Prerelease version 0.0.2a. Refactors the complexity metric from a sim…

…milarity metric to a dissimilarity metric, which better captures the uniqueness of a group of NGS reads.

lib/NGSCI/calculator.rb

@@ -126,7 +126,7 @@ def sci(reads)
       reads = reads.group_by(&:start).map{|k,v| v.max{|x,y| x.length <=> y.length}}
       d = summed_dissimilarity(reads)
       uniquereads = reads.size
-      return [numreads,uniquereads,(d.to_f/@read_length).round(4),(uniquereads*d/@denominator).round(4)]
+      return [numreads,uniquereads,(d.to_f/@read_length).round(4),(100*uniquereads*d/@denominator).round(4)]
     end
 
 
@@ -187,7 +187,7 @@ def max_summed_dissimilarity(read_length)
     # Denomiator = read length * max_summed_dissimilarity / (read_length * read_length)
     #
     # @param [Integer] read_length The read length
-    # @return [Float] denominator The denominator including normalization factors for the complexity index
+    # @return [Float] denominator The denominator including normalization factors for the complexity index 349184
     def denominator_calc(read_length)
       read_length*max_summed_dissimilarity(read_length)
     end    

lib/NGSCI/version.rb

@@ -1,3 +1,3 @@
 module NGSCI
-  VERSION = "0.0.1.a"
+  VERSION = "0.0.2.a"
 end

spec/lib/calculator_spec.rb

@@ -165,7 +165,7 @@
       @read_length = 76
       @calc = NGSCI::Calculator.new(testbam,testfasta)
     end
-    it "returns a float denominator" do
+    it "returns a float" do
       expect(@calc.max_summed_dissimilarity(@read_length)).to be_kind_of Integer
     end
   end
@@ -200,10 +200,9 @@ def tri(x,n=0)
   end
   context "when averaging per read" do
     it "is equal to 1/3 times (read_length - 1)" do
-      (32..200).each do |x|
-        read_length = x
-        @calc = NGSCI::Calculator.new(testbam,testfasta)
-        calculated_max_summed_dissimilarity = @calc.max_summed_dissimilarity(read_length)/read_length/(read_length)
+      @calc = NGSCI::Calculator.new(testbam,testfasta)
+      (32..200).each do |read_length|
+        calculated_max_summed_dissimilarity = @calc.max_summed_dissimilarity(read_length)/(read_length*read_length)
         expect(calculated_max_summed_dissimilarity).to eq((read_length-1)/3)
       end
     end
@@ -220,11 +219,12 @@ def tri(x,n=0)
       read_length = 76
       expect(@calc.denominator_calc(read_length)).to be_kind_of Integer
     end
-    it "returns the correct denominator for each read length" do
-      (32..200).each do |x|
-        read_length = x
-        expect(@calc.denominator_calc(read_length)).to eq((read_length^3)*(read_length-1)/3 )
-      end
+  end
+  it "is the max_summed_dissimilarity * read length" do
+    @calc = NGSCI::Calculator.new(testbam,testfasta)
+    (32..200).each do |read_length|
+      max_sum_dissim = @calc.max_summed_dissimilarity(read_length)
+      expect(@calc.denominator_calc(read_length)).to eq(read_length*max_sum_dissim)
     end
   end
 end