JAL-838 ScoreMatrix now respects SimilarityParams!

[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 efde30d..41d7383 100644 (file)
--- a/src/jalview/analysis/scoremodels/ScoreMatrix.java
+++ b/src/jalview/analysis/scoremodels/ScoreMatrix.java
@@ -21,10 +21,12 @@
 package jalview.analysis.scoremodels;
 
 import jalview.api.analysis.PairwiseScoreModelI;
+import jalview.api.analysis.SimilarityParamsI;
 import jalview.api.analysis.SimilarityScoreModelI;
 import jalview.datamodel.AlignmentView;
 import jalview.math.Matrix;
 import jalview.math.MatrixI;
+import jalview.util.Comparison;
 
 import java.util.Arrays;
 
@@ -72,8 +74,6 @@ public class ScoreMatrix implements SimilarityScoreModelI,
    */
   private boolean peptide;
 
-  private boolean symmetric;
-
   /**
    * 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
@@ -347,18 +347,23 @@ public class ScoreMatrix implements SimilarityScoreModelI,
    * </ul>
    */
   @Override
-  public MatrixI findSimilarities(AlignmentView seqstrings)
+  public MatrixI findSimilarities(AlignmentView seqstrings,
+          SimilarityParamsI options)
   {
     char gapChar = scoreGapAsAny ? (seqstrings.isNa() ? 'N' : 'X') : ' ';
     String[] seqs = seqstrings.getSequenceStrings(gapChar);
-    return findSimilarities(seqs);
+    return findSimilarities(seqs, options);
   }
 
   /**
+   * Computes pairwise similarities of a set of sequences using the given
+   * parameters
+   * 
    * @param seqs
+   * @param params
    * @return
    */
-  protected MatrixI findSimilarities(String[] seqs)
+  protected MatrixI findSimilarities(String[] seqs, SimilarityParamsI params)
   {
     double[][] values = new double[seqs.length][];
     for (int row = 0; row < seqs.length; row++)
@@ -366,15 +371,7 @@ public class ScoreMatrix implements SimilarityScoreModelI,
       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);
-          float score = getPairwiseScore(c1, c2);
-          total += score;
-        }
+        double total = computeSimilarity(seqs[row], seqs[col], params);
         values[row][col] = total;
       }
     }
@@ -382,6 +379,68 @@ public class ScoreMatrix implements SimilarityScoreModelI,
   }
 
   /**
+   * Calculates the pairwise similarity of two strings using the given
+   * calculation parameters
+   * 
+   * @param seq1
+   * @param seq2
+   * @param params
+   * @return
+   */
+  protected double computeSimilarity(String seq1, String seq2,
+          SimilarityParamsI params)
+  {
+    int len1 = seq1.length();
+    int len2 = seq2.length();
+    double total = 0;
+
+    int width = Math.max(len1, len2);
+    for (int i = 0; i < width; i++)
+    {
+      if (i >= len1 || i >= len2)
+      {
+        /*
+         * off the end of one sequence; stop if we are only matching
+         * on the shorter sequence length, else treat as trailing gap
+         */
+        if (params.denominateByShortestLength())
+        {
+          break;
+        }
+      }
+      // Change GAP_SPACE to GAP_DASH if we adopt - for gap in matrices
+      char c1 = i >= len1 ? Comparison.GAP_SPACE : seq1.charAt(i);
+      char c2 = i >= len2 ? Comparison.GAP_SPACE : seq2.charAt(i);
+      boolean gap1 = Comparison.isGap(c1);
+      boolean gap2 = Comparison.isGap(c2);
+
+      if (gap1 && gap2)
+      {
+        /*
+         * gap-gap: include if options say so, else ignore
+         */
+        if (!params.includeGappedColumns())
+        {
+          continue;
+        }
+      }
+      else if (gap1 || gap2)
+      {
+        /*
+         * gap-residue: score if options say so
+         */
+        if (!params.includesGaps())
+        {
+          continue;
+        }
+      }
+      float score = getPairwiseScore(c1, c2);
+      total += score;
+    }
+    return total;
+  }
+
+  /**
    * Answers a hashcode computed from the symbol alphabet and the matrix score
    * values
    */
@@ -416,11 +475,6 @@ public class ScoreMatrix implements SimilarityScoreModelI,
     return false;
   }
 
-  public boolean isSymmetric()
-  {
-    return symmetric;
-  }
-
   /**
    * Returns the alphabet the matrix scores for, as a string of characters
    *