X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=code_ruby%2Flib%2Fmaasha%2Fseq.rb;h=c22ecec8948a0230079aacd05af1e552a168819e;hb=eb15b4040e0ceba1c3df40b7f0a597c469771cee;hp=73bf0552938cef82fa937aa43ff7be23b190c94b;hpb=0ab290c76c4e253ea809f9a7c2321a28de72b286;p=biopieces.git

diff --git a/code_ruby/lib/maasha/seq.rb b/code_ruby/lib/maasha/seq.rb
index 73bf055..c22ecec 100644
--- a/code_ruby/lib/maasha/seq.rb
+++ b/code_ruby/lib/maasha/seq.rb
@@ -30,6 +30,7 @@ require 'narray'
 autoload :BackTrack,   'maasha/seq/backtrack'
 autoload :Dynamic,     'maasha/seq/dynamic'
 autoload :Homopolymer, 'maasha/seq/homopolymer'
+autoload :Hamming,     'maasha/seq/hamming'
 autoload :Levenshtein, 'maasha/seq/levenshtein'
 autoload :Ambiguity,   'maasha/seq/ambiguity'
 
@@ -92,7 +93,7 @@ class Seq
     type     = record[:SEQ_TYPE].to_sym if record[:SEQ_TYPE]
     qual     = record[:SCORES]
 
-    self.new(seq_name, seq, type, qual)
+    self.new(seq_name: seq_name, seq: seq, type: type, qual: qual)
   end
 
   # Class method that generates all possible oligos of a specifed length and type.
@@ -124,16 +125,20 @@ class Seq
     oligos
   end
 
-  # Initialize a sequence object with the following arguments:
-  # - seq_name: Name of the sequence.
-  # - seq: The sequence.
-  # - type: The sequence type - DNA, RNA, or protein
-  # - qual: An Illumina type quality scores string.
-  def initialize(seq_name = nil, seq = nil, type = nil, qual = nil)
-    @seq_name = seq_name
-    @seq      = seq
-    @type     = type
-    @qual     = qual
+  # Initialize a sequence object with the following options:
+  # - :seq_name   Name of the sequence.
+  # - :seq        The sequence.
+  # - :type       The sequence type - DNA, RNA, or protein
+  # - :qual       An Illumina type quality scores string.
+  def initialize(options = {})
+    @seq_name = options[:seq_name]
+    @seq      = options[:seq]
+    @type     = options[:type]
+    @qual     = options[:qual]
+
+    if @seq and @qual and @seq.length != @qual.length
+      raise SeqError, "Sequence length and score length mismatch: #{@seq.length} != #{@qual.length}"
+    end
   end
 
   # Method that guesses and returns the sequence type
@@ -289,8 +294,8 @@ class Seq
     raise SeqError, "Missing seq_name" if self.seq_name.nil? or self.seq_name == ''
     raise SeqError, "Missing seq"      if self.seq.nil?      or self.seq.empty?
 
-    seq_name = self.seq_name.to_s
-    seq      = self.seq.to_s
+    seq_name = self.seq_name
+    seq      = self.seq.dup
 
     unless wrap.nil?
       seq.gsub!(/(.{#{wrap}})/) do |match|
@@ -338,7 +343,14 @@ class Seq
 
   # Method to reverse the sequence.
   def reverse
-    Seq.new(self.seq_name, self.seq.reverse, self.type, self.qual ? self.qual.reverse : self.qual)
+    entry = Seq.new(
+      seq_name: self.seq_name,
+      seq:      self.seq.reverse,
+      type:     self.type,
+      qual:     (self.qual ? self.qual.reverse : self.qual)
+    )
+
+    entry
   end
 
   # Method to reverse the sequence.
@@ -352,10 +364,11 @@ class Seq
   def complement
     raise SeqError, "Cannot complement 0 length sequence" if self.length == 0
 
-    entry          = Seq.new
-    entry.seq_name = self.seq_name
-    entry.type     = self.type
-    entry.qual     = self.qual
+    entry = Seq.new(
+      seq_name: self.seq_name,
+      type:     self.type,
+      qual:     self.qual
+    )
 
     if self.is_dna?
       entry.seq = self.seq.tr('AGCUTRYWSMKHDVBNagcutrywsmkhdvbn', 'TCGAAYRWSKMDHBVNtcgaayrwskmdhbvn')
@@ -387,7 +400,7 @@ class Seq
   # two Sequence objects (case insensitive).
   def hamming_distance(entry, options = nil)
     if options and options[:ambiguity]
-      Ambiguity.hamming_distance(self.seq, entry.seq)
+      Hamming.distance(self.seq, entry.seq)
     else
       self.seq.upcase.hamming_distance(entry.seq.upcase)
     end
@@ -419,7 +432,12 @@ class Seq
 
   # Method to return a new Seq object with shuffled sequence.
   def shuffle
-    Seq.new(self.seq_name, self.seq.split('').shuffle!.join, self.type, self.qual)
+    Seq.new(
+      seq_name: self.seq_name,
+      seq:      self.seq.split('').shuffle!.join,
+      type:     self.type,
+      qual:     self.qual
+    )
   end
 
   # Method to shuffle a sequence randomly inline.
@@ -486,7 +504,7 @@ class Seq
 
     seq_name = self.seq_name.nil? ? nil : self.seq_name.dup
 
-    Seq.new(seq_name, seq, self.type, qual)
+    Seq.new(seq_name: seq_name, seq: seq, type: self.type, qual: qual)
   end
 
   # Method that replaces a sequence with a subsequence from a given start position
@@ -651,22 +669,26 @@ class Seq
     regex_start = Regexp.new(start_codons.join('|'), true)
     regex_stop  = Regexp.new(stop_codons.join('|'), true)
 
-    while pos_beg and pos_beg < self.length - size_min
-      if pos_beg = self.seq.index(regex_start, pos_beg)
-        if pos_end = self.seq.index(regex_stop, pos_beg)
-          length = (pos_end - pos_beg) + 3
+    while pos_beg = self.seq.index(regex_start, pos_beg)
+      pos_end = pos_beg + 3
+       
+      while pos_end = self.seq.index(regex_stop, pos_end)
+        length = (pos_end - pos_beg) + 3
 
-          if (length % 3) == 0
-            if size_min <= length and length <= size_max
-              subseq = self.subseq(pos_beg, length)
+        if (length % 3) == 0
+          if size_min <= length and length <= size_max
+            subseq = self.subseq(pos_beg, length)
 
-              orfs << [subseq, pos_beg, pos_end + 3]
-            end
+            orfs << [subseq, pos_beg, pos_end + 3]
           end
+
+          break
         end
 
-        pos_beg += 1
+        pos_end += 1
       end
+
+      pos_beg += 1
     end
 
     if pick_longest