changed layout of ruby source

[biopieces.git] / bp_bin / read_fastq

diff --git a/bp_bin/read_fastq b/bp_bin/read_fastq
index 8aeedf9e1ca52b6fe9ced2334ddee3a12379b4e4..063edd6ce2883bbe5b46f286c59175c71cf55456 100755 (executable)
--- a/bp_bin/read_fastq
+++ b/bp_bin/read_fastq
@@ -1,6 +1,6 @@
-#!/usr/bin/env perl
+#!/usr/bin/env ruby
 
-# Copyright (C) 2007-2009 Martin A. Hansen.
+# Copyright (C) 2007-2010 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
@@ -18,112 +18,56 @@
 
 # http://www.gnu.org/copyleft/gpl.html
 
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-# Read FASTQ entries from one or more files.
-
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# This program is part of the Biopieces framework (www.biopieces.org).
 
-use warnings;
-use strict;
-use Maasha::Biopieces;
-use Maasha::Filesys;
-use Maasha::Fastq;
-
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-
-my ( $options, $in, $out, $record, $data_in, $num, $entry );
-
-$options = Maasha::Biopieces::parse_options(
-    [
-        { long => 'data_in', short => 'i', type => 'files!', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'num',     short => 'n', type => 'uint',   mandatory => 'no', default => undef, allowed => undef, disallowed => 0     },
-        { long => 'quality', short => 'q', type => 'uint',   mandatory => 'no', default => 20,    allowed => undef, disallowed => undef },
-    ]   
-);
-
-$in  = Maasha::Biopieces::read_stream( $options->{ "stream_in" } );
-$out = Maasha::Biopieces::write_stream( $options->{ "stream_out" } );
-
-while ( $record = Maasha::Biopieces::get_record( $in ) ) {
-    Maasha::Biopieces::put_record( $record, $out );
-}
-
-if ( $options->{ 'data_in' } )
-{
-    $data_in = Maasha::Filesys::files_read_open( $options->{ 'data_in' } );
-
-    $num = 1;
-
-    while ( $entry = Maasha::Fastq::get_entry( $data_in ) ) 
-    {
-        if ( $record = Maasha::Fastq::fastq2biopiece( $entry ) )
-        {
-            lowercase_low_scores( $record, $options->{ 'quality' } );
-
-            Maasha::Biopieces::put_record( $record, $out );
-        }
-
-        last if $options->{ "num" } and $num == $options->{ "num" };
-
-        $num++;
-    }
-
-    close $data_in;
-}
-
-Maasha::Biopieces::close_stream( $in );
-Maasha::Biopieces::close_stream( $out );
+# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# Read FASTQ entries from one or more files.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+require 'maasha/biopieces'
+require 'maasha/fastq'
 
-sub lowercase_low_scores
-{
-    # Martin A. Hansen, July 2009.
+casts = []
+casts << {:long=>'data_in', :short=>'i', :type=>'files!', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'num',     :short=>'n', :type=>'uint',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>'0'}
+casts << {:long=>'solexa',  :short=>'s', :type=>'flag',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
 
-    # Lowercase residues in sequence that are below a given score threshold.
+PHRED_SCORES = Regexp.new('[!"#$%&\'()*+,-./0123456789:]')
 
-    # This one would be lovely to have written in C.
+bp = Biopieces.new
 
-    my ( $record,    # Biopiece record
-         $quality,   # quality threshold
-       ) = @_;
+options = bp.parse(ARGV, casts)
 
-    # Returns nothing.
-
-    my ( @seqs, @scores, $i );
-
-    @seqs   = split //, $record->{ 'SEQ' };
-    @scores = split /;/, $record->{ 'SCORES' };
-
-    for ( $i = 0; $i < @scores; $i++ ) {
-        $seqs[ $i ] = lc $seqs[ $i ] if $scores[ $i ] < $quality;
-    }
-
-    $record->{ 'SEQ' }    = join "",  @seqs;
-    $record->{ 'SCORES' } = join ";", @scores;
-}
-
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+bp.each_record do |record|
+  bp.puts record
+end
 
+num  = 0
+last = false
 
-BEGIN
-{
-    Maasha::Biopieces::status_set();
-}
+if options.has_key? :data_in
+  options[:data_in].each do |file|
+    Fastq.open(file, mode='r') do |fastq|
+      fastq.each do |entry|
+        entry.convert_phred2illumina!  if entry.qual.match PHRED_SCORES
+        entry.convert_solexa2illumina! if options[:solexa]
+        bp.puts entry.to_bp
+        num += 1
 
+        if options.has_key? :num and options[:num] == num
+          last = true
+          break
+        end
+      end
+    end
 
-END
-{
-    Maasha::Biopieces::status_log();
-}
+    break if last
+  end
+end
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<