bacterial-annotator 0.1.7 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 2620736bb062558a4f44ee488f0c670945f42c85
-  data.tar.gz: 0ebfcfe4f935415818ade300927148a6cdcccff1
+  metadata.gz: 4dbc3b5e016a3f24ba82ab525a0c8daa9617afec
+  data.tar.gz: 7c8ba281632d84d7131cf1a27e143f596e76f414
 SHA512:
-  metadata.gz: ea81aa3caad269501af23e31b5d6fa9157b1a291da6467b31b74faaf207038255c5f42cd50f8c2c560304c84754af00bbdaf710120c0a8c9c03b9f029134fb18
-  data.tar.gz: a89e07dc87b7c219f664be43fb17154ae7c40d19ffc23fdd84ae0ff89dc3d66867ba8cd8121c01b7c2a4b8fa14db7f2fbf65f03ac04e2c502bca180a31babd1d
+  metadata.gz: 73f72513dc9964c46c73ef39477a9b5148b010d68449ff270f5b15f556a5e56e027a9ddb80149b2ce44e12c09466ac11267c118c97c8ea8892eb582d220588cf
+  data.tar.gz: 0a92c0ceec70a12500a8dcf4433d76eff7edf2e4fe7ca180151efc016ee81a3e7a6279cc633a6adaaec5936425908c3a4299742ae1571bc1970017a058a8dedb

data/bin/bacterial-annotator

@@ -64,7 +64,7 @@ def parseOptions
   options[:pidentity] = 70
   options[:minlength] = 500
   options[:meta] = 0
-  
+
   while x = ARGV.shift
 
     case x.downcase
@@ -113,7 +113,7 @@ if ARGV.size > 1
     abort "exiting blat is missing"
   end
 
-  # Check Options 
+  # Check Options
   if ! options.has_key? :refgenome and ! options.has_key? :remote_db and ! options.has_key? :external_db
     puts "You didn't provide a reference genome or a database for the annotation !"
   elsif ! options.has_key? :input

data/lib/bacterial-annotator.rb

@@ -106,7 +106,46 @@ class BacterialAnnotator
       # dump foreign proteins to file
       foreign_cds_file = dump_cds
 
-    else
+      # 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
+
+    else                        # no reference genome
 
       # no reference genome .. will process all the CDS
       foreign_cds_file = @fasta.prodigal_files[:proteins]
@@ -122,12 +161,14 @@ class BacterialAnnotator
     puts "\nPrinting Statistics.."
     print_stats "#{@outdir}/Annotation-Stats.txt"
 
+
   end                           # end of method
 
 
   # Finishing the annotation of the remaining CDS
   def finish_annotation remaining_cds_file
 
+    # only finish the annotation with an external DB
     if @options.has_key? :external_db	# from an external DB
 
       db_file = @options[:external_db]

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

@@ -20,6 +20,7 @@ class FastaManip
     @meta = meta
     @prodigal_files = nil
     @single_fasta = nil
+    @seq_info = nil
 
     if @fasta_flat.dbclass != Bio::FastaFormat
       abort "Aborting : The input sequence is not a fasta file !"
@@ -61,7 +62,7 @@ class FastaManip
       file_name = seq.definition.chomp.split(" ")[0]
       @prodigal_files[:contigs] << "#{file_name}"
       @prodigal_files[:contigs_length] << seq.seq.length
-      File.open("#{outdir}/single-fasta/#{file_name}.fasta", "w") do |fwrite| 
+      File.open("#{outdir}/single-fasta/#{file_name}.fasta", "w") do |fwrite|
         fwrite.write(seq)
       end
       @single_fasta[file_name] = seq
@@ -132,17 +133,19 @@ class FastaManip
       end
 
     end
-    
+
     return outseq, sequence.length
 
   end
 
 
-  # extract protein and gene names
+  # extract protein and gene names from prodigal... with contig numbering
   def extract_cds_names
 
     prot_ids = {}
+    prot_length = {}
     flatfile = Bio::FlatFile.auto(@prodigal_files[:proteins])
+
     flatfile.each_entry do |entry|
       prot_id = entry.definition.split(" ")[0]
       contig = prot_id.split("_")[0..-2].join("_")
@@ -150,6 +153,10 @@ class FastaManip
         prot_ids[contig] = []
       end
       prot_ids[contig] << prot_id
+
+      # puts "Prodigal length : " + entry.seq.length.to_s
+      prot_length[prot_id] = entry.seq.length-1 # minus the stop codon
+
     end
 
     prot_ids.each do |k,prot_array|
@@ -157,10 +164,10 @@ class FastaManip
     end
 
     @prodigal_files[:prot_ids_by_contig] = prot_ids
+    @prodigal_files[:prot_ids_length] = prot_length
 
   end
 
-
   private :extract_cds_names # :split_fasta, :split_genbank
 
 end

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

@@ -25,7 +25,7 @@ class GenbankManip
     flat_gbk = Bio::FlatFile.auto(@gbk_file)
 
     # Check if gbk is valid
-    if flat_gbk.dbclass != Bio::GenBank    
+    if flat_gbk.dbclass != Bio::GenBank
       abort "Aborting : The input #{@gbk_file} is not a valid genbank file !"
     else
       @gbk = flat_gbk.next_entry
@@ -67,10 +67,11 @@ class GenbankManip
         pepBioSeq = Bio::Sequence.auto(pep)
 
         if protId.strip == ""
-          protId = locustag          
+          protId = locustag
         end
 
-        @coding_seq[protId] = {location: loc,
+        @coding_seq[protId] = {protId: protId,
+                               location: loc,
                                locustag: locustag,
                                gene: gene[0],
                                product: product[0],
@@ -113,7 +114,7 @@ class GenbankManip
 
     contig = @gbk.definition
 
-    # iterate through 
+    # iterate through
     @gbk.features.each_with_index do |cds, ft_index|
 
       next if cds.feature != "CDS"
@@ -150,7 +151,7 @@ class GenbankManip
 
         # 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', "correspond to #{locus} locus (#{pId}% identity) from #{reference_locus.entry_id}")
+          qNote = Bio::Feature::Qualifier.new('note', "corresponds to #{locus} locus (#{pId}% identity) from #{reference_locus.entry_id}")
           ftArray.push(qNote)
         end
 
@@ -180,7 +181,7 @@ class GenbankManip
   ###################
   # Private Methods #
   ###################
-  
+
   # Fct: Get dna sequence
   def get_DNA (cds, seq)
     loc = cds.locations
@@ -208,5 +209,5 @@ class GenbankManip
 
   private :fetch_ncbi_genome, :get_DNA
 
-  
+
 end                             # end of Class

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

@@ -83,7 +83,7 @@ class RemoteNCBI
           end
 
         end
-       
+
       rescue
         try += 1
         puts "#{try} POST try for #{f}"

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

@@ -24,7 +24,7 @@ class SyntenyManip
   def run_blat root, outdir
     system("#{root}/blat.linux -out=blast8 -minIdentity=#{@pidentity} -prot #{@subject_file} #{@query_file} #{outdir}/#{@name}.blat8.tsv")
     @aln_file = "#{outdir}/#{@name}.blat8.tsv"
-    # extract_hits 
+    # extract_hits
   end                           # end of method
 
   # Extract Hit from blast8 file and save it in hash
@@ -45,21 +45,28 @@ class SyntenyManip
           next if lA[2].to_f < @pidentity
           @aln_hits[key] = {
             pId: lA[2].to_f.round(2),
-            length: lA[3].to_i,
             evalue: lA[10],
             score: lA[11].to_f,
-            hits: [hit]
+            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]]
           }
         elsif lA[11].to_f > @aln_hits[key][:score]
           @aln_hits[key] = {
             pId: lA[2].to_f.round(2),
-            length: lA[3].to_i,
             evalue: lA[10],
             score: lA[11].to_f,
-            hits: [hit]
+            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]]
           }
         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]
         end
       end
     end
@@ -102,6 +109,7 @@ class SyntenyManip
         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

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: bacterial-annotator
 version: !ruby/object:Gem::Version
-  version: 0.1.7
+  version: 0.2.0
 platform: ruby
 authors:
 - Maxime Deraspe
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-10-27 00:00:00.000000000 Z
+date: 2016-09-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bio
@@ -70,7 +70,7 @@ files:
 - lib/bacterial-annotator/synteny-manip.rb
 homepage: http://rubygems.org/gems/bacterial-annotator
 licenses:
-- GPLv3
+- GPL-3.0
 metadata: {}
 post_install_message: 
 rdoc_options: []
@@ -88,9 +88,8 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.4.5
+rubygems_version: 2.5.1
 signing_key: 
 specification_version: 4
 summary: Bacterial Annotator
 test_files: []
-has_rdoc: