seqtrimnext 2.0.46 → 2.0.48

data/History.txt

@@ -1,3 +1,11 @@
+=== 2.0.48 2012-05-16
+
+Select biggest sequence on cluster 
+
+=== 2.0.47 2012-05-14
+
+Added option for illumina qualities based on phred+64 scores
+
 === 2.0.46 2012-04-13
 
 Checkpointing activated. Jobs can be restarted where stopped.

data/bin/seqtrimnext

@@ -206,6 +206,11 @@ optparse = OptionParser.new do |opts|
   opts.on( '-c', '--check_databases', 'Check Blast databases and reformat if necessary' ) do
     options[:check_db] = true
   end
+
+  options[:use_checkpoint] = false
+  opts.on( '-C', '--use_checkpoint', 'Restore at checkpoint if scbi_mapreduce_checkpoint file is available' ) do
+    options[:use_checkpoint] = true
+  end
   
   # options[:skip_initial_stats] = false
   # opts.on( '-k', '--skip_initial_stats', 'Skip initial stats' ) do
@@ -227,6 +232,16 @@ optparse = OptionParser.new do |opts|
   opts.on( '-Q', '--fastq FILE', 'Fastq input file. Use - for <STDIN>' ) do |file|
     options[:fastq] = file
   end
+  
+  options[:format] = nil
+  opts.on( '-F', '--fastq_quality_format FORMAT', 'Fastq input quality format use sanger or illumina18 for phred+33 based scores. Use illumina15 for phred+64 based scores (default is sanger) file. Use - for <STDIN>' ) do |value|
+    options[:format] = value
+    if !['sanger','illumina15', 'illumina18'].include?(value)
+      STDERR.puts "ERROR: Invalid FASTQ format parameter #{value}"
+      exit
+    end
+  end
+  
 
   options[:fasta] = nil
   opts.on( '-f', '--fasta FILE', 'Fasta input file' ) do |file|

data/lib/seqtrimnext/classes/params.rb

@@ -123,11 +123,16 @@ class Params
         #puts line,line[0]
         # en ruby19 line[0] da el caracter, no el chr
         #if (line[0]!=62) && (line[0]!=48)
-        if (line[0]!='>'[0]) && (line[0]!='0'[0])
+        # if (line[0]!='>'[0]) && (line[0]!='0'[0])
+
+        # line doesn't finish in *
+        if (line[0]!='>'[0]) && (!(line =~ /\*$/))
+          
           #puts line
           # puts line,line[0]
           if line =~ />([^\.]+)\.\.\.\s/
             #puts 'ok'
+            # puts $1
             @clusters[$1]=1
           end
         end

data/lib/seqtrimnext/classes/seqtrim.rb

@@ -160,6 +160,16 @@ class Seqtrim
     only_workers=options[:only_workers]
     chunk_size = options[:chunk_size]
     use_json = options[:json]
+    
+    # check for checkpoint
+    
+    if File.exists?(ScbiMapreduce::CHECKPOINT_FILE)
+      if !options[:use_checkpoint]
+        STDERR.puts "ERROR: A checkpoint file exists, either delete it or provide -C flag to use it"
+        exit
+      end
+    end
+    
        
     
     # it is the server part
@@ -172,13 +182,28 @@ class Seqtrim
     
     # open sequence reader and expand input files paths
     if options[:fastq]
+      
       if options[:fastq]=='-'
         seqs_path = STDIN
       else
         seqs_path = File.expand_path(options[:fastq])
       end
+      
       cd_hit_input_file = seqs_path
-      sequence_reader = FastqFile.new(seqs_path,'r',:sanger, true)
+      
+      # choose fastq quality format
+      format=:sanger
+      
+      case options[:format]
+      when 'sanger'
+        format = :sanger
+      when 'illumina15'
+        format = :ilumina
+      when 'illumina18'
+        format = :sanger
+      end
+      
+      sequence_reader = FastqFile.new(seqs_path,'r',format, true)
       # cd_hit_input_file = 'cd-hit-input.fasta'
       cd_hit_input_file = seqs_path
       # $LOG.info "Converting input file for cd-hit-454"

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 = '2.0.46'
+   VERSION = '2.0.48'
   
   SEQTRIM_VERSION = VERSION
   

metadata

@@ -2,7 +2,7 @@
 name: seqtrimnext
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 2.0.46
+  version: 2.0.48
 platform: ruby
 authors: 
 - Dario Guerrero & Almudena Bocinos
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2012-04-13 00:00:00 Z
+date: 2012-05-16 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: narray