bio-polyploid-tools 0.10.1 → 1.2.0

data/bio-polyploid-tools.gemspec

@@ -2,29 +2,28 @@
 # DO NOT EDIT THIS FILE DIRECTLY
 # Instead, edit Juwelier::Tasks in Rakefile, and run 'rake gemspec'
 # -*- encoding: utf-8 -*-
-# stub: bio-polyploid-tools 0.10.1 ruby lib
+# stub: bio-polyploid-tools 1.2.0 ruby lib
 
 Gem::Specification.new do |s|
   s.name = "bio-polyploid-tools".freeze
-  s.version = "0.10.1"
+  s.version = "1.2.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0".freeze) if s.respond_to? :required_rubygems_version=
   s.require_paths = ["lib".freeze]
   s.authors = ["Ricardo H.  Ramirez-Gonzalez".freeze]
-  s.date = "2019-03-28"
+  s.date = "2020-10-28"
   s.description = "Repository of tools developed at Crop Genetics in JIC to work with polyploid wheat".freeze
   s.email = "ricardo.ramirez-gonzalez@jic.ac.uk".freeze
-  s.executables = ["bfr.rb".freeze, "blast_triads.rb".freeze, "blast_triads_promoters.rb".freeze, "count_variations.rb".freeze, "filter_blat_by_target_coverage.rb".freeze, "filter_exonerate_by_identity.rb".freeze, "find_best_blat_hit.rb".freeze, "find_best_exonerate.rb".freeze, "find_homoeologue_variations.rb".freeze, "get_longest_hsp_blastx_triads.rb".freeze, "hexaploid_primers.rb".freeze, "homokaryot_primers.rb".freeze, "mafft_triads.rb".freeze, "mafft_triads_promoters.rb".freeze, "map_markers_to_contigs.rb".freeze, "marker_to_vcf.rb".freeze, "markers_in_region.rb".freeze, "mask_triads.rb".freeze, "polymarker.rb".freeze, "polymarker_capillary.rb".freeze, "snp_position_to_polymarker.rb".freeze, "snps_between_bams.rb".freeze, "tag_stats.rb".freeze, "vcfLineToTable.rb".freeze, "vcfToPolyMarker.rb".freeze]
+  s.executables = ["bfr.rb".freeze, "blast_triads.rb".freeze, "blast_triads_promoters.rb".freeze, "count_variations.rb".freeze, "filter_blat_by_target_coverage.rb".freeze, "filter_exonerate_by_identity.rb".freeze, "find_best_blat_hit.rb".freeze, "find_best_exonerate.rb".freeze, "get_longest_hsp_blastx_triads.rb".freeze, "hexaploid_primers.rb".freeze, "homokaryot_primers.rb".freeze, "mafft_triads.rb".freeze, "mafft_triads_promoters.rb".freeze, "map_markers_to_contigs.rb".freeze, "marker_to_vcf.rb".freeze, "markers_in_region.rb".freeze, "mask_triads.rb".freeze, "polymarker.rb".freeze, "polymarker_capillary.rb".freeze, "polymarker_deletions.rb".freeze, "snp_position_to_polymarker.rb".freeze, "snps_between_bams.rb".freeze, "tag_stats.rb".freeze, "vcfLineToTable.rb".freeze, "vcfToPolyMarker.rb".freeze]
   s.extra_rdoc_files = [
-    "README",
     "README.md"
   ]
   s.files = [
     ".travis.yml",
     "Gemfile",
-    "README",
     "README.md",
     "Rakefile",
+    "SECURITY.md",
     "VERSION",
     "bin/bfr.rb",
     "bin/blast_triads.rb",
@@ -34,7 +33,6 @@ Gem::Specification.new do |s|
     "bin/filter_exonerate_by_identity.rb",
     "bin/find_best_blat_hit.rb",
     "bin/find_best_exonerate.rb",
-    "bin/find_homoeologue_variations.rb",
     "bin/get_longest_hsp_blastx_triads.rb",
     "bin/hexaploid_primers.rb",
     "bin/homokaryot_primers.rb",
@@ -46,6 +44,7 @@ Gem::Specification.new do |s|
     "bin/mask_triads.rb",
     "bin/polymarker.rb",
     "bin/polymarker_capillary.rb",
+    "bin/polymarker_deletions.rb",
     "bin/snp_position_to_polymarker.rb",
     "bin/snps_between_bams.rb",
     "bin/tag_stats.rb",
@@ -102,6 +101,17 @@ Gem::Specification.new do |s|
     "lib/bio/db/exonerate.rb",
     "lib/bio/db/primer3.rb",
     "lib/bioruby-polyploid-tools.rb",
+    "test/data/7B_amplicon_test.fa",
+    "test/data/7B_amplicon_test.fa.fai",
+    "test/data/7B_amplicon_test_reference.fa",
+    "test/data/7B_amplicon_test_reference.fa.fai",
+    "test/data/7B_amplicon_test_reference.fa.ndb",
+    "test/data/7B_amplicon_test_reference.fa.nhr",
+    "test/data/7B_amplicon_test_reference.fa.nin",
+    "test/data/7B_amplicon_test_reference.fa.not",
+    "test/data/7B_amplicon_test_reference.fa.nsq",
+    "test/data/7B_amplicon_test_reference.fa.ntf",
+    "test/data/7B_amplicon_test_reference.fa.nto",
     "test/data/BS00068396_51.fa",
     "test/data/BS00068396_51_blast.tab",
     "test/data/BS00068396_51_contigs.aln",
@@ -185,29 +195,21 @@ Gem::Specification.new do |s|
   ]
   s.homepage = "http://github.com/tgac/bioruby-polyploid-tools".freeze
   s.licenses = ["MIT".freeze]
-  s.rubygems_version = "2.7.7".freeze
+  s.rubygems_version = "3.1.2".freeze
   s.summary = "Tool to work with polyploids, NGS and molecular biology".freeze
 
   if s.respond_to? :specification_version then
     s.specification_version = 4
+  end
 
-    if Gem::Version.new(Gem::VERSION) >= Gem::Version.new('1.2.0') then
-      s.add_runtime_dependency(%q<bio>.freeze, [">= 1.5.1"])
-      s.add_runtime_dependency(%q<bio-samtools>.freeze, [">= 2.6.2"])
-      s.add_runtime_dependency(%q<descriptive_statistics>.freeze, [">= 0"])
-      s.add_runtime_dependency(%q<systemu>.freeze, [">= 2.5.2"])
-      s.add_development_dependency(%q<shoulda>.freeze, [">= 2.10"])
-      s.add_development_dependency(%q<test-unit>.freeze, [">= 0"])
-      s.add_development_dependency(%q<juwelier>.freeze, [">= 0"])
-    else
-      s.add_dependency(%q<bio>.freeze, [">= 1.5.1"])
-      s.add_dependency(%q<bio-samtools>.freeze, [">= 2.6.2"])
-      s.add_dependency(%q<descriptive_statistics>.freeze, [">= 0"])
-      s.add_dependency(%q<systemu>.freeze, [">= 2.5.2"])
-      s.add_dependency(%q<shoulda>.freeze, [">= 2.10"])
-      s.add_dependency(%q<test-unit>.freeze, [">= 0"])
-      s.add_dependency(%q<juwelier>.freeze, [">= 0"])
-    end
+  if s.respond_to? :add_runtime_dependency then
+    s.add_runtime_dependency(%q<bio>.freeze, [">= 1.5.1"])
+    s.add_runtime_dependency(%q<bio-samtools>.freeze, [">= 2.6.2"])
+    s.add_runtime_dependency(%q<descriptive_statistics>.freeze, [">= 0"])
+    s.add_runtime_dependency(%q<systemu>.freeze, [">= 2.5.2"])
+    s.add_development_dependency(%q<shoulda>.freeze, [">= 2.10"])
+    s.add_development_dependency(%q<test-unit>.freeze, [">= 0"])
+    s.add_development_dependency(%q<juwelier>.freeze, [">= 0"])
   else
     s.add_dependency(%q<bio>.freeze, [">= 1.5.1"])
     s.add_dependency(%q<bio-samtools>.freeze, [">= 2.6.2"])

data/lib/bio/BIOExtensions.rb

@@ -64,7 +64,7 @@ end
 
 class Bio::NucleicAcid
 
-  IUPAC_CODES = {
+  IUPAC_CODES ||= {
 
     'y' => 'ct',
     'r' => 'ag',

data/lib/bio/PolyploidTools/ExonContainer.rb

@@ -18,7 +18,7 @@ module Bio::PolyploidTools
     end
 
     def gene_models(path)
-      @gene_models_db = Bio::DB::Fasta::FastaFile.new({:fasta=>path})
+      @gene_models_db = Bio::DB::Fasta::FastaFile.new(fasta: path)
       @gene_models_db.index
       @gene_models_path = path
     end
@@ -48,7 +48,7 @@ module Bio::PolyploidTools
 
     #Sets the reference file for the gene models
     def chromosomes(path)
-      @chromosomes_db = Bio::DB::Fasta::FastaFile.new({:fasta=>path})
+      @chromosomes_db = Bio::DB::Fasta::FastaFile.new(fasta: path)
       @chromosomes_path = path
     end
 
@@ -72,11 +72,10 @@ module Bio::PolyploidTools
       name = opts[:name]
       path = opts[:reference_path]
       path = opts[:alig_path]
-      chromosomes[name] = Bio::DB::Fasta::FastaFile.new({:fasta=>path})
+      chromosomes[name] = Bio::DB::Fasta::FastaFile.new(fasta: path)
     end
     
     def add_snp(snp)
-      #TODO: add to the snp the maximum number of hits? 
       snp.max_hits = self.max_hits
       @snp_map[snp.gene] = Array.new unless   @snp_map[snp.gene] 
       @snp_map[snp.gene] << snp
@@ -110,7 +109,7 @@ module Bio::PolyploidTools
       ret_str = ""
       @parents.each  do |name, bam|
         ret_str << ">#{gene_region.id}_SNP-#{snp.position}_#{name} Overlapping_exons:#{gene_region.to_s} localSNPpo:#{local_pos_in_gene+1}\n" 
-        to_print =  bam.consensus_with_ambiguities({:region=>gene_region}).to_s
+        to_print =  bam.consensus_with_ambiguities(region: gene_region).to_s
         to_print[local_pos_in_gene] = to_print[local_pos_in_gene].upcase
         ret_str << to_print << "\n"
       end
@@ -141,6 +140,7 @@ module Bio::PolyploidTools
           begin 
             file.puts snp.aligned_sequences_fasta
           rescue Exception=>e
+            #puts snp.inspect
             @missing_exons << snp.to_s
             $stderr.puts "print_fasta_snp_exones:" + snp.to_s + ":" + e.to_s
             $stderr.puts "Local position: #{snp.local_position}"
@@ -151,7 +151,7 @@ module Bio::PolyploidTools
       end
     end
 
-    def print_primer_3_exons (file, target_chromosome , parental )
+    def print_primer_3_exons (file, target_chromosome , parental,  max_specific_primers: 20 )
       added = 0
       
       @snp_map.each do | gene, snp_array|
@@ -159,9 +159,9 @@ module Bio::PolyploidTools
           string = ""
           begin 
             primer_3_min_seq_length
-            string = snp.primer_3_string( snp.chromosome, parental )
-            #TODO: add tan error to the SNP this snp has more than max_hits. Or maybe inside the SNP file. 
-            #puts "print_primer_3_exons: #{string.size}"
+            string = snp.primer_3_string( snp.chromosome, parental,  max_specific_primers: max_specific_primers )
+            #TODO: add tan error to the SNP this snp has more than max_hits. 
+            #Or maybe inside the SNP file. 
             if string.size > 0
               file.puts string
               added += 1

data/lib/bio/PolyploidTools/NoSNPSequence.rb

@@ -55,11 +55,15 @@ module Bio::PolyploidTools
 
      def mask_aligned_chromosomal_snp(chromosome)
       return nil if  aligned_sequences.values.size == 0
-      names = exon_sequences.keys
+      names = aligned_sequences.keys
+      parentals =  parental_sequences.keys
+      names = names - parentals
+
+
+      best_target = get_target_sequence(names, chromosome)
+      masked_snps = aligned_sequences[best_target].downcase if aligned_sequences[best_target]
+      masked_snps = "-" * aligned_sequences.values[0].size  unless aligned_sequences[best_target]
 
-      masked_snps = aligned_sequences[chromosome].downcase if aligned_sequences[chromosome]
-   
-      masked_snps = "-" * aligned_sequences.values[0].size  unless aligned_sequences[chromosome]
       #TODO: Make this chromosome specific, even when we have more than one alignment going to the region we want.
       i = 0
       while i < masked_snps.size
@@ -105,26 +109,23 @@ module Bio::PolyploidTools
 
         aligned_sequences.each_pair do |name, val|  
           has_del = true if val[i] == '-'
-          print "#{val[i]}\t"
+          #print "#{val[i]}\t"
         end
         count += 1 if has_del
-        print "#{count}\n"
+        #print "#{count}\n"
       end
       return count
     end
 
     def primer_region(target_chromosome, parental_chr )
       chromosome_seq = aligned_sequences[target_chromosome]
-      #chromosome_seq = "-" * parental.size unless chromosome_seq
-      if aligned_sequences.size == 0
-        #puts aligned_sequences.inspect
-        #puts surrounding_exon_sequences.inspect
-        #puts self.inspect
-        chromosome_seq = surrounding_exon_sequences[target_chromosome]
-
-      end
+      names = aligned_sequences.keys
+      target_chromosome = get_target_sequence(names, target_chromosome)
+      chromosome_seq = aligned_sequences[target_chromosome]
+      chromosome_seq = surrounding_exon_sequences[target_chromosome ]if aligned_sequences.size == 0
+      chromosome_seq = "-" * sequence_original.size unless chromosome_seq
       chromosome_seq = chromosome_seq.downcase
-
+      #puts chromosome_seq
       mask = mask_aligned_chromosomal_snp(target_chromosome)
     
       pr = PrimerRegion.new
@@ -146,7 +147,7 @@ module Bio::PolyploidTools
               pr.crhomosome_specific_intron << position_in_region
             elsif Bio::NucleicAcid.is_valid(parental[i], mask[i])
               parental[i] = mask[i]
-              pr.chromosome_specific << position_in_region if count_deletions_around(1,target_chromosome) < 3
+              pr.chromosome_specific << position_in_region #if count_deletions_around(1,target_chromosome) < 3
               pr.chromosome_specific_in_mask << i
             end
 
@@ -165,16 +166,15 @@ module Bio::PolyploidTools
           position_in_region += 1
         end #Closes region with bases 
       end         
-
       pr.sequence=parental.gsub('-','')
       pr
     end
 
-    def return_primer_3_string_test(opts={})
-
-      left = opts[:right_pos]
+    def return_primer_3_string(opts={})
+      #puts "return_primer_3_string #{opts.inspect}"
+      left = opts[:left_pos]
       right = opts[:right_pos]
-      sequence =  opts[:sequence]
+      sequence =  opts[:sequence].clone
       orientation = "forward"
       if opts[:right_pos]
         orientation = "forward"
@@ -201,7 +201,7 @@ module Bio::PolyploidTools
 
       #In case that we don't have a right primer, we do both orientations
       unless opts[:right_pos]
-        sequence =  opts[:sequence]    
+        sequence =  opts[:sequence].clone    
         left = sequence.size - left - 1
         orientation = "reverse"
         sequence = reverse_complement_string(sequence)
@@ -222,8 +222,10 @@ module Bio::PolyploidTools
         end
     end
 
-    def primer_3_all_strings(target_chromosome, parental) 
+    def primer_3_all_strings(target_chromosome, parental, max_specific_primers: nil) 
+      #puts "primer_3_all_strings: #{target_chromosome} #{parental}"
       pr = primer_region(target_chromosome, parental )
+      #puts pr.inspect
       primer_3_propertes = Array.new
 
       seq_original = String.new(pr.sequence)
@@ -236,24 +238,28 @@ module Bio::PolyploidTools
         snp_type = "non-homoeologous"
       end
 
-      pr.chromosome_specific.each do |pos|
-        
-        seq_snp =  String.new(pr.sequence)
-        orgiginal_base = seq_snp[pos]
-        other_chromosome_base = get_base_in_different_chromosome(pos, target_chromosome)
+      pr.chromosome_specific.each_with_index do |pos , i|
+        seq_snp =  seq_original.clone
+        #original_base = seq_snp[pos]
+        #puts "___"
+        #puts aligned_sequences.keys.inspect
+        #puts target_chromosome
+        t_chr =  get_target_sequence(aligned_sequences.keys, target_chromosome)
+        other_chromosome_base = get_base_in_different_chromosome(pr.chromosome_specific_in_mask[i], t_chr)
 
         args = {
           :name =>"#{gene} A chromosome_specific exon #{snp_type} #{chromosome}", 
           :left_pos => pos,  
-          :sequence=>seq_original
+          :sequence=>seq_snp
         }
         
-        
+        seq_snp =  seq_original.clone
         primer_3_propertes << return_primer_3_string(args)
+        
         args[:name] = "#{gene} B chromosome_specific exon #{snp_type} #{chromosome}"
-        args[:sequence] = seq_snp
-        #TODO: Find base from another chromosome
         seq_snp[pos] =  other_chromosome_base.upcase
+        args[:sequence] = seq_snp
+        
         
         primer_3_propertes << return_primer_3_string(args)
       end
@@ -265,7 +271,7 @@ module Bio::PolyploidTools
     def aligned_sequences
      
       return @aligned_sequences if @aligned_sequences
-      if sequences_to_align.size == 1
+      if sequences_to_align.size <= 1
         @aligned_sequences = sequences_to_align
         return @aligned_sequences
       end

data/lib/bio/PolyploidTools/SNP.rb

@@ -162,6 +162,7 @@ module Bio::PolyploidTools
     end
 
     def add_exon(exon, arm, filter_best: true)
+      exon_list[arm] = Array.new unless exon_list[arm]
       if filter_best and exon_list[arm].size > 0
         current = exon_list[arm].first
         exon_list[arm] = [exon] if exon.record.score > current.record.score 
@@ -370,7 +371,7 @@ module Bio::PolyploidTools
     end
 
 
-    def primer_3_all_strings(target_chromosome, parental) 
+    def primer_3_all_strings(target_chromosome, parental, max_specific_primers: 20 ) 
 
       pr = primer_region(target_chromosome, parental )
       primer_3_propertes = Array.new
@@ -401,9 +402,16 @@ module Bio::PolyploidTools
       else
         @snp_type = "non-homoeologous"
       end
+      
+      total_candidates  = pr.chromosome_specific.size
+      total_candidates += pr.crhomosome_specific_intron.size 
+      total_candidates += pr.almost_chromosome_specific.size
+      total_candidates += pr.almost_crhomosome_specific_intron.size
 
-      pr.chromosome_specific.each do |pos|
+      skip_specific = total_candidates > max_specific_primers
 
+      pr.chromosome_specific.each do |pos|
+        break if skip_specific
         args = {:name =>"#{gene}:#{original}#{position}#{snp} #{original_name} chromosome_specific exon #{@snp_type} #{chromosome}", :left_pos => pr.snp_pos, :right_pos => pos, :sequence=>seq_original}
         primer_3_propertes << return_primer_3_string(args)
         args[:name] = "#{gene}:#{original}#{position}#{snp} #{snp_in} chromosome_specific exon #{@snp_type} #{chromosome}"
@@ -411,41 +419,38 @@ module Bio::PolyploidTools
         primer_3_propertes << return_primer_3_string(args)
       end
 
-      pr.almost_chromosome_specific.each do |pos|
-        args = {:name =>"#{gene}:#{original}#{position}#{snp} #{original_name} chromosome_semispecific exon #{@snp_type} #{chromosome}", :left_pos => pr.snp_pos, :right_pos => pos, :sequence=>seq_original}
+      pr.crhomosome_specific_intron.each do |pos|
+        break if skip_specific
+        args = {:name =>"#{gene}:#{original}#{position}#{snp} #{original_name} chromosome_specific intron #{@snp_type} #{chromosome}", :left_pos => pr.snp_pos, :right_pos => pos, :sequence=>seq_original}
         primer_3_propertes << return_primer_3_string(args)
-        args[:name] = "#{gene}:#{original}#{position}#{snp} #{snp_in} chromosome_semispecific exon #{@snp_type} #{chromosome}"
+        args[:name] = "#{gene}:#{original}#{position}#{snp} #{snp_in} chromosome_specific exon #{@snp_type} #{chromosome}"
         args[:sequence] = seq_snp
         primer_3_propertes << return_primer_3_string(args)
-
       end
 
-      pr.crhomosome_specific_intron.each do |pos|
-
-        args = {:name =>"#{gene}:#{original}#{position}#{snp} #{original_name} chromosome_specific intron #{@snp_type} #{chromosome}", :left_pos => pr.snp_pos, :right_pos => pos, :sequence=>seq_original}
+      pr.almost_chromosome_specific.each do |pos|
+        break if skip_specific
+        args = {:name =>"#{gene}:#{original}#{position}#{snp} #{original_name} chromosome_semispecific exon #{@snp_type} #{chromosome}", :left_pos => pr.snp_pos, :right_pos => pos, :sequence=>seq_original}
         primer_3_propertes << return_primer_3_string(args)
-        args[:name] = "#{gene}:#{original}#{position}#{snp} #{snp_in} chromosome_specific exon #{@snp_type} #{chromosome}"
+        args[:name] = "#{gene}:#{original}#{position}#{snp} #{snp_in} chromosome_semispecific exon #{@snp_type} #{chromosome}"
         args[:sequence] = seq_snp
         primer_3_propertes << return_primer_3_string(args)
       end
 
       pr.almost_crhomosome_specific_intron.each do |pos|
+        break if skip_specific
         args = {:name =>"#{gene}:#{original}#{position}#{snp} #{original_name} chromosome_semispecific intron #{@snp_type} #{chromosome}", :left_pos => pr.snp_pos, :right_pos => pos, :sequence=>seq_original}
         primer_3_propertes << return_primer_3_string(args)
         args[:name] = "#{gene}:#{original}#{position}#{snp} #{snp_in} chromosome_semispecific exon #{@snp_type} #{chromosome}"
         args[:sequence] = seq_snp
         primer_3_propertes << return_primer_3_string(args)
-
       end
 
-
       args = {:name =>"#{gene}:#{original}#{position}#{snp} #{original_name} chromosome_nonspecific all #{@snp_type} #{chromosome}", :left_pos => pr.snp_pos, :sequence=>seq_original}
       primer_3_propertes << return_primer_3_string(args)
       args[:name] = "#{gene}:#{original}#{position}#{snp} #{snp_in} chromosome_nonspecific all #{@snp_type} #{chromosome}"
       args[:sequence] = seq_snp
       primer_3_propertes << return_primer_3_string(args)
-
-
       primer_3_propertes
     end
 
@@ -457,8 +462,8 @@ module Bio::PolyploidTools
       "#{original}#{position}#{snp}".upcase
     end
 
-    def primer_3_string(target_chromosome, parental) 
-      strings = primer_3_all_strings(target_chromosome, parental) 
+    def primer_3_string(target_chromosome, parental, max_specific_primers: 20) 
+      strings = primer_3_all_strings(target_chromosome, parental, max_specific_primers: max_specific_primers) 
       strings.join
     end
 
@@ -558,7 +563,7 @@ module Bio::PolyploidTools
     def aligned_sequences
      
       return @aligned_sequences if @aligned_sequences
-     
+      return Hash.new if sequences_to_align.size == 0
       
       options = ['--maxiterate', '1000', '--localpair', '--quiet']
       mafft = Bio::MAFFT.new( "mafft" , options)
@@ -756,13 +761,13 @@ module Bio::PolyploidTools
       self.exon_list.each do |chromosome, exon_arr| 
         exon_arr.each do |exon|
           exon_start_offset = exon.query_region.start - gene_region.start
-          flanquing_region  = exon.target_flanking_region_from_position(position,flanking_size)
+          flanking_region  = exon.target_flanking_region_from_position(position,flanking_size)
           #TODO: Padd when the exon goes over the regions... 
-          #puts flanquing_region.inspect
+          #puts flanking_region.inspect
           #Ignoring when the exon is in a gap
           unless exon.snp_in_gap 
-            exon_seq = container.chromosome_sequence(flanquing_region)
-            @surrounding_exon_sequences["#{chromosome}_#{flanquing_region.start}_#{exon.record.score}"] = exon_seq
+            exon_seq = container.chromosome_sequence(flanking_region)
+            @surrounding_exon_sequences["#{chromosome}_#{flanking_region.start}_#{exon.record.score}"] = exon_seq
           end
         end
       end