SpectralEntropy as measure of SpectralSimilarity (#633)

* First commit

* Improved test for SpectralEntropy

* Fixed what Shortreed broke

* Deleted redundant check in SpectralEntropy method

Co-authored-by: trishorts <mshort@chem.wisc.edu>

Author: Alexander-Sol

mzLib/MassSpectrometry/MzSpectra/SpectralSimilarity.cs

@@ -15,6 +15,7 @@ public SpectralSimilarity(MzSpectrum experimentalSpectrum, MzSpectrum theoretica
             TheoreticalYArray = Normalize(FilterOutIonsBelowThisMz(theoreticalSpectrum.XArray, theoreticalSpectrum.YArray, filterOutBelowThisMz).Select(p => p.Item2).ToArray(), scheme);
             TheoreticalXArray = FilterOutIonsBelowThisMz(theoreticalSpectrum.XArray, theoreticalSpectrum.YArray, filterOutBelowThisMz).Select(p => p.Item1).ToArray();
             LocalPpmTolerance = toleranceInPpm;
+            normalizationScheme = scheme;
             _intensityPairs = IntensityPairs(allPeaks);
         }
 
@@ -25,6 +26,7 @@ public SpectralSimilarity(MzSpectrum experimentalSpectrum, double[] theoreticalX
             TheoreticalYArray = Normalize(FilterOutIonsBelowThisMz(theoreticalX, theoreticalY, filterOutBelowThisMz).Select(p => p.Item2).ToArray(), scheme);
             TheoreticalXArray = FilterOutIonsBelowThisMz(theoreticalX, theoreticalY, filterOutBelowThisMz).Select(p => p.Item1).ToArray();
             LocalPpmTolerance = toleranceInPpm;
+            normalizationScheme = scheme;
             _intensityPairs = IntensityPairs(allPeaks);
         }
 
@@ -42,6 +44,8 @@ public SpectralSimilarity(double[] P_XArray, double[] P_YArray, double[] Q_XArra
         public double[] TheoreticalYArray { get; private set; }
         public double[] TheoreticalXArray { get; private set; }
 
+        private SpectrumNormalizationScheme normalizationScheme;
+
         private double LocalPpmTolerance;
 
         private readonly List<(double, double)> _intensityPairs = new();
@@ -315,6 +319,33 @@ private double[] NormalizeSpectrumSum(double[] spectrum)
             return sum;
         }
 
+        // This method should only be used with the SpectrumSum normalization method
+        // This method should only be used when allPeaks is set to true
+        public double? SpectralEntropy()
+        {
+            double theoreticalEntropy = 0;
+            foreach (double intensity in TheoreticalYArray)
+            {
+                theoreticalEntropy += -1 * intensity * Math.Log(intensity);
+            }
+            double experimentalEntropy = 0;
+            foreach (double intensity in ExperimentalYArray)
+            {
+                experimentalEntropy += -1 * intensity * Math.Log(intensity);
+            }
+
+            double combinedEntropy = 0;
+            foreach ( (double,double) intensityPair in _intensityPairs)
+            {
+                double combinedIntensity = intensityPair.Item1 / 2 + intensityPair.Item2 / 2;
+                combinedEntropy += -1* combinedIntensity * Math.Log(combinedIntensity);
+            }
+
+            double similarityScore = 1 - (2 * combinedEntropy - theoreticalEntropy - experimentalEntropy) / Math.Log(4);
+            return similarityScore;
+        }
+      
+        
         public double? KullbackLeiblerDivergence_P_Q()
         {
             double divergence = 0;

mzLib/Test/TestSpectralSimilarity.cs

@@ -28,6 +28,8 @@ public void TestAllSpectrumSimilaritiesWithoutMzFilter()
             Assert.That(s.PearsonsCorrelation(), Is.EqualTo(0.42).Within(0.01));
             Assert.That(s.DotProduct(), Is.EqualTo(0.17).Within(0.01));
             Assert.That(s.SearleSimilarity(), Is.EqualTo(2.4391).Within(0.01));
+            s = new SpectralSimilarity(primary, secondary, SpectralSimilarity.SpectrumNormalizationScheme.spectrumSum, ppmTolerance, true, 0);
+            Assert.That(s.SpectralEntropy(), Is.EqualTo(0.79).Within(0.01));
 
             //Test all normalization schemes
             primary = new MzSpectrum(new double[] { 1, 2, 3 }, new double[] { 2, 4, 6 }, false);