fixed seq qual length check

[biopieces.git] / bp_bin / write_fastq_files

diff --git a/bp_bin/write_fastq_files b/bp_bin/write_fastq_files
index eab5a870aedb7a5622efc4e68b8200e79ddf5cf9..b475b121eeec6cead74d03a9c767023d640870ef 100755 (executable)
--- a/bp_bin/write_fastq_files
+++ b/bp_bin/write_fastq_files
@@ -32,13 +32,21 @@ require 'maasha/biopieces'
 require 'maasha/fastq'
 require 'pp'
 
+allowed_enc = 'base_33,base_64'
+
 casts = []
-casts << {:long=>'key',       :short=>'k', :type=>'string', :mandatory=>true,  :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'dir',       :short=>'d', :type=>'dir!',   :mandatory=>true,  :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'no_stream', :short=>'x', :type=>'flag',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {long: 'key',       short: 'k', type: 'string', mandatory: true,  default: nil,       allowed: nil,               disallowed: nil}
+casts << {long: 'dir',       short: 'd', type: 'dir!',   mandatory: true,  default: nil,       allowed: nil,               disallowed: nil}
+casts << {long: 'prefix',    short: 'p', type: 'string', mandatory: false, default: nil,       allowed: nil,               disallowed: nil}
+casts << {long: 'no_stream', short: 'x', type: 'flag',   mandatory: false, default: nil,       allowed: nil,               disallowed: nil}
+casts << {long: 'compress',  short: 'Z', type: 'string', mandatory: false, default: nil,       allowed: "gzip,bzip,bzip2", disallowed: nil}
+casts << {long: 'encoding',  short: 'e', type: 'string', mandatory: false, default: 'base_33', allowed: allowed_enc,       disallowed: nil}
 
 options = Biopieces.options_parse(ARGV, casts)
 
+encoding = options[:encoding].to_sym
+compress = options[:compress] ? options[:compress].to_sym : nil
+
 key = options[:key].to_sym
 
 fh_hash = {}
@@ -46,17 +54,25 @@ fh_hash = {}
 Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
   input.each_record do |record|
     if record[:SEQ_NAME] and record[:SEQ] and record[:SCORES] and record[key]
-      seq = Seq.new_bp(record)
+      entry = Seq.new_bp(record)
+      entry.qual_convert!(:base_33, encoding)
 
-      if fh_hash.has_key? record[key].to_sym
+      if fh_hash[record[key].to_sym]
         fastq_io = fh_hash[record[key].to_sym]
       else
-        fastq_file = File.join(options[:dir], record[key] + ".fastq")
-        fastq_io   = Fastq.open(fastq_file, "w")
+        if options[:prefix]
+          fastq_file = File.join(options[:dir], [options[:prefix], record[key]].join("_") + ".fastq")
+        else
+          fastq_file = File.join(options[:dir], record[key] + ".fastq")
+        end
+
+        fastq_file << ".gz"  if compress == :gzip
+        fastq_file << ".bz2" if compress == :bzip or compress == :bzip2
+        fastq_io   = Fastq.open(fastq_file, "w", compress: compress)
         fh_hash[record[key].to_sym] = fastq_io
       end
 
-      fastq_io.puts seq.to_fastq
+      fastq_io.puts entry.to_fastq
     end
 
     output.puts record unless options[:no_stream]