X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2Fscoremodels%2FScoreMatrix.java;h=b206339805cf4d1f026c15eb34f3987a21a543b9;hb=f5e62b7ae1c880870d2c162bcd3aa3097f799b41;hp=84835a4d1065f04266901e34b857d454ceb30d51;hpb=8717834368bd00d8adfa47ee099288acd34363ef;p=jalview.git diff --git a/src/jalview/analysis/scoremodels/ScoreMatrix.java b/src/jalview/analysis/scoremodels/ScoreMatrix.java index 84835a4..b206339 100644 --- a/src/jalview/analysis/scoremodels/ScoreMatrix.java +++ b/src/jalview/analysis/scoremodels/ScoreMatrix.java @@ -20,18 +20,34 @@ */ 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; +import jalview.util.Comparison; import java.util.Arrays; -public class ScoreMatrix implements SimilarityScoreModelI, - PairwiseScoreModelI +/** + * A class that models a substitution score matrix for any given alphabet of + * symbols. Instances of this class are immutable and thread-safe, so the same + * object is returned from calls to getInstance(). + */ +public class ScoreMatrix extends SimilarityScoreModel + implements PairwiseScoreModelI { + private static final char GAP_CHARACTER = Comparison.GAP_DASH; + + /* + * 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; + /* * Jalview 2.10.1 treated gaps as X (peptide) or N (nucleotide) * for pairwise scoring; 2.10.2 uses gap score (last column) in @@ -48,10 +64,16 @@ 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; @@ -73,26 +95,52 @@ 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 * 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) + 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( "score matrix size must match alphabet size"); } - for (float[] row : matrix) + for (float[] row : values) { if (row.length != alphabet.length) { @@ -101,12 +149,17 @@ public class ScoreMatrix implements SimilarityScoreModelI, } } - this.matrix = matrix; - this.name = name; + this.matrix = values; + this.name = theName; + this.description = theDescription; this.symbols = alphabet; symbolIndex = buildSymbolIndex(alphabet); + findMinMax(); + + symmetric = checkSymmetry(); + /* * crude heuristic for now... */ @@ -114,6 +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. @@ -123,11 +222,14 @@ public class ScoreMatrix implements SimilarityScoreModelI, * Mappings are added automatically for lower case symbols (for non case * sensitive scoring), unless they are explicitly present in the alphabet (are * scored separately in the score matrix). + *
+ * the gap character (space, dash or dot) included in the alphabet (if any) is + * recorded in a field * * @param alphabet * @return */ - static short[] buildSymbolIndex(char[] alphabet) + short[] buildSymbolIndex(char[] alphabet) { short[] index = new short[MAX_ASCII + 1]; Arrays.fill(index, UNMAPPED); @@ -162,6 +264,12 @@ public class ScoreMatrix implements SimilarityScoreModelI, } @Override + public String getDescription() + { + return description; + } + + @Override public boolean isDNA() { return !peptide; @@ -214,8 +322,9 @@ 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) @@ -237,7 +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(); } /** @@ -274,7 +390,6 @@ public class ScoreMatrix implements SimilarityScoreModelI, else { sb.append("ScoreMatrix ").append(getName()).append("\n"); - sb.append(symbols).append("\n"); } for (char sym : symbols) { @@ -344,46 +459,108 @@ public class ScoreMatrix implements SimilarityScoreModelI, *