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       while (qset.hasMoreTokens() && ((query = qset.nextToken())!=null))
 215       {
 216         StringBuffer respart = getPDBFile(query.toUpperCase());
 217         if(respart!=null)
 218           result.append(respart);
 219       }
 220
 221
 222       if (result.length()>0)
 223         parseResult(result.toString(), null);
 224     }
 225     else if( database.getSelectedItem().equals("PFAM"))
 226     {
 227       try{
 228         result.append(new FastaFile(
 229            "http://www.sanger.ac.uk/cgi-bin/Pfam/getalignment.pl?format=fal&acc="
 230            +  textfield.getText().toUpperCase(), "URL").print()
 231            );
 232
 233          if(result.length()>0)
 234            parseResult( result.toString(), textfield.getText().toUpperCase() );
 235
 236       }catch(java.io.IOException ex)
 237       {   result = null;    }
 238     }
 239
 240     if (result == null || result.length() == 0)
 241       showErrorMessage("Error retrieving " + textfield.getText()
 242                        + " from " + database.getSelectedItem());
 243
 244     resetDialog();
 245     return;
 246   }
 247
 248   void getUniprotFile(String id)
 249   {
 250     EBIFetchClient ebi = new EBIFetchClient();
 251     File file = ebi.fetchDataAsFile("uniprot:" + id, "xml", null);
 252
 253     DBRefFetcher dbref = new DBRefFetcher();
 254     Vector entries = dbref.getUniprotEntries(file);
 255
 256     if (entries != null)
 257     {
 258       //First, make the new sequences
 259       Enumeration en = entries.elements();
 260       while (en.hasMoreElements())
 261       {
 262         UniprotEntry entry = (UniprotEntry) en.nextElement();
 263
 264         StringBuffer name = new StringBuffer(">UniProt/Swiss-Prot");
 265         Enumeration en2 = entry.getAccession().elements();
 266         while (en2.hasMoreElements())
 267         {
 268           name.append("|");
 269           name.append(en2.nextElement());
 270         }
 271         en2 = entry.getName().elements();
 272         while (en2.hasMoreElements())
 273         {
 274           name.append("|");
 275           name.append(en2.nextElement());
 276         }
 277
 278         if (entry.getProtein() != null)
 279         {
 280            name.append(" " + entry.getProtein().getName().elementAt(0));
 281         }
 282
 283         result.append(name + "\n" + entry.getUniprotSequence().getContent() +
 284                       "\n");
 285
 286       }
 287
 288       //Then read in the features and apply them to the dataset
 289       SequenceI[] sequence = parseResult(result.toString(), null);
 290       for (int i = 0; i < entries.size(); i++)
 291       {
 292         UniprotEntry entry = (UniprotEntry) entries.elementAt(i);
 293         Enumeration e = entry.getDbReference().elements();
 294         Vector onlyPdbEntries = new Vector();
 295         while (e.hasMoreElements())
 296         {
 297           PDBEntry pdb = (PDBEntry) e.nextElement();
 298           if (!pdb.getType().equals("PDB"))
 299             continue;
 300
 301           onlyPdbEntries.addElement(pdb);
 302         }
 303
 304         Enumeration en2 = entry.getAccession().elements();
 305         while (en2.hasMoreElements())
 306         {
 307           sequence[i].getDatasetSequence().addDBRef(new DBRefEntry(DBRefSource.UNIPROT,
 308                     "0",
 309                     en2.nextElement().toString()));
 310         }
 311
 312
 313
 314
 315         sequence[i].getDatasetSequence().setPDBId(onlyPdbEntries);
 316         if (entry.getFeature() != null)
 317         {
 318           e = entry.getFeature().elements();
 319           while (e.hasMoreElements())
 320           {
 321             SequenceFeature sf = (SequenceFeature) e.nextElement();
 322             sf.setFeatureGroup("Uniprot");
 323             sequence[i].getDatasetSequence().addSequenceFeature( sf );
 324           }
 325         }
 326       }
 327     }
 328   }
 329
 330   StringBuffer getPDBFile(String id)
 331   {
 332     StringBuffer result = new StringBuffer();
 333     String chain = null;
 334     if (id.indexOf(":") > -1)
 335     {
 336       chain = id.substring(id.indexOf(":") + 1);
 337       id = id.substring(0, id.indexOf(":"));
 338     }
 339
 340     EBIFetchClient ebi = new EBIFetchClient();
 341     String file = ebi.fetchDataAsFile("pdb:" + id, "pdb", "raw").getAbsolutePath();
 342     if (file == null)
 343       return null;
 344     try
 345     {
 346       PDBfile pdbfile = new PDBfile(file, jalview.io.AppletFormatAdapter.FILE);
 347       for (int i = 0; i < pdbfile.chains.size(); i++)
 348       {
 349         if (chain == null ||
 350             ( (PDBChain) pdbfile.chains.elementAt(i)).id.
 351             toUpperCase().equals(chain))
 352
 353           result.append("\n>PDB|" + id + "|" +
 354                         ( (PDBChain) pdbfile.chains.elementAt(i)).sequence.
 355                         getName() +
 356                         "\n"
 357                         +
 358                         ( (PDBChain) pdbfile.chains.elementAt(i)).sequence.
 359                         getSequence());
 360       }
 361     }
 362     catch (Exception ex) // Problem parsing PDB file
 363     {
 364       jalview.bin.Cache.log.warn("Exception when retrieving " +
 365                                  textfield.getText() + " from " +
 366                                  database.getSelectedItem(), ex);
 367       return null;
 368     }
 369
 370     return result;
 371   }
 372
 373   SequenceI[] parseResult(String result, String title)
 374   {
 375     String format = new IdentifyFile().Identify(result, "Paste");
 376     SequenceI[] sequences = null;
 377
 378     if (FormatAdapter.isValidFormat(format))
 379     {
 380       sequences = null;
 381       try{ sequences = new FormatAdapter().readFile(result.toString(), "Paste",
 382                                                format);}
 383       catch(Exception ex){}
 384
 385       if (sequences != null && sequences.length > 0)
 386       {
 387         if (alignFrame == null)
 388         {
 389           AlignFrame af = new AlignFrame(new Alignment(sequences),
 390                                            AlignFrame.DEFAULT_WIDTH,
 391                                            AlignFrame.DEFAULT_HEIGHT
 392 );
 393           af.currentFileFormat = format;
 394           if(title==null)
 395             title = "Retrieved from " + database.getSelectedItem();
 396           Desktop.addInternalFrame(af,
 397                                    title,
 398                                    AlignFrame.DEFAULT_WIDTH,
 399                                    AlignFrame.DEFAULT_HEIGHT);
 400           af.statusBar.setText("Successfully pasted alignment file");
 401
 402           try
 403           {
 404             af.setMaximum(jalview.bin.Cache.getDefault("SHOW_FULLSCREEN", false));
 405           }
 406           catch (Exception ex)
 407           {}
 408         }
 409         else
 410         {
 411           for (int i = 0; i < sequences.length; i++)
 412           {
 413             alignFrame.viewport.alignment.addSequence(sequences[i]);
 414
 415             ////////////////////////////
 416             //Dataset needs extension;
 417             /////////////////////////////
 418             Sequence ds = new Sequence(sequences[i].getName(),
 419                                        AlignSeq.extractGaps("-. ",
 420                 sequences[i].getSequenceAsString()),
 421                                        sequences[i].getStart(),
 422                                        sequences[i].getEnd());
 423             sequences[i].setDatasetSequence(ds);
 424             alignFrame.viewport.alignment.getDataset().addSequence(ds);
 425           }
 426           alignFrame.viewport.setEndSeq(alignFrame.viewport.alignment.
 427                                         getHeight());
 428           alignFrame.viewport.alignment.getWidth();
 429           alignFrame.viewport.firePropertyChange("alignment", null,
 430                                                  alignFrame.viewport.
 431                                                  getAlignment().getSequences());
 432
 433         }
 434
 435         if (database.getSelectedItem().equals("PDB"))
 436         {
 437           // Parse out the ids from the structured names
 438           boolean errors = false;
 439           for (int i = 0; i < sequences.length; i++)
 440           {
 441             PDBEntry entry = new PDBEntry();
 442             com.stevesoft.pat.Regex idbits = new com.stevesoft.pat.Regex(
 443                 "PDB\\|([0-9A-z]{4})\\|(.)");
 444             if (idbits.search(sequences[i].getName()))
 445             {
 446               String pdbid = idbits.substring(1);
 447               String pdbccode = idbits.substring(2);
 448               // Construct the PDBEntry
 449               entry.setId(pdbid);
 450               if (entry.getProperty() == null)
 451                 entry.setProperty(new Hashtable());
 452               entry.getProperty().put("chains",
 453                                       pdbccode
 454                                       + "=" + sequences[i].getStart()
 455                                       + "-" + sequences[i].getEnd());
 456               sequences[i].getDatasetSequence().addPDBId(entry);
 457
 458               // We make a DBRefEtntry because we have obtained the PDB file from a verifiable source
 459               // JBPNote - PDB DBRefEntry should also carry the chain and mapping information
 460               DBRefEntry dbentry = new DBRefEntry(jalview.datamodel.DBRefSource.PDB,"0",pdbid);
 461               sequences[i].getDatasetSequence().addDBRef(dbentry);
 462             }
 463             else
 464             {
 465               // don't add an entry for this chain, but this is probably a bug
 466               // that the user should know about.
 467               jalview.bin.Cache.log.warn(
 468                   "No PDBEntry constructed for sequence " + i + " : " +
 469                   sequences[i].getName());
 470               errors = true;
 471             }
 472           }
 473           if (errors)
 474             jalview.bin.Cache.log.warn(
 475                 "Query string that resulted in PDBEntry construction failure was :\n" +
 476                 textfield.getText());
 477         }
 478
 479       }
 480       else
 481         showErrorMessage("Error retrieving " + textfield.getText()
 482                          + " from " + database.getSelectedItem());
 483     }
 484
 485     return sequences;
 486
 487   }
 488
 489   void showErrorMessage(final String error)
 490   {
 491     resetDialog();
 492     javax.swing.SwingUtilities.invokeLater(new Runnable()
 493     {
 494       public void run()
 495       {
 496         JOptionPane.showInternalMessageDialog(Desktop.desktop,
 497                                               error, "Error Retrieving Data",
 498                                           JOptionPane.WARNING_MESSAGE);
 499       }
 500     });
 501   }
 502 }
 503