seqtrimnext 2.0.42 → 2.0.45

data/History.txt

@@ -1,3 +1,15 @@
+=== 2.0.45 2012-03-05
+
+Improved LowComplexity plugin to ignore low complexity regions inside low qual regions
+
+=== 2.0.44 2012-02-23
+
+Skip output
+
+=== 2.0.43 2012-02-22
+
+Lower logger levels
+
 === 2.0.42 2012-02-01
 
 Added custom cd-hit cmd parameters

data/bin/seqtrimnext

@@ -265,6 +265,16 @@ optparse = OptionParser.new do |opts|
   opts.on( '-j', '--json', 'Save results in json file' ) do
     options[:json] = true
   end
+  
+  options[:skip_output] = false
+  opts.on( '-K', '--no-verbose', 'Change to no verbose mode. Every sequence will not be written to output log' ) do
+    options[:skip_output] = true
+  end
+  
+  options[:skip_report] = false
+  opts.on( '-R', '--no-report', 'Change to no verbose mode. Every sequence will not be written to output log' ) do
+    options[:skip_report] = true
+  end
 
   # This displays the help screen, all programs are
   # assumed to have this option.
@@ -300,6 +310,8 @@ $LOG.datetime_format = "%Y-%m-%d %H:%M:%S"
 
 #DEBUG < INFO < WARN < ERROR < FATAL < UNKNOWN
 
+$LOG.info("SeqTrimNext version #{Seqtrimnext::SEQTRIM_VERSION}")
+
 
 $LOG.info("Using BLASTDB: "+ $FORMATTED_DB_PATH)
 $LOG.info("Using options: "+ options.to_json)
@@ -370,13 +382,20 @@ end
 
 s = Seqtrim.new(options)
 
-#generate report
 
 
-if system("which generate_report.rb > /dev/null ")
+#generate report
+
+if !options[:skip_report] && system("which generate_report.rb > /dev/null ")
   cmd="generate_report.rb output_files 2> report_generation_errors.log"
   $LOG.info "Generating report #{cmd}"
   `#{cmd}`
 else
-  $LOG.info "If you want a detailed report in PDF format, install the optional seqtrimnext_report gem (gem install seqtrimnext_report)"
+  skip_text='.'
+  
+  if options[:skip_report]
+    skip_text=' and remove the -R option from the command line.'
+  end
+  
+  $LOG.info "If you want a detailed report in PDF format, be sure you have installed the optional seqtrimnext_report gem (gem install seqtrimnext_report)#{skip_text}"
 end

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 = '2.0.42'
+   VERSION = '2.0.45'
   
   SEQTRIM_VERSION = VERSION
   

data/lib/seqtrimnext/actions/action_low_quality.rb

@@ -11,8 +11,10 @@ class ActionLowQuality < SeqtrimAction
   
   def initialize(start_pos,end_pos)
      super(start_pos,end_pos)    
+     # esto es cut=false porque al principio el plugin lowqual estaba al inicio del pipeline y habia que dejar 
+     # la secuencia larga para que se encontrasen los contaminantes y vectores
+     # Tambien esta por si un linker tiene baja calidad que pueda encontrarlo
      @cut =false
-
    end
    
 # def apply_to(seq)

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

@@ -13,11 +13,13 @@ STATS_PATH=File.join(OUTPUT_PATH,'stats.json')
 
 class SeqtrimWorkManager < ScbiMapreduce::WorkManager
 
-  def self.init_work_manager(sequence_reader, params, chunk_size = 100, use_json=false)
+  def self.init_work_manager(sequence_reader, params, chunk_size = 100, use_json=false, skip_output=false)
     @@full_stats={}
     @@params= params
     @@exit = false
 
+    @@skip_output=skip_output
+
     @@chunk_size = chunk_size
 
 
@@ -173,17 +175,17 @@ class SeqtrimWorkManager < ScbiMapreduce::WorkManager
 
     # reset count stats since they are repeated by checkpointing
 
-    # if @@full_stats['sequences'] && @@full_stats['repeated']
-    #   @@full_stats['sequences']['count']['repeated']=0
-    # end
-    #
-    # if @@full_stats['sequences'] && @@full_stats['processed']
-    #   @@full_stats['sequences']['processed']['count']=0
-    # end
-    #
-    # if @@full_stats['sequences'] && @@full_stats['total']
-    #   @@full_stats['sequences']['total']['count']=0
-    # end
+    if @@full_stats['sequences'] && @@full_stats['repeated']
+      @@full_stats['sequences']['count']['repeated']=0
+    end
+    
+    if @@full_stats['sequences'] && @@full_stats['processed']
+      @@full_stats['sequences']['processed']['count']=0
+    end
+    
+    if @@full_stats['sequences'] && @@full_stats['total']
+      @@full_stats['sequences']['total']['count']=0
+    end
 
     super
     # return checkpoint
@@ -288,14 +290,16 @@ class SeqtrimWorkManager < ScbiMapreduce::WorkManager
   #   end
 
   def work_received(obj)
-
+    
     res = obj
 
     # collect stats
     @@full_stats.add_stats(obj.stats)
 
     # print output in screen
-    puts obj.output_text
+    if !@@skip_output
+      puts obj.output_text
+    end
 
     # save results to files
     save_files(obj)

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

@@ -74,8 +74,11 @@ class SeqtrimWorker <  ScbiMapreduce::Worker
     end
 
 		def starting_worker
+
         # $WORKER_LOG.level = Logger::ERROR
+        $WORKER_LOG.level = Logger::WARN
 				$WORKER_LOG.info "Loading actions"
+
 				@action_manager = ActionManager.new
 
 				$WORKER_LOG.info "Loading plugins"

data/lib/seqtrimnext/classes/seqtrim.rb

@@ -137,6 +137,7 @@ class Seqtrim
 		comment='Seqtrim version'
 		default_value=Seqtrimnext::SEQTRIM_VERSION
 		params.check_param(errors,'seqtrim_version','String',default_value,comment)
+
 		
 		if !errors.empty?
           $LOG.error 'Please, define the following global parameters in params file:'
@@ -308,7 +309,7 @@ class Seqtrim
       else
   			$LOG.info 'Starting server'
       	        
- 				SeqtrimWorkManager.init_work_manager(sequence_reader, params,chunk_size,use_json)
+ 				SeqtrimWorkManager.init_work_manager(sequence_reader, params,chunk_size,use_json,options[:skip_output])
 				
         begin
   				cpus=1

data/lib/seqtrimnext/plugins/plugin_low_complexity.rb

@@ -54,17 +54,33 @@ class PluginLowComplexity < Plugin
       # puts found_dust.to_json
       total_dust=0
       if !dust_query.nil?
+        # low_quals=seq.get_actions(ActionLowQuality)
+        
         dust_query.dust.each do |dust|
           start=dust[0]
           stop=dust[1]
           dust_size=dust[1]-dust[0]+1
-          total_dust+=dust_size
+          
 
           if (dust_size)>=MIN_DUST_SIZE
           
-            a = seq.new_action(start,stop,'ActionLowComplexity')
-            # a.left_action=true
-            actions.push a
+            # check if low complexity is inside a lowqual region
+            # in_low_qual=false
+            # low_quals.each do |lq|
+            #   if lq.contains_action?(start,stop,0)
+            #     in_low_qual=true
+            #     break
+            #   end
+            # end
+            
+            if !seq.range_inside_action_type?(start,stop,ActionLowQuality)
+              
+              total_dust+=dust_size    
+              a = seq.new_action(start,stop,'ActionLowComplexity')
+              # a.left_action=true
+              actions.push a
+              
+            end
             # break
           end
         end

data/lib/seqtrimnext/templates/amplicons.txt

@@ -2,7 +2,7 @@
 # General parameters to extract Amplicons
 # ======================================
 
-plugin_list = PluginLowHighSize,PluginKey,PluginMids,PluginIndeterminations,PluginAbAdapters,PluginLowQuality,PluginAmplicons
+plugin_list = PluginLowHighSize,PluginKey,PluginMids,PluginIndeterminations,PluginAbAdapters,PluginAmplicons,PluginLowQuality
 
 # do not remove cloned sequences
 remove_clonality=false

metadata

@@ -2,7 +2,7 @@
 name: seqtrimnext
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 2.0.42
+  version: 2.0.45
 platform: ruby
 authors: 
 - Dario Guerrero & Almudena Bocinos
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2012-02-01 00:00:00 Z
+date: 2012-03-05 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: narray