X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bp_bin%2Fread_fastq;h=44a69ed871cd40f686d4f9be675258377b4c6ebf;hb=32a9953ec9526fd800a1d564f9bc7c70965689d6;hp=eb926618efdeae2dff7bd5ddc50fab0fd1771b43;hpb=d3e2e8686941aa5a790e981fb1e631c37fcc3517;p=biopieces.git

diff --git a/bp_bin/read_fastq b/bp_bin/read_fastq
index eb92661..44a69ed 100755
--- 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) 2007-2013 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
@@ -18,84 +18,73 @@
 
 # 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 => 'convert2dec', short => 'c', type => 'flag',   mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'cutoff',      short => 'C', type => 'int',    mandatory => 'no', default => 20,    allowed => undef, disallowed => undef },
-        { long => 'soft_mask',   short => 's', type => 'flag',   mandatory => 'no', default => undef, 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 ) )
-        {
-            Maasha::Fastq::softmask_phred_str( $record->{ 'SEQ' }, $record->{ 'SCORES' }, $options->{ 'cutoff' } ) if $options->{ 'soft_mask' };
-            $record->{ 'SCORES' } = Maasha::Fastq::phred_str2dec_str( $record->{ 'SCORES' } ) if $options->{ 'convert2dec' };
-
-            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.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
-
-BEGIN
-{
-    Maasha::Biopieces::status_set();
-}
-
-
-END
-{
-    Maasha::Biopieces::status_log();
-}
+require 'maasha/biopieces'
+require 'maasha/fastq'
+
+allowed_enc = 'auto,sanger,solexa,illumina1.3,illumina1.5,illumina1.8'
+
+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=>'encoding', :short=>'e', :type=>'string', :mandatory=>false, :default=>'auto', :allowed=>allowed_enc, :disallowed=>nil}
+
+options = Biopieces.options_parse(ARGV, casts)
+
+num   = 0
+first = true
+last  = false
+
+Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
+  unless options[:data_in] and options[:data_in].first == '-'
+    input.each_record do |record|
+      output.puts record
+    end
+  end
+
+  if options[:data_in]
+    options[:data_in].each do |file|
+      encoding = options[:encoding].downcase.delete('.')
+
+      Fastq.open(file, mode='r') do |fastq|
+        fastq.each do |entry|
+          if encoding == 'auto'
+            if entry.qual_base33? # sanger or illumina18
+              encoding = 'illumina18'
+            elsif entry.qual_base64? # solexa or illumina13 or illumina15
+              encoding = 'illumina13'
+            else
+              raise SeqError, "Could not auto-detect quality score encoding"
+            end
+          elsif first
+            #raise SeqError, "Quality score outside valid range" unless entry.qual_valid?(encoding)
+            first = false
+          end
+
+          entry.convert_scores!(encoding, 'illumina13')
+          output.puts entry.to_bp
+          num += 1
+
+          if options[:num] == num
+            last = true
+            break
+          end
+        end
+      end
+
+      break if last
+    end
+  end
+end
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<