X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bp_bin%2Fread_fastq;h=e4fd75f257357988657a205f984a41f356d9eeb4;hb=5de6112b70b59420b245ce636a8b2e3c90acbe00;hp=8fec43bb6724288b527378f53c1cafde228443dc;hpb=144447a01acd669161794a3ca0d034bf99323840;p=biopieces.git

diff --git a/bp_bin/read_fastq b/bp_bin/read_fastq
index 8fec43b..e4fd75f 100755
--- a/bp_bin/read_fastq
+++ b/bp_bin/read_fastq
@@ -1,6 +1,6 @@
 #!/usr/bin/env ruby
 
-# Copyright (C) 2007-2011 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
@@ -31,35 +31,102 @@
 require 'maasha/biopieces'
 require 'maasha/fastq'
 
-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}
+MAX_TEST = 1_000
+
+allowed_enc = 'auto,base_33,base_64'
 
-PHRED_SCORES = Regexp.new('[!"#$%&\'()*+,-./0123456789:]')
+casts = []
+casts << {:long=>'data_in',  :short=>'i', :type=>'files!', :mandatory=>false, :default=>nil,    :allowed=>nil,         :disallowed=>nil}
+casts << {:long=>'data_in2', :short=>'j', :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)
 
+encoding = options[:encoding].to_sym
+
 num  = 0
 last = false
 
 Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
-  unless options[:data_in].first == '-'
+  unless options[:data_in] and options[:data_in].first == '-'
     input.each_record do |record|
       output.puts record
     end
   end
 
-  if options.has_key? :data_in
+  if options[:data_in] and options[:data_in2]
+    raise "data_in files differ" if options[:data_in].size != options[:data_in2].size
+
+    (0 ... options[:data_in].size).each do |i|
+      file1 = options[:data_in][i]
+      file2 = options[:data_in2][i]
+
+      io1 = Fastq.open(file1, 'r')
+      io2 = Fastq.open(file2, 'r')
+
+      while entry1 = io1.get_entry and entry2 = io2.get_entry
+        if encoding == :auto
+          if entry1.qual_base33? or entry2.qual_base33?
+            encoding = :base_33
+          elsif entry1.qual_base64? or entry2.qual_base64?
+            encoding = :base_64
+          else
+            raise SeqError, "Could not auto-detect quality score encoding"
+          end
+        end
+
+        entry1.qual_convert!(encoding, :base_33)
+        entry2.qual_convert!(encoding, :base_33)
+        entry1.qual_coerce!(:base_33)
+        entry2.qual_coerce!(:base_33)
+
+        if num < MAX_TEST
+          raise SeqError, "Quality score outside valid range" unless entry1.qual_valid?(:base_33)
+          raise SeqError, "Quality score outside valid range" unless entry2.qual_valid?(:base_33)
+        end
+
+        output.puts entry1.to_bp
+        output.puts entry2.to_bp
+
+        num += 2
+
+        if options[:num] and num >= options[:num]
+          last = true
+          break
+        end
+      end
+
+      io1.close
+      io2.close
+
+      break if last
+    end
+  elsif options[:data_in]
     options[:data_in].each do |file|
-      Fastq.open(file, mode='r') do |fastq|
+      Fastq.open(file, 'r') do |fastq|
         fastq.each do |entry|
-          entry.convert_phred2illumina!  if entry.qual.match PHRED_SCORES
-          entry.convert_solexa2illumina! if options[:solexa]
+          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.has_key? :num and options[:num] == num
+          if options[:num] == num
             last = true
             break
           end