X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;ds=inline;f=src%2Fjalview%2Fanalysis%2Fscoremodels%2FScoreMatrix.java;h=b206339805cf4d1f026c15eb34f3987a21a543b9;hb=f55ab79ffe7987b4efa091a28a393019c69adfbe;hp=f7da9f3f8845db6fc737532a85f34aaa68299621;hpb=81f1e29b00bc98403d45d2258da71ac19c425c57;p=jalview.git
diff --git a/src/jalview/analysis/scoremodels/ScoreMatrix.java b/src/jalview/analysis/scoremodels/ScoreMatrix.java
index f7da9f3..b206339 100644
--- a/src/jalview/analysis/scoremodels/ScoreMatrix.java
+++ b/src/jalview/analysis/scoremodels/ScoreMatrix.java
@@ -20,9 +20,10 @@
*/
package jalview.analysis.scoremodels;
+import jalview.api.AlignmentViewPanel;
import jalview.api.analysis.PairwiseScoreModelI;
+import jalview.api.analysis.ScoreModelI;
import jalview.api.analysis.SimilarityParamsI;
-import jalview.api.analysis.SimilarityScoreModelI;
import jalview.datamodel.AlignmentView;
import jalview.math.Matrix;
import jalview.math.MatrixI;
@@ -32,16 +33,20 @@ import java.util.Arrays;
/**
* A class that models a substitution score matrix for any given alphabet of
- * symbols
+ * symbols. Instances of this class are immutable and thread-safe, so the same
+ * object is returned from calls to getInstance().
*/
-public class ScoreMatrix implements SimilarityScoreModelI,
- PairwiseScoreModelI
+public class ScoreMatrix extends SimilarityScoreModel
+ implements PairwiseScoreModelI
{
+ private static final char GAP_CHARACTER = Comparison.GAP_DASH;
+
/*
- * this fields records which gap character (if any) is used in the alphabet;
- * space, dash or dot are recognised as gap symbols
+ * 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 char gapCharacter = '0';
+ private static final int UNKNOWN_IDENTITY_SCORE = 1;
/*
* Jalview 2.10.1 treated gaps as X (peptide) or N (nucleotide)
@@ -90,6 +95,12 @@ public class ScoreMatrix implements SimilarityScoreModelI,
*/
private boolean peptide;
+ private float minValue;
+
+ private float maxValue;
+
+ 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
@@ -104,6 +115,26 @@ public class ScoreMatrix implements SimilarityScoreModelI,
*/
public ScoreMatrix(String theName, char[] alphabet, float[][] values)
{
+ this(theName, null, alphabet, values);
+ }
+
+ /**
+ * Constructor given a name, description, symbol alphabet, and matrix of
+ * scores for pairs of symbols. The matrix should be square and of the same
+ * size as the alphabet, for example 20x20 for a 20 symbol alphabet.
+ *
+ * @param theName
+ * Unique, human readable name for the matrix
+ * @param theDescription
+ * descriptive display name suitable for use in menus
+ * @param alphabet
+ * the symbols to which scores apply
+ * @param values
+ * Pairwise scores indexed according to the symbol alphabet
+ */
+ public ScoreMatrix(String theName, String theDescription, char[] alphabet,
+ float[][] values)
+ {
if (alphabet.length != values.length)
{
throw new IllegalArgumentException(
@@ -120,10 +151,15 @@ public class ScoreMatrix implements SimilarityScoreModelI,
this.matrix = values;
this.name = theName;
+ this.description = theDescription;
this.symbols = alphabet;
symbolIndex = buildSymbolIndex(alphabet);
+ findMinMax();
+
+ symmetric = checkSymmetry();
+
/*
* crude heuristic for now...
*/
@@ -131,6 +167,52 @@ public class ScoreMatrix implements SimilarityScoreModelI,
}
/**
+ * Answers true if the matrix is symmetric, else false. Usually, substitution
+ * matrices are symmetric, which allows calculations to be short cut.
+ *
+ * @return
+ */
+ private boolean checkSymmetry()
+ {
+ for (int i = 0; i < matrix.length; i++)
+ {
+ for (int j = i; j < matrix.length; j++)
+ {
+ if (matrix[i][j] != matrix[j][i])
+ {
+ return false;
+ }
+ }
+ }
+ return true;
+ }
+
+ /**
+ * 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.
@@ -154,11 +236,6 @@ public class ScoreMatrix implements SimilarityScoreModelI,
short pos = 0;
for (char c : alphabet)
{
- if (Comparison.isGap(c))
- {
- gapCharacter = c;
- }
-
if (c <= MAX_ASCII)
{
index[c] = pos;
@@ -245,21 +322,9 @@ public class ScoreMatrix implements SimilarityScoreModelI,
}
/**
- * Answers the matrix index for the gap character, or -1 if unmapped in the
- * matrix. Use this method only if using getMatrix
in order to
- * compute scores directly (without symbol lookup) for efficiency.
- *
- * @return
- * @see #getMatrix()
- */
- public int getGapIndex()
- {
- return getMatrixIndex(gapCharacter);
- }
-
- /**
- * 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 +346,14 @@ 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 +390,6 @@ public class ScoreMatrix implements SimilarityScoreModelI,
else
{
sb.append("ScoreMatrix ").append(getName()).append("\n");
- sb.append(symbols).append("\n");
}
for (char sym : symbols)
{
@@ -388,13 +459,14 @@ public class ScoreMatrix implements SimilarityScoreModelI,
*