X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;ds=sidebyside;f=src%2Fjalview%2Fanalysis%2Fscoremodels%2FScoreMatrix.java;h=e2c14e91e590cff8a69b98a08ff588702a228b31;hb=ae783be4cfa1df7fd59a4d0895df8bd1534d334c;hp=41d73838b98e6b83c690c7d0d009816d40d1d5b9;hpb=baad2f0ba2b171dd3d52c17afa46ef800334ea5e;p=jalview.git diff --git a/src/jalview/analysis/scoremodels/ScoreMatrix.java b/src/jalview/analysis/scoremodels/ScoreMatrix.java index 41d7383..e2c14e9 100644 --- a/src/jalview/analysis/scoremodels/ScoreMatrix.java +++ b/src/jalview/analysis/scoremodels/ScoreMatrix.java @@ -30,10 +30,27 @@ import jalview.util.Comparison; import java.util.Arrays; +/** + * A class that models a substitution score matrix for any given alphabet of + * symbols + */ public class ScoreMatrix implements SimilarityScoreModelI, PairwiseScoreModelI { /* + * 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; + + /* + * this fields records which gap character (if any) is used in the alphabet; + * space, dash or dot are recognised as gap symbols + */ + private char gapCharacter = '0'; + + /* * 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) @@ -49,10 +66,16 @@ 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; @@ -74,26 +97,30 @@ public class ScoreMatrix implements SimilarityScoreModelI, */ 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 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) { @@ -102,12 +129,14 @@ public class ScoreMatrix implements SimilarityScoreModelI, } } - this.matrix = matrix; - this.name = name; + this.matrix = values; + this.name = theName; this.symbols = alphabet; symbolIndex = buildSymbolIndex(alphabet); + findMinMax(); + /* * crude heuristic for now... */ @@ -115,6 +144,31 @@ public class ScoreMatrix implements SimilarityScoreModelI, } /** + * 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. @@ -124,17 +178,25 @@ 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);
short pos = 0;
for (char c : alphabet)
{
+ if (Comparison.isGap(c))
+ {
+ gapCharacter = c;
+ }
+
if (c <= MAX_ASCII)
{
index[c] = pos;
@@ -163,6 +225,12 @@ public class ScoreMatrix implements SimilarityScoreModelI,
}
@Override
+ public String getDescription()
+ {
+ return description;
+ }
+
+ @Override
public boolean isDNA()
{
return !peptide;
@@ -215,8 +283,22 @@ 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
+ * 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. 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)
@@ -238,7 +320,14 @@ 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();
}
/**
@@ -275,7 +364,6 @@ public class ScoreMatrix implements SimilarityScoreModelI,
else
{
sb.append("ScoreMatrix ").append(getName()).append("\n");
- sb.append(symbols).append("\n");
}
for (char sym : symbols)
{
@@ -350,7 +438,8 @@ public class ScoreMatrix implements SimilarityScoreModelI,
public MatrixI findSimilarities(AlignmentView seqstrings,
SimilarityParamsI options)
{
- char gapChar = scoreGapAsAny ? (seqstrings.isNa() ? 'N' : 'X') : ' ';
+ char gapChar = scoreGapAsAny ? (seqstrings.isNa() ? 'N' : 'X')
+ : gapCharacter;
String[] seqs = seqstrings.getSequenceStrings(gapChar);
return findSimilarities(seqs, options);
}
@@ -408,9 +497,9 @@ public class ScoreMatrix implements SimilarityScoreModelI,
break;
}
}
- // Change GAP_SPACE to GAP_DASH if we adopt - for gap in matrices
- char c1 = i >= len1 ? Comparison.GAP_SPACE : seq1.charAt(i);
- char c2 = i >= len2 ? Comparison.GAP_SPACE : seq2.charAt(i);
+
+ char c1 = i >= len1 ? gapCharacter : seq1.charAt(i);
+ char c2 = i >= len2 ? gapCharacter : seq2.charAt(i);
boolean gap1 = Comparison.isGap(c1);
boolean gap2 = Comparison.isGap(c2);
@@ -429,7 +518,7 @@ public class ScoreMatrix implements SimilarityScoreModelI,
/*
* gap-residue: score if options say so
*/
- if (!params.includesGaps())
+ if (!params.includeGaps())
{
continue;
}
@@ -480,8 +569,23 @@ public class ScoreMatrix implements SimilarityScoreModelI,
*
* @return
*/
- public String getSymbols()
+ String getSymbols()
{
return new String(symbols);
}
+
+ public void setDescription(String desc)
+ {
+ description = desc;
+ }
+
+ public float getMinimumScore()
+ {
+ return minValue;
+ }
+
+ public float getMaximumScore()
+ {
+ return maxValue;
+ }
}