X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bp_bin%2Ftrim_seq;h=c89eefb6d6ef95c18c0f89279a6370956e7d3874;hb=48bea5c28b89dc5586d0bddb338ccd6ba23aa1f9;hp=4cda52483cd036d6de09be3146e22aa3df502d42;hpb=d2b48a034defcacac6e9134894560818b2790426;p=biopieces.git

diff --git a/bp_bin/trim_seq b/bp_bin/trim_seq
index 4cda524..c89eefb 100755
--- a/bp_bin/trim_seq
+++ b/bp_bin/trim_seq
@@ -1,6 +1,6 @@
-#!/usr/bin/env perl
+#!/usr/bin/env ruby
 
-# Copyright (C) 2007-2010 Martin A. Hansen.
+# Copyright (C) 2007-2012 Martin A. Hansen.
 
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License
@@ -18,85 +18,46 @@
 
 # http://www.gnu.org/copyleft/gpl.html
 
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-# Trim sequence ends for residues with a low quality score.
-
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# This program is part of the Biopieces framework (www.biopieces.org).
 
-use warnings;
-use strict;
-use Maasha::Biopieces;
-use Maasha::Fastq;
+# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# Trim sequence ends removing residues with a low quality score.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+require 'maasha/biopieces'
+require 'maasha/seq'
+require 'pp'
 
-my ( $options, $in, $out, $record, $left, $right, $pos_l, $pos_r );
-
-$options = Maasha::Biopieces::parse_options(
-    [
-        { long => 'min',  short => 'm', type => 'uint',   mandatory => 'no', default => 15,     allowed => undef,             disallowed => 0 },
-        { long => 'trim', short => 't', type => 'string', mandatory => 'no', default => 'both', allowed => 'left,right,both', disallowed => undef },
-    ]   
-);
-
-$right = 1 if $options->{ 'trim' } eq 'right';
-$left  = 1 if $options->{ 'trim' } eq 'left';
-$right = 1 if $options->{ 'trim' } eq 'both';
-$left  = 1 if $options->{ 'trim' } eq 'both';
+casts = []
+casts << {:long=>'min_qual', :short=>'m', :type=>'uint',   :mandatory=>true, :default=>20,     :allowed=>nil,               :disallowed=>'0'}
+casts << {:long=>'min_len',  :short=>'l', :type=>'uint',   :mandatory=>true, :default=>3,      :allowed=>nil,               :disallowed=>'0'}
+casts << {:long=>'trim',     :short=>'t', :type=>'string', :mandatory=>true, :default=>'both', :allowed=>'left,right,both', :disallowed=>nil}
 
-$in  = Maasha::Biopieces::read_stream( $options->{ "stream_in" } );
-$out = Maasha::Biopieces::write_stream( $options->{ "stream_out" } );
+options = Biopieces.options_parse(ARGV, casts)
 
-while ( $record = Maasha::Biopieces::get_record( $in ) )
-{
-    if ( exists $record->{ 'SEQ' } and exists $record->{ 'SCORES' } )
-    {
-        if ( $right )
-        {
-            $pos_r = Maasha::Fastq::trim_right( $record->{ 'SCORES' }, $options->{ 'min' } );
-        
-            $record->{ 'SEQ' }       = substr $record->{ 'SEQ' }, 0, $pos_r + 1;
-            $record->{ 'SCORES' }    = substr $record->{ 'SCORES' }, 0, $pos_r + 1; 
-            $record->{ 'TRIM_RIGHT' } = $pos_r;
-        }
-
-        if ( $left )
-        {
-            $pos_l = Maasha::Fastq::trim_left( $record->{ 'SCORES' }, $options->{ 'min' } );
-        
-            $record->{ 'SEQ' }      = substr $record->{ 'SEQ' }, $pos_l;
-            $record->{ 'SCORES' }   = substr $record->{ 'SCORES' }, $pos_l;
-            $record->{ 'TRIM_LEFT' } = $pos_l;
-        }
-
-        $record->{ 'SEQ_LEN' } = length $record->{ 'SEQ' };
-    }
-
-    Maasha::Biopieces::put_record( $record, $out );
-}
-
-Maasha::Biopieces::close_stream( $in );
-Maasha::Biopieces::close_stream( $out );
-
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+trim = options[:trim].to_sym
 
+Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
+  input.each_record do |record|
+    if record[:SEQ] and record[:SCORES]
+      entry = Seq.new_bp(record)
 
-BEGIN
-{
-    Maasha::Biopieces::status_set();
-}
+      case trim
+      when :both  then entry.quality_trim!(options[:min_qual], options[:min_len])
+      when :left  then entry.quality_trim_left!(options[:min_qual], options[:min_len])
+      when :right then entry.quality_trim_right!(options[:min_qual], options[:min_len])
+      end
 
+      record.merge! entry.to_bp
+    end
 
-END
-{
-    Maasha::Biopieces::status_log();
-}
+    output.puts record
+  end
+end
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<