X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;ds=sidebyside;f=src%2Fjalview%2Fanalysis%2Fscoremodels%2FScoreMatrix.java;h=7f71d0fc7858dfdf5d9c9c64876727a482276a61;hb=993b6a26f415bf05ce6dce0b93e96eaf95733d4b;hp=eb94817f4d441f322bc7df25145a78a2a41eacfe;hpb=7c5c70f4bbe69a722dd6e8bd269d7d01e1089034;p=jalview.git

diff --git a/src/jalview/analysis/scoremodels/ScoreMatrix.java b/src/jalview/analysis/scoremodels/ScoreMatrix.java
index eb94817..7f71d0f 100644
--- a/src/jalview/analysis/scoremodels/ScoreMatrix.java
+++ b/src/jalview/analysis/scoremodels/ScoreMatrix.java
@@ -20,10 +20,25 @@
  */
 package jalview.analysis.scoremodels;
 
+import jalview.api.analysis.PairwiseScoreModelI;
+import jalview.api.analysis.SimilarityScoreModelI;
+import jalview.datamodel.AlignmentView;
+import jalview.math.Matrix;
+import jalview.math.MatrixI;
+
 import java.util.Arrays;
 
-public class ScoreMatrix implements PairwiseScoreModelI
+public class ScoreMatrix implements SimilarityScoreModelI,
+        PairwiseScoreModelI
 {
+  /*
+   * Jalview 2.10.1 treated gaps as X (peptide) or N (nucleotide)
+   * for pairwise scoring; 2.10.2 uses gap score (last column) in
+   * score matrix (JAL-2397)
+   * Set this flag to true (via Groovy) for 2.10.1 behaviour
+   */
+  private static boolean scoreGapAsAny = false;
+
   public static final short UNMAPPED = (short) -1;
 
   private static final String BAD_ASCII_ERROR = "Unexpected character %s in getPairwiseScore";
@@ -158,16 +173,22 @@ public class ScoreMatrix implements PairwiseScoreModelI
   }
 
   /**
-   * Returns the score matrix as used in getPairwiseScore. If using this matrix
-   * directly, callers <em>must</em> also call <code>getMatrixIndex</code> in
-   * order to get the matrix index for each character (symbol).
+   * Returns a copy of the score matrix as used in getPairwiseScore. If using
+   * this matrix directly, callers <em>must</em> also call
+   * <code>getMatrixIndex</code> in order to get the matrix index for each
+   * character (symbol).
    * 
    * @return
    * @see #getMatrixIndex(char)
    */
   public float[][] getMatrix()
   {
-    return matrix;
+    float[][] v = new float[matrix.length][matrix.length];
+    for (int i = 0; i < matrix.length; i++)
+    {
+      v[i] = Arrays.copyOf(matrix[i], matrix[i].length);
+    }
+    return v;
   }
 
   /**
@@ -228,7 +249,12 @@ public class ScoreMatrix implements PairwiseScoreModelI
   }
 
   /**
-   * Print the score matrix, optionally formatted as html, with the alphabet symbols as column headings and at the start of each row
+   * Print the score matrix, optionally formatted as html, with the alphabet
+   * symbols as column headings and at the start of each row.
+   * <p>
+   * The non-html format should give an output which can be parsed as a score
+   * matrix file
+   * 
    * @param html
    * @return
    */
@@ -244,6 +270,11 @@ public class ScoreMatrix implements PairwiseScoreModelI
       sb.append("<table border=\"1\">");
       sb.append(html ? "<tr><th></th>" : "");
     }
+    else
+    {
+      sb.append("ScoreMatrix ").append(getName()).append("\n");
+      sb.append(symbols).append("\n");
+    }
     for (char sym : symbols)
     {
       if (html)
@@ -292,4 +323,104 @@ public class ScoreMatrix implements PairwiseScoreModelI
   {
     return symbols.length;
   }
+
+  /**
+   * 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>
+   */
+  @Override
+  public MatrixI findSimilarities(AlignmentView seqstrings)
+  {
+    char gapChar = scoreGapAsAny ? (seqstrings.isNa() ? 'N' : 'X') : ' ';
+    String[] seqs = seqstrings.getSequenceStrings(gapChar);
+    return findSimilarities(seqs);
+  }
+
+  /**
+   * @param seqs
+   * @return
+   */
+  protected MatrixI findSimilarities(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);
+          float score = getPairwiseScore(c1, c2);
+          total += score;
+        }
+        values[row][col] = total;
+      }
+    }
+    return new Matrix(values);
+  }
+
+  /**
+   * Answers a hashcode computed from the symbol alphabet and the matrix score
+   * values
+   */
+  @Override
+  public int hashCode()
+  {
+    int hs = Arrays.hashCode(symbols);
+    for (float[] row : matrix)
+    {
+      hs = hs * 31 + Arrays.hashCode(row);
+    }
+    return hs;
+  }
+
+  /**
+   * Answers true if the argument is a ScoreMatrix with the same symbol alphabet
+   * and score values, else false
+   */
+  @Override
+  public boolean equals(Object obj)
+  {
+    if (!(obj instanceof ScoreMatrix))
+    {
+      return false;
+    }
+    ScoreMatrix sm = (ScoreMatrix) obj;
+    if (Arrays.equals(symbols, sm.symbols)
+            && Arrays.deepEquals(matrix, sm.matrix))
+    {
+      return true;
+    }
+    return false;
+  }
+
+  /**
+   * Returns the alphabet the matrix scores for, as a string of characters
+   * 
+   * @return
+   */
+  public String getSymbols()
+  {
+    return new String(symbols);
+  }
 }