seqtrimnext 2.0.33 → 2.0.35

data/History.txt

@@ -1,3 +1,11 @@
+=== 2.0.35 2011-06-22
+
+FarAdapters improvement
+
+=== 2.0.34 2011-06-21
+
+Added some additional FAR AB_ADAPTER restrictions
+
 === 2.0.33 2011-06-17
 
 Added SEQTRIMNEXT_INIT environment variable

data/bin/create_graphs.rb

@@ -1,14 +1,16 @@
+#!/usr/bin/env ruby
 require 'stringio'
 # require 'test/unit'
+require 'seqtrimnext'
 require 'json'
 require 'gnuplot'
 
-ROOT_PATH=File.dirname(File.dirname(__FILE__))
+# ROOT_PATH=File.dirname(File.dirname(__FILE__))
 
 # $: << File.expand_path(File.join(ROOT_PATH,'test'))
-$: << File.expand_path(File.join(ROOT_PATH,'classes'))
-$: << File.expand_path(File.join(ROOT_PATH,'plugins'))
-$: << File.expand_path(File.join(ROOT_PATH,'utils'))
+# $: << File.expand_path(File.join(ROOT_PATH,'classes'))
+# $: << File.expand_path(File.join(ROOT_PATH,'plugins'))
+# $: << File.expand_path(File.join(ROOT_PATH,'utils'))
 
 if ARGV.empty?
 puts "Usage: #{$0} stats.json initial_stats.json"

data/lib/seqtrimnext/classes/em_classes/seqtrim_worker.rb

@@ -1,26 +1,29 @@
 
 #finds the classes that were in the folder 'classes'
 
-ROOT_PATH=File.dirname(File.dirname(File.dirname(__FILE__)))
-
-$: << File.expand_path(File.join(ROOT_PATH, 'classes'))
-$: << File.expand_path(File.join(ROOT_PATH, 'classes','blast'))
-
-#finds the classes that were in the folder 'plugins'
-$: << File.expand_path(File.join(ROOT_PATH, 'plugins'))
-
-#finds the classes that were in the folder 'plugins'
-$: << File.expand_path(File.join(ROOT_PATH, 'actions'))
-
-#finds the classes that were in the folder 'utils'
-$: << File.expand_path(File.join(ROOT_PATH, 'utils'))
-
-$: << File.expand_path(File.join(ROOT_PATH, 'classes','em_classes'))
+# ROOT_PATH=File.dirname(File.dirname(File.dirname(__FILE__)))
+# 
+# $: << File.expand_path(File.join(ROOT_PATH, 'classes'))
+# $: << File.expand_path(File.join(ROOT_PATH, 'classes','blast'))
+# 
+# #finds the classes that were in the folder 'plugins'
+# $: << File.expand_path(File.join(ROOT_PATH, 'plugins'))
+# 
+# #finds the classes that were in the folder 'plugins'
+# $: << File.expand_path(File.join(ROOT_PATH, 'actions'))
+# 
+# #finds the classes that were in the folder 'utils'
+# $: << File.expand_path(File.join(ROOT_PATH, 'utils'))
+# 
+# $: << File.expand_path(File.join(ROOT_PATH, 'classes','em_classes'))
+# 
+# $: << File.expand_path(ROOT_PATH)
 
-$: << File.expand_path(ROOT_PATH)
+require 'seqtrimnext'
 
 $SEQTRIM_PATH = ROOT_PATH
 
+
 if ENV['BLASTDB']# && Dir.exists?(ENV['BLASTDB'])
   $FORMATTED_DB_PATH = ENV['BLASTDB']
   $DB_PATH = File.dirname($FORMATTED_DB_PATH)

data/lib/seqtrimnext/classes/params.rb

@@ -12,6 +12,7 @@ class Params
     @comments = {}
     # @param_order={}
     @mids = {}
+    @ab_adapters={}
     @linkers = {}
     @clusters = {}
 
@@ -54,28 +55,36 @@ class Params
     end
   end# end def
 
-  # Load mid's file
-  def load_mids(path_file)
+  def load_db_fastas(path_file)
+    res={}
     if File.exists?(path_file)
       ff = FastaFile.new(path_file)
       ff.each {|n,f|
         # @mid_sizes[n]=f.size
-        @mids[n]=f
+        res[n]=f
       }
 
       ff.close
     end
+    return res
+  end
+
+  # Load mid's file
+  def load_mids(path_file)
+    @mids=load_db_fastas(path_file)
+    # puts @mids
+  end
+
+  # Load ab_adapters file
+  def load_ab_adapters(path_file)
+    @ab_adapters=load_db_fastas(path_file)
+        # puts @ab_adapters
   end
 
   # Load mid's file
   def load_linkers(path_file)
-    if File.exists?(path_file)
-      ff = FastaFile.new(path_file)
-      ff.each {|n,f|
-        @linkers[n]=f
-      }
-      ff.close
-    end
+    @linkers=load_db_fastas(path_file)
+        # puts @linkers
   end
 
   def load_repeated_seqs(file_path)
@@ -90,7 +99,7 @@ class Params
         #if (line[0]!=62) && (line[0]!=48)
         if (line[0]!='>'[0]) && (line[0]!='0'[0])
           #puts line
-          #puts line,line[0]
+          # puts line,line[0]
           if line =~ />([^\.]+)\.\.\.\s/
             #puts 'ok'
             @clusters[$1]=1
@@ -157,6 +166,12 @@ class Params
     return @linkers[linker]
   end
 
+  # Return the ab of param
+  def get_ab_adapter(adapter)
+    return @ab_adapters[adapter]
+  end
+
+
   def get_plugin
     plugin='General'
     # puts caller(2)[1]

data/lib/seqtrimnext/classes/seqtrim.rb

@@ -223,6 +223,7 @@ class Seqtrim
       
       # leer mids
       params.load_mids(File.join($FORMATTED_DB_PATH,'mids.fasta'))
+      params.load_ab_adapters(File.join($FORMATTED_DB_PATH,'adapters_ab.fasta'))
       params.load_linkers(File.join($FORMATTED_DB_PATH,'linkers.fasta'))
 
       

data/lib/seqtrimnext/plugins/plugin.rb

@@ -66,7 +66,7 @@ class Plugin
     #     puts r2_end.class
     #     puts "-------"
     #puts "overlap? (#{r1_start}<=#{r2_end}) and (#{r1_end}>=#{r2_start})"
-    return ((r1_start<=r2_end) and (r1_end>=r2_start) )
+    return ((r1_start<=r2_end+1) and (r1_end>=r2_start-1) )
   end
 
   def merge_hits(hits,merged_hits,merged_ids=nil)

data/lib/seqtrimnext/plugins/plugin_ab_adapters.rb

@@ -43,6 +43,18 @@ class PluginAbAdapters < Plugin
     return blast_table_results
  end
  
+ # filter hits that are far the extreme and do not have a valid length
+ def filter_hits(hits,end_pos)
+   hits.reverse_each do |hit|
+    if (hit.q_end < (end_pos-40)) && ((hit.q_end-hit.q_beg+1)<(@params.get_ab_adapter(hit.subject_id).length*0.80).to_i)
+      hits.delete(hit)
+    #   puts "DELETE #{hit.inspect}"
+    # else
+    #   puts "ACCEPTED #{hit.inspect}, >= #{(@params.get_ab_adapter(hit.subject_id).length*0.2).to_i}"
+    end
+   end
+   
+ end
  
  def exec_seq(seq,blast_query)
    if blast_query.query_id != seq.seq_name
@@ -62,8 +74,10 @@ class PluginAbAdapters < Plugin
     # blast_table_results = blast.do_blast(seq.seq_fasta)             #rise seq to adapterss  executing over blast
     
      #BlastTableResult.new(res)
-     # puts blast.get_blast_cmd
+     # puts blast_query.inspect
      # puts blast_table_results.inspect
+     
+     filter_hits(blast_query.hits, seq.seq_fasta.length)
     
     adapters=[]
     # blast_table_results.querys.each do |query|     # first round to save adapters without overlap
@@ -76,6 +90,10 @@ class PluginAbAdapters < Plugin
       merge_hits(adapters2,adapters)
     end until (adapters2.count == adapters.count)
 
+    # puts "MERGED"
+    # puts "="*50
+    # adapters.each {|a| puts a.inspect}
+    
     max_to_end=@params.get_param('max_ab_to_end').to_i
     # type = 'ActionAbAdapter'
     actions=[]   

data/lib/seqtrimnext.rb

@@ -30,7 +30,7 @@ module Seqtrimnext
   # SEQTRIM_VERSION_STAGE = 'b'
   # SEQTRIM_VERSION = "2.0.0#{SEQTRIM_VERSION_STAGE}#{SEQTRIM_VERSION_REVISION}"
   
-   VERSION = '2.0.33'
+   VERSION = '2.0.35'
   
   SEQTRIM_VERSION = VERSION
   

metadata

@@ -2,7 +2,7 @@
 name: seqtrimnext
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 2.0.33
+  version: 2.0.35
 platform: ruby
 authors: 
 - Dario Guerrero & Almudena Bocinos
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-06-17 00:00:00 Z
+date: 2011-06-22 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: narray