rewrite of FASTQ internals

[biopieces.git] / bp_bin / read_fastq

diff --git a/bp_bin/read_fastq b/bp_bin/read_fastq
index 088f8d6f06adf21dd5b109fe6097a3a34ecbfa15..78a3fcceaf689bcf198e2544fb45316e8c92cb4b 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) 2007-2013 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 +18,77 @@
 
 # http://www.gnu.org/copyleft/gpl.html
 
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-# Read FASTQ entries from one or more files.
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-
-use warnings;
-use strict;
-use Maasha::Biopieces;
-use Maasha::Filesys;
-use Maasha::Fastq;
-
-
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# This program is part of the Biopieces framework (www.biopieces.org).
 
-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 ) )
-        {
-            Maasha::Fastq::lowercase_low_scores( $record->{ 'SEQ' }, $record->{ 'SCORES' }, $options->{ 'quality' } );
-
-            $record->{ 'SCORES' }     =~ s/(.)/ord( $1 ) - 33 . ";"/ge; # http://maq.sourceforge.net/fastq.shtml
-            $record->{ 'SCORE_MEAN' } = sprintf( "%.2f", Maasha::Calc::mean( [ split /;/, $record->{ 'SCORES' } ] ) );
-
-            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,base_33,base_64'
+
+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)
+
+MAX_TEST = 1_000
+
+num  = 0
+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.to_sym
+
+      Fastq.open(file, mode='r') do |fastq|
+        fastq.each do |entry|
+          if encoding == :auto
+            if entry.qual_base33?
+              encoding = :base_33
+            elsif entry.qual_base64?
+              encoding = :base_64
+            else
+              raise SeqError, "Could not auto-detect quality score encoding"
+            end
+          end
+
+          entry.qual_convert!(encoding, :base_33)
+          entry.qual_coerce!(:base_33)
+
+          if num < MAX_TEST
+            raise SeqError, "Quality score outside valid range" unless entry.qual_valid?(:base_33)
+          end
+
+          output.puts entry.to_bp
+          num += 1
+
+          if options[:num] == num
+            last = true
+            break
+          end
+        end
+      end
+
+      break if last
+    end
+  end
+end
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<