rewrote assemble_seq_velvet

[biopieces.git] / bp_bin / read_fastq

diff --git a/bp_bin/read_fastq b/bp_bin/read_fastq
index 923a3b057c7198b9792a799fa82e4b2102d26f8a..c2f74cd7bb7a07219ac43ecfd574ad726f5b8f00 100755 (executable)
--- a/bp_bin/read_fastq
+++ b/bp_bin/read_fastq
@@ -31,12 +31,12 @@
 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}
+allowed_enc = 'auto,sanger,solexa,illumina1.3,illumina1.5,illumina1.8'
 
-PHRED_SCORES = Regexp.new('[!"#$%&\'()*+,-./0123456789:]')
+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)
 
@@ -44,20 +44,33 @@ num  = 0
 last = false
 
 Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
-  input.each_record do |record|
-    output.puts record
+  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]
     options[:data_in].each do |file|
+      encoding = options[:encoding].downcase.delete('.')
+
       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]
+          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
+          end
+
+          entry.convert_scores!(encoding, 'illumina13')
           output.puts entry.to_bp
           num += 1
 
-          if options.has_key? :num and options[:num] == num
+          if options[:num] == num
             last = true
             break
           end