seqtrimnext 2.0.41 → 2.0.42

data/History.txt

@@ -1,3 +1,7 @@
+=== 2.0.42 2012-02-01
+
+Added custom cd-hit cmd parameters
+
 === 2.0.41 2011-11-04
 
 Added extra adapters plugin for transcriptomics

data/lib/seqtrimnext/classes/seqtrim.rb

@@ -27,12 +27,29 @@ class Seqtrim
   # reads the sequences from files 'fasta' , until now without qualities,
   # and executes the plugins over the sequences in the pool of threads
   
+
+  def get_custom_cdhit(cd_hit_input_file,params)
+    cmd=''
+    begin
+      cdhit_custom_parameters=params.get_param('cdhit_custom_parameters').strip
+      
+      if !cdhit_custom_parameters.nil? and !cdhit_custom_parameters.empty?
+        cmd = "cd-hit-454 -i #{cd_hit_input_file} -o clusters.fasta #{cdhit_custom_parameters} > cd-hit-454.out"
+      end
+    
+    rescue Exception => exception #not an integer, send via ssh to other machine
+      cmd=''
+    end
+    
+    return cmd
+  end
   
   def get_cd_hit_cmd(cd_hit_input_file,workers,init_file_path)
     
     num_cpus_cdhit=1
     cmd=''
     
+    
     # if workers is an integer, reduce it by one in the server
 		begin
 		  Integer(workers)
@@ -72,6 +89,8 @@ class Seqtrim
 	    end
     end
     
+    
+    
     return cmd
   end
   
@@ -94,6 +113,10 @@ class Seqtrim
 	  default_value='true'
 	  params.check_param(errors,'remove_clonality','String',default_value,comment)
 
+    comment='Custom parameters used by CD-HIT-454 (leave empty to let seqtrimnext decide). Execute "cd-hit-454 help" in command line to see a list of parameters'
+	  default_value=''
+	  params.check_param(errors,'cdhit_custom_parameters','String',default_value,comment)
+
     comment='Generate initial stats'
 	  default_value='true'
 	  params.check_param(errors,'generate_initial_stats','String',default_value,comment)
@@ -241,8 +264,10 @@ class Seqtrim
       
       #execute cd-hit
       if params.get_param('remove_clonality')=='true'
-        
-        cmd=get_cd_hit_cmd(cd_hit_input_file,workers,$SEQTRIMNEXT_INIT)
+        cmd=get_custom_cdhit(cd_hit_input_file,params)
+        if cmd.empty?
+          cmd=get_cd_hit_cmd(cd_hit_input_file,workers,$SEQTRIMNEXT_INIT)
+        end
         
         $LOG.info "Executing cd-hit-454: #{cmd}"
         

data/lib/seqtrimnext/templates/transcriptomics_plants.txt

@@ -3,6 +3,6 @@
 # ======================================
 
 
-plugin_list = PluginLowHighSize,PluginMids,PluginIndeterminations,PluginAbAdapters,PluginFindPolyAt,PluginContaminants,PluginVectors,PluginLowQuality,PluginLowComplexity
+plugin_list = PluginLowHighSize,PluginMids,PluginIndeterminations,PluginAbAdapters,PluginAdapters,PluginFindPolyAt,PluginContaminants,PluginVectors,PluginLowQuality,PluginLowComplexity
 
 contaminants_db="contaminants.fasta cont_ribosome.fasta cont_mitochondrias.fasta cont_plastids.fasta"

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 = '2.0.41'
+   VERSION = '2.0.42'
   
   SEQTRIM_VERSION = VERSION
   

metadata

@@ -2,7 +2,7 @@
 name: seqtrimnext
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 2.0.41
+  version: 2.0.42
 platform: ruby
 authors: 
 - Dario Guerrero & Almudena Bocinos
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-11-07 00:00:00 Z
+date: 2012-02-01 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: narray