From 7d123b6259738869c16529b8c28cfd8931da4205 Mon Sep 17 00:00:00 2001
From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Tue, 7 Feb 2017 12:43:19 +0000
Subject: [PATCH] JAL-2397 PCA gap cost uses last column of score matrix

---
 src/jalview/schemes/ScoreMatrix.java      |   38 +++--------------------------
 test/jalview/schemes/ScoreMatrixTest.java |   17 ++++++-------
 2 files changed, 11 insertions(+), 44 deletions(-)

diff --git a/src/jalview/schemes/ScoreMatrix.java b/src/jalview/schemes/ScoreMatrix.java
index b5e3ef1..d82f54c 100644
--- a/src/jalview/schemes/ScoreMatrix.java
+++ b/src/jalview/schemes/ScoreMatrix.java
@@ -93,7 +93,7 @@ public class ScoreMatrix extends PairwiseSeqScoreModel
   @Override
   public int getPairwiseScore(char c, char d)
   {
-    int pog = 0;
+    int score = 0;
 
     try
     {
@@ -101,45 +101,13 @@ public class ScoreMatrix extends PairwiseSeqScoreModel
               : ResidueProperties.nucleotideIndex[c];
       int b = (type == 0) ? ResidueProperties.aaIndex[d]
               : ResidueProperties.nucleotideIndex[d];
-
-      /*
-       * FIXME: 2.10.1 PCA treats gap as [22] or 'X', but Tree
-       * calculation treats as [23]; which is correct?
-       */
-      /*
-       * hack to convert unassigned / unknown (including gap)
-       * to index of unknown (X for amino acids, N for nucleotide)
-       * TODO: statically assign gap characters to this index?
-       */
-//      if (type == 0)
-//      {
-//        if (a == ResidueProperties.maxProteinIndex)
-//        {
-//          a = ResidueProperties.aaIndex['X'];
-//        }
-//        if (b == ResidueProperties.maxProteinIndex)
-//        {
-//          b = ResidueProperties.aaIndex['X'];
-//        }
-//      }
-//      if (type != 0)
-//      {
-//        if (a == ResidueProperties.maxNucleotideIndex)
-//        {
-//          a = ResidueProperties.nucleotideIndex['N'];
-//        }
-//        if (b == ResidueProperties.maxNucleotideIndex)
-//        {
-//          b = ResidueProperties.nucleotideIndex['N'];
-//        }
-//      }
-      pog = matrix[a][b];
+      score = matrix[a][b];
     } catch (Exception e)
     {
       // System.out.println("Unknown residue in " + A1 + " " + A2);
     }
 
-    return pog;
+    return score;
   }
 
   /**
diff --git a/test/jalview/schemes/ScoreMatrixTest.java b/test/jalview/schemes/ScoreMatrixTest.java
index 99af601..e15dd41 100644
--- a/test/jalview/schemes/ScoreMatrixTest.java
+++ b/test/jalview/schemes/ScoreMatrixTest.java
@@ -143,26 +143,25 @@ public class ScoreMatrixTest
     }
     /*
      * verify expected BLOSUM dot product scores
-     * Note: gap is treated like 'X' [22] in the matrix
      */
     // F.F + K.K + L.L = 6 + 5 + 4 = 15
     assertEquals(pairwise.getValue(0, 0), 15d);
-    // R.R + X.X + D.D = 5 + -1 + 6 = 10
-    assertEquals(pairwise.getValue(1, 1), 10d);
+    // R.R + -.- + D.D = 5 + 1 + 6 = 12
+    assertEquals(pairwise.getValue(1, 1), 12d);
     // Q.Q + I.I + A.A = 5 + 4 + 4 = 13
     assertEquals(pairwise.getValue(2, 2), 13d);
     // G.G + W.W + C.C = 6 + 11 + 9 = 26
     assertEquals(pairwise.getValue(3, 3), 26d);
-    // F.R + K.X + L.D = -3 + -1 + -3 = -8
-    assertEquals(pairwise.getValue(0, 1), -8d);
+    // F.R + K.- + L.D = -3 + -4 + -4 = -11
+    assertEquals(pairwise.getValue(0, 1), -11d);
     // F.Q + K.I + L.A = -3 + -3 + -1 = -7
     assertEquals(pairwise.getValue(0, 2), -7d);
     // F.G + K.W + L.C = -3 + -3 + -1 = -7
     assertEquals(pairwise.getValue(0, 3), -7d);
-    // R.Q + X.I + D.A = 1 + -1 + -2 = -2
-    assertEquals(pairwise.getValue(1, 2), -2d);
-    // R.G + X.W + D.C = -2 + -2 + -3 = -7
-    assertEquals(pairwise.getValue(1, 3), -7d);
+    // R.Q + -.I + D.A = 1 + -4 + -2 = -5
+    assertEquals(pairwise.getValue(1, 2), -5d);
+    // R.G + -.W + D.C = -2 + -4 + -3 = -9
+    assertEquals(pairwise.getValue(1, 3), -9d);
     // Q.G + I.W + A.C = -2 + -3 + 0 = -5
     assertEquals(pairwise.getValue(2, 3), -5d);
   }
-- 
1.7.10.2