Fixed forward specs

pjotrp · Jan 21, 2011 · 97d2d0d · 97d2d0d
1 parent 138e5f8
commit 97d2d0d
Show file tree

Hide file tree

Showing 2 changed files with 26 additions and 18 deletions.
diff --git a/lib/bigbio/db/emitters/orf_emitter.rb b/lib/bigbio/db/emitters/orf_emitter.rb
@@ -83,7 +83,12 @@ class ShortFrameState
       def initialize seq, ntseq_pos, ntmin_size
         # @seq = seq.upcase  
         @seq = seq
-        @min_size_codons = (ntmin_size/3).to_i
+        @min_size_codons = if ntmin_size > 3
+                             (ntmin_size/3).to_i
+                           else
+                             2  # otherwise we get single STOP codons
+                           end
+
         @codons = FrameCodonSequence.new(seq,ntseq_pos)
         @ntseq_pos = ntseq_pos # nucleotides
         # @codons.track_sequence_pos = seq_pos
@@ -105,16 +110,15 @@ def get_startstop_orfs
       # the first sequence is returned when incomplete. +strip_leading+
       # is used to remove the shared codon with the last sequence.
       #
-      def get_codon_orfs1 splitter_func,do_include_leftmost,do_strip_leading 
+      def get_codon_orfs1 splitter_func,do_include_leftmost_orf,do_strip_leading_codon
         orfs = split(@codons,splitter_func)
         return [] if orfs.size == 0
         # Drop the first sequence, if there is no match on the first position
-        orfs.shift if !do_include_leftmost and !splitter_func.call(orfs.first[0])
+        orfs.shift if !do_include_leftmost_orf and !splitter_func.call(orfs.first[0])
         orfs = orfs.map { |codons| 
-          codons = codons.shift if do_strip_leading and splitter_func.call(codons[0])
+          codons = codons.shift if do_strip_leading_codon and splitter_func.call(codons[0])
           codons
         }
-        orfs = orfs.find_all { | orf | orf.size > 0 }
         TrackSequenceTrait.update_sequence_pos(orfs,@ntseq_pos) # nail against parent
       end
 
@@ -135,10 +139,11 @@ def split codons, is_splitter_func
           if is_splitter_func.call(c)
             node.push c
             size = node.size
+            # p node
             list.push FrameCodonSequence.new(node,pos+1-size) if size > @min_size_codons
             node = []
           end
-          node.push c
+          node.push c  # always push boundary codon
         end
         list
       end

diff --git a/spec/emitter_spec.rb b/spec/emitter_spec.rb
@@ -44,8 +44,8 @@
   it "should handle min_size" do
     fr = ShortFrameState.new "atggattaaatgtaatggatttaatgtaaa",0,9
     orfs = fr.get_stopstop_orfs
-    orfs.map{ | orf | orf.pos }.should == [ 5 ]
     orfs.map{ | orf | orf.to_seq }.should == [ "TGGATTTAA"]
+    orfs.map{ | orf | orf.pos }.should == [ 5 ]
     fr.get_startstop_orfs.should == []
   end
   it "should find ORFs in" do
@@ -147,36 +147,39 @@
 
   it "should combine a forward frame without ORFs in first seq" do
     s1 = "atggattaaatgta"
-    s2 = "atggatttaatgtaaa"
+    #     ......---===xx
+    s2 = "atggatttaattattataaa"
+    #     x======xxx======xxx.
     fr = ShortFrameState.new s1,0,0
     fr.ntseq_pos.should == 0
     orfs = fr.get_stopstop_orfs
     orfs.size == 0 # in codons
     fr3 = FrameCodonHelpers::CreateShortFrame.create_right(fr,orfs,s2)
-    # p fr3
-    fr3.ntseq_pos.should == 9
-    fr3.codons.to_seq.should == "ATGTAATGGATTTAATGTAAA"
+    fr3.ntseq_pos.should == 0
+    fr3.codons.to_seq.should == "ATGGATTAAATGTAATGGATTTAATTATTATAA"
     norfs = fr3.get_stopstop_orfs
-    orfs += norfs
-    orfs.map{ | orf | orf.to_seq }.should == ["ATGGATTAA", "TGGATTTAA"]
-    orfs.map{ | orf | orf.track_sequence_pos }.should == [0,11]
+    orfs = norfs
+    orfs.map{ | orf | orf.to_seq }.should == ["ATGTAA","TGGATTTAA","TTATTATAA"]
+    orfs.map{ | orf | orf.track_ntseq_pos }.should == [9,9+6,9+6+9]
   end
 
   it "should combine a forward frame without ORFs in first seq" do
     s1 = "atggattaaatgta"
+    #     ......---===xx
     s2 = "atggatttaatgtaaa"
+    #     x======xxx
     fr = ShortFrameState.new s1,0,0
     fr.ntseq_pos.should == 0
     orfs = fr.get_stopstop_orfs
     orfs.size == 0 # in codons
     fr3 = FrameCodonHelpers::CreateShortFrame.create_right(fr,orfs,s2)
     # p fr3
-    fr3.ntseq_pos.should == 9
-    fr3.codons.to_seq.should == "ATGTAATGGATTTAATGTAAA"
+    fr3.ntseq_pos.should == 0 # on the combined sequences
+    fr3.codons.to_seq.should == "ATGGATTAAATGTAATGGATTTAATGTAAA"
     norfs = fr3.get_stopstop_orfs
     orfs += norfs
-    orfs.map{ | orf | orf.to_seq }.should == ["ATGGATTAA", "TGGATTTAA"]
-    orfs.map{ | orf | orf.track_sequence_pos }.should == [0,11]
+    orfs.map{ | orf | orf.to_seq }.should == ["ATGTAA", "TGGATTTAA"]
+    orfs.map{ | orf | orf.track_ntseq_pos }.should == [9,9+6]
   end
 
   it "should combine a reverse frame" do