bacterial-annotator 0.9.2 → 0.9.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b90fb0d1c0bc3d82a10706a7d6d2c0554099b94ce8ec27f63d4ebcad0ec11f04
-  data.tar.gz: c98732d57ad75290b6f0584ee559de98f8676ed0d94e5f4e9886f152269bf75e
+  metadata.gz: 6873b45085524788e9ea2c82b29c5e3cfb6998c6cd7ff1f16d56dc64378c910e
+  data.tar.gz: 7b02f34f8ffef086ade18526db05b5fffdeabee339a4808ec09181a7cb108b53
 SHA512:
-  metadata.gz: eab03371ef5f9fbf75b694f58ba4108adb38f6316131a96efc38a0c938fed727f3583f80e3f97604103c5c25f0d271fa9bcb397670b16bfd1bb2b930ca47be83
-  data.tar.gz: 6e48a1c67bd2add06ca3772bab0f734050a28476e05a9879f23fbf6df1cd88a314aee21c469a8226623978fefcd1b0d62cd240960b4097c55802395a891d5ae9
+  metadata.gz: 4b01f90fe7bc53e79a4245fbd64830a8fa967447e4a609715feef1ec21a68583531b9d1120631234204215bfa3bf6cfdd0b311704f030cf4cdf08d670fcabcf8
+  data.tar.gz: 15f9a395c37249bf937052d1917c89a42122a720bb1beddd1c4b2bddef3883111cb7e2e883da7757ccd9170eb3675499bbb70ac8dcfc62d4e722b9434549c92c

data/bin/ba_diamond

@@ -23,7 +23,6 @@ def installDiamond
     Dir.chdir("#{ROOT_path}/")
     `tar xvf diamond-linux64.tar.gz; rm diamond-linux64.tar.gz`
     `mv diamond ./diamond.linux`
-    `rm diamond_manual.pdf`
     File.chmod(0755, "#{ROOT_path}/diamond.linux")
   rescue
     abort "Problem installing Diamond, aborting"

data/lib/bacterial-annotator/sequence-annotation.rb

@@ -3,7 +3,7 @@
 # email:	maximilien1er@gmail.com
 # date:    	15-02-24
 # version: 	0.0.1
-# licence:  	
+# licence:
 
 require 'json'
 require 'zlib'
@@ -104,7 +104,16 @@ class SequenceAnnotation
     if flat_gbk.dbclass != Bio::GenBank
       abort "Aborting : The input #{gbk_file} is not a valid genbank file !"
     else
-      @gbk = flat_gbk.next_entry
+
+      @gbk_all = []
+      @bioseq_all = []
+      flat_gbk.each_entry do |e|
+        @gbk_all << e
+        @bioseq_all << e.to_biosequence
+      end
+
+      @gbk = @gbk_all[0]
+
     end
 
     @bioseq = @gbk.to_biosequence
@@ -221,41 +230,45 @@ class SequenceAnnotation
 
     if @coding_seq.empty?
 
-      # Iterate over each CDS
-      @gbk.each_cds do |ft|
-        ftH = ft.to_hash
-        loc = ft.locations
-        gene = []
-        product = []
-        protId = ""
-        if ftH.has_key? "pseudo"
-          next
-        end
-        gene = ftH["gene"] if !ftH["gene"].nil?
-        product = ftH["product"] if !ftH["product"].nil?
-        protId = ftH["protein_id"][0] if !ftH["protein_id"].nil?
-        locustag = ftH["locus_tag"][0] if !ftH["locus_tag"].nil?
-
-        dna = get_DNA(ft,@bioseq)
-        pep = dna.translate
-        pepBioSeq = Bio::Sequence.auto(pep)
-        dnaBioSeq = Bio::Sequence.auto(dna)
-
-        if protId.strip == ""
-          protId = locustag
-        end
+      @gbk_all.each_with_index do |gb, gb_i|
+
+        # Iterate over each CDS
+        gb.each_cds do |ft|
+          ftH = ft.to_hash
+          loc = ft.locations
+          gene = []
+          product = []
+          protId = ""
+          if ftH.has_key? "pseudo"
+            next
+          end
 
-        @coding_seq[protId] = {
-          protId: protId,
-          location: loc,
-          locustag: locustag,
-          gene: gene[0],
-          product: product[0],
-          bioseq: pepBioSeq,
-          bioseq_gene: dnaBioSeq,
-          length: pepBioSeq.length
-        }
+          gene = ftH["gene"] if !ftH["gene"].nil?
+          product = ftH["product"] if !ftH["product"].nil?
+          protId = ftH["protein_id"][0] if !ftH["protein_id"].nil?
+          locustag = ftH["locus_tag"][0] if !ftH["locus_tag"].nil?
+
+          dna = get_DNA(ft,@bioseq_all[gb_i])
+          pep = dna.translate
+          pepBioSeq = Bio::Sequence.auto(pep)
+          dnaBioSeq = Bio::Sequence.auto(dna)
+
+          if protId.strip == ""
+            protId = locustag
+          end
 
+          @coding_seq[protId] = {
+            protId: protId,
+            location: loc,
+            locustag: locustag,
+            gene: gene[0],
+            product: product[0],
+            bioseq: pepBioSeq,
+            bioseq_gene: dnaBioSeq,
+            length: pepBioSeq.length
+          }
+
+        end
       end
 
     end
@@ -270,43 +283,45 @@ class SequenceAnnotation
     if @rna_seq.empty?
 
       @rna_seq = {}
-      @gbk.features do |ft|
 
-        next if ! ft.feature.to_s.include? "RNA"
+      @gbk_all.each_with_index do |gb, gb_i|
+        gb.features do |ft|
 
-        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 = ""
+          next if ! ft.feature.to_s.include? "RNA"
 
-        if !ftH["product"].nil?
-          product = ftH["product"][0]
-          # puts ftH["product"].join(",") + "---" + ftH["product"][0]
-        end
+          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 = ""
 
-        locustag = ftH["locus_tag"][0] if !ftH["locus_tag"].nil?
+          if !ftH["product"].nil?
+            product = ftH["product"][0]
+            # puts ftH["product"].join(",") + "---" + ftH["product"][0]
+          end
 
-        # 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)
+          locustag = ftH["locus_tag"][0] if !ftH["locus_tag"].nil?
 
-        @rna_seq[locustag] = {
-          type: ft.feature.to_s,
-          location: loc,
-          locustag: locustag,
-          product: product,
-          bioseq_gene: dnaBioSeq
-        }
+          # puts "#{@accession}\t#{seqBeg}\t#{seqEnd}\t#{strand}\t#{protId}\t#{locustag}\t#{gene[0]}\t#{product[0]}"
+          dna = get_DNA(ft,@bioseq_all[gb_i])
+          dnaBioSeq = Bio::Sequence.auto(dna)
 
-      end
+          @rna_seq[locustag] = {
+            type: ft.feature.to_s,
+            location: loc,
+            locustag: locustag,
+            product: product,
+            bioseq_gene: dnaBioSeq
+          }
 
+        end
+      end
     end
 
     @rna_seq
@@ -622,4 +637,3 @@ class SequenceAnnotation
 
 
 end                             # end of Class
-

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

@@ -1,10 +1,10 @@
 # -*- coding: utf-8 -*-
 # author:  	maxime déraspe
 # email:	maximilien1er@gmail.com
-# review:  	
+# review:
 # date:    	15-02-24
 # version: 	0.0.1
-# licence:  	
+# licence:
 
 require 'json'
 require 'zlib'
@@ -111,7 +111,7 @@ class SequenceSynteny
         system("#{@root}/diamond.linux makedb --db #{subject_file} --in #{subject_file} > /dev/null 2>&1")
         db_file = subject_file
       end
-      system("#{@root}/diamond.linux blastp --db #{db_file} -q #{query_file} -o #{@outdir}/#{@name}.diamond.tsv -f 6 > /dev/null 2>&1")
+      system("#{@root}/diamond.linux blastp --masking none --db #{db_file} -q #{query_file} -o #{@outdir}/#{@name}.diamond.tsv -f 6 > /dev/null 2>&1")
     else
       # system("#{@root}/glsearch36.linux -b 3 -E 1e-25 -m 8 #{@subject_file} #{@query_file} > #{@outdir}/#{@name}.fasta36.tsv")
     end

data/lib/bacterial-comparator.rb

@@ -45,6 +45,9 @@ class BacterialComparator
       min_pid = min_pid/100
     end
 
+    @min_cov = min_cov
+    @min_pid = min_pid
+
     @aln_opt = options[:align].downcase
     @run_phylo = 0
     if options[:phylogeny] == 1
@@ -592,8 +595,8 @@ class BacterialComparator
                                            query_prot_file,
                                            ref_prot_file,
                                            "Prot-Ref",
-                                           0.80,
-                                           0.80,
+                                           @min_cov,
+                                           @min_cov,
                                            "prot")
 
     print "# Running alignment with Reference Genome CDS (diamond).."
@@ -613,6 +616,7 @@ class BacterialComparator
       next if ! syn_val.has_key? :homology
       next if syn_val[:homology][:assert_cutoff].inject(:+) < 3
       next if ref_annotated.has_key? syn_val[:homology][:hits][0] and ref_annotated[syn_val[:homology][:hits][0]][:partial] == 0
+
       ref_annotated[syn_val[:homology][:hits][0]] = {
         key: prot,
         pId: syn_val[:homology][:pId],
@@ -622,6 +626,7 @@ class BacterialComparator
         length: syn_val[:homology][:length][0],
         partial: (syn_val[:partial] ? 1 : 0)
       }
+
       # ref_annotated[syn_val[:homology][:hits][0]] = {
       #   key: prot,
       #   pId: syn_val[:homology][:pId],
@@ -631,6 +636,7 @@ class BacterialComparator
       #   length: syn_val[:homology][:length][0],
       #   partial: (syn_val[:partial] ? 1 : 0)
       # }
+
     end
 
     # print ref_annotated
@@ -642,13 +648,21 @@ class BacterialComparator
       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]][:cov_subject]
-        query_length = query_lengths[ref_annotated[ref_v[:protId]][:key]]
-        coverage_query = ref_annotated[ref_v[:protId]][:cov_query]
-        pId = ref_annotated[ref_v[:protId]][:pId]
-        partial = ref_annotated[ref_v[:protId]][:partial]
+
+        if ref_annotated[ref_v[:protId]][:pId] >= @min_pid and
+          ref_annotated[ref_v[:protId]][:cov_query] >= @min_cov and
+          ref_annotated[ref_v[:protId]][:cov_subject] >= @min_cov
+
+          gene = ref_annotated[ref_v[:protId]][:key]
+          coverage_ref = ref_annotated[ref_v[:protId]][:cov_subject]
+          query_length = query_lengths[ref_annotated[ref_v[:protId]][:key]]
+          coverage_query = ref_annotated[ref_v[:protId]][:cov_query]
+          pId = ref_annotated[ref_v[:protId]][:pId]
+          partial = ref_annotated[ref_v[:protId]][:partial]
+        end
+
       end
 
       _locus_tag = ref_v[:locustag] || ""

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: bacterial-annotator
 version: !ruby/object:Gem::Version
-  version: 0.9.2
+  version: 0.9.4
 platform: ruby
 authors:
 - Maxime Deraspe
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2022-09-05 00:00:00.000000000 Z
+date: 2023-07-17 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bio