JAL-2416 explicit constant for unknown/unknown identity score (*

[jalview.git] / src / jalview / analysis / scoremodels / ScoreMatrix.java

diff --git a/src/jalview/analysis/scoremodels/ScoreMatrix.java b/src/jalview/analysis/scoremodels/ScoreMatrix.java
index f7da9f3..e2c14e9 100644 (file)
--- a/src/jalview/analysis/scoremodels/ScoreMatrix.java
+++ b/src/jalview/analysis/scoremodels/ScoreMatrix.java
@@ -38,6 +38,13 @@ public class ScoreMatrix implements SimilarityScoreModelI,
         PairwiseScoreModelI
 {
   /*
+   * an arbitrary score to assign for identity of an unknown symbol
+   * (this is the value on the diagonal in the * column of the NCBI matrix)
+   * (though a case could be made for using the minimum diagonal value)
+   */
+  private static final int UNKNOWN_IDENTITY_SCORE = 1;
+
+  /*
    * this fields records which gap character (if any) is used in the alphabet;
    * space, dash or dot are recognised as gap symbols
    */
@@ -90,6 +97,10 @@ public class ScoreMatrix implements SimilarityScoreModelI,
    */
   private boolean peptide;
 
+  private float minValue;
+
+  private float maxValue;
+
   /**
    * Constructor given a name, symbol alphabet, and matrix of scores for pairs
    * of symbols. The matrix should be square and of the same size as the
@@ -124,6 +135,8 @@ public class ScoreMatrix implements SimilarityScoreModelI,
 
     symbolIndex = buildSymbolIndex(alphabet);
 
+    findMinMax();
+
     /*
      * crude heuristic for now...
      */
@@ -131,6 +144,31 @@ public class ScoreMatrix implements SimilarityScoreModelI,
   }
 
   /**
+   * Record the minimum and maximum score values
+   */
+  protected void findMinMax()
+  {
+    float min = Float.MAX_VALUE;
+    float max = -Float.MAX_VALUE;
+    if (matrix != null)
+    {
+      for (float[] row : matrix)
+      {
+        if (row != null)
+        {
+          for (float f : row)
+          {
+            min = Math.min(min, f);
+            max = Math.max(max, f);
+          }
+        }
+      }
+    }
+    minValue = min;
+    maxValue = max;
+  }
+
+  /**
    * Returns an array A where A[i] is the position in the alphabet array of the
    * character whose value is i. For example if the alphabet is { 'A', 'D', 'X'
    * } then A['D'] = A[68] = 1.
@@ -258,8 +296,9 @@ public class ScoreMatrix implements SimilarityScoreModelI,
   }
 
   /**
-   * Returns the pairwise score for substituting c with d, or zero if c or d is
-   * an unscored or unexpected character
+   * Returns the pairwise score for substituting c with d. If either c or d is
+   * an unexpected character, returns 1 for identity (c == d), else the minimum
+   * score value in the matrix.
    */
   @Override
   public float getPairwiseScore(char c, char d)
@@ -281,7 +320,14 @@ public class ScoreMatrix implements SimilarityScoreModelI,
     {
       return matrix[cIndex][dIndex];
     }
-    return 0;
+
+    /*
+     * one or both symbols not found in the matrix
+     * currently scoring as 1 (for identity) or the minimum
+     * matrix score value (otherwise)
+     * (a case could be made for using minimum row/column value instead)
+     */
+    return c == d ? UNKNOWN_IDENTITY_SCORE : getMinimumScore();
   }
 
   /**
@@ -318,7 +364,6 @@ public class ScoreMatrix implements SimilarityScoreModelI,
     else
     {
       sb.append("ScoreMatrix ").append(getName()).append("\n");
-      sb.append(symbols).append("\n");
     }
     for (char sym : symbols)
     {
@@ -524,7 +569,7 @@ public class ScoreMatrix implements SimilarityScoreModelI,
    * 
    * @return
    */
-  public String getSymbols()
+  String getSymbols()
   {
     return new String(symbols);
   }
@@ -533,4 +578,14 @@ public class ScoreMatrix implements SimilarityScoreModelI,
   {
     description = desc;
   }
+
+  public float getMinimumScore()
+  {
+    return minValue;
+  }
+
+  public float getMaximumScore()
+  {
+    return maxValue;
+  }
 }