X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2Fscoremodels%2FScoreMatrix.java;h=6cdfacbe56695b322b074d05cf90fef8565b37ea;hb=d25ed8f9fb513fe5c33823afbeeb557c73a8c27b;hp=f0115bfca2874ae34f8a9d2c18623802c98838e1;hpb=c6e8e8ccd10f21698226ae37196cd9680e6804a0;p=jalview.git diff --git a/src/jalview/analysis/scoremodels/ScoreMatrix.java b/src/jalview/analysis/scoremodels/ScoreMatrix.java index f0115bf..6cdfacb 100644 --- a/src/jalview/analysis/scoremodels/ScoreMatrix.java +++ b/src/jalview/analysis/scoremodels/ScoreMatrix.java @@ -20,23 +20,43 @@ */ 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.datamodel.AlignmentView; +import jalview.math.Matrix; +import jalview.math.MatrixI; +import jalview.util.Comparison; -import java.io.BufferedReader; -import java.io.IOException; -import java.io.InputStream; -import java.io.InputStreamReader; import java.util.Arrays; -import java.util.StringTokenizer; -public class ScoreMatrix extends PairwiseSeqScoreModel implements - ScoreModelI +/** + * 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 { - public static final short UNMAPPED = (short) -1; + 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; - private static final String DELIMITERS = " ,\t"; + /* + * 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; - private static final String COMMENT_CHAR = "#"; + public static final short UNMAPPED = (short) -1; private static final String BAD_ASCII_ERROR = "Unexpected character %s in getPairwiseScore"; @@ -44,10 +64,16 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements /* * 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; @@ -69,24 +95,67 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements */ private boolean peptide; + private float minValue; + + private float maxValue; + + /** + * 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 theName + * Unique, human readable name for the matrix + * @param alphabet + * the symbols to which scores apply + * @param values + * Pairwise scores indexed according to the symbol alphabet + */ + public ScoreMatrix(String theName, char[] alphabet, float[][] values) + { + this(theName, null, alphabet, values); + } + /** - * Constructor + * 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 name + * @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 matrix + * @param values * Pairwise scores indexed according to the symbol alphabet */ - public ScoreMatrix(String name, char[] alphabet, float[][] matrix) + public ScoreMatrix(String theName, String theDescription, char[] alphabet, + float[][] values) { - this.matrix = matrix; - this.name = name; + if (alphabet.length != values.length) + { + throw new IllegalArgumentException( + "score matrix size must match alphabet size"); + } + for (float[] row : values) + { + if (row.length != alphabet.length) + { + throw new IllegalArgumentException( + "score matrix size must be square"); + } + } + + this.matrix = values; + this.name = theName; + this.description = theDescription; this.symbols = alphabet; symbolIndex = buildSymbolIndex(alphabet); + findMinMax(); + /* * crude heuristic for now... */ @@ -94,6 +163,31 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements } /** + * 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. @@ -103,11 +197,14 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements * 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);
@@ -142,6 +239,12 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements
}
@Override
+ public String getDescription()
+ {
+ return description;
+ }
+
+ @Override
public boolean isDNA()
{
return !peptide;
@@ -153,25 +256,60 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements
return peptide;
}
- @Override
+ /**
+ * Returns a copy of the score matrix as used in getPairwiseScore. If using
+ * this matrix directly, callers must also call
+ * getMatrixIndex
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;
}
/**
- * Returns the pairwise score for substituting c with d, or zero if c or d is
- * an unscored or unexpected character
+ * Answers the matrix index for a given 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.
+ *
+ * @param c
+ * @return
+ * @see #getMatrix()
+ */
+ public int getMatrixIndex(char c)
+ {
+ if (c < symbolIndex.length)
+ {
+ return symbolIndex[c];
+ }
+ else
+ {
+ return UNMAPPED;
+ }
+ }
+
+ /**
+ * 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)
{
- if (c > MAX_ASCII)
+ if (c >= symbolIndex.length)
{
System.err.println(String.format(BAD_ASCII_ERROR, c));
return 0;
}
- if (d > MAX_ASCII)
+ if (d >= symbolIndex.length)
{
System.err.println(String.format(BAD_ASCII_ERROR, d));
return 0;
@@ -183,7 +321,14 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements
{
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();
}
/**
@@ -196,7 +341,12 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements
}
/**
- * 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.
+ *
+ * The non-html format should give an output which can be parsed as a score + * matrix file + * * @param html * @return */ @@ -212,6 +362,10 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements sb.append("
" : ""); } + else + { + sb.append("ScoreMatrix ").append(getName()).append("\n"); + } for (char sym : symbols) { if (html) @@ -251,138 +405,191 @@ public class ScoreMatrix extends PairwiseSeqScoreModel implements } /** - * Parse a score matrix from the given input stream and returns a ScoreMatrix - * object. If parsing fails, error messages are written to syserr and null is - * returned. It is the caller's responsibility to close the input stream. + * Answers the number of symbols coded for (also equal to the number of rows + * and columns of the score matrix) * - * @param is * @return */ - public static ScoreMatrix parse(InputStream is) + public int getSize() { - ScoreMatrix sm = null; - BufferedReader br = new BufferedReader(new InputStreamReader(is)); - int lineNo = 0; - String name = null; - String alphabet = null; - float[][] scores = null; - int size = 0; - int row = 0; - - try - { - String data; + return symbols.length; + } - while ((data = br.readLine()) != null) + /** + * 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 + * |
---|