X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2Fscoremodels%2FScoreMatrix.java;h=b73f826e40a51c8f4fb745c7234fbfacb328261c;hb=dd97032c11b883017a9c8125ff30a58dfb2144bd;hp=7f71d0fc7858dfdf5d9c9c64876727a482276a61;hpb=993b6a26f415bf05ce6dce0b93e96eaf95733d4b;p=jalview.git

diff --git a/src/jalview/analysis/scoremodels/ScoreMatrix.java b/src/jalview/analysis/scoremodels/ScoreMatrix.java
index 7f71d0f..b73f826 100644
--- a/src/jalview/analysis/scoremodels/ScoreMatrix.java
+++ b/src/jalview/analysis/scoremodels/ScoreMatrix.java
@@ -21,16 +21,20 @@
 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;
 
 public class ScoreMatrix implements SimilarityScoreModelI,
         PairwiseScoreModelI
 {
+  private static final char GAP_CHARACTER = Comparison.GAP_DASH;
+
   /*
    * Jalview 2.10.1 treated gaps as X (peptide) or N (nucleotide)
    * for pairwise scoring; 2.10.2 uses gap score (last column) in
@@ -47,10 +51,16 @@ public class ScoreMatrix implements SimilarityScoreModelI,
 
   /*
    * the name of the model as shown in menus
+   * each score model in use should have a unique name
    */
   private String name;
 
   /*
+   * a description for the model as shown in tooltips
+   */
+  private String description;
+
+  /*
    * the characters that the model provides scores for
    */
   private char[] symbols;
@@ -77,21 +87,21 @@ public class ScoreMatrix implements SimilarityScoreModelI,
    * of symbols. The matrix should be square and of the same size as the
    * alphabet, for example 20x20 for a 20 symbol alphabet.
    * 
-   * @param name
+   * @param theName
    *          Unique, human readable name for the matrix
    * @param alphabet
    *          the symbols to which scores apply
-   * @param matrix
+   * @param values
    *          Pairwise scores indexed according to the symbol alphabet
    */
-  public ScoreMatrix(String name, char[] alphabet, float[][] matrix)
+  public ScoreMatrix(String theName, char[] alphabet, float[][] values)
   {
-    if (alphabet.length != matrix.length)
+    if (alphabet.length != values.length)
     {
       throw new IllegalArgumentException(
               "score matrix size must match alphabet size");
     }
-    for (float[] row : matrix)
+    for (float[] row : values)
     {
       if (row.length != alphabet.length)
       {
@@ -100,8 +110,8 @@ public class ScoreMatrix implements SimilarityScoreModelI,
       }
     }
 
-    this.matrix = matrix;
-    this.name = name;
+    this.matrix = values;
+    this.name = theName;
     this.symbols = alphabet;
 
     symbolIndex = buildSymbolIndex(alphabet);
@@ -161,6 +171,12 @@ public class ScoreMatrix implements SimilarityScoreModelI,
   }
 
   @Override
+  public String getDescription()
+  {
+    return description;
+  }
+
+  @Override
   public boolean isDNA()
   {
     return !peptide;
@@ -345,18 +361,24 @@ 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') : ' ';
+    char gapChar = scoreGapAsAny ? (seqstrings.isNa() ? 'N' : 'X')
+            : Comparison.GAP_DASH;
     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++)
@@ -364,15 +386,7 @@ 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;
       }
     }
@@ -380,6 +394,68 @@ 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;
+        }
+      }
+
+      char c1 = i >= len1 ? GAP_CHARACTER : seq1.charAt(i);
+      char c2 = i >= len2 ? GAP_CHARACTER : 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.includeGaps())
+        {
+          continue;
+        }
+      }
+      float score = getPairwiseScore(c1, c2);
+      total += score;
+    }
+    return total;
+  }
+
+  /**
    * Answers a hashcode computed from the symbol alphabet and the matrix score
    * values
    */
@@ -423,4 +499,9 @@ public class ScoreMatrix implements SimilarityScoreModelI,
   {
     return new String(symbols);
   }
+
+  public void setDescription(String desc)
+  {
+    description = desc;
+  }
 }