Aritalab:Lecture/Programming/Java/Alignment

From Metabolomics.JP
Jump to: navigation, search

配列アライメントアルゴリズムの全体を以下に示します。このページにあるコード部分を全てつなげて Alignment.java としてコンパイル、実行してください。このファイルだけで動きます。


Contents

クラスおよびペナルティの定義

public class Alignment {
  private int LOCAL = 0, GLOBAL = 1;
  private int AlignType = -1;
  private int MatchScore = 1;
  private int GapPenalty = -2;
  private int MismatchPenalty = -1;
  private int[][] scoreTable;
  private String seqX, seqY;

  public Alignment(String x, String y) {
    seqX = x;
    seqY = y;
    scoreTable = new int[x.length() + 1][y.length() + 1];
  }

  private void initializeTable() {
    scoreTable[0][0] = 0;
    for (int i = 1; i < scoreTable.length; i++)
      scoreTable[i][0] = scoreTable[i - 1][0] + AlignType * GapPenalty;
    for (int i = 1; i < scoreTable[0].length; i++)
      scoreTable[0][i] = scoreTable[0][i - 1] + AlignType * GapPenalty;
  }

  private int getMax(int x, int y, int z, int w) {
    int t = x > y ? x : y;
    int s = z > w ? z : w;
    return (t > s ? t : s);
  }

  private int getMax(int x, int y, int z) {
    int t = x > y ? x : y;
    return (t > z ? t : z);
  }

トレースバック関数

重要なのは traceBack関数の do { ... } while 文の中身だけで、後はトレースを印刷するための準備です。

  private String[] traceBack() {
    int x = 0, y = 0;
    if (AlignType == LOCAL) {
      // Search for the maximum score brute-force (very inefficient).
      // The max score can be recorded when computing scoreTable.
      int max = 0;
      for (int i = 0; i < scoreTable.length; i)
          if (scoreTable[i][j] >= max) {
            x = i;
            y = j;
            max = scoreTable[i][j];
          }
      }
    } else if (AlignType == GLOBAL) {
      x = scoreTable.length - 1;
      y = scoreTable[0].length - 1;
    }
    StringBuffer line1 = new StringBuffer();
    StringBuffer line2 = new StringBuffer();
    line1.append(':');
    line1.append(x);
    line2.append(':');
    line2.append(y);
    do {
      if (scoreTable[x - 1][y] + GapPenalty == scoreTable[x][y]) {
        line1.insert(0, seqX.charAt(x-- - 1));
        line2.insert(0, '-');
      } else if (scoreTable[x][y - 1] + GapPenalty == scoreTable[x][y]) {
        line1.insert(0, '-');
        line2.insert(0, seqY.charAt(y-- - 1));
      } else {
        line1.insert(0, seqX.charAt(x-- - 1));
        line2.insert(0, seqY.charAt(y-- - 1));
      }
    } while (x != 0 && y != 0
        && (AlignType == GLOBAL ? true : scoreTable[x][y] != 0));
    line1.insert(0, ':');
    line1.insert(0, String.format("%1$3d", x + 1));
    line2.insert(0, ':');
    line2.insert(0, String.format("%1$3d", y + 1));
    return new String[] { line1.toString(), line2.toString() };
  }

  public void printTrace() {
    String[] trace = traceBack();
    String x = trace[0];
    String y = trace[1];
    StringBuffer sb = new StringBuffer();
    sb.append("    ");
    for (int i = 4; x.charAt(i) != ':'; ii * width < x.length()) {
      System.out.println(x.substring(i * width,
          Math.min(x.length(), (i + 1) * width)));
      System.out.println(z.substring(i * width,
          Math.min(z.length(), (i + 1) * width)));
      System.out.println(y.substring(i * width,
          Math.min(y.length(), (i + 1) * width)));
    }
  }

アルゴリズム

getMax関数で最大値を取る漸化式を for ループで二重に回しているだけです。大域アライメントと局所アライメントの違いは、漸化式の候補に 0 を加えるだけです。

  public void NeedlemanWunsch() {
    AlignType = GLOBAL;
    initializeTable();
    for (int i = 1; i < scoreTable.length; i)
        scoreTable[i][j] = getMax(
            scoreTable[i - 1][j] + GapPenalty,
            scoreTable[i][j - 1] + GapPenalty,
            scoreTable[i - 1][j - 1]
                + (seqX.charAt(i - 1) == seqY.charAt(j - 1) ? MatchScore
                    : MismatchPenalty));
  }

  public void SmithWaterman() {
    AlignType = LOCAL;
    initializeTable();
    for (int i = 1; i < scoreTable.length; i)
        scoreTable[i][j] = getMax(
            0,
            scoreTable[i - 1][j] + GapPenalty,
            scoreTable[i][j - 1] + GapPenalty,
            scoreTable[i - 1][j - 1]
                + (seqX.charAt(i - 1) == seqY.charAt(j - 1) ? MatchScore
                    : MismatchPenalty));
  }

メイン関数

  public void printTable() {
    if (scoreTable == null)
      return;
    StringBuffer sb = new StringBuffer();
    for (int i = 0; i < scoreTable.length; i)
        sb.append(String.format("%1$4d", scoreTable[i][j]));
      sb.append("\n");
    }
    System.out.println(sb.toString());
  }

  public static void main(String[] args) {
    Alignment A = new Alignment("aattgaagg", "gctagg");
    A.NeedlemanWunsch();
    //A.SmithWaterman();
    A.printTable();
    A.printTrace();
  }
}

実行結果

> java Alignment

   0  -2  -4  -6  -8 -10 -12
  -2  -1  -3  -5  -5  -7  -9
  -4  -3  -2  -4  -4  -6  -8
  -6  -5  -4  -1  -3  -5  -7
  -8  -7  -6  -3  -2  -4  -6
 -10  -7  -8  -5  -4  -1  -3
 -12  -9  -8  -7  -4  -3  -2
 -14 -11 -10  -9  -6  -5  -4
 -16 -13 -12 -11  -8  -5  -4
 -18 -15 -14 -13 -10  -7  -4

  1:aattgaagg:9
      |  | ||    
  1:gct--a-gg:6
Personal tools
Namespaces

Variants
Actions
Navigation
metabolites
Toolbox