seqtrimnext 2.0.59 → 2.0.60

data/lib/seqtrimnext/plugins_old/plugin_low_quality_old.rb

@@ -0,0 +1,382 @@
+require "plugin"  
+
+########################################################
+# Author: Almudena Bocinos Rioboo                      
+# 
+# Defines the main methods that are necessary to execute PluginLowQuality. See the main method called execute.
+
+#                                                     
+# Inherit: Plugin
+########################################################
+
+class PluginLowQuality < Plugin   
+  
+    
+      
+      def create_sum_window(qual,ini,index_window_end)
+      
+        # puts "--------index w #{index_window_end}" 
+        sum=[] 
+        i=ini  
+        # puts "#{i} #{index_window_end}"
+        while (i<=index_window_end) # initialize sum                                  
+          sum[i]=0 
+          i += 1
+        end
+        # puts " contenido de sum" + sum.join.to_s  + " i index_window_end  window #{i} #{index_window_end} #{@window}" 
+      
+        i=ini
+        while (i<ini+@window)
+      
+          sum[ini] += qual[i] 
+          i+=1
+        end                                           
+      
+      
+        i=ini+1 
+      
+        while (i<=index_window_end)            
+      
+          sum[i]=sum[i-1]-qual[i-1]+qual[i+@window-1]
+          i+=1
+      
+        end   
+      
+        # puts '2____' + sum.join(',') + 'pos sum' + ini.to_s    
+      
+        return sum 
+      
+      end   
+      
+      def find_bounds_high_quality(sum,ini,index_window_end) 
+      
+        new_start = -1
+        new_end = -1
+        
+      # puts " ini #{ini} iwe #{index_window_end}"
+      # puts "ini #{ini} index_window_end #{index_window_end} sum[ini] #{sum[ini]} cut_off #{@cut_off} suma #{sum.size} " 
+       if (ini>index_window_end) 
+           temp_start= ini
+           # new_start, new_end = temp_start, index_window_end 
+           new_end = index_window_end # para que no crea que no hay alta calidad, sino que hemos sobrepasado el indice final de la ventana
+             # new_start, new_end = index_window_end, index_window_end 
+       end   
+      # puts " temp_start #{temp_start}" if (ini>index_window_end)
+      temp_start=((ini<=index_window_end) && (sum[ini]>=@cut_off))? ini : -1    
+      
+        i=ini+1
+        while (i<=index_window_end)
+          if (sum[i]>=@cut_off)  
+            if (temp_start<0)
+               temp_start=i  #just in! 
+               # puts "just in ---- #{sum[i]}>= cut off #{@cut_off} pos #{temp_start}"   
+            end
+      
+          else 
+              # puts "sum #{sum[i]} < cut off "
+              if(temp_start>=0)              #just out!   
+                # puts "update #{sum[i]}< cut off #{@cut_off} pos #{i}.if #{i-1} - #{temp_start} > #{new_end} - #{new_start}"
+                if (((i-1-temp_start)>=(new_end-new_start)))   
+                  new_start,new_end=temp_start,i-1 
+                  # puts "just out ---- new start,new_end = #{temp_start}, #{i-1}  index_window_end = #{index_window_end}"   
+                end
+                temp_start= -1 
+              end
+          end
+          i+=1  
+      
+      
+        end 
+        # puts "4 temp_start #{temp_start} new_start #{new_start} new-end #{new_end}"  
+      
+        if (temp_start != -1)   # finished while ok           
+          # puts "4 #{index_window_end} - #{temp_start} > #{new_end} - #{new_start}"
+            if ((index_window_end- temp_start) >= (new_end-new_start)) #put the end of the window at the end of sequence
+                new_start, new_end = temp_start, index_window_end     #-1
+            end
+        end  
+      
+        # puts "5 temp_start #{temp_start} new_start #{new_start} new-end #{new_end}"   
+        
+        # puts  " newstart  #{new_start} newend #{new_end}" 
+       
+        return new_start,new_end 
+       
+      
+      end  
+      
+      def cut_fine_bounds_short(qual,new_start,new_end)
+      
+          i=0                    
+          # puts " qual[new_start+i] new_start #{new_start} i #{i} = #{new_start+i} qual.size #{qual.size}"
+          while (i<@window)
+            if (qual[new_start+i]>=@low)
+              break
+            end    
+            i+=1
+          end  
+          new_start +=i 
+          # puts "#{new_start} ***********"
+      
+          i=@window -1
+          while (i>=0)  
+            if (qual[new_end+i]>=@low)    
+              break            
+            end
+            i-=1            
+          end     
+          new_end += i
+          # puts "6a new_start #{new_start} new-end #{new_end}"     
+          
+           # puts " #{new_start} #{new_end} .o.o.o.o.o.o.o.o2 short"    
+          return new_start, new_end  
+      
+      end  
+      
+      
+      # cuts fine the high quality bounds
+      def cut_fine_bounds(qual,new_start,new_end)   
+        # puts "  ççççççççççççççç #{new_start+@window} >= #{new_end} " 
+        # puts " #{new_start} #{new_end} .o.o.o.o.o.o.o.o1"
+        # cut it fine
+      
+         one_ok = 0         
+      
+          i=@window-1
+          # puts " qual[new_start+i] new_start #{new_start} i #{i} = #{new_start+i} qual.size #{qual.size}"
+          while (i>=0) 
+              if (qual[new_start+i] < @low) 
+                  break if one_ok
+              else 
+                  one_ok = 1
+              end    
+              i-=1
+          end
+          new_start += i+1
+          oneOk = 0  
+          i=0
+          while (i<@window) 
+              if (qual[new_end+i] < @low) 
+                  break if oneOk
+              else 
+                  oneOk = 1
+              end  
+              i+=1
+          end
+          new_end += i-1 
+          # puts "6b  new_start #{new_start} new-end #{new_end}"  
+      
+        # puts " #{new_start} #{new_end} .o.o.o.o.o.o.o.o2"
+        return new_start, new_end
+      
+      end
+      
+      def find_high_quality(qual,ini=0)  
+      
+        # puts qual.class.to_s + qual.size.to_s + 'size,' + @window.to_s + ' window, '+ qual.join(',')  + 'size' + qual.size.to_s
+        
+        update=false
+        # if @window>qual.length-ini     #search in the last window although has a low size
+        #     @window=qual.length-ini   
+        #      # puts ' UPDATE WINDOW  Y CUT OFF ' + @window.to_s
+        #      @cut_off=@window*@low   
+        #      update=true
+        #   end          
+                   
+        if (ini==0 or update)
+          #index_window_start = ini
+          @index_window_end = qual.size- @window #don't sub 1, or will lost the last nucleotide of the sequence -1;
+          #TODO En seqtrim de Juan iwe, que en nuestro seqtrim se llama index_window_end, está perdiendo 2 nucleótidos de la última ventana calculada 
+      
+      
+          @sum = create_sum_window(qual,ini,@index_window_end) 
+          # puts "SUMA #{@sum.join(' ')}"   
+        end              
+              
+        new_start, new_end = find_bounds_high_quality(@sum,ini,@index_window_end) 
+        # puts " #{new_start} #{new_end} .o.o.o.o.o.o.o.o1"
+      
+        if (new_start>=0)
+          if (new_start+@window >= new_end)
+             # puts "cfs"     
+            new_start, new_end = cut_fine_bounds_short(qual,new_start,new_end)
+            # puts "cfs"
+      
+          else  
+            # puts "cf"
+            new_start, new_end = cut_fine_bounds(qual,new_start,new_end) 
+            # puts "cf"
+          end 
+        end 
+        
+         # puts " #{new_start} #{new_end} .o.o.o.o.o.o.o.o2" 
+      
+        return new_start,new_end #+1
+      
+      
+      end
+      
+      
+      def add_action_before_high_qual(p_begin,p_end,actions,seq,start)
+      
+        action_size = p_begin-1
+        if action_size>=(@window/2)  
+      
+      
+          # puts "action_SIZE1 #{action_size} > #{@window/2}"
+      
+          if ( (p_begin>0) && (action_size>0) )  #if there is action before the high qual part 
+            # it's created an action before of the high quality part
+            a = seq.new_action(start ,p_begin-1,"ActionLowQuality") # adds the ActionInsert to the sequence before adding the actionMid
+            # puts " new low qual start: #{start}  = #{a.start_pos} end: #{p_begin} -1 = #{a.end_pos}"
+            actions.push a   
+          end 
+        end             
+      end  
+      
+      def add_action_after_high_qual(p_begin,p_end,actions,seq)
+      
+        action_size = seq.insert_end-p_end
+        if action_size>=(@window/2)
+      
+      
+           # puts "action_SIZE2 #{action_size} > #{@window/2}"
+      
+           if ((p_end<seq.seq_fasta.size-1) && (action_size>0) )  #if there is action before the high qual part 
+             # it's created an action before of the high quality part
+             a = seq.new_action(p_end-seq.insert_start+1,seq.seq_fasta.size-1,"ActionLowQuality") # adds the ActionInsert to the sequence before adding the actionMid
+      
+             actions.push a   
+           end 
+         end 
+      end
+       
+                                                                                   
+
+
+  
+  ######################################################################
+  #---------------------------------------------------------------------
+ 
+  # Begins the plugin1's execution whit the sequence "seq"
+  # Creates an action by each subsequence with low quality to eliminate it 
+  # A subsequence has low quality if (the add of all its qualitis < subsequence_size*20)  
+  # Creates the qualities windows from the sequence, looks for the subsequence with high quality 
+  # and mark, with an action, the before part to the High Quality Subsequence like a low quality part 
+  # Finally  mark, with an action, the after part to the High Quality Subsequence like a low quality part 
+  #----------------------------------------------------------------- 
+   
+  def exec_seq(seq,blast_query)
+
+     if ((self.class.to_s=='PluginLowQuality') && seq.seq_qual.nil? ) 
+       $LOG.error " Quality File haven't been provided. It's impossible to execute " + self.class.to_s     
+     elsif (seq.seq_qual.size>0)
+       $LOG.debug "[#{self.class.to_s}, seq: #{seq.seq_name}]: checking low quality of the sequence"    
+
+       @low=@params.get_param('min_quality').to_i
+
+       if @params.get_param('window_width').to_i>seq.seq_fasta.length   
+         @window=seq.seq_fasta.length   
+       
+       else 
+         @window=@params.get_param('window_width').to_i
+       end 
+       @cut_off=@window*@low   
+                                           
+       type='ActionLowQuality' 
+       low_qual=0
+       actions=[] 
+       
+       p_begin,p_end =0,-1 # positions from high quality bounds    
+                                
+       
+       
+       while ((p_begin>=0)  && (p_end + 1 < seq.seq_qual.size) ) 
+         
+         
+         p_begin_old,p_end_old= p_begin, p_end
+         p_begin,p_end = find_high_quality(seq.seq_qual,p_end+1)  
+         
+         if ((p_begin>0) && (p_begin-p_end_old-1>=@window/2)) #if we have found the high quality part, and  the low quality part has enough size 
+            # it's created an action before of the high quality part 
+            add_action_before_high_qual(p_begin,p_end,actions,seq,p_end_old+1) 
+
+            # puts "low1 ini fin  #{p_end_old+1} #{p_begin-1} = #{p_begin-1-p_end_old-1+1}"     
+            low_qual = p_begin-1-p_end_old-1 + 1 
+            
+            add_stats('low_qual',low_qual)
+            # @stats[:low_qual]={low_qual => 1} 
+             
+         end
+         
+         # puts "-----ññññ----- high quality  #{p_begin}   #{p_end}+#{seq.insert_start} seq size #{seq.seq_fasta.size}"
+
+       end
+        
+       # puts "high [#{p_begin}, #{p_end}] old [#{p_begin_old}, #{p_end_old}] size #{seq.seq_qual.size}"   
+       if ((p_begin>=0) && (p_end+1<seq.seq_qual.size))  #if we have found the high quality part 
+          
+          # it's created an action after of the high quality part     
+          add_action_after_high_qual(p_begin,p_end,actions,seq) 
+          # puts "low2 ini fin #{p_end+1} #{seq.seq_fasta.size-1}  = #{seq.seq_fasta.size-1-p_end-1+1}"
+          low_qual = seq.seq_fasta.size-1 - p_end-seq.insert_start-1 + 1
+          # if @stats[:low_qual][low_qual].nil?
+          #              @stats[:low_qual][low_qual] = 0
+          #           end
+          #           @stats[:low_qual][low_qual] += 1  
+          add_stats('low_qual',low_qual) 
+          # @stats[:low_qual]={low_qual => 1}    
+       end                                     
+
+       # puts "-----ññññ----- high quality  #{p_begin}   #{p_end}"  
+   
+       
+       if p_end<0 and p_end_old #add action low qual to all the part      
+         a = seq.new_action(p_end_old+1 ,seq.seq_fasta.size-1,"ActionLowQuality") # adds the ActionInsert to the sequence before adding the actionMid
+         # puts "new low qual start: #{p_end_old+1} end: #{seq.seq_fasta.size-1} = #{seq.seq_fasta.size-1 - p_end_old-1 + 1}" 
+         low_qual = seq.seq_fasta.size-1 - p_end_old-1 + 1 
+         
+          # if @stats[:low_qual][low_qual].nil?
+          #             @stats[:low_qual][low_qual] = 0
+          #          end
+          #          @stats[:low_qual][low_qual] += 1   
+         add_stats('low_qual',low_qual) 
+         # @stats[:low_qual]={'low_qual' => 1} 
+          
+         actions.push a
+       end
+       
+       # puts "------- ADDING ACTIONs LOW QUAL #{actions.size}"    
+       seq.add_actions(actions)
+     end       
+
+   end 
+   
+  #-----------------------------------------------------------------
+  
+  
+  ######################################################################
+  #---------------------------------------------------------------------
+
+  #Returns an array with the errors due to parameters are missing 
+  def self.check_params(params) 
+    
+    errors=[]
+    
+   	comment='Minimum quality value for every nucleotide'
+		default_value = 20
+		params.check_param(errors,'min_quality','Integer',default_value,comment)
+		
+   	comment='Quality window for scanning low quality segments'
+		default_value = 15
+		params.check_param(errors,'window_width','Integer',default_value,comment)
+   
+
+    
+    return errors
+  end
+  
+  
+  private :find_high_quality
+  
+end

data/lib/seqtrimnext/plugins_old/plugin_rem_adit_artifacts.rb

@@ -0,0 +1,234 @@
+require "plugin"
+
+
+########################################################
+# Author: Almudena Bocinos Rioboo                      
+# 
+# Defines the main methods that are necessary to execute PluginRemAditArtifacts
+
+#                                                     
+# Inherit: Plugin
+########################################################
+
+class PluginRemAditArtifacts < Plugin
+  
+  
+  
+  def exec_seq(seq,blast_query)
+       
+    $LOG.debug "[#{self.class.to_s}, seq: #{seq.seq_name}]: removing artifacts into the sequence"    
+    seq2 = seq.seq_fasta    
+    first = 0
+    last = seq2.size-1    
+    old_first=first
+    old_last=last
+      
+
+    while (seq2 =~ /^(GCGGGG|CCCCGC)/i)
+      first += 6
+      seq2.slice!(0..5)
+    end
+    
+    
+    while (seq2 =~ /(GCGGGG|CCCCGC)$/i)
+      last -= 6
+      seq2.slice!(seq2.size-1-5..seq2.size-1)
+      
+    end
+   
+    
+    #is_forward, is_cDNA, 
+    #TrimExtremeNXs(first,last)
+    is_forward = @params.get_param('is_forward')=='true'
+    is_cDNA = @params.get_param('is_cDNA')=='true'
+     
+    previous_first,previous_last =0,0
+    
+    until ((previous_first == first) && (previous_last == last))
+      previous_first,previous_last = first, last      
+    
+      if (is_cDNA)
+        if (is_forward)
+           
+          nTs = 0
+          nTs = $1.length if (seq2 =~ /^(T+)/i) 
+          
+          if (nTs > 3)
+            seq2.slice!(0..nTs -1)            
+            first += nTs #-1
+           
+          end
+          
+          nAs = 0
+          nAs = $1.length if (seq2 =~ /(A+)$/i)
+        
+          if (nAs > 3)            
+            seq2.slice!(seq2.size - nAs..seq2.size - 1)
+            last -= nAs
+            
+          end
+        else #si es backward
+           
+          nTs = 0
+          nTs = $1.length if (seq2 =~ /(T+)$/i) 
+          
+          if (nTs > 3)            
+            seq2.slice!(seq2.size-nTs..seq2.size-1)
+            last -= nTs
+            
+          end
+    
+          nAs = 0
+          nAs = $1.length if (seq2 =~ /^(A+)/i)
+          
+          if (nAs > 3)            
+            seq2.slice!(0..nAs -1)
+            first += nAs
+          
+          end
+        end    
+      end
+    end 
+    
+      
+    if (((first>=0) && (first>old_first)) || ((last>=0) && (last<old_last)))
+      type='ActionRemAditArtifacts'           
+      actions = []          
+      # seq.add_action(first,last,type)  
+      a=seq.new_action(first,last,type) 
+      actions.push a
+      seq.add_actions(actions)    
+    end
+   
+    
+  end
+  ######################################################################
+  #---------------------------------------------------------------------
+  def execute_old(seq)
+    seq2 = seq.seq_fasta   
+    #seq2 = 'dGCGGGG' 
+    first = 0
+    last = seq2.size-1    
+    old_first=first
+    old_last=last
+      
+    # puts '1 '+seq2
+    # puts 'POS '+first.to_s
+    # puts 'POS '+last.to_s
+    while (seq2 =~ /^(GCGGGG|CCCCGC)/i)
+      first += 6
+      seq2.slice!(0..5)
+     # puts '2 '+seq2
+     # already = true
+    end
+    
+    
+    while (seq2 =~ /(GCGGGG|CCCCGC)$/i)
+      last -= 6
+      seq2.slice!(seq2.size-1-5..seq2.size-1)
+      # puts '3 '+seq2
+     # already = true
+    end
+   
+    
+    #is_forward, is_cDNA, 
+    #TrimExtremeNXs(first,last)
+    is_forward = @params.get_param('is_forward')
+    is_cDNA = @params.get_param('is_cDNA')
+     # puts '4 '+seq2
+     previous_first,previous_last =0,0
+    
+    until ((previous_first == first) && (previous_last == last))
+      previous_first,previous_last = first, last
+      # puts 'POS5-F '+first.to_s
+      # puts 'POS5-L '+last.to_s
+    
+      if (is_cDNA)
+        if (is_forward)
+           # puts '5 '+seq2
+          nTs = 0
+          nTs = $1.length if (seq2 =~ /^(T+)/i) 
+          if (nTs > 3)
+            seq2.slice!(0..nTs -1)
+             # puts '6 '+seq2
+            first += nTs #-1
+            # puts 'POS6-F '+first.to_s
+          end
+          nAs = 0
+          nAs = $1.length if (seq2 =~ /(A+)$/i)
+          # puts '6-7 '+seq2 + nAs.to_s
+          if (nAs > 3)
+            # puts '7 '+seq2
+            seq2.slice!(seq2.size - nAs..seq2.size - 1)
+            last -= nAs#seq2.size-nAs-2
+            # puts 'POS7-L '+last.to_s
+          end
+        else #si es backward
+           # puts '5b '+seq2
+          nTs = 0
+          nTs = $1.length if (seq2 =~ /(T+)$/i) 
+          if (nTs > 3)
+             # puts '6b '+seq2
+            seq2.slice!(seq2.size-nTs..seq2.size-1)
+            last -= nTs#seq2.size-nTs -2
+            # puts 'POS6b-L '+last.to_s
+          end
+    
+          nAs = 0
+          nAs = $1.length if (seq2 =~ /^(A+)/i) 
+          if (nAs > 3)
+             # puts '7b '+seq2
+            seq2.slice!(0..nAs -1)
+            first += nAs#nAs -1
+            # puts 'POS7b-f '+first.to_s
+          end
+        end    
+      end
+    end 
+    
+       #first -= 1 if (old_first!= first) 
+        #last += 1 if (old_last!= last)
+    
+    # puts 'POS7-8 '+first.to_s
+    # puts 'POS7-8 '+last.to_s
+      
+    if (((first>=0) && (first>old_first)) || ((last>=0) && (last<old_last)))
+      type='ActionRemAditArtifacts'
+            
+       # puts '8 '+seq2
+      seq.add_action(first,last,type)    
+    end
+     # puts '9 '+seq2
+    
+  end
+  
+ 
+
+  ######################################################################
+  #---------------------------------------------------------------------
+
+  #Returns an array with the errors due to parameters are missing 
+  def self.check_params(params)
+    errors=[]
+    
+    
+    
+#    if !params.exists?('ta')
+#      errors.push " The param <ta> doesn't exist"
+#    end
+    
+#    if !params.exists?('poly_at_length')
+#      errors.push " The param <poly_at_length> doesn't exist"
+#    end
+    
+
+    
+    return errors
+  end
+  
+  
+  
+  
+  
+  
+end

data/lib/seqtrimnext/version.rb

@@ -0,0 +1,4 @@
+module Seqtrimnext
+  VERSION = "2.0.60"
+  SEQTRIM_VERSION = VERSION
+end

data/lib/seqtrimnext.rb

@@ -1,8 +1,4 @@
-$:.unshift(File.dirname(__FILE__)) unless
-  $:.include?(File.dirname(__FILE__)) || $:.include?(File.expand_path(File.dirname(__FILE__)))
-
-
-# setup REQUIRE_PATH
+require "seqtrimnext/version"
 
 ROOT_PATH=File.join(File.dirname(__FILE__),'seqtrimnext')
 
@@ -22,16 +18,6 @@ $: << File.expand_path(File.join(ROOT_PATH, 'classes','em_classes'))
 
 $: << File.expand_path(File.join(ROOT_PATH, 'latex','classes'))
 
-# puts $:
-
 module Seqtrimnext
-  
-  # SEQTRIM_VERSION_REVISION=27
-  # SEQTRIM_VERSION_STAGE = 'b'
-  # SEQTRIM_VERSION = "2.0.0#{SEQTRIM_VERSION_STAGE}#{SEQTRIM_VERSION_REVISION}"
-  
-   VERSION = '2.0.59'
-  
-  SEQTRIM_VERSION = VERSION
-  
+  # Your code goes here...
 end

data/seqtrimnext.gemspec

@@ -0,0 +1,38 @@
+# coding: utf-8
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'seqtrimnext/version'
+
+Gem::Specification.new do |spec|
+  spec.name          = "seqtrimnext"
+  spec.version       = Seqtrimnext::VERSION
+  spec.authors       = ["Dario Guerrero", "Almudena Bocinos"]
+  spec.email         = ["dariogf@gmail.com", "alkoke@gmail.com
+t"]
+  spec.summary       = %q{Sequences preprocessing and cleaning software}
+  spec.description   = %q{Seqtrimnext is a plugin based system to preprocess and clean sequences from multiple NGS sequencing platforms}
+  spec.homepage      = ""
+  spec.license       = "MIT"
+
+  spec.files         = `git ls-files -z`.split("\x0")
+  spec.executables   = spec.files.grep(%r{^bin/}) { |f| File.basename(f) }
+  spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
+  spec.require_paths = ["lib"]
+
+  spec.add_development_dependency "bundler", "~> 1.7"
+  spec.add_development_dependency "rake", "~> 10.0"
+
+  spec.add_runtime_dependency 'narray','>=0'
+  spec.add_runtime_dependency 'gnuplot','>=0'
+  spec.add_runtime_dependency 'term-ansicolor','>=1.0.5'
+  spec.add_runtime_dependency 'xml-simple','>=1.0.12'
+  spec.add_runtime_dependency 'scbi_blast','>=0.0.34'
+  spec.add_runtime_dependency 'scbi_mapreduce','>=0.0.38'
+  spec.add_runtime_dependency 'scbi_fasta','>=0.1.7'
+  spec.add_runtime_dependency 'scbi_fastq','>=0.0.18'
+  spec.add_runtime_dependency 'scbi_plot','>=0.0.6'
+  spec.add_runtime_dependency 'scbi_math','>=0.0.1'
+  spec.add_runtime_dependency 'scbi_headers','>=0.0.2'
+
+
+end