X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bp_bin%2Fmask_seq;h=3cd162a264aae007105932017a4a59636e19ddfb;hb=44f810b56c353bfe302469a8313b0aada285b27e;hp=fe70a64a5d554612bbcff68ef8d44cafee280092;hpb=43d1572d3b70aee8426471eaf6d786a456cb2d3e;p=biopieces.git

diff --git a/bp_bin/mask_seq b/bp_bin/mask_seq
index fe70a64..3cd162a 100755
--- a/bp_bin/mask_seq
+++ b/bp_bin/mask_seq
@@ -24,47 +24,33 @@
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
-# Soft mask sequences in the stream based on Solexa/Illumina/Phred type quality scores.
+# Mask sequences in the stream based on quality scores.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+require 'maasha/biopieces'
+require 'maasha/seq'
 
-require 'biopieces'
+casts = []
+casts << {:long=>'cutoff',   :short=>'c', :type=>'int',  :mandatory=>false, :default=>20,  :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'hardmask', :short=>'h', :type=>'flag', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+
+options = Biopieces.options_parse(ARGV, casts)
 
-# Expading class Hash with possibly evil monkey patch.
-class Hash
-  # Soft masks sequence residues where the corresponding quality score
-  # is below a given cutoff.
-  def mask_seq!(cutoff, base)
-    if self.has_key? :SEQ and self.has_key? :SCORES
-      seq    = self[:SEQ].upcase
-      scores = self[:SCORES]
-      i      = 0
+Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
+  input.each_record do |record|
+    if record.has_key? :SEQ
+      entry = Seq.new_bp(record)
 
-      scores.each_char do |score|
-        seq[i] = seq[i].downcase if score.ord - base < cutoff
-        i += 1
-      end
+      options[:hardmask] ? entry.mask_seq_hard!(options[:cutoff]) : entry.mask_seq_soft!(options[:cutoff])
 
-      self[:SEQ] = seq
+      record[:SEQ] = entry.seq
     end
 
-    self
+    output.puts record
   end
 end
 
-casts = []
-casts << {:long=>'cutoff', :short=>'c', :type=>'int',  :mandatory=>false, :default=>20, :allowed=>nil,        :disallowed=>nil}
-casts << {:long=>'base',   :short=>'b', :type=>'uint', :mandatory=>false, :default=>64, :allowed=>"33,59,64", :disallowed=>nil}
-
-bp = Biopieces.new
-
-options = bp.parse(ARGV, casts)
-
-bp.each_record do |record|
-  bp.puts record.mask_seq!(options[:cutoff], options[:base])
-end
-
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<