+            pId: lA[2].to_f.round(2),
+            evalue: lA[10],
+            score: lA[11].to_f,
+            hits: [hit],
+            product: [product],
+            length: [lA[3].to_i],
+            query_location: [[lA[6].to_i,lA[7].to_i]],
+            subject_location: [[lA[8].to_i,lA[9].to_i]],
+            feature: [feature]
+          }
+        elsif lA[11].to_f == @aln_hits[key][:score]
+          @aln_hits[key][:hits] << hit
+          @aln_hits[key][:length] << lA[3].to_i
+          @aln_hits[key][:query_location] << [lA[6].to_i,lA[7].to_i]
+          @aln_hits[key][:subject_location] << [lA[8].to_i,lA[9].to_i]
+          @aln_hits[key][:feature] << feature
+          @aln_hits[key][:product] << product
+        end
+      end
+    end
 
-  # Get the annotations for a contig for RerenceGenome
-  def get_annotation_for_contig prots_to_annotate, ref_cds
+    prune_aln_hits @aln_hits
+
+  end                           # end of method
 
-    return {} if prots_to_annotate == nil
 
-    contig_to_annotate = prots_to_annotate[0].split("_")[0..-2].join("_")
+  # Get the annotations for a contig for RerenceGenome
+  def get_annotation_for_contig contig_to_annotate, prots_to_annotate=nil, ref_cds=nil
+
     annotations = {}
-    prots = []
 
-    @aln_hits.each_key do |k|
-      contig = k.split("_")[0..-2].join("_")
-      if contig == contig_to_annotate
-        prots << k
+    if prots_to_annotate != nil
+
+      # contig_to_annotate = prots_to_annotate[0].split("_")[0..-2].join("_")
+      prots = []
+
+      @aln_hits.each_key do |k|
+        contig = k.split("_")[0..-2].join("_")
+        if contig == contig_to_annotate
+          prots << k
+        end
       end
-    end
 
-    # sorting the prot by their appearance in the contig
-    prots.sort! { |a,b| a.split("_")[-1].to_i <=> b.split("_")[-1].to_i }
+      # sorting the prot by their appearance in the contig
+      prots.sort! { |a,b| a.split("_")[-1].to_i <=> b.split("_")[-1].to_i }
 
-    i = 0
-    prots_to_annotate.each do |p|
+      i = 0
+      prots_to_annotate.each do |p|
 
-      if @aln_hits.has_key? p
+        if @aln_hits.has_key? p
 
-        hit_index = 0
+          hit_index = 0
 
-        if @aln_hits[p][:hits].length > 1
-          hit_index = choose_best_hit i, prots, ref_cds
-        end
+          if @aln_hits[p][:hits].length > 1
+            hit_index = choose_best_hit i, prots, ref_cds
+          end
 
-        h = @aln_hits[p][:hits][hit_index]
-        hit = ref_cds[h]
-        annotations[p] = hit
-        annotations[p][:pId] = @aln_hits[p][:pId]
-        annotations[p][:length] = @aln_hits[p][:length][hit_index]
-        i+=1
+          h = @aln_hits[p][:hits][hit_index]
+          hit = ref_cds[h]
+          annotations[p] = hit
+          annotations[p][:pId] = @aln_hits[p][:pId]
+          annotations[p][:length] = @aln_hits[p][:length][hit_index]
+          i+=1
 
-      else
+        else
+
+          annotations[p] = nil
+
+        end
 
-        annotations[p] = nil
+      end
+
+    elsif ! @aln_hits.empty?
 
+      @aln_hits.each_key do |k|
+        contig = k.split("_")[0..-3].join("_")
+        if contig == contig_to_annotate
+          annotations[k] = @aln_hits[k]
+        end
       end
 
     end
@@ -192,6 +269,64 @@ class SyntenyManip
 
   end                           # end of method
 
+  def prune_aln_hits aln_hits
+
+    # @aln_hits[key] = {
+    #   pId: lA[2].to_f.round(2),
+    #   evalue: lA[10],
+    #   score: lA[11].to_f,
+    #   hits: [hit],
+    #   length: [lA[3].to_i],
+    #   query_location: [[lA[6].to_i,lA[7].to_i]],
+    #   subject_location: [[lA[8].to_i,lA[9].to_i]],
+    #   feature: [feature]
+    # }
+
+    keys_to_delete = []
+
+    aln_hits.each do |key1,val1|
+
+      aln_hits.each do |key2,val2|
+
+        next if key1==key2
+        next if keys_to_delete.include? key1
+        next if keys_to_delete.include? key2
+
+        if val1[:query_location][0][0] >= val2[:query_location][0][0] and
+          val1[:query_location][0][0] < val2[:query_location][0][1]
+          overlap_len = val2[:query_location][0][1] - val1[:query_location][0][0]
+          val1_len = val1[:query_location][0][1]-val1[:query_location][0][0]
+          val2_len = val2[:query_location][0][1]-val2[:query_location][0][0]
+          if overlap_len.to_f/val1_len > 0.2 and overlap_len.to_f/val2_len > 0.2
+            if val1[:score] < val2[:score]
+              keys_to_delete << key1
+            else
+              keys_to_delete << key2
+            end
+          end
+        elsif val2[:query_location][0][0] >= val1[:query_location][0][0] and
+             val2[:query_location][0][0] < val1[:query_location][0][1]
+          overlap_len = val1[:query_location][0][1] - val2[:query_location][0][0]
+          val1_len = val1[:query_location][0][1]-val1[:query_location][0][0]
+          val2_len = val2[:query_location][0][1]-val2[:query_location][0][0]
+          if overlap_len.to_f/val1_len > 0.2 and overlap_len.to_f/val2_len > 0.2
+            if val1[:score] < val2[:score]
+              keys_to_delete << key1
+            else
+              keys_to_delete << key2
+            end
+          end
+        end
+
+      end
+
+    end
+
+    keys_to_delete.each do |k|
+      aln_hits.delete(k)
+    end
+
+  end                           # end of method
 
 
 end                             # end of class

data/lib/bacterial-annotator.rb

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.0.1
@@ -33,7 +33,7 @@ class BacterialAnnotator
         abort "Output directory already exist ! Choose another one or use -f to overwrite"
       else
         puts "Overwriting output directory #{@outdir}"
-        FileUtils.remove_dir(@outdir, force=true)
+        FileUtils.remove_dir(@outdir, :force=>true)
       end
     end
     Dir.mkdir(@outdir)
@@ -67,6 +67,7 @@ class BacterialAnnotator
     puts "Prodigal done."
     if @with_refence_genome
       @refgenome.write_cds_to_file @outdir
+      @refgenome.write_rna_to_file @outdir
       puts "Successfully loaded #{@refgenome.gbk.definition}"
     end
   end                           # end of method
@@ -77,10 +78,11 @@ class BacterialAnnotator
     # process reference genome synteny
     if @with_refence_genome        # Annotation with the Reference Genome
 
-      @prot_synteny = SyntenyManip.new(@fasta.prodigal_files[:proteins], @refgenome.cds_file, "Prot-Ref", @pidentity)
-      puts "\nRunning BLAT alignment with Reference Genome.."
+      # run CDS annotation
+      puts "\nRunning BLAT alignment with Reference Genome CDS.."
+      @prot_synteny = SyntenyManip.new(@fasta.prodigal_files[:proteins], @refgenome.cds_file, "Prot-Ref", @pidentity, "prot")
       @prot_synteny.run_blat @root, @outdir
-      @prot_synteny.extract_hits :refgenome
+      @prot_synteny.extract_hits_prodigal :refgenome
 
       @fasta.prodigal_files[:contigs].each_with_index do |contig, contig_index|
 
@@ -91,9 +93,9 @@ class BacterialAnnotator
         end
 
         contig_prots = @fasta.prodigal_files[:prot_ids_by_contig][contig]
-
+        # contig_to_annotate = contig_prots[0].split("_")[0..-2].join("_")
         # contig_prot_annotations = @prot_synteny.get_annotation_for_contig contig_prots, @refgenome.coding_seq
-        @contig_annotations[contig] = @prot_synteny.get_annotation_for_contig contig_prots, @refgenome.coding_seq
+        @contig_annotations[contig] = @prot_synteny.get_annotation_for_contig contig, contig_prots, @refgenome.coding_seq
 
         remaining_cds = cumulate_annotation_stats_reference contig, @contig_annotations[contig]
 
@@ -106,44 +108,20 @@ class BacterialAnnotator
       # dump foreign proteins to file
       foreign_cds_file = dump_cds
 
-      # Iterate over each Ref protein and print syntheny
-      synteny_file = File.open("#{@outdir}/Prot-Synteny.tsv","w")
-      synteny_file.write("RefLocusTag\tRefProtID\tRefLength\tRefCoverage\tIdentity\tQueryGene\tQueryLength\tQueryCoverage\n")
-      ref_annotated = {}
-      @contig_annotations.each do |contig,prot_annotations|
-        prot_annotations.each do |key,prot|
-          # p key
-          # p prot
-          ref_annotated[prot[:protId]] = {key: key, length: prot[:length], pId: prot[:pId]} if prot != nil
-        end
-      end
-
-      @refgenome.coding_seq.each do |ref_k, ref_v|
-        gene = ""
-        coverage_ref = ""
-        coverage_query = ""
-        query_length = ""
-        pId = ""
-        if ref_annotated[ref_v[:protId]] != nil
-          gene = ref_annotated[ref_v[:protId]][:key]
-          coverage_ref = (ref_annotated[ref_v[:protId]][:length].to_f/ref_v[:bioseq].seq.length.to_f).round(2)
-          query_length = @fasta.prodigal_files[:prot_ids_length][gene]
-          coverage_query = (ref_annotated[ref_v[:protId]][:length].to_f/query_length.to_f).round(2)
-          pId = ref_annotated[ref_v[:protId]][:pId]
-        end
-
-        synteny_file.write(ref_v[:protId])
-        synteny_file.write("\t"+ref_v[:locustag])
-        synteny_file.write("\t"+ref_v[:bioseq].seq.length.to_s)
-        synteny_file.write("\t"+coverage_ref.to_s)
-        synteny_file.write("\t"+pId.to_s)
-        synteny_file.write("\t"+gene)
-        synteny_file.write("\t"+query_length.to_s)
-        synteny_file.write("\t"+coverage_query.to_s)
-        synteny_file.write("\n")
+      # dump reference CDS synteny to file
+      dump_ref_synteny_to_file
 
+      # run RNA annotation
+      puts "\nRunning BLAT alignment with Reference Genome RNA.."
+      @rna_synteny = SyntenyManip.new(@fasta.fasta_file, @refgenome.rna_file, "RNA-Ref", @pidentity, "dna")
+      @rna_synteny.run_blat @root, @outdir
+      @rna_synteny.extract_hits_dna :rna
+      @contig_annotations_rna = {}
+      @fasta.prodigal_files[:contigs].each_with_index do |contig, contig_index|
+        puts "adding rna_annotation for contig #{contig}"
+        @contig_annotations_rna[contig] = @rna_synteny.get_annotation_for_contig contig
+        p @contig_annotations_rna[contig]
       end
-      synteny_file.close
 
     else                        # no reference genome
 
@@ -156,7 +134,7 @@ class BacterialAnnotator
     finish_annotation foreign_cds_file
 
     # Parse annotations to genbank files
-    parsing_genbank_files
+    parse_genbank_files
 
     puts "\nPrinting Statistics.."
     print_stats "#{@outdir}/Annotation-Stats.txt"
@@ -177,7 +155,7 @@ class BacterialAnnotator
       externaldb_synteny = SyntenyManip.new(remaining_cds_file, db_file, "Prot-ExternalDB", @pidentity)
       puts "\nRunning BLAT alignment with External Database.."
       externaldb_synteny.run_blat @root, @outdir
-      externaldb_synteny.extract_hits :externaldb
+      externaldb_synteny.extract_hits_prodigal :externaldb
 
       externaldb_synteny.aln_hits.each do |k,v|
         contig_of_protein = k.split("_")[0..-2].join("_")
@@ -188,15 +166,14 @@ class BacterialAnnotator
 
         hit_gi = v[:hits][0]
 
-        note = "Protein homology (#{v[:pId]}% identity) with gi:#{hit_gi}"
-
-        # p v
-        # p ref_cds[hit_gi]
+        # note = "Protein homology (#{v[:pId]}% identity) with gi:#{hit_gi}"
+        note = "Protein homology (#{v[:pId]}% identity) with #{hit_gi}"
 
         if ref_cds[hit_gi][:org] != ""
           note +=  " from #{ref_cds[hit_gi][:org]}"
         end
         @contig_annotations[contig_of_protein][k] = {product: ref_cds[hit_gi][:product],
+                                                     feature: "cds",
                                                      gene: nil,
                                                      locustag: nil,
                                                      note: note}
@@ -237,17 +214,16 @@ class BacterialAnnotator
           end
           ncbiblast.aln_hits.each do |k,v|
             contig_of_protein = k.split("_")[0..-2].join("_")
-            # @contig_annotations[contig_of_protein][k][:product] = v[:hits][0][:product]
             if ! @contig_annotations.has_key? contig_of_protein
               @contig_annotations[contig_of_protein] = {}
             end
-
-            note = "Protein homology (#{v[:pId]}% identity) with gi:#{v[:hits][0][:gi]}"
-            # note = "correspond to gi:#{v[:hits][0][:gi]}"
+            # note = "Protein homology (#{v[:pId]}% identity) with gi:#{v[:hits][0][:gi]}"
+            note = "Protein homology (#{v[:pId]}% identity) with gi:#{v[:hits][0][:accession]}"
             if v[:hits][0][:org] != ""
               note +=  " from #{v[:hits][0][:org]}"
             end
             @contig_annotations[contig_of_protein][k] = {product: v[:hits][0][:product],
+                                                         feature: "cds",
                                                          gene: nil,
                                                          locustag: nil,
                                                          note: note}
@@ -263,7 +239,7 @@ class BacterialAnnotator
 
 
   # parse all genbank files
-  def parsing_genbank_files
+  def parse_genbank_files
 
     puts "\nParsing annotation into genbank files.."
     @contig_annotations.each do |contig, contig_prot_annotations|
@@ -271,7 +247,15 @@ class BacterialAnnotator
       gbk_to_annotate = GenbankManip.new("#{gbk_path}/#{contig}.gbk", "#{gbk_path}")
       reference_locus = nil
       reference_locus = @refgenome.gbk.locus if @with_refence_genome
-      gbk_to_annotate.add_annotation contig_prot_annotations, gbk_path, 0, reference_locus
+      gbk_to_annotate.add_annotations contig_prot_annotations, "inplace", reference_locus
+
+      if @contig_annotations_rna.has_key? contig
+        puts "Trying RNA annotation"
+        gbk_to_annotate.add_annotations @contig_annotations_rna[contig], "new"
+      end
+
+      gbk_to_annotate.save_genbank_to_file gbk_path
+
     end
 
   end                           # end of method
@@ -314,6 +298,7 @@ class BacterialAnnotator
     p_cds_annotated = @annotation_stats[:annotated_cds].to_f/@annotation_stats[:total_cds].to_f
 
     File.open(file, "w") do |fopen|
+
       fopen.write("#Contigs annotation based on reference genomes\n")
       fopen.write("Short Contigs (< #{@minlength}) :\t\t" + @annotation_stats[:short_contigs].length.to_s + "\n")
       fopen.write("Foreign Contigs :\t\t" + @annotation_stats[:foreign_contigs].length.to_s + "\n")
@@ -446,6 +431,51 @@ class BacterialAnnotator
 
   end                           # end of method
 
-  private :dump_cds, :split_remaining_cds_file
+  # will reference CDS synteny to file
+  def dump_ref_synteny_to_file
+
+    # Iterate over each Ref protein and print syntheny
+    synteny_file = File.open("#{@outdir}/Prot-Synteny.tsv","w")
+    synteny_file.write("RefLocusTag\tRefProtID\tRefLength\tRefCoverage\tIdentity\tQueryGene\tQueryLength\tQueryCoverage\n")
+    ref_annotated = {}
+    @contig_annotations.each do |contig,prot_annotations|
+      prot_annotations.each do |key,prot|
+        # p key
+        # p prot
+        ref_annotated[prot[:protId]] = {key: key, length: prot[:length], pId: prot[:pId]} if prot != nil
+      end
+    end
+
+    @refgenome.coding_seq.each do |ref_k, ref_v|
+
+      gene = ""
+      coverage_ref = ""
+      coverage_query = ""
+      query_length = ""
+      pId = ""
+      if ref_annotated[ref_v[:protId]] != nil
+        gene = ref_annotated[ref_v[:protId]][:key]
+        coverage_ref = (ref_annotated[ref_v[:protId]][:length].to_f/ref_v[:bioseq].seq.length.to_f).round(2)
+        query_length = @fasta.prodigal_files[:prot_ids_length][gene]
+        coverage_query = (ref_annotated[ref_v[:protId]][:length].to_f/query_length.to_f).round(2)
+        pId = ref_annotated[ref_v[:protId]][:pId]
+      end
+
+      synteny_file.write(ref_v[:protId])
+      synteny_file.write("\t"+ref_v[:locustag])
+      synteny_file.write("\t"+ref_v[:bioseq].seq.length.to_s)
+      synteny_file.write("\t"+coverage_ref.to_s)
+      synteny_file.write("\t"+pId.to_s)
+      synteny_file.write("\t"+gene)
+      synteny_file.write("\t"+query_length.to_s)
+      synteny_file.write("\t"+coverage_query.to_s)
+      synteny_file.write("\n")
+
+    end
+    synteny_file.close
+
+  end
+
+  private :dump_cds, :split_remaining_cds_file, :dump_ref_synteny_to_file
 
 end                             # end of class

data/lib/bacterial-comparator.rb

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
-# email:	maxime@deraspe.net
+# email:	maximilien1er@gmail.com
 # review:  	
 # date:    	15-02-24
 # version: 	0.0.1

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: bacterial-annotator
 version: !ruby/object:Gem::Version
-  version: 0.3.7
+  version: 0.3.9
 platform: ruby
 authors:
 - Maxime Deraspe
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-01-25 00:00:00.000000000 Z
+date: 2017-02-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bio
@@ -72,7 +72,7 @@ dependencies:
         version: 1.9.0
 description: GEM to annotate bacterial genome sequence based on a reference genome
   and complete the annotation with an external database or a remote database.
-email: maxime@deraspe.net
+email: maximilien1er@gmail.com
 executables:
 - bacterial-annotator
 - ba_prodigal