moss.java

A program to find frequent molecular substructures and discriminative fragments in a database of mol

    int    rgmin, rgmax;        /* minimum and maximum ring size */

    if (this.running) {         /* if running, stop computations */
      if (this.stopped) return; /* if already stopped, abort */
      this.stat.setText("aborting search ...");
      this.stopped = true; this.miner.stop(); return;
    }                           /* stop the miner and abort */
    this.running = true;        /* set the status to running */
    this.stopped = false;       /* and not stopped */
    this.exec.setText("Abort"); /* change the button text */
    this.start = System.currentTimeMillis();
    this.miner = new Miner();   /* create a substructure miner */
    try {                       /* and try to initialize it */
      fmt = fmtnames[fmt_seed.getSelectedIndex()];
      this.miner.setSeed    (fmt, seed.getText());
      this.miner.setExcluded(fmt, excl.getText(), exseed.getText());
      try { spth = Float.parseFloat(thresh.getText()); }
      catch (Exception e) { spth = 0.5F; }
      thresh.setText("" +spth); /* get the threshold value */
      fmt = fmtnames[fmt_in.getSelectedIndex()];
      this.miner.setInput(fmt, fn_in.getText(), spth,
                          invert.isSelected());
      if (kekule.isSelected())  /* set the input data set and */
        this.miner.aromatize(); /* convert Kekul\'e representations */
      fmt = fmtnames[fmt_out.getSelectedIndex()];
      this.miner.setOutput(fmt, fn_sub.getText(), fn_ids.getText());
      mode = Miner.BONDEXT;     /* set default search mode */
      if (exttype.getSelectedIndex() > 0) mode |= Miner.RIGHTEXT;
      if (chains.isSelected())            mode |= Miner.CHAINEXT;
      if (canonic.isSelected())           mode |= Miner.PR_CANONIC;
      if (equiv.isSelected())             mode |= Miner.PR_EQUIV;
      if (normal.isSelected())            mode |= Miner.NORMFORM;
      if (verbose.isSelected())           mode |= Miner.VERBOSE;
      switch (perfect.getSelectedIndex()) {
        case 0: mode |= Miner.PR_PERFECT; break;
        case 1: mode |= Miner.PR_PARTIAL; break;
      }                         /* set pruning and other flags */
      try { supp = Float.parseFloat(minsupp.getText()); }
      catch (Exception e) { supp = 10.0F; }
      minsupp.setText(""+supp); /* get minimal support in focus */
      try { comp = Float.parseFloat(maxsupp.getText()); }
      catch (Exception e) { supp = 2.0F; }
      maxsupp.setText(""+comp); /* get maximal support in complement */
      if (supp >= 0) supp *= 0.01F;  /* change support values from */
      if (comp >= 0) comp *= 0.01F;  /* percentages to fractions */
      min = ((Integer)minsize.getValue()).intValue();
      max = ((Integer)maxsize.getValue()).intValue();
      if (!rings.isSelected())  /* if not to distinguish ring bonds, */
        rgmin = rgmax = 0;      /* clear the range of ring sizes */
      else {                    /* if to distinguish ring bonds */
        switch (ringext.getSelectedIndex()) {
          case 1: mode |= Miner.RINGEXT;                     break;
          case 2: mode |= Miner.CLOSERINGS|Miner.PR_UNCLOSE; break;
          case 3: mode |= Miner.RINGEXT   |Miner.MERGERINGS
                       |  Miner.CLOSERINGS|Miner.PR_UNCLOSE; break;
        }                       /* set ring extension flags */
        rgmin = ((Integer)minring.getValue()).intValue();
        rgmax = ((Integer)maxring.getValue()).intValue();
      }                         /* get ring sizes */
      this.miner.setParams(mode, supp, comp, min, max, rgmin, rgmax);
      matom = mrgat = Atom.TYPEMASK; /* set default masks */
      mbond = mrgbd = Bond.TYPEMASK|Bond.RINGBOND;
      switch (bdarom.getSelectedIndex()) {
        case 1: mbond &= Bond.DOWNGRADE;
                mrgbd &= Bond.DOWNGRADE; break;
        case 2: mbond &= Bond.UPGRADE;
                mrgbd &= Bond.UPGRADE;   break;
      }                         /* treatment of aromatic bonds */
      switch (bdtype.getSelectedIndex()) {
        case 1: mrgbd &= ~Bond.TYPEMASK; break;
        case 2: mbond &= ~Bond.TYPEMASK;
                mrgbd &= ~Bond.TYPEMASK; break;
      }                         /* where to ignore the bond type */
      switch (attype.getSelectedIndex()) {
        case 1: mrgat &= ~Atom.TYPEMASK; break;
        case 2: matom &= ~Atom.TYPEMASK;
                mrgat &= ~Atom.TYPEMASK; break;
      }                         /* where to ignore the atom type */
      if (charge.isSelected()) matom |= Atom.CHARGEMASK;
      if (atarom.isSelected()) matom |= Atom.AROMATIC;
      this.miner.setMasks(matom, mbond, mrgat, mrgbd);
      this.miner.setMaxEmbs(((Integer)maxemb.getValue()).intValue()); }
    catch (IOException ioe) {   /* check successful initialization */
      System.out.println("\nerror: " +ioe.getMessage());
      JOptionPane.showMessageDialog(this,
        "Initialization failed:\n" +ioe.getMessage(),
        "Error", JOptionPane.ERROR_MESSAGE);
      this.exec.setText("Run"); /* reset the button text */
      this.running = false;     /* if the initialization failed, */
      this.miner   = null;      /* show an error message, */
      return;                   /* set the status to not running, */
    }                           /* and "delete" the miner */
    this.thread = new Thread(this.miner);
    this.thread.start();        /* start computations as a thread */
    this.timer = new Timer(200, new ActionListener () {
      private int cnt = 0;      /* counter for status text update */
      public void actionPerformed (ActionEvent e) {
        if (!MoSS.this.thread.isAlive()) {
          MoSS.this.timer.stop(); MoSS.this.result();
          return;               /* stop the status update timer */
        }                       /* and show the computation result */
        if (--this.cnt <= 0) {  /* on every 5th check (once a second) */
          this.cnt = 5; MoSS.this.update(); }
      } } );                    /* update the status text */
    this.timer.start();         /* start the status update timer */
  }  /* execute() */

  /*------------------------------------------------------------------*/

  public void stateChanged (ChangeEvent e)
  {                             /* --- handle change of checkbox */
    boolean state = this.rings.isSelected();
    this.minring.setEnabled(state);
    this.maxring.setEnabled(state);
    this.ringext.setEnabled(state);
    for (int i = 3; --i >= 0; )
      this.labels[i].setEnabled(state);
  }  /* stateChanged() */

  /*------------------------------------------------------------------*/

  public void run ()
  {                             /* --- create toolbox dialog */
    GridBagLayout      g  = new GridBagLayout();
    GridBagConstraints lc = new GridBagConstraints();
    GridBagConstraints rc = new GridBagConstraints();
    JPanel             grid;
    JPanel             bbar;
    JLabel             lbl;
    JTextArea          help;
    JButton            btn;

    this.tab = new JTabbedPane(JTabbedPane.TOP,
                               JTabbedPane.SCROLL_TAB_LAYOUT);
    lc.fill      =              /* fill fields in both directions */
    rc.fill      = GridBagConstraints.HORIZONTAL;
    rc.weightx   = 1.0;         /* resize only the input fields, */
    lc.weightx   = 0.0;         /* but not the labels */
    lc.ipadx     = 10;          /* gap between labels and inputs */
    lc.ipady     = 10;          /* make all lines of the same height */
    rc.gridwidth = GridBagConstraints.REMAINDER;

    /* --- files and formats --- */
    grid = new JPanel(g);
    grid.setBorder(BorderFactory.createEmptyBorder(5, 5, 5, 5));
    tab.addTab("Formats and Files", null, grid, null);

    lbl = new JLabel("Molecule input format:");
    g.setConstraints(lbl, lc);         grid.add(lbl);
    fmt_in = new JComboBox(fmtnames);  fmt_in.setFont(font);
    g.setConstraints(fmt_in, rc);      grid.add(fmt_in);
 
    lbl = new JLabel("Substructure output format:       ");
    g.setConstraints(lbl, lc);         grid.add(lbl);
    fmt_out = new JComboBox(fmtnames); fmt_out.setFont(font);
    g.setConstraints(fmt_out, rc);     grid.add(fmt_out);
 
    lbl = new JLabel("Seed format:");
    g.setConstraints(lbl, lc);          grid.add(lbl);
    fmt_seed = new JComboBox(fmtnames); fmt_seed.setFont(font);
    g.setConstraints(fmt_seed, rc);     grid.add(fmt_seed);
 
    lbl   = new JLabel("Input data file:");
    g.setConstraints(lbl,   lc); grid.add(lbl);
    btn   = new JButton("Browse...");
    g.setConstraints(btn,   rc); grid.add(btn);
    fn_in = new JTextField("moss.dat");
    g.setConstraints(fn_in, rc); grid.add(fn_in);
    btn.addActionListener(new AbstractAction () {
      public void actionPerformed (ActionEvent e) {
        if (chooser == null) chooser = createChooser();
        chooser.setDialogType(JFileChooser.OPEN_DIALOG);
        chooser.setDialogTitle("Choose Input Data File");
        int r = chooser.showDialog(MoSS.this, null);
        if (r == JFileChooser.APPROVE_OPTION)
          fn_in.setText(chooser.getSelectedFile().getPath());
      } } );

    lbl    = new JLabel("Substructure output file:");
    g.setConstraints(lbl,   lc); grid.add(lbl);
    btn    = new JButton("Browse...");
    g.setConstraints(btn,   rc); grid.add(btn);
    fn_sub = new JTextField("moss.sub");
    g.setConstraints(fn_sub, rc); grid.add(fn_sub);
    btn.addActionListener(new AbstractAction () {
      public void actionPerformed (ActionEvent e) {
        if (chooser == null) chooser = createChooser();
        chooser.setDialogType(JFileChooser.OPEN_DIALOG);
        chooser.setDialogTitle("Choose Substructure Output File");
        int r = chooser.showDialog(MoSS.this, null);
        if (r == JFileChooser.APPROVE_OPTION)
          fn_sub.setText(chooser.getSelectedFile().getPath());
      } } );

    lbl    = new JLabel("Molecule identifier output file:");
    g.setConstraints(lbl,   lc); grid.add(lbl);
    btn    = new JButton("Browse...");
    g.setConstraints(btn,   rc); grid.add(btn);
    fn_ids = new JTextField("");
    g.setConstraints(fn_ids, rc); grid.add(fn_ids);
    btn.addActionListener(new AbstractAction () {
      public void actionPerformed (ActionEvent e) {
        if (chooser == null) chooser = createChooser();
        chooser.setDialogType(JFileChooser.OPEN_DIALOG);
        chooser.setDialogTitle("Choose Molecule Identifier File");
        int r = chooser.showDialog(MoSS.this, null);
        if (r == JFileChooser.APPROVE_OPTION)
          fn_ids.setText(chooser.getSelectedFile().getPath());
      } } );

    lbl = new JLabel("Seed description:");
    g.setConstraints(lbl, rc);          grid.add(lbl);
    seed = new JTextField("*");         seed.setFont(font);
    g.setConstraints(seed, rc);         grid.add(seed);

    help = new JTextArea(
       "The seed is a substructure from which to start the search.\n"
      +"A star (*) represents an empty seed (no restriction).");
    help.setFont(small);
    help.setBackground(this.getBackground());
    g.setConstraints(help, rc); grid.add(help);

    help = new JTextArea((String)null);
    help.setFont(small);
    help.setBackground(this.getBackground());
    help.setPreferredSize(new Dimension(0,0));
    rc.weighty = 1.0;
    g.setConstraints(help, rc); grid.add(help);
    rc.weighty = 0.0;

    /* --- basic parameters --- */
    grid = new JPanel(g);
    grid.setBorder(BorderFactory.createEmptyBorder(5, 5, 5, 5));
    tab.addTab("Basic Parameters", null, grid, null);

    lbl = new JLabel("Threshold for split:");
    g.setConstraints(lbl, lc);          grid.add(lbl);
    thresh = new JTextField("0.5");     thresh.setFont(font);
    g.setConstraints(thresh, rc);       grid.add(thresh);

    lbl = new JLabel("Invert split:");
    g.setConstraints(lbl, lc);          grid.add(lbl);
    invert = new JCheckBox("", false);
    g.setConstraints(invert, rc);       grid.add(invert);

    help = new JTextArea(
       "The molecules in the input file are split "
      +"into a focus set and\nits complement. "
      +"By default molecules with a value less than\n"
      +"or equal to the threshold are in the focus, "
      +"all other molecules\nare in the complement. "
      +"If the split is inverted, focus set and\n"
      +"complement are exchanged.");
    help.setFont(small);
    help.setBackground(this.getBackground());
    g.setConstraints(help, rc); grid.add(help);

    lbl = new JLabel("Minimal support in focus:");
    g.setConstraints(lbl, lc);          grid.add(lbl);
    minsupp = new JTextField("10.0");   minsupp.setFont(font);
    g.setConstraints(minsupp, rc);      grid.add(minsupp);

    lbl = new JLabel("Maximal support in complement:");
    g.setConstraints(lbl, lc);          grid.add(lbl);
    maxsupp = new JTextField("2.0");    maxsupp.setFont(font);
    g.setConstraints(maxsupp, rc);      grid.add(maxsupp);

    help = new JTextArea(
       "The support of a substructure is measured "
      +"in percent of the\nsize (number of molecules) "
      +"of the corresponding subset.");
    help.setFont(small);
    help.setBackground(this.getBackground());
    g.setConstraints(help, rc); grid.add(help);

    lbl = new JLabel("Minimal substructure size:");
    g.setConstraints(lbl, lc);          grid.add(lbl);
    minsize = new JSpinner(new SpinnerNumberModel(1, 1, 999999, 1));
    g.setConstraints(minsize, rc);      grid.add(minsize);

    lbl = new JLabel("Maximal substructure size:");
    g.setConstraints(lbl, lc);          grid.add(lbl);
    maxsize = new JSpinner(new SpinnerNumberModel(0, 0, 999999, 1));
    g.setConstraints(maxsize, rc);      grid.add(maxsize);

    help = new JTextArea(
       "The size of a substructure is measured "
      +"as the number of atoms.\n"
      +"A maximal size of zero means "
      +"that there is no size limit.");
    help.setFont(small);
    help.setBackground(this.getBackground());
    g.setConstraints(help, rc); grid.add(help);

    help = new JTextArea((String)null);
    help.setFont(small);
    help.setBackground(this.getBackground());
    help.setPreferredSize(new Dimension(0,0));
    rc.weighty = 1.0;
    g.setConstraints(help, rc); grid.add(help);
    rc.weighty = 0.0;

    /* --- matching --- */
    grid = new JPanel(g);
    grid.setBorder(BorderFactory.createEmptyBorder(5, 5, 5, 5));
    tab.addTab("Matching", null, grid, null);

    lbl = new JLabel("Aromatic bonds:");
    g.setConstraints(lbl, lc);         grid.add(lbl);
    bdarom = new JComboBox(aromnames); bdarom.setFont(font);
    g.setConstraints(bdarom, rc);      grid.add(bdarom);
 
    help = new JTextArea(
       "Downgrading aromatic bonds means treating them as single,\n"
      +"upgrading means treating them as double bonds.");
    help.setFont(small);
    help.setBackground(this.getBackground());
    g.setConstraints(help, rc); grid.add(help);

    lbl = new JLabel("Ignore type of bonds:");
    g.setConstraints(lbl, lc);         grid.add(lbl);
    bdtype = new JComboBox(typenames);