*/
package jalview.analysis.scoremodels;
-import jalview.api.analysis.ScoreModelI;
-import jalview.schemes.ResidueProperties;
+import jalview.api.analysis.PairwiseScoreModelI;
+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;
-public class ScoreMatrix extends PairwiseSeqScoreModel implements
- ScoreModelI
+import java.util.Arrays;
+
+/**
+ * A class that models a substitution score matrix for any given alphabet of
+ * symbols
+ */
+public class ScoreMatrix implements SimilarityScoreModelI,
+ PairwiseScoreModelI
{
- String name;
+ /*
+ * 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';
- @Override
- public String getName()
- {
- return name;
- }
+ /*
+ * 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;
- /**
- * reference to integer score matrix
+ public static final short UNMAPPED = (short) -1;
+
+ private static final String BAD_ASCII_ERROR = "Unexpected character %s in getPairwiseScore";
+
+ private static final int MAX_ASCII = 127;
+
+ /*
+ * the name of the model as shown in menus
+ * each score model in use should have a unique name
*/
- int[][] matrix;
+ private String name;
- /**
- * 0 for Protein Score matrix. 1 for dna score matrix
+ /*
+ * a description for the model as shown in tooltips
+ */
+ private String description;
+
+ /*
+ * the characters that the model provides scores for
+ */
+ private char[] symbols;
+
+ /*
+ * the score matrix; both dimensions must equal the number of symbols
+ * matrix[i][j] is the substitution score for replacing symbols[i] with symbols[j]
+ */
+ private float[][] matrix;
+
+ /*
+ * quick lookup to convert from an ascii character value to the index
+ * of the corresponding symbol in the score matrix
+ */
+ private short[] symbolIndex;
+
+ /*
+ * true for Protein Score matrix, false for dna score matrix
*/
- int type;
+ private boolean peptide;
/**
+ * 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 matrix
- * Pairwise scores indexed according to appropriate symbol alphabet
- * @param type
- * 0 for Protein, 1 for NA
+ * @param alphabet
+ * the symbols to which scores apply
+ * @param values
+ * Pairwise scores indexed according to the symbol alphabet
*/
- public ScoreMatrix(String name, int[][] matrix, int type)
+ public ScoreMatrix(String theName, char[] alphabet, float[][] values)
{
- this.matrix = matrix;
- this.type = type;
- 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.symbols = alphabet;
+
+ symbolIndex = buildSymbolIndex(alphabet);
+
+ /*
+ * crude heuristic for now...
+ */
+ peptide = alphabet.length >= 20;
+ }
+
+ /**
+ * 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.
+ * <p>
+ * Unmapped characters (not in the alphabet) get an index of -1.
+ * <p>
+ * 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).
+ * <p>
+ * the gap character (space, dash or dot) included in the alphabet (if any) is
+ * recorded in a field
+ *
+ * @param alphabet
+ * @return
+ */
+ 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;
+ }
+
+ /*
+ * also map lower-case character (unless separately mapped)
+ */
+ if (c >= 'A' && c <= 'Z')
+ {
+ short lowerCase = (short) (c + ('a' - 'A'));
+ if (index[lowerCase] == UNMAPPED)
+ {
+ index[lowerCase] = pos;
+ }
+ }
+ pos++;
+ }
+ return index;
+ }
+
+ @Override
+ public String getName()
+ {
+ return name;
+ }
+
+ @Override
+ public String getDescription()
+ {
+ return description;
}
@Override
public boolean isDNA()
{
- return type == 1;
+ return !peptide;
}
@Override
public boolean isProtein()
{
- return type == 0;
+ return peptide;
}
- @Override
- public int[][] getMatrix()
+ /**
+ * Returns a copy of the score matrix as used in getPairwiseScore. If using
+ * this matrix directly, callers <em>must</em> also call
+ * <code>getMatrixIndex</code> in order to get the matrix index for each
+ * character (symbol).
+ *
+ * @return
+ * @see #getMatrixIndex(char)
+ */
+ public float[][] getMatrix()
+ {
+ 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;
+ }
+
+ /**
+ * Answers the matrix index for a given character, or -1 if unmapped in the
+ * matrix. Use this method only if using <code>getMatrix</code> in order to
+ * compute scores directly (without symbol lookup) for efficiency.
+ *
+ * @param c
+ * @return
+ * @see #getMatrix()
+ */
+ public int getMatrixIndex(char c)
{
- return matrix;
+ if (c < symbolIndex.length)
+ {
+ return symbolIndex[c];
+ }
+ else
+ {
+ return UNMAPPED;
+ }
}
/**
+ * Answers the matrix index for the gap character, or -1 if unmapped in the
+ * matrix. Use this method only if using <code>getMatrix</code> in order to
+ * compute scores directly (without symbol lookup) for efficiency.
*
- * @param A1
- * @param A2
- * @return score for substituting first char in A1 with first char in A2
+ * @return
+ * @see #getMatrix()
*/
- public int getPairwiseScore(String A1, String A2)
+ public int getGapIndex()
{
- return getPairwiseScore(A1.charAt(0), A2.charAt(0));
+ return getMatrixIndex(gapCharacter);
}
+ /**
+ * Returns the pairwise score for substituting c with d, or zero if c or d is
+ * an unscored or unexpected character
+ */
@Override
- public int getPairwiseScore(char c, char d)
+ public float getPairwiseScore(char c, char d)
{
- int pog = 0;
-
- try
+ if (c >= symbolIndex.length)
{
- int a = (type == 0) ? ResidueProperties.aaIndex[c]
- : ResidueProperties.nucleotideIndex[c];
- int b = (type == 0) ? ResidueProperties.aaIndex[d]
- : ResidueProperties.nucleotideIndex[d];
-
- pog = matrix[a][b];
- } catch (Exception e)
+ System.err.println(String.format(BAD_ASCII_ERROR, c));
+ return 0;
+ }
+ if (d >= symbolIndex.length)
{
- // System.out.println("Unknown residue in " + A1 + " " + A2);
+ System.err.println(String.format(BAD_ASCII_ERROR, d));
+ return 0;
}
- return pog;
+ int cIndex = symbolIndex[c];
+ int dIndex = symbolIndex[d];
+ if (cIndex != UNMAPPED && dIndex != UNMAPPED)
+ {
+ return matrix[cIndex][dIndex];
+ }
+ return 0;
}
/**
return outputMatrix(false);
}
+ /**
+ * Print the score matrix, optionally formatted as html, with the alphabet
+ * symbols as column headings and at the start of each row.
+ * <p>
+ * The non-html format should give an output which can be parsed as a score
+ * matrix file
+ *
+ * @param html
+ * @return
+ */
public String outputMatrix(boolean html)
{
- StringBuffer sb = new StringBuffer();
- int[] symbols = (type == 0) ? ResidueProperties.aaIndex
- : ResidueProperties.nucleotideIndex;
- int symMax = (type == 0) ? ResidueProperties.maxProteinIndex
- : ResidueProperties.maxNucleotideIndex;
- boolean header = true;
+ StringBuilder sb = new StringBuilder(512);
+
+ /*
+ * heading row with alphabet
+ */
if (html)
{
sb.append("<table border=\"1\">");
+ sb.append(html ? "<tr><th></th>" : "");
+ }
+ else
+ {
+ sb.append("ScoreMatrix ").append(getName()).append("\n");
+ sb.append(symbols).append("\n");
+ }
+ for (char sym : symbols)
+ {
+ if (html)
+ {
+ sb.append("<th> ").append(sym).append(" </th>");
+ }
+ else
+ {
+ sb.append("\t").append(sym);
+ }
+ }
+ sb.append(html ? "</tr>\n" : "\n");
+
+ /*
+ * table of scores
+ */
+ for (char c1 : symbols)
+ {
+ if (html)
+ {
+ sb.append("<tr><td>");
+ }
+ sb.append(c1).append(html ? "</td>" : "");
+ for (char c2 : symbols)
+ {
+ sb.append(html ? "<td>" : "\t")
+ .append(matrix[symbolIndex[c1]][symbolIndex[c2]])
+ .append(html ? "</td>" : "");
+ }
+ sb.append(html ? "</tr>\n" : "\n");
+ }
+ if (html)
+ {
+ sb.append("</table>");
+ }
+ return sb.toString();
+ }
+
+ /**
+ * Answers the number of symbols coded for (also equal to the number of rows
+ * and columns of the score matrix)
+ *
+ * @return
+ */
+ public int getSize()
+ {
+ return symbols.length;
+ }
+
+ /**
+ * 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
+ * <ul>
+ * <li>Sequences:</li>
+ * <li>FKL</li>
+ * <li>R-D</li>
+ * <li>QIA</li>
+ * <li>GWC</li>
+ * <li>Score matrix is BLOSUM62</li>
+ * <li>Gaps treated same as X (unknown)</li>
+ * <li>product [0, 0] = F.F + K.K + L.L = 6 + 5 + 4 = 15</li>
+ * <li>product [1, 1] = R.R + -.- + D.D = 5 + -1 + 6 = 10</li>
+ * <li>product [2, 2] = Q.Q + I.I + A.A = 5 + 4 + 4 = 13</li>
+ * <li>product [3, 3] = G.G + W.W + C.C = 6 + 11 + 9 = 26</li>
+ * <li>product[0, 1] = F.R + K.- + L.D = -3 + -1 + -3 = -8
+ * <li>and so on</li>
+ * </ul>
+ */
+ @Override
+ public MatrixI findSimilarities(AlignmentView seqstrings,
+ SimilarityParamsI options)
+ {
+ char gapChar = scoreGapAsAny ? (seqstrings.isNa() ? 'N' : 'X')
+ : gapCharacter;
+ String[] seqs = seqstrings.getSequenceStrings(gapChar);
+ return findSimilarities(seqs, options);
+ }
+
+ /**
+ * Computes pairwise similarities of a set of sequences using the given
+ * parameters
+ *
+ * @param seqs
+ * @param params
+ * @return
+ */
+ protected MatrixI findSimilarities(String[] seqs, SimilarityParamsI params)
+ {
+ double[][] values = new double[seqs.length][];
+ for (int row = 0; row < seqs.length; row++)
+ {
+ values[row] = new double[seqs.length];
+ for (int col = 0; col < seqs.length; col++)
+ {
+ double total = computeSimilarity(seqs[row], seqs[col], params);
+ values[row][col] = total;
+ }
}
- for (char sym = 'A'; sym <= 'Z'; sym++)
+ return new Matrix(values);
+ }
+
+ /**
+ * Calculates the pairwise similarity of two strings using the given
+ * calculation parameters
+ *
+ * @param seq1
+ * @param seq2
+ * @param params
+ * @return
+ */
+ protected double computeSimilarity(String seq1, String seq2,
+ SimilarityParamsI params)
+ {
+ int len1 = seq1.length();
+ int len2 = seq2.length();
+ double total = 0;
+
+ int width = Math.max(len1, len2);
+ for (int i = 0; i < width; i++)
{
- if (symbols[sym] >= 0 && symbols[sym] < symMax)
+ if (i >= len1 || i >= len2)
{
- if (header)
+ /*
+ * off the end of one sequence; stop if we are only matching
+ * on the shorter sequence length, else treat as trailing gap
+ */
+ if (params.denominateByShortestLength())
{
- sb.append(html ? "<tr><td></td>" : "");
- for (char sym2 = 'A'; sym2 <= 'Z'; sym2++)
- {
- if (symbols[sym2] >= 0 && symbols[sym2] < symMax)
- {
- sb.append((html ? "<td> " : "\t") + sym2
- + (html ? " </td>" : ""));
- }
- }
- header = false;
- sb.append(html ? "</tr>\n" : "\n");
+ break;
}
- if (html)
+ }
+
+ 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);
+
+ if (gap1 && gap2)
+ {
+ /*
+ * gap-gap: include if options say so, else ignore
+ */
+ if (!params.includeGappedColumns())
{
- sb.append("<tr>");
+ continue;
}
- sb.append((html ? "<td>" : "") + sym + (html ? "</td>" : ""));
- for (char sym2 = 'A'; sym2 <= 'Z'; sym2++)
+ }
+ else if (gap1 || gap2)
+ {
+ /*
+ * gap-residue: score if options say so
+ */
+ if (!params.includeGaps())
{
- if (symbols[sym2] >= 0 && symbols[sym2] < symMax)
- {
- sb.append((html ? "<td>" : "\t")
- + matrix[symbols[sym]][symbols[sym2]]
- + (html ? "</td>" : ""));
- }
+ continue;
}
- sb.append(html ? "</tr>\n" : "\n");
}
+ float score = getPairwiseScore(c1, c2);
+ total += score;
}
- if (html)
+ return total;
+ }
+
+ /**
+ * Answers a hashcode computed from the symbol alphabet and the matrix score
+ * values
+ */
+ @Override
+ public int hashCode()
+ {
+ int hs = Arrays.hashCode(symbols);
+ for (float[] row : matrix)
{
- sb.append("</table>");
+ hs = hs * 31 + Arrays.hashCode(row);
}
- return sb.toString();
+ return hs;
+ }
+
+ /**
+ * Answers true if the argument is a ScoreMatrix with the same symbol alphabet
+ * and score values, else false
+ */
+ @Override
+ public boolean equals(Object obj)
+ {
+ if (!(obj instanceof ScoreMatrix))
+ {
+ return false;
+ }
+ ScoreMatrix sm = (ScoreMatrix) obj;
+ if (Arrays.equals(symbols, sm.symbols)
+ && Arrays.deepEquals(matrix, sm.matrix))
+ {
+ return true;
+ }
+ return false;
+ }
+
+ /**
+ * Returns the alphabet the matrix scores for, as a string of characters
+ *
+ * @return
+ */
+ public String getSymbols()
+ {
+ return new String(symbols);
+ }
+
+ public void setDescription(String desc)
+ {
+ description = desc;
}
}
git://source.jalview.org / jalview.git / blobdiff