JAL-2434 use a constant for StructureMapping.NO_CHAIN (space)

[jalview.git] / src / jalview / schemes / ScoreMatrix.java

diff --git a/src/jalview/schemes/ScoreMatrix.java b/src/jalview/schemes/ScoreMatrix.java
index c47654a..d82f54c 100644 (file)
--- a/src/jalview/schemes/ScoreMatrix.java
+++ b/src/jalview/schemes/ScoreMatrix.java
@@ -21,6 +21,8 @@
 package jalview.schemes;
 
 import jalview.analysis.scoremodels.PairwiseSeqScoreModel;
+import jalview.math.Matrix;
+import jalview.math.MatrixI;
 
 public class ScoreMatrix extends PairwiseSeqScoreModel
 {
@@ -77,10 +79,11 @@ public class ScoreMatrix extends PairwiseSeqScoreModel
   }
 
   /**
+   * Answers the score for substituting first char in A1 with first char in A2
    * 
    * @param A1
    * @param A2
-   * @return score for substituting first char in A1 with first char in A2
+   * @return
    */
   public int getPairwiseScore(String A1, String A2)
   {
@@ -90,7 +93,7 @@ public class ScoreMatrix extends PairwiseSeqScoreModel
   @Override
   public int getPairwiseScore(char c, char d)
   {
-    int pog = 0;
+    int score = 0;
 
     try
     {
@@ -98,41 +101,13 @@ public class ScoreMatrix extends PairwiseSeqScoreModel
               : ResidueProperties.nucleotideIndex[c];
       int b = (type == 0) ? ResidueProperties.aaIndex[d]
               : ResidueProperties.nucleotideIndex[d];
-
-      /*
-       * 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;
   }
 
   /**
@@ -197,4 +172,47 @@ public class ScoreMatrix extends PairwiseSeqScoreModel
     }
     return sb.toString();
   }
+
+  /**
+   * Computes an NxN matrix where N is the number of sequences, and entry [i, j]
+   * is sequence[i] pairwise multiplied with sequence[j], as a sum of scores
+   * computed using the current score matrix. For example
+   * <ul>
+   * <li>Sequences:</li>
+   * <li>FKL</li>
+   * <li>R-D</li>
+   * <li>QIA</li>
+   * <li>GWC</li>
+   * <li>Score matrix is BLOSUM62</li>
+   * <li>Gaps treated same as X (unknown)</li>
+   * <li>product [0, 0] = F.F + K.K + L.L = 6 + 5 + 4 = 15</li>
+   * <li>product [1, 1] = R.R + -.- + D.D = 5 + -1 + 6 = 10</li>
+   * <li>product [2, 2] = Q.Q + I.I + A.A = 5 + 4 + 4 = 13</li>
+   * <li>product [3, 3] = G.G + W.W + C.C = 6 + 11 + 9 = 26</li>
+   * <li>product[0, 1] = F.R + K.- + L.D = -3 + -1 + -3 = -8
+   * <li>and so on</li>
+   * </ul>
+   */
+  public MatrixI computePairwiseScores(String[] seqs)
+  {
+    double[][] values = new double[seqs.length][];
+    for (int row = 0; row < seqs.length; row++)
+    {
+      values[row] = new double[seqs.length];
+      for (int col = 0; col < seqs.length; col++)
+      {
+        int total = 0;
+        int width = Math.min(seqs[row].length(), seqs[col].length());
+        for (int i = 0; i < width; i++)
+        {
+          char c1 = seqs[row].charAt(i);
+          char c2 = seqs[col].charAt(i);
+          int score = getPairwiseScore(c1, c2);
+          total += score;
+        }
+        values[row][col] = total;
+      }
+    }
+    return new Matrix(values);
+  }
 }