ExonContainer.rb in bio-polyploid-tools-0.8.5

- old
+ new

@@ -3,20 +3,21 @@
   class ExonContainer
     attr_reader :parental_1_sam,  :parental_2_sam  
     attr_reader :parental_1_name, :parental_2_name, :gene_models_db
     attr_reader :chromosomes, :snp_map
     attr_reader :parents
-    attr_accessor :flanking_size , :primer_3_min_seq_length
+    attr_accessor :flanking_size , :primer_3_min_seq_length, :max_hits
 
     BASES = [:A, :C, :G, :T]
     #Sets the reference file for the gene models
 
     def initialize
       @parents=Hash.new
       @snp_map = Hash.new 
       @snp_contigs
       @primer_3_min_seq_length = 50
+      @max_hits = 10
     end
 
     def gene_models(path)
       @gene_models_db = Bio::DB::Fasta::FastaFile.new({:fasta=>path})
       @gene_models_db.index
@@ -74,12 +75,15 @@
       path = opts[:alig_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
+
     end
 
     def add_snp_file(filename, chromosome, snp_in, original_name)
 
       File.open(filename) do | f |
@@ -155,10 +159,11 @@
         snp_array.each do |snp|
           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}"
             if string.size > 0
               file.puts string
               added += 1
             end 
@@ -203,9 +208,23 @@
                     $stderr.puts "Failed for the range #{record.query_start}-#{record.query_end} for position #{snp.position}"
                   end
                 end
               end
             end
+          end
+        end
+      end
+      remove_alignments_over_max_hits
+    end
+
+    def remove_alignments_over_max_hits
+      @snp_map.each_pair do | gene, snp_array| 
+        snp_array.each do |snp|
+          if snp.exon_list.size > max_hits
+            total_hits = snp.exon_list.size
+            snp.exon_list = {} 
+            snp.repetitive = true
+            snp.errors << "The marker is in a repetitive region (#{total_hits} hits to reference)"
           end
         end
       end
     end
 
\ No newline at end of file