*/
package jalview.analysis.scoremodels;
-import jalview.api.analysis.ScoreModelI;
+import jalview.math.Matrix;
+import jalview.math.MatrixI;
-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
+public class ScoreMatrix implements PairwiseScoreModelI
{
public static final short UNMAPPED = (short) -1;
- private static final String DELIMITERS = " ,\t";
-
- private static final String COMMENT_CHAR = "#";
-
private static final String BAD_ASCII_ERROR = "Unexpected character %s in getPairwiseScore";
private static final int MAX_ASCII = 127;
private boolean peptide;
/**
- * Constructor
+ * 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
* Unique, human readable name for the matrix
*/
public ScoreMatrix(String name, char[] alphabet, float[][] matrix)
{
+ if (alphabet.length != matrix.length)
+ {
+ throw new IllegalArgumentException(
+ "score matrix size must match alphabet size");
+ }
+ for (float[] row : matrix)
+ {
+ if (row.length != alphabet.length)
+ {
+ throw new IllegalArgumentException(
+ "score matrix size must be square");
+ }
+ }
+
this.matrix = matrix;
this.name = name;
this.symbols = alphabet;
return peptide;
}
- @Override
+ /**
+ * Returns 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()
{
return matrix;
}
/**
+ * 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)
+ {
+ if (c < symbolIndex.length)
+ {
+ return symbolIndex[c];
+ }
+ else
+ {
+ return UNMAPPED;
+ }
+ }
+
+ /**
* Returns the pairwise score for substituting c with d, or zero if c or d is
* an unscored or unexpected character
*/
@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;
}
/**
- * 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.
- * Expected format:
+ * Answers the number of symbols coded for (also equal to the number of rows
+ * and columns of the score matrix)
*
- * <pre>
- * ScoreMatrix displayName
- * # comment lines begin with hash sign
- * # symbol alphabet should be the next non-comment line
- * ARNDCQEGHILKMFPSTWYVBZX *
- * # scores matrix, with space, comma or tab delimited values
- * # [i, j] = score for substituting symbol[i] with symbol[j]
- * # first column in each row is optionally the 'substituted' symbol
- * A 4 -1 -2 -2 0 -1 -1 0 -2 -1 -1 -1 -1 -2 -1 1 0 -3 -2 0 -2 -1 0 -4 -4
- * ..etc..
- * </pre>
- *
- * @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;
+ return symbols.length;
+ }
- try
+ /**
+ * 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>
+ */
+ public MatrixI computePairwiseScores(String[] seqs)
+ {
+ double[][] values = new double[seqs.length][];
+ for (int row = 0; row < seqs.length; row++)
{
- String data;
-
- while ((data = br.readLine()) != null)
+ values[row] = new double[seqs.length];
+ for (int col = 0; col < seqs.length; col++)
{
- lineNo++;
- data = data.trim();
- if (data.startsWith(COMMENT_CHAR))
- {
- continue;
- }
- if (data.toLowerCase().startsWith("scorematrix"))
- {
- /*
- * Parse name from ScoreMatrix <name>
- */
- if (name != null)
- {
- System.err
- .println("Warning: 'ScoreMatrix' repeated in file at line "
- + lineNo);
- }
- StringTokenizer nameLine = new StringTokenizer(data, DELIMITERS);
- if (nameLine.countTokens() != 2)
- {
- System.err
- .println("Format error: expected 'ScoreMatrix <name>', found '"
- + data + "' at line " + lineNo);
- return null;
- }
- nameLine.nextToken();
- name = nameLine.nextToken();
- continue;
- }
- else if (name == null)
- {
- System.err
- .println("Format error: 'ScoreMatrix <name>' should be the first non-comment line");
- return null;
- }
-
- /*
- * next line after ScoreMatrix should be the alphabet of scored symbols
- */
- if (alphabet == null)
- {
- alphabet = data;
- size = alphabet.length();
- scores = new float[size][];
- continue;
- }
-
- /*
- * too much information?
- */
- if (row >= size && data.length() > 0) {
- System.err
- .println("Unexpected extra input line in score model file "
- + data);
- return null;
- }
-
- /*
- * subsequent lines should be the symbol scores
- * optionally with the symbol as the first column for readability
- */
- StringTokenizer scoreLine = new StringTokenizer(data, DELIMITERS);
- if (scoreLine.countTokens() == size + 1)
+ int total = 0;
+ int width = Math.min(seqs[row].length(), seqs[col].length());
+ for (int i = 0; i < width; i++)
{
- /*
- * check 'guide' symbol is the row'th letter of the alphabet
- */
- String symbol = scoreLine.nextToken();
- if (symbol.length() > 1
- || symbol.charAt(0) != alphabet.charAt(row))
- {
- System.err
- .println(String
- .format("Error parsing score matrix at line %d, expected %s but found %s",
- lineNo, alphabet.charAt(row), symbol));
- return null;
- }
+ char c1 = seqs[row].charAt(i);
+ char c2 = seqs[col].charAt(i);
+ float score = getPairwiseScore(c1, c2);
+ total += score;
}
- if (scoreLine.countTokens() != size)
- {
- System.err.println(String.format(
- "Expected %d scores at line %d but found %d", size,
- lineNo, scoreLine.countTokens()));
- return null;
- }
- scores[row] = new float[size];
- int col = 0;
- String value = null;
- while (scoreLine.hasMoreTokens())
- {
- try
- {
- value = scoreLine.nextToken();
- scores[row][col] = Float.valueOf(value);
- col++;
- } catch (NumberFormatException e)
- {
- System.err.println(String.format(
- "Invalid score value %s at line %d column %d", value,
- lineNo, col));
- return null;
- }
- }
- row++;
+ values[row][col] = total;
}
- } catch (IOException e)
- {
- System.err.println("Error reading score matrix file: "
- + e.getMessage() + " at line " + lineNo);
}
-
- /*
- * out of data - check we found enough
- */
- if (row < size)
- {
- System.err
- .println(String
- .format("Expected %d rows of score data in score matrix but only found %d",
- size, row));
- return null;
- }
-
- /*
- * If we get here, then name, alphabet and scores have been parsed successfully
- */
- sm = new ScoreMatrix(name, alphabet.toCharArray(), scores);
- return sm;
+ return new Matrix(values);
}
}