2 * Jalview - A Sequence Alignment Editor and Viewer
\r
3 * Copyright (C) 2007 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle
\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
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
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
19 package jalview.gui;
\r
25 import java.awt.event.*;
\r
26 import javax.swing.*;
\r
29 import jalview.datamodel.*;
\r
30 import jalview.datamodel.xdb.embl.*;
\r
31 import java.io.File;
\r
32 import jalview.io.*;
\r
33 import java.awt.Rectangle;
\r
34 import java.awt.BorderLayout;
\r
35 import java.awt.Dimension;
\r
37 public class SequenceFetcher
\r
38 extends JPanel implements Runnable
\r
40 JInternalFrame frame;
\r
41 AlignFrame alignFrame;
\r
42 StringBuffer result;
\r
43 final String noDbSelected = "-- Select Database --";
\r
44 public SequenceFetcher(AlignFrame af)
\r
47 database.addItem(noDbSelected);
\r
48 database.addItem("Uniprot");
\r
49 database.addItem("EMBL");
\r
50 database.addItem("EMBLCDS");
\r
51 database.addItem("PDB");
\r
52 database.addItem("PFAM");
\r
58 catch (Exception ex)
\r
60 ex.printStackTrace();
\r
63 frame = new JInternalFrame();
\r
64 frame.setContentPane(this);
\r
65 if (System.getProperty("os.name").startsWith("Mac"))
\r
67 Desktop.addInternalFrame(frame, getFrameTitle(), 400, 140);
\r
71 Desktop.addInternalFrame(frame, getFrameTitle(), 400, 125);
\r
75 private String getFrameTitle()
\r
77 return ( (alignFrame == null) ? "New " : "Additional ") +
\r
81 private void jbInit()
\r
84 this.setLayout(borderLayout2);
\r
86 database.setFont(new java.awt.Font("Verdana", Font.PLAIN, 11));
\r
87 jLabel1.setFont(new java.awt.Font("Verdana", Font.ITALIC, 11));
\r
88 jLabel1.setHorizontalAlignment(SwingConstants.CENTER);
\r
90 "Separate multiple accession ids with semi colon \";\"");
\r
92 ok.addActionListener(new ActionListener()
\r
94 public void actionPerformed(ActionEvent e)
\r
96 ok_actionPerformed();
\r
99 close.setText("Close");
\r
100 close.addActionListener(new ActionListener()
\r
102 public void actionPerformed(ActionEvent e)
\r
104 close_actionPerformed(e);
\r
107 textArea.setFont(new java.awt.Font("Verdana", Font.PLAIN, 11));
\r
108 textArea.setLineWrap(true);
\r
109 textArea.addKeyListener(new KeyAdapter()
\r
111 public void keyPressed(KeyEvent e)
\r
113 if(e.getKeyCode()==KeyEvent.VK_ENTER)
\r
114 ok_actionPerformed();
\r
117 jPanel3.setLayout(borderLayout1);
\r
118 borderLayout1.setVgap(5);
\r
120 jPanel1.add(close);
\r
121 jPanel3.add(jPanel2, java.awt.BorderLayout.WEST);
\r
122 jPanel2.add(database);
\r
123 jPanel3.add(jScrollPane1, java.awt.BorderLayout.CENTER);
\r
124 jPanel3.add(jLabel1, java.awt.BorderLayout.NORTH);
\r
125 this.add(jPanel1, java.awt.BorderLayout.SOUTH);
\r
126 this.add(jPanel3, java.awt.BorderLayout.CENTER);
\r
127 jScrollPane1.getViewport().add(textArea);
\r
131 JComboBox database = new JComboBox();
\r
132 JLabel jLabel1 = new JLabel();
\r
133 JButton ok = new JButton();
\r
134 JButton close = new JButton();
\r
135 JPanel jPanel1 = new JPanel();
\r
136 JTextArea textArea = new JTextArea();
\r
137 JScrollPane jScrollPane1 = new JScrollPane();
\r
138 JPanel jPanel2 = new JPanel();
\r
139 JPanel jPanel3 = new JPanel();
\r
140 BorderLayout borderLayout1 = new BorderLayout();
\r
141 BorderLayout borderLayout2 = new BorderLayout();
\r
142 public void close_actionPerformed(ActionEvent e)
\r
146 frame.setClosed(true);
\r
148 catch (Exception ex)
\r
152 public void ok_actionPerformed()
\r
154 database.setEnabled(false);
\r
155 textArea.setEnabled(false);
\r
156 ok.setEnabled(false);
\r
157 close.setEnabled(false);
\r
159 Thread worker = new Thread(this);
\r
163 private void resetDialog()
\r
165 database.setEnabled(true);
\r
166 textArea.setEnabled(true);
\r
167 ok.setEnabled(true);
\r
168 close.setEnabled(true);
\r
174 if (database.getSelectedItem().equals(noDbSelected))
\r
176 error += "Please select the source database\n";
\r
178 com.stevesoft.pat.Regex empty = new com.stevesoft.pat.Regex("\\s+", "");
\r
179 textArea.setText(empty.replaceAll(textArea.getText()));
\r
180 if (textArea.getText().length() == 0)
\r
182 error += "Please enter a (semi-colon separated list of) database id(s)";
\r
184 if (error.length() > 0)
\r
186 showErrorMessage(error);
\r
191 result = new StringBuffer();
\r
192 if (database.getSelectedItem().equals("Uniprot"))
\r
194 getUniprotFile(textArea.getText());
\r
196 else if (database.getSelectedItem().equals("EMBL")
\r
197 || database.getSelectedItem().equals("EMBLCDS"))
\r
199 String DBRefSource = database.getSelectedItem().equals("EMBLCDS")
\r
200 ? jalview.datamodel.DBRefSource.EMBLCDS
\r
201 : jalview.datamodel.DBRefSource.EMBL;
\r
203 StringTokenizer st = new StringTokenizer(textArea.getText(), ";");
\r
204 SequenceI[] seqs = null;
\r
205 while(st.hasMoreTokens())
\r
207 EBIFetchClient dbFetch = new EBIFetchClient();
\r
209 File reply = dbFetch.fetchDataAsFile(
\r
210 database.getSelectedItem().toString().toLowerCase(
\r
211 ) + ":" + st.nextToken(),
\r
214 jalview.datamodel.xdb.embl.EmblFile efile=null;
\r
215 if (reply != null && reply.exists())
\r
217 efile = jalview.datamodel.xdb.embl.EmblFile.getEmblFile(reply);
\r
220 for (Iterator i=efile.getEntries().iterator(); i.hasNext(); ) {
\r
221 EmblEntry entry = (EmblEntry) i.next();
\r
222 SequenceI[] seqparts = entry.getSequences(false,true, DBRefSource);
\r
223 if (seqparts!=null) {
\r
224 SequenceI[] newseqs = null;
\r
227 newseqs = new SequenceI[seqparts.length];
\r
229 newseqs = new SequenceI[seqs.length+seqparts.length];
\r
231 for (;si<seqs.length; si++) {
\r
232 newseqs[si] = seqs[si];
\r
236 for (int j=0;j<seqparts.length; si++, j++) {
\r
237 newseqs[si] = seqparts[j].deriveSequence(); // place DBReferences on dataset and refer
\r
247 if (seqs!=null && seqs.length>0) {
\r
248 if (parseResult(new Alignment(seqs), null, null)!=null)
\r
249 result.append("# Successfully parsed the "+database.getSelectedItem()+" Queries into an Alignment");
\r
252 else if (database.getSelectedItem().equals("PDB"))
\r
254 StringTokenizer qset = new StringTokenizer(textArea.getText(), ";");
\r
256 SequenceI[] seqs = null;
\r
257 while (qset.hasMoreTokens() && ((query = qset.nextToken())!=null))
\r
259 SequenceI[] seqparts = getPDBFile(query.toUpperCase());
\r
260 if (seqparts != null)
\r
268 SequenceI[] newseqs = new SequenceI[seqs.length+seqparts.length];
\r
270 for (; i < seqs.length; i++)
\r
272 newseqs[i] = seqs[i];
\r
275 for (int j=0;j<seqparts.length; i++, j++)
\r
277 newseqs[i] = seqparts[j];
\r
281 result.append("# Success for "+query.toUpperCase()+"\n");
\r
284 if (seqs != null && seqs.length > 0)
\r
286 if (parseResult(new Alignment(seqs), null, null)!=null)
\r
289 "# Successfully parsed the PDB File Queries into an Alignment");
\r
293 else if( database.getSelectedItem().equals("PFAM"))
\r
297 result.append(new FastaFile(
\r
298 "http://www.sanger.ac.uk/cgi-bin/Pfam/getalignment.pl?format=fal&acc="
\r
299 + textArea.getText().toUpperCase(), "URL").print()
\r
302 if(result.length()>0)
\r
304 parseResult( result.toString(), textArea.getText().toUpperCase() );
\r
308 catch (java.io.IOException ex)
\r
314 if (result == null || result.length() == 0)
\r
316 showErrorMessage("Error retrieving " + textArea.getText()
\r
317 + " from " + database.getSelectedItem());
\r
324 void getUniprotFile(String id)
\r
326 EBIFetchClient ebi = new EBIFetchClient();
\r
327 File file = ebi.fetchDataAsFile("uniprot:" + id, "xml", null);
\r
329 DBRefFetcher dbref = new DBRefFetcher();
\r
330 Vector entries = dbref.getUniprotEntries(file);
\r
332 if (entries != null)
\r
334 //First, make the new sequences
\r
335 Enumeration en = entries.elements();
\r
336 while (en.hasMoreElements())
\r
338 UniprotEntry entry = (UniprotEntry) en.nextElement();
\r
340 StringBuffer name = new StringBuffer(">UniProt/Swiss-Prot");
\r
341 Enumeration en2 = entry.getAccession().elements();
\r
342 while (en2.hasMoreElements())
\r
345 name.append(en2.nextElement());
\r
347 en2 = entry.getName().elements();
\r
348 while (en2.hasMoreElements())
\r
351 name.append(en2.nextElement());
\r
354 if (entry.getProtein() != null)
\r
356 name.append(" " + entry.getProtein().getName().elementAt(0));
\r
359 result.append(name + "\n" + entry.getUniprotSequence().getContent() +
\r
364 //Then read in the features and apply them to the dataset
\r
365 Alignment al = parseResult(result.toString(), null);
\r
366 for (int i = 0; i < entries.size(); i++)
\r
368 UniprotEntry entry = (UniprotEntry) entries.elementAt(i);
\r
369 Enumeration e = entry.getDbReference().elements();
\r
370 Vector onlyPdbEntries = new Vector();
\r
371 while (e.hasMoreElements())
\r
373 PDBEntry pdb = (PDBEntry) e.nextElement();
\r
374 if (!pdb.getType().equals("PDB"))
\r
379 onlyPdbEntries.addElement(pdb);
\r
382 Enumeration en2 = entry.getAccession().elements();
\r
383 while (en2.hasMoreElements())
\r
385 al.getSequenceAt(i).getDatasetSequence().addDBRef(new DBRefEntry(
\r
386 DBRefSource.UNIPROT,
\r
388 en2.nextElement().toString()));
\r
394 al.getSequenceAt(i).getDatasetSequence().setPDBId(onlyPdbEntries);
\r
395 if (entry.getFeature() != null)
\r
397 e = entry.getFeature().elements();
\r
398 while (e.hasMoreElements())
\r
400 SequenceFeature sf = (SequenceFeature) e.nextElement();
\r
401 sf.setFeatureGroup("Uniprot");
\r
402 al.getSequenceAt(i).getDatasetSequence().addSequenceFeature( sf );
\r
409 SequenceI[] getPDBFile(String id)
\r
411 Vector result = new Vector();
\r
412 String chain = null;
\r
413 if (id.indexOf(":") > -1)
\r
415 chain = id.substring(id.indexOf(":") + 1);
\r
416 id = id.substring(0, id.indexOf(":"));
\r
419 EBIFetchClient ebi = new EBIFetchClient();
\r
420 String file = ebi.fetchDataAsFile("pdb:" + id, "pdb", "raw").
\r
428 PDBfile pdbfile = new PDBfile(file, jalview.io.AppletFormatAdapter.FILE);
\r
429 for (int i = 0; i < pdbfile.chains.size(); i++)
\r
431 if (chain == null ||
\r
432 ( (PDBChain) pdbfile.chains.elementAt(i)).id.
\r
433 toUpperCase().equals(chain))
\r
435 PDBChain pdbchain = (PDBChain) pdbfile.chains.elementAt(i);
\r
436 // Get the Chain's Sequence - who's dataset includes any special features added from the PDB file
\r
437 SequenceI sq = pdbchain.sequence;
\r
438 // Specially formatted name for the PDB chain sequences retrieved from the PDB
\r
439 sq.setName("PDB|"+id+"|"+sq.getName());
\r
440 // Might need to add more metadata to the PDBEntry object
\r
443 * PDBEntry entry = new PDBEntry();
\r
444 // Construct the PDBEntry
\r
446 if (entry.getProperty() == null)
\r
447 entry.setProperty(new Hashtable());
\r
448 entry.getProperty().put("chains",
\r
450 + "=" + sq.getStart()
\r
451 + "-" + sq.getEnd());
\r
452 sq.getDatasetSequence().addPDBId(entry);
\r
455 // We make a DBRefEtntry because we have obtained the PDB file from a verifiable source
\r
456 // JBPNote - PDB DBRefEntry should also carry the chain and mapping information
\r
457 DBRefEntry dbentry = new DBRefEntry(jalview.datamodel.DBRefSource.PDB,
\r
458 "0", id + pdbchain.id);
\r
459 sq.addDBRef(dbentry);
\r
460 // and add seuqence to the retrieved set
\r
461 result.addElement(sq.deriveSequence());
\r
465 if (result.size() < 1)
\r
467 throw new Exception("WsDBFetch for PDB id resulted in zero result size");
\r
470 catch (Exception ex) // Problem parsing PDB file
\r
472 jalview.bin.Cache.log.warn("Exception when retrieving " +
\r
473 textArea.getText() + " from " +
\r
474 database.getSelectedItem(), ex);
\r
479 SequenceI[] results = new SequenceI[result.size()];
\r
480 for (int i = 0, j = result.size(); i < j; i++)
\r
482 results[i] = (SequenceI) result.elementAt(i);
\r
483 result.setElementAt(null,i);
\r
487 Alignment parseResult(String result, String title)
\r
489 String format = new IdentifyFile().Identify(result, "Paste");
\r
490 Alignment sequences = null;
\r
491 if (FormatAdapter.isValidFormat(format))
\r
496 sequences = new FormatAdapter().readFile(result.toString(), "Paste",
\r
499 catch (Exception ex)
\r
502 if (sequences!=null)
\r
504 return parseResult(sequences, title, format);
\r
509 showErrorMessage("Error retrieving " + textArea.getText()
\r
510 + " from " + database.getSelectedItem());
\r
516 Alignment parseResult(Alignment al, String title, String currentFileFormat)
\r
519 if (al != null && al.getHeight() > 0)
\r
521 if (alignFrame == null)
\r
523 AlignFrame af = new AlignFrame(al,
\r
524 AlignFrame.DEFAULT_WIDTH,
\r
525 AlignFrame.DEFAULT_HEIGHT);
\r
526 if (currentFileFormat!=null)
\r
528 af.currentFileFormat = currentFileFormat; // WHAT IS THE DEFAULT FORMAT FOR NON-FormatAdapter Sourced Alignments?
\r
533 title = "Retrieved from " + database.getSelectedItem();
\r
536 Desktop.addInternalFrame(af,
\r
538 AlignFrame.DEFAULT_WIDTH,
\r
539 AlignFrame.DEFAULT_HEIGHT);
\r
541 af.statusBar.setText("Successfully pasted alignment file");
\r
545 af.setMaximum(jalview.bin.Cache.getDefault("SHOW_FULLSCREEN", false));
\r
547 catch (Exception ex)
\r
552 for (int i = 0; i < al.getHeight(); i++)
\r
554 alignFrame.viewport.alignment.addSequence(al.getSequenceAt(i)); // this also creates dataset sequence entries
\r
556 alignFrame.viewport.setEndSeq(alignFrame.viewport.alignment.
\r
558 alignFrame.viewport.alignment.getWidth();
\r
559 alignFrame.viewport.firePropertyChange("alignment", null,
\r
560 alignFrame.viewport.
\r
561 getAlignment().getSequences());
\r
567 void showErrorMessage(final String error)
\r
570 javax.swing.SwingUtilities.invokeLater(new Runnable()
\r
574 JOptionPane.showInternalMessageDialog(Desktop.desktop,
\r
575 error, "Error Retrieving Data",
\r
576 JOptionPane.WARNING_MESSAGE);
\r