src/jalview/gui/SequenceFetcher.java

   1 /*
   2  * Jalview - A Sequence Alignment Editor and Viewer
   3  * Copyright (C) 2006 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public License
   7  * as published by the Free Software Foundation; either version 2
   8  * of the License, or (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA
  18  */
  19 package jalview.gui;
  20
  21 import javax.swing.*;
  22 import java.awt.*;
  23 import java.awt.event.*;
  24 import jalview.io.EBIFetchClient;
  25 import MCview.*;
  26 import jalview.datamodel.*;
  27 import jalview.analysis.AlignSeq;
  28 import java.io.File;
  29 import jalview.io.*;
  30 import java.util.*;
  31
  32 public class SequenceFetcher
  33     extends JPanel implements Runnable
  34 {
  35   JInternalFrame frame;
  36   AlignFrame alignFrame;
  37   StringBuffer result;
  38   final String noDbSelected = "-- Select Database --";
  39   public SequenceFetcher(AlignFrame af)
  40   {
  41     alignFrame = af;
  42     database.addItem(noDbSelected);
  43     database.addItem("Uniprot");
  44     database.addItem("EMBL");
  45     database.addItem("EMBLCDS");
  46     database.addItem("PDB");
  47     database.addItem("PFAM");
  48
  49     try
  50     {
  51       jbInit();
  52     }
  53     catch (Exception ex)
  54     {
  55       ex.printStackTrace();
  56     }
  57
  58     frame = new JInternalFrame();
  59     frame.setContentPane(this);
  60     if (System.getProperty("os.name").startsWith("Mac"))
  61       Desktop.addInternalFrame(frame, getFrameTitle(), 400, 140);
  62     else
  63       Desktop.addInternalFrame(frame, getFrameTitle(), 400, 125);
  64   }
  65
  66   private String getFrameTitle()
  67   {
  68     return ( (alignFrame == null) ? "New " : "Additional ") + "Sequence Fetcher";
  69   }
  70
  71   private void jbInit()
  72       throws Exception
  73   {
  74     this.setLayout(gridBagLayout1);
  75
  76     database.setFont(new java.awt.Font("Verdana", Font.PLAIN, 11));
  77     database.setMinimumSize(new Dimension(160, 21));
  78     database.setPreferredSize(new Dimension(160, 21));
  79     jLabel1.setFont(new java.awt.Font("Verdana", Font.ITALIC, 11));
  80     jLabel1.setText(
  81         "Separate multiple accession ids with semi colon \";\"");
  82     ok.setText("OK");
  83     ok.addActionListener(new ActionListener()
  84     {
  85       public void actionPerformed(ActionEvent e)
  86       {
  87         ok_actionPerformed(e);
  88       }
  89     });
  90     close.setText("Close");
  91     close.addActionListener(new ActionListener()
  92     {
  93       public void actionPerformed(ActionEvent e)
  94       {
  95         close_actionPerformed(e);
  96       }
  97     });
  98     textfield.setFont(new java.awt.Font("Verdana", Font.PLAIN, 11));
  99     textfield.addActionListener(new ActionListener()
 100     {
 101       public void actionPerformed(ActionEvent e)
 102       {
 103         ok_actionPerformed(e);
 104       }
 105     });
 106     jPanel1.add(ok);
 107     jPanel1.add(close);
 108     this.add(jLabel1, new GridBagConstraints(0, 0, 2, 1, 0.0, 0.0
 109                                              , GridBagConstraints.WEST,
 110                                              GridBagConstraints.NONE,
 111                                              new Insets(7, 4, 0, 6), 77, 6));
 112     this.add(jPanel1, new GridBagConstraints(0, 2, 2, 1, 1.0, 1.0
 113                                              , GridBagConstraints.WEST,
 114                                              GridBagConstraints.BOTH,
 115                                              new Insets(7, -2, 7, 12), 241, -2));
 116     this.add(database, new GridBagConstraints(0, 1, 1, 1, 1.0, 0.0
 117                                               , GridBagConstraints.WEST,
 118                                               GridBagConstraints.NONE,
 119                                               new Insets(0, 4, 0, 0), 1, 0));
 120     this.add(textfield, new GridBagConstraints(1, 1, 1, 1, 1.0, 0.0
 121                                                , GridBagConstraints.CENTER,
 122                                                GridBagConstraints.NONE,
 123                                                new Insets(0, 0, 0, 6), 211, 1));
 124   }
 125
 126   JComboBox database = new JComboBox();
 127   JLabel jLabel1 = new JLabel();
 128   JButton ok = new JButton();
 129   JButton close = new JButton();
 130   JPanel jPanel1 = new JPanel();
 131   JTextField textfield = new JTextField();
 132   GridBagLayout gridBagLayout1 = new GridBagLayout();
 133   public void close_actionPerformed(ActionEvent e)
 134   {
 135     try
 136     {
 137       frame.setClosed(true);
 138     }
 139     catch (Exception ex)
 140     {}
 141   }
 142
 143   public void ok_actionPerformed(ActionEvent e)
 144   {
 145     database.setEnabled(false);
 146     textfield.setEnabled(false);
 147     ok.setEnabled(false);
 148     close.setEnabled(false);
 149
 150     Thread worker = new Thread(this);
 151     worker.start();
 152   }
 153
 154   private void resetDialog()
 155   {
 156     database.setEnabled(true);
 157     textfield.setEnabled(true);
 158     ok.setEnabled(true);
 159     close.setEnabled(true);
 160   }
 161
 162   public void run()
 163   {
 164     String error = "";
 165     if (database.getSelectedItem().equals(noDbSelected))
 166       error += "Please select the source database\n";
 167     com.stevesoft.pat.Regex empty = new com.stevesoft.pat.Regex("\\s+", "");
 168     textfield.setText(empty.replaceAll(textfield.getText()));
 169     if (textfield.getText().length() == 0)
 170       error += "Please enter a (semi-colon separated list of) database id(s)";
 171     if (error.length() > 0)
 172     {
 173       showErrorMessage(error);
 174       resetDialog();
 175       return;
 176     }
 177
 178     result = new StringBuffer();
 179     if (database.getSelectedItem().equals("Uniprot"))
 180     {
 181       getUniprotFile(textfield.getText());
 182     }
 183     else if (database.getSelectedItem().equals("EMBL")
 184              || database.getSelectedItem().equals("EMBLCDS"))
 185     {
 186       StringTokenizer st = new StringTokenizer(textfield.getText(), ";");
 187       while(st.hasMoreTokens())
 188       {
 189         EBIFetchClient dbFetch = new EBIFetchClient();
 190
 191         String[] reply = dbFetch.fetchData(
 192             database.getSelectedItem().toString().toLowerCase(
 193             ) + ":" + st.nextToken(),
 194             "fasta", "raw");
 195 //
 196         if (reply != null)
 197         {
 198           for (int i = 0; i < reply.length; i++)
 199             result.append(reply[i] + "\n");
 200         }
 201       }
 202
 203       if(result!=null && result.length()>1) // arbitrary minimum length for a seuqence file
 204       {
 205         System.out.println(result.toString());
 206
 207         parseResult(result.toString(), null);
 208       }
 209     }
 210     else if (database.getSelectedItem().equals("PDB"))
 211     {
 212       StringTokenizer qset = new StringTokenizer(textfield.getText(), ";");
 213       String query;
 214       SequenceI[] seqs = null;
 215       while (qset.hasMoreTokens() && ((query = qset.nextToken())!=null))
 216       {
 217         SequenceI[] seqparts = getPDBFile(query.toUpperCase());
 218           if (seqparts!=null) {
 219               if (seqs==null) {
 220                   seqs = seqparts;
 221               } else {
 222                   SequenceI[] newseqs = new SequenceI[seqs.length+seqparts.length];
 223                   int i=0;
 224                   for (;i<seqs.length; i++) {
 225                       newseqs[i] = seqs[i];
 226                       seqs[i] = null;
 227                   }
 228                   for (int j=0;j<seqparts.length; i++, j++)
 229                       newseqs[i] = seqparts[j];
 230                   seqs=newseqs;
 231               }
 232               result.append("# Success for "+query.toUpperCase()+"\n");
 233           }
 234       }
 235       if (seqs!=null && seqs.length>0) {
 236           if (parseResult(new Alignment(seqs), null, null)!=null)
 237               result.append("# Successfully parsed the PDB File Queries into an Alignment");
 238       }
 239     }
 240     else if( database.getSelectedItem().equals("PFAM"))
 241     {
 242       try{
 243         result.append(new FastaFile(
 244            "http://www.sanger.ac.uk/cgi-bin/Pfam/getalignment.pl?format=fal&acc="
 245            +  textfield.getText().toUpperCase(), "URL").print()
 246            );
 247
 248          if(result.length()>0)
 249            parseResult( result.toString(), textfield.getText().toUpperCase() );
 250
 251       }catch(java.io.IOException ex)
 252       {   result = null;    }
 253     }
 254
 255     if (result == null || result.length() == 0)
 256       showErrorMessage("Error retrieving " + textfield.getText()
 257                        + " from " + database.getSelectedItem());
 258
 259     resetDialog();
 260     return;
 261   }
 262
 263   void getUniprotFile(String id)
 264   {
 265     EBIFetchClient ebi = new EBIFetchClient();
 266     File file = ebi.fetchDataAsFile("uniprot:" + id, "xml", null);
 267
 268     DBRefFetcher dbref = new DBRefFetcher();
 269     Vector entries = dbref.getUniprotEntries(file);
 270
 271     if (entries != null)
 272     {
 273       //First, make the new sequences
 274       Enumeration en = entries.elements();
 275       while (en.hasMoreElements())
 276       {
 277         UniprotEntry entry = (UniprotEntry) en.nextElement();
 278
 279         StringBuffer name = new StringBuffer(">UniProt/Swiss-Prot");
 280         Enumeration en2 = entry.getAccession().elements();
 281         while (en2.hasMoreElements())
 282         {
 283           name.append("|");
 284           name.append(en2.nextElement());
 285         }
 286         en2 = entry.getName().elements();
 287         while (en2.hasMoreElements())
 288         {
 289           name.append("|");
 290           name.append(en2.nextElement());
 291         }
 292
 293         if (entry.getProtein() != null)
 294         {
 295            name.append(" " + entry.getProtein().getName().elementAt(0));
 296         }
 297
 298         result.append(name + "\n" + entry.getUniprotSequence().getContent() +
 299                       "\n");
 300
 301       }
 302
 303       //Then read in the features and apply them to the dataset
 304       Alignment al = parseResult(result.toString(), null);
 305       for (int i = 0; i < entries.size(); i++)
 306       {
 307         UniprotEntry entry = (UniprotEntry) entries.elementAt(i);
 308         Enumeration e = entry.getDbReference().elements();
 309         Vector onlyPdbEntries = new Vector();
 310         while (e.hasMoreElements())
 311         {
 312           PDBEntry pdb = (PDBEntry) e.nextElement();
 313           if (!pdb.getType().equals("PDB"))
 314             continue;
 315
 316           onlyPdbEntries.addElement(pdb);
 317         }
 318
 319         Enumeration en2 = entry.getAccession().elements();
 320         while (en2.hasMoreElements())
 321         {
 322           al.getSequenceAt(i).getDatasetSequence().addDBRef(new DBRefEntry(DBRefSource.UNIPROT,
 323                     "0",
 324                     en2.nextElement().toString()));
 325         }
 326
 327
 328
 329
 330         al.getSequenceAt(i).getDatasetSequence().setPDBId(onlyPdbEntries);
 331         if (entry.getFeature() != null)
 332         {
 333           e = entry.getFeature().elements();
 334           while (e.hasMoreElements())
 335           {
 336             SequenceFeature sf = (SequenceFeature) e.nextElement();
 337             sf.setFeatureGroup("Uniprot");
 338             al.getSequenceAt(i).getDatasetSequence().addSequenceFeature( sf );
 339           }
 340         }
 341       }
 342     }
 343   }
 344
 345   SequenceI[] getPDBFile(String id)
 346   {
 347     Vector result = new Vector();
 348     String chain = null;
 349     if (id.indexOf(":") > -1)
 350     {
 351       chain = id.substring(id.indexOf(":") + 1);
 352       id = id.substring(0, id.indexOf(":"));
 353     }
 354
 355     EBIFetchClient ebi = new EBIFetchClient();
 356     String file = ebi.fetchDataAsFile("pdb:" + id, "pdb", "raw").getAbsolutePath();
 357     if (file == null)
 358       return null;
 359     try
 360     {
 361       PDBfile pdbfile = new PDBfile(file, jalview.io.AppletFormatAdapter.FILE);
 362       for (int i = 0; i < pdbfile.chains.size(); i++)
 363       {
 364         if (chain == null ||
 365             ( (PDBChain) pdbfile.chains.elementAt(i)).id.
 366             toUpperCase().equals(chain)) {
 367             PDBChain pdbchain = (PDBChain) pdbfile.chains.elementAt(i);
 368             // Get the Chain's Sequence - who's dataset includes any special features added from the PDB file
 369             SequenceI sq = pdbchain.sequence;
 370             // Specially formatted name for the PDB chain sequences retrieved from the PDB
 371             sq.setName("PDB|"+id+"|"+sq.getName());
 372             // Might need to add more metadata to the PDBEntry object
 373             // like below
 374             /*
 375              * PDBEntry entry = new PDBEntry();
 376             // Construct the PDBEntry
 377             entry.setId(id);
 378             if (entry.getProperty() == null)
 379                 entry.setProperty(new Hashtable());
 380             entry.getProperty().put("chains",
 381                         pdbchain.id
 382                         + "=" + sq.getStart()
 383                         + "-" + sq.getEnd());
 384             sq.getDatasetSequence().addPDBId(entry);
 385             */
 386             // Add PDB DB Refs
 387             // We make a DBRefEtntry because we have obtained the PDB file from a verifiable source
 388             // JBPNote - PDB DBRefEntry should also carry the chain and mapping information
 389             DBRefEntry dbentry = new DBRefEntry(jalview.datamodel.DBRefSource.PDB,"0",id+pdbchain.id);
 390             sq.addDBRef(dbentry);
 391             // and add seuqence to the retrieved set
 392             result.addElement(sq.deriveSequence());
 393         }
 394       }
 395     }
 396     catch (Exception ex) // Problem parsing PDB file
 397     {
 398       jalview.bin.Cache.log.warn("Exception when retrieving " +
 399                                  textfield.getText() + " from " +
 400                                  database.getSelectedItem(), ex);
 401       return null;
 402     }
 403     SequenceI[] results = new SequenceI[result.size()];
 404     for (int i=0,j=result.size();i<j; i++) {
 405         results[i] = (SequenceI) result.elementAt(i);
 406         result.setElementAt(null,i);
 407     }
 408     return results;
 409   }
 410   Alignment parseResult(String result, String title)
 411   {
 412       String format = new IdentifyFile().Identify(result, "Paste");
 413       Alignment sequences = null;
 414       if (FormatAdapter.isValidFormat(format))
 415       {
 416         sequences = null;
 417         try{ sequences = new FormatAdapter().readFile(result.toString(), "Paste",
 418                                                  format);}
 419         catch(Exception ex){}
 420
 421         if (sequences!=null)
 422             return parseResult(sequences, title, format);
 423       } else {
 424               showErrorMessage("Error retrieving " + textfield.getText()
 425                       + " from " + database.getSelectedItem());
 426       }
 427
 428       return null;
 429   }
 430   Alignment parseResult(Alignment al, String title, String currentFileFormat) {
 431
 432       if (al != null && al.getHeight() > 0)
 433       {
 434         if (alignFrame == null)
 435         {
 436           AlignFrame af = new AlignFrame(al,
 437                                            AlignFrame.DEFAULT_WIDTH,
 438                                            AlignFrame.DEFAULT_HEIGHT);
 439           if (currentFileFormat!=null)
 440               af.currentFileFormat = currentFileFormat; // WHAT IS THE DEFAULT FORMAT FOR NON-FormatAdapter Sourced Alignments?
 441
 442           if(title==null)
 443             title = "Retrieved from " + database.getSelectedItem();
 444
 445           Desktop.addInternalFrame(af,
 446                                    title,
 447                                    AlignFrame.DEFAULT_WIDTH,
 448                                    AlignFrame.DEFAULT_HEIGHT);
 449
 450           af.statusBar.setText("Successfully pasted alignment file");
 451
 452           try
 453           {
 454             af.setMaximum(jalview.bin.Cache.getDefault("SHOW_FULLSCREEN", false));
 455           }
 456           catch (Exception ex)
 457           {}
 458         }
 459         else
 460         {
 461           for (int i = 0; i < al.getHeight(); i++)
 462           {
 463             alignFrame.viewport.alignment.addSequence(al.getSequenceAt(i)); // this also creates dataset sequence entries
 464           }
 465           alignFrame.viewport.setEndSeq(alignFrame.viewport.alignment.
 466                                         getHeight());
 467           alignFrame.viewport.alignment.getWidth();
 468           alignFrame.viewport.firePropertyChange("alignment", null,
 469                                                  alignFrame.viewport.
 470                                                  getAlignment().getSequences());
 471         }
 472       }
 473     return al;
 474   }
 475
 476   void showErrorMessage(final String error)
 477   {
 478     resetDialog();
 479     javax.swing.SwingUtilities.invokeLater(new Runnable()
 480     {
 481       public void run()
 482       {
 483         JOptionPane.showInternalMessageDialog(Desktop.desktop,
 484                                               error, "Error Retrieving Data",
 485                                           JOptionPane.WARNING_MESSAGE);
 486       }
 487     });
 488   }
 489 }
 490