src/jalview/gui/SequenceFetcher.java

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