2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.analysis.scoremodels;
23 import jalview.api.analysis.PairwiseScoreModelI;
24 import jalview.api.analysis.SimilarityParamsI;
25 import jalview.api.analysis.SimilarityScoreModelI;
26 import jalview.datamodel.AlignmentView;
27 import jalview.math.Matrix;
28 import jalview.math.MatrixI;
29 import jalview.util.Comparison;
31 import java.util.Arrays;
34 * A class that models a substitution score matrix for any given alphabet of
37 public class ScoreMatrix implements SimilarityScoreModelI,
40 private static final char GAP_CHARACTER = Comparison.GAP_DASH;
43 * an arbitrary score to assign for identity of an unknown symbol
44 * (this is the value on the diagonal in the * column of the NCBI matrix)
45 * (though a case could be made for using the minimum diagonal value)
47 private static final int UNKNOWN_IDENTITY_SCORE = 1;
50 * Jalview 2.10.1 treated gaps as X (peptide) or N (nucleotide)
51 * for pairwise scoring; 2.10.2 uses gap score (last column) in
52 * score matrix (JAL-2397)
53 * Set this flag to true (via Groovy) for 2.10.1 behaviour
55 private static boolean scoreGapAsAny = false;
57 public static final short UNMAPPED = (short) -1;
59 private static final String BAD_ASCII_ERROR = "Unexpected character %s in getPairwiseScore";
61 private static final int MAX_ASCII = 127;
64 * the name of the model as shown in menus
65 * each score model in use should have a unique name
70 * a description for the model as shown in tooltips
72 private String description;
75 * the characters that the model provides scores for
77 private char[] symbols;
80 * the score matrix; both dimensions must equal the number of symbols
81 * matrix[i][j] is the substitution score for replacing symbols[i] with symbols[j]
83 private float[][] matrix;
86 * quick lookup to convert from an ascii character value to the index
87 * of the corresponding symbol in the score matrix
89 private short[] symbolIndex;
92 * true for Protein Score matrix, false for dna score matrix
94 private boolean peptide;
96 private float minValue;
98 private float maxValue;
101 * Constructor given a name, symbol alphabet, and matrix of scores for pairs
102 * of symbols. The matrix should be square and of the same size as the
103 * alphabet, for example 20x20 for a 20 symbol alphabet.
106 * Unique, human readable name for the matrix
108 * the symbols to which scores apply
110 * Pairwise scores indexed according to the symbol alphabet
112 public ScoreMatrix(String theName, char[] alphabet, float[][] values)
114 if (alphabet.length != values.length)
116 throw new IllegalArgumentException(
117 "score matrix size must match alphabet size");
119 for (float[] row : values)
121 if (row.length != alphabet.length)
123 throw new IllegalArgumentException(
124 "score matrix size must be square");
128 this.matrix = values;
130 this.symbols = alphabet;
132 symbolIndex = buildSymbolIndex(alphabet);
137 * crude heuristic for now...
139 peptide = alphabet.length >= 20;
143 * Record the minimum and maximum score values
145 protected void findMinMax()
147 float min = Float.MAX_VALUE;
148 float max = -Float.MAX_VALUE;
151 for (float[] row : matrix)
157 min = Math.min(min, f);
158 max = Math.max(max, f);
168 * Returns an array A where A[i] is the position in the alphabet array of the
169 * character whose value is i. For example if the alphabet is { 'A', 'D', 'X'
170 * } then A['D'] = A[68] = 1.
172 * Unmapped characters (not in the alphabet) get an index of -1.
174 * Mappings are added automatically for lower case symbols (for non case
175 * sensitive scoring), unless they are explicitly present in the alphabet (are
176 * scored separately in the score matrix).
178 * the gap character (space, dash or dot) included in the alphabet (if any) is
179 * recorded in a field
184 short[] buildSymbolIndex(char[] alphabet)
186 short[] index = new short[MAX_ASCII + 1];
187 Arrays.fill(index, UNMAPPED);
189 for (char c : alphabet)
197 * also map lower-case character (unless separately mapped)
199 if (c >= 'A' && c <= 'Z')
201 short lowerCase = (short) (c + ('a' - 'A'));
202 if (index[lowerCase] == UNMAPPED)
204 index[lowerCase] = pos;
213 public String getName()
219 public String getDescription()
225 public boolean isDNA()
231 public boolean isProtein()
237 * Returns a copy of the score matrix as used in getPairwiseScore. If using
238 * this matrix directly, callers <em>must</em> also call
239 * <code>getMatrixIndex</code> in order to get the matrix index for each
240 * character (symbol).
243 * @see #getMatrixIndex(char)
245 public float[][] getMatrix()
247 float[][] v = new float[matrix.length][matrix.length];
248 for (int i = 0; i < matrix.length; i++)
250 v[i] = Arrays.copyOf(matrix[i], matrix[i].length);
256 * Answers the matrix index for a given character, or -1 if unmapped in the
257 * matrix. Use this method only if using <code>getMatrix</code> in order to
258 * compute scores directly (without symbol lookup) for efficiency.
264 public int getMatrixIndex(char c)
266 if (c < symbolIndex.length)
268 return symbolIndex[c];
277 * Returns the pairwise score for substituting c with d. If either c or d is
278 * an unexpected character, returns 1 for identity (c == d), else the minimum
279 * score value in the matrix.
282 public float getPairwiseScore(char c, char d)
284 if (c >= symbolIndex.length)
286 System.err.println(String.format(BAD_ASCII_ERROR, c));
289 if (d >= symbolIndex.length)
291 System.err.println(String.format(BAD_ASCII_ERROR, d));
295 int cIndex = symbolIndex[c];
296 int dIndex = symbolIndex[d];
297 if (cIndex != UNMAPPED && dIndex != UNMAPPED)
299 return matrix[cIndex][dIndex];
303 * one or both symbols not found in the matrix
304 * currently scoring as 1 (for identity) or the minimum
305 * matrix score value (otherwise)
306 * (a case could be made for using minimum row/column value instead)
308 return c == d ? UNKNOWN_IDENTITY_SCORE : getMinimumScore();
312 * pretty print the matrix
315 public String toString()
317 return outputMatrix(false);
321 * Print the score matrix, optionally formatted as html, with the alphabet
322 * symbols as column headings and at the start of each row.
324 * The non-html format should give an output which can be parsed as a score
330 public String outputMatrix(boolean html)
332 StringBuilder sb = new StringBuilder(512);
335 * heading row with alphabet
339 sb.append("<table border=\"1\">");
340 sb.append(html ? "<tr><th></th>" : "");
344 sb.append("ScoreMatrix ").append(getName()).append("\n");
346 for (char sym : symbols)
350 sb.append("<th> ").append(sym).append(" </th>");
354 sb.append("\t").append(sym);
357 sb.append(html ? "</tr>\n" : "\n");
362 for (char c1 : symbols)
366 sb.append("<tr><td>");
368 sb.append(c1).append(html ? "</td>" : "");
369 for (char c2 : symbols)
371 sb.append(html ? "<td>" : "\t")
372 .append(matrix[symbolIndex[c1]][symbolIndex[c2]])
373 .append(html ? "</td>" : "");
375 sb.append(html ? "</tr>\n" : "\n");
379 sb.append("</table>");
381 return sb.toString();
385 * Answers the number of symbols coded for (also equal to the number of rows
386 * and columns of the score matrix)
392 return symbols.length;
396 * Computes an NxN matrix where N is the number of sequences, and entry [i, j]
397 * is sequence[i] pairwise multiplied with sequence[j], as a sum of scores
398 * computed using the current score matrix. For example
400 * <li>Sequences:</li>
405 * <li>Score matrix is BLOSUM62</li>
406 * <li>Gaps treated same as X (unknown)</li>
407 * <li>product [0, 0] = F.F + K.K + L.L = 6 + 5 + 4 = 15</li>
408 * <li>product [1, 1] = R.R + -.- + D.D = 5 + -1 + 6 = 10</li>
409 * <li>product [2, 2] = Q.Q + I.I + A.A = 5 + 4 + 4 = 13</li>
410 * <li>product [3, 3] = G.G + W.W + C.C = 6 + 11 + 9 = 26</li>
411 * <li>product[0, 1] = F.R + K.- + L.D = -3 + -1 + -3 = -8
416 public MatrixI findSimilarities(AlignmentView seqstrings,
417 SimilarityParamsI options)
419 char gapChar = scoreGapAsAny ? (seqstrings.isNa() ? 'N' : 'X')
421 String[] seqs = seqstrings.getSequenceStrings(gapChar);
422 return findSimilarities(seqs, options);
426 * Computes pairwise similarities of a set of sequences using the given
433 protected MatrixI findSimilarities(String[] seqs, SimilarityParamsI params)
435 double[][] values = new double[seqs.length][];
436 for (int row = 0; row < seqs.length; row++)
438 values[row] = new double[seqs.length];
439 for (int col = 0; col < seqs.length; col++)
441 double total = computeSimilarity(seqs[row], seqs[col], params);
442 values[row][col] = total;
445 return new Matrix(values);
449 * Calculates the pairwise similarity of two strings using the given
450 * calculation parameters
457 protected double computeSimilarity(String seq1, String seq2,
458 SimilarityParamsI params)
460 int len1 = seq1.length();
461 int len2 = seq2.length();
464 int width = Math.max(len1, len2);
465 for (int i = 0; i < width; i++)
467 if (i >= len1 || i >= len2)
470 * off the end of one sequence; stop if we are only matching
471 * on the shorter sequence length, else treat as trailing gap
473 if (params.denominateByShortestLength())
479 char c1 = i >= len1 ? GAP_CHARACTER : seq1.charAt(i);
480 char c2 = i >= len2 ? GAP_CHARACTER : seq2.charAt(i);
481 boolean gap1 = Comparison.isGap(c1);
482 boolean gap2 = Comparison.isGap(c2);
487 * gap-gap: include if options say so, else ignore
489 if (!params.includeGappedColumns())
494 else if (gap1 || gap2)
497 * gap-residue: score if options say so
499 if (!params.includeGaps())
504 float score = getPairwiseScore(c1, c2);
511 * Answers a hashcode computed from the symbol alphabet and the matrix score
515 public int hashCode()
517 int hs = Arrays.hashCode(symbols);
518 for (float[] row : matrix)
520 hs = hs * 31 + Arrays.hashCode(row);
526 * Answers true if the argument is a ScoreMatrix with the same symbol alphabet
527 * and score values, else false
530 public boolean equals(Object obj)
532 if (!(obj instanceof ScoreMatrix))
536 ScoreMatrix sm = (ScoreMatrix) obj;
537 if (Arrays.equals(symbols, sm.symbols)
538 && Arrays.deepEquals(matrix, sm.matrix))
546 * Returns the alphabet the matrix scores for, as a string of characters
552 return new String(symbols);
555 public void setDescription(String desc)
560 public float getMinimumScore()
565 public float getMaximumScore()