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;
33 public class ScoreMatrix implements SimilarityScoreModelI,
36 private static final char GAP_CHARACTER = Comparison.GAP_DASH;
39 * Jalview 2.10.1 treated gaps as X (peptide) or N (nucleotide)
40 * for pairwise scoring; 2.10.2 uses gap score (last column) in
41 * score matrix (JAL-2397)
42 * Set this flag to true (via Groovy) for 2.10.1 behaviour
44 private static boolean scoreGapAsAny = false;
46 public static final short UNMAPPED = (short) -1;
48 private static final String BAD_ASCII_ERROR = "Unexpected character %s in getPairwiseScore";
50 private static final int MAX_ASCII = 127;
53 * the name of the model as shown in menus
58 * the characters that the model provides scores for
60 private char[] symbols;
63 * the score matrix; both dimensions must equal the number of symbols
64 * matrix[i][j] is the substitution score for replacing symbols[i] with symbols[j]
66 private float[][] matrix;
69 * quick lookup to convert from an ascii character value to the index
70 * of the corresponding symbol in the score matrix
72 private short[] symbolIndex;
75 * true for Protein Score matrix, false for dna score matrix
77 private boolean peptide;
80 * Constructor given a name, symbol alphabet, and matrix of scores for pairs
81 * of symbols. The matrix should be square and of the same size as the
82 * alphabet, for example 20x20 for a 20 symbol alphabet.
85 * Unique, human readable name for the matrix
87 * the symbols to which scores apply
89 * Pairwise scores indexed according to the symbol alphabet
91 public ScoreMatrix(String theName, char[] alphabet, float[][] values)
93 if (alphabet.length != values.length)
95 throw new IllegalArgumentException(
96 "score matrix size must match alphabet size");
98 for (float[] row : values)
100 if (row.length != alphabet.length)
102 throw new IllegalArgumentException(
103 "score matrix size must be square");
107 this.matrix = values;
109 this.symbols = alphabet;
111 symbolIndex = buildSymbolIndex(alphabet);
114 * crude heuristic for now...
116 peptide = alphabet.length >= 20;
120 * Returns an array A where A[i] is the position in the alphabet array of the
121 * character whose value is i. For example if the alphabet is { 'A', 'D', 'X'
122 * } then A['D'] = A[68] = 1.
124 * Unmapped characters (not in the alphabet) get an index of -1.
126 * Mappings are added automatically for lower case symbols (for non case
127 * sensitive scoring), unless they are explicitly present in the alphabet (are
128 * scored separately in the score matrix).
133 static short[] buildSymbolIndex(char[] alphabet)
135 short[] index = new short[MAX_ASCII + 1];
136 Arrays.fill(index, UNMAPPED);
138 for (char c : alphabet)
146 * also map lower-case character (unless separately mapped)
148 if (c >= 'A' && c <= 'Z')
150 short lowerCase = (short) (c + ('a' - 'A'));
151 if (index[lowerCase] == UNMAPPED)
153 index[lowerCase] = pos;
162 public String getName()
168 public boolean isDNA()
174 public boolean isProtein()
180 * Returns a copy of the score matrix as used in getPairwiseScore. If using
181 * this matrix directly, callers <em>must</em> also call
182 * <code>getMatrixIndex</code> in order to get the matrix index for each
183 * character (symbol).
186 * @see #getMatrixIndex(char)
188 public float[][] getMatrix()
190 float[][] v = new float[matrix.length][matrix.length];
191 for (int i = 0; i < matrix.length; i++)
193 v[i] = Arrays.copyOf(matrix[i], matrix[i].length);
199 * Answers the matrix index for a given character, or -1 if unmapped in the
200 * matrix. Use this method only if using <code>getMatrix</code> in order to
201 * compute scores directly (without symbol lookup) for efficiency.
207 public int getMatrixIndex(char c)
209 if (c < symbolIndex.length)
211 return symbolIndex[c];
220 * Returns the pairwise score for substituting c with d, or zero if c or d is
221 * an unscored or unexpected character
224 public float getPairwiseScore(char c, char d)
226 if (c >= symbolIndex.length)
228 System.err.println(String.format(BAD_ASCII_ERROR, c));
231 if (d >= symbolIndex.length)
233 System.err.println(String.format(BAD_ASCII_ERROR, d));
237 int cIndex = symbolIndex[c];
238 int dIndex = symbolIndex[d];
239 if (cIndex != UNMAPPED && dIndex != UNMAPPED)
241 return matrix[cIndex][dIndex];
247 * pretty print the matrix
250 public String toString()
252 return outputMatrix(false);
256 * Print the score matrix, optionally formatted as html, with the alphabet
257 * symbols as column headings and at the start of each row.
259 * The non-html format should give an output which can be parsed as a score
265 public String outputMatrix(boolean html)
267 StringBuilder sb = new StringBuilder(512);
270 * heading row with alphabet
274 sb.append("<table border=\"1\">");
275 sb.append(html ? "<tr><th></th>" : "");
279 sb.append("ScoreMatrix ").append(getName()).append("\n");
280 sb.append(symbols).append("\n");
282 for (char sym : symbols)
286 sb.append("<th> ").append(sym).append(" </th>");
290 sb.append("\t").append(sym);
293 sb.append(html ? "</tr>\n" : "\n");
298 for (char c1 : symbols)
302 sb.append("<tr><td>");
304 sb.append(c1).append(html ? "</td>" : "");
305 for (char c2 : symbols)
307 sb.append(html ? "<td>" : "\t")
308 .append(matrix[symbolIndex[c1]][symbolIndex[c2]])
309 .append(html ? "</td>" : "");
311 sb.append(html ? "</tr>\n" : "\n");
315 sb.append("</table>");
317 return sb.toString();
321 * Answers the number of symbols coded for (also equal to the number of rows
322 * and columns of the score matrix)
328 return symbols.length;
332 * Computes an NxN matrix where N is the number of sequences, and entry [i, j]
333 * is sequence[i] pairwise multiplied with sequence[j], as a sum of scores
334 * computed using the current score matrix. For example
336 * <li>Sequences:</li>
341 * <li>Score matrix is BLOSUM62</li>
342 * <li>Gaps treated same as X (unknown)</li>
343 * <li>product [0, 0] = F.F + K.K + L.L = 6 + 5 + 4 = 15</li>
344 * <li>product [1, 1] = R.R + -.- + D.D = 5 + -1 + 6 = 10</li>
345 * <li>product [2, 2] = Q.Q + I.I + A.A = 5 + 4 + 4 = 13</li>
346 * <li>product [3, 3] = G.G + W.W + C.C = 6 + 11 + 9 = 26</li>
347 * <li>product[0, 1] = F.R + K.- + L.D = -3 + -1 + -3 = -8
352 public MatrixI findSimilarities(AlignmentView seqstrings,
353 SimilarityParamsI options)
355 char gapChar = scoreGapAsAny ? (seqstrings.isNa() ? 'N' : 'X')
356 : Comparison.GAP_DASH;
357 String[] seqs = seqstrings.getSequenceStrings(gapChar);
358 return findSimilarities(seqs, options);
362 * Computes pairwise similarities of a set of sequences using the given
369 protected MatrixI findSimilarities(String[] seqs, SimilarityParamsI params)
371 double[][] values = new double[seqs.length][];
372 for (int row = 0; row < seqs.length; row++)
374 values[row] = new double[seqs.length];
375 for (int col = 0; col < seqs.length; col++)
377 double total = computeSimilarity(seqs[row], seqs[col], params);
378 values[row][col] = total;
381 return new Matrix(values);
385 * Calculates the pairwise similarity of two strings using the given
386 * calculation parameters
393 protected double computeSimilarity(String seq1, String seq2,
394 SimilarityParamsI params)
396 int len1 = seq1.length();
397 int len2 = seq2.length();
400 int width = Math.max(len1, len2);
401 for (int i = 0; i < width; i++)
403 if (i >= len1 || i >= len2)
406 * off the end of one sequence; stop if we are only matching
407 * on the shorter sequence length, else treat as trailing gap
409 if (params.denominateByShortestLength())
415 char c1 = i >= len1 ? GAP_CHARACTER : seq1.charAt(i);
416 char c2 = i >= len2 ? GAP_CHARACTER : seq2.charAt(i);
417 boolean gap1 = Comparison.isGap(c1);
418 boolean gap2 = Comparison.isGap(c2);
423 * gap-gap: include if options say so, else ignore
425 if (!params.includeGappedColumns())
430 else if (gap1 || gap2)
433 * gap-residue: score if options say so
435 if (!params.includeGaps())
440 float score = getPairwiseScore(c1, c2);
447 * Answers a hashcode computed from the symbol alphabet and the matrix score
451 public int hashCode()
453 int hs = Arrays.hashCode(symbols);
454 for (float[] row : matrix)
456 hs = hs * 31 + Arrays.hashCode(row);
462 * Answers true if the argument is a ScoreMatrix with the same symbol alphabet
463 * and score values, else false
466 public boolean equals(Object obj)
468 if (!(obj instanceof ScoreMatrix))
472 ScoreMatrix sm = (ScoreMatrix) obj;
473 if (Arrays.equals(symbols, sm.symbols)
474 && Arrays.deepEquals(matrix, sm.matrix))
482 * Returns the alphabet the matrix scores for, as a string of characters
486 public String getSymbols()
488 return new String(symbols);