2 * Jalview - A Sequence Alignment Editor and Viewer
\r
3 * Copyright (C) 2005 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
21 import jalview.datamodel.*;
\r
23 import jalview.gui.*;
\r
29 import org.exolab.castor.mapping.Mapping;
\r
31 import org.exolab.castor.xml.*;
\r
32 import jalview.analysis.AlignSeq;
\r
40 * @version $Revision$
\r
42 public class DBRefFetcher implements Runnable
\r
47 ArrayList unknownSequences;
\r
48 CutAndPasteTransfer output = new CutAndPasteTransfer();
\r
49 StringBuffer sbuffer = new StringBuffer();
\r
50 boolean uniprotFlag = false;
\r
52 public DBRefFetcher()
\r
55 public Vector getUniprotEntries(File file)
\r
57 UniprotFile uni = new UniprotFile();
\r
60 // 1. Load the mapping information from the file
\r
61 Mapping map = new Mapping(uni.getClass().getClassLoader());
\r
62 java.net.URL url = getClass().getResource("/uniprot_mapping.xml");
\r
63 map.loadMapping(url);
\r
65 // 2. Unmarshal the data
\r
66 Unmarshaller unmar = new Unmarshaller(uni);
\r
67 unmar.setIgnoreExtraElements(true);
\r
68 unmar.setMapping(map);
\r
69 // unmar.setDebug(true);
\r
71 uni = (UniprotFile) unmar.unmarshal(new FileReader(file));
\r
75 System.out.println("Error getUniprotEntries() "+e);
\r
79 return uni.getUniprotEntries();
\r
83 * Creates a new SequenceFeatureFetcher object.
\r
85 * @param align DOCUMENT ME!
\r
86 * @param ap DOCUMENT ME!
\r
88 public DBRefFetcher(AlignmentI align, AlignFrame af)
\r
91 unknownSequences = new ArrayList();
\r
93 this.dataset = align.getDataset();
\r
95 Thread thread = new Thread(this);
\r
104 long startTime = System.currentTimeMillis();
\r
105 af.setProgressBar("Fetching db refs", startTime);
\r
110 Vector sequences = dataset.getSequences();
\r
112 while (seqIndex < sequences.size())
\r
114 Vector ids = new Vector();
\r
116 for (int i = 0; (seqIndex < sequences.size()) && (i < 50);
\r
119 Sequence sequence = (Sequence) sequences.get(seqIndex);
\r
120 Vector uprefs = jalview.util.DBRefUtils.selectRefs(sequence.getDBRef(), new String[] {
\r
121 jalview.datamodel.DBRefSource.UNIPROT});
\r
124 // we know the id for this entry, so don't note its ID in the unknownSequences list
\r
125 for (int j=0,k=uprefs.size(); j<k; j++)
\r
126 ids.add(((DBRefEntry) uprefs.get(j)).getAccessionId());
\r
127 unknownSequences.add(sequence);
\r
129 if (!ids.contains(sequence.getName()))
\r
131 ids.add(sequence.getName());
\r
132 unknownSequences.add(sequence);
\r
137 ///////////////////////////////////
\r
139 if (ids.size() > 0)
\r
141 StringBuffer remainingIds = new StringBuffer("uniprot:");
\r
142 for (int i = 0; i < ids.size(); i++)
\r
144 if(ids.get(i).toString().indexOf("|")>-1)
\r
146 remainingIds.append(ids.get(i).toString().substring(
\r
147 ids.get(i).toString().lastIndexOf("|") + 1));
\r
148 uniprotFlag = true;
\r
151 remainingIds.append(ids.get(i));
\r
153 remainingIds.append(";");
\r
156 EBIFetchClient ebi = new EBIFetchClient();
\r
157 File file = ebi.fetchDataAsFile(remainingIds.toString(),
\r
164 ReadUniprotFile(file, ids);
\r
169 catch (Exception ex)
\r
171 ex.printStackTrace();
\r
174 if (sbuffer.length() > 0)
\r
177 "Your sequences have been matched to Uniprot. Some of the ids have been\n" +
\r
178 "altered, most likely the start/end residue will have been updated.\n" +
\r
179 "Save your alignment to maintain the updated id.\n\n" +
\r
180 sbuffer.toString());
\r
181 Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);
\r
182 // The above is the dataset, we must now find out the index
\r
183 // of the viewed sequence
\r
187 af.setProgressBar("DBRef search completed", startTime);
\r
188 // promptBeforeBlast();
\r
193 void promptBeforeBlast()
\r
195 // This must be outside the run() body as java 1.5
\r
196 // will not return any value from the OptionPane to the expired thread.
\r
197 if (unknownSequences.size() > 0)
\r
199 // int reply = javax.swing.JOptionPane.showConfirmDialog(
\r
200 // Desktop.desktop, "Couldn't find a match for "+unknownSequences.size()+" sequences."
\r
201 // +"\nPerform blast for unknown sequences?",
\r
202 // "Blast for Unidentified Sequences",
\r
203 // javax.swing.JOptionPane.YES_NO_OPTION, javax.swing.JOptionPane.QUESTION_MESSAGE);
\r
204 javax.swing.JOptionPane.showMessageDialog(
\r
205 Desktop.desktop, "Couldn't find a match for "+unknownSequences.size()+" sequences.",
\r
206 "Unidentified Sequences",
\r
207 javax.swing.JOptionPane.WARNING_MESSAGE);
\r
210 // if(reply == javax.swing.JOptionPane.YES_OPTION)
\r
211 // new WSWUBlastClient(ap, align, unknownSequences);
\r
218 * @param result DOCUMENT ME!
\r
219 * @param out DOCUMENT ME!
\r
220 * @param align DOCUMENT ME!
\r
222 void ReadUniprotFile(File file, Vector ids)
\r
227 SequenceI sequence = null;
\r
229 Vector entries = getUniprotEntries(file);
\r
231 int i, iSize = entries==null?0:entries.size();
\r
232 UniprotEntry entry;
\r
233 for (i = 0; i < iSize; i++)
\r
235 entry = (UniprotEntry) entries.elementAt(i);
\r
236 String idmatch = entry.getAccession().elementAt(0).toString();
\r
237 sequence = dataset.findName(idmatch);
\r
239 if (sequence == null)
\r
241 //Sequence maybe Name, not Accession
\r
242 idmatch = entry.getName().elementAt(0).toString();
\r
243 sequence = dataset.findName(idmatch);
\r
247 ids.remove(sequence.getName());
\r
249 else if (sequence == null && uniprotFlag)
\r
251 StringBuffer upid = new StringBuffer("UniProt/Swiss-Prot|");
\r
252 for(int u=0; u<entry.getAccession().size(); u++)
\r
253 upid.append(entry.getAccession().elementAt(u)+"|");
\r
255 sequence = dataset.findName(upid+idmatch);
\r
256 ids.remove(idmatch);
\r
259 if(sequence ==null)
\r
261 System.out.println(idmatch+" not found");
\r
266 sequence.addDBRef(new DBRefEntry(DBRefSource.UNIPROT,
\r
268 entry.getAccession().elementAt(0).toString()));
\r
270 System.out.println("Adding dbref to "+sequence.getName()+" : "+
\r
271 entry.getAccession().elementAt(0).toString());
\r
273 String nonGapped = AlignSeq.extractGaps("-. ", sequence.getSequence());
\r
275 int absStart = entry.getUniprotSequence().getContent().indexOf(
\r
276 nonGapped.toString());
\r
278 if (absStart == -1)
\r
280 // Is UniprotSequence contained in dataset sequence?
\r
281 absStart = nonGapped.toString().indexOf(entry.getUniprotSequence().getContent());
\r
284 sbuffer.append(sequence.getName() +
\r
285 " SEQUENCE NOT %100 MATCH \n");
\r
292 if(entry.getFeature()!=null)
\r
294 Enumeration e = entry.getFeature().elements();
\r
295 while (e.hasMoreElements())
\r
297 SequenceFeature sf = (SequenceFeature) e.nextElement();
\r
298 sf.setBegin(sf.getBegin() + absStart + 1);
\r
299 sf.setEnd(sf.getEnd() + absStart + 1);
\r
303 sbuffer.append(sequence.getName() +
\r
304 " HAS "+absStart+" PREFIXED RESIDUES COMPARED TO UNIPROT - ANY SEQUENCE FEATURES"
\r
305 +" HAVE BEEN ADJUSTED ACCORDINGLY \n");
\r
311 unknownSequences.remove(sequence);
\r
313 int absEnd = absStart + nonGapped.toString().length();
\r
316 Enumeration e = entry.getDbReference().elements();
\r
317 Vector onlyPdbEntries = new Vector();
\r
318 while(e.hasMoreElements())
\r
320 PDBEntry pdb = (PDBEntry)e.nextElement();
\r
321 if(!pdb.getType().equals("PDB"))
\r
324 onlyPdbEntries.addElement(pdb);
\r
327 sequence.setPDBId(onlyPdbEntries);
\r
329 if (entry.getFeature()!=null) {
\r
330 //e = entry.getFeature().elements();
\r
331 // while (e.hasMoreElements())
\r
333 // SequenceFeature sf = (SequenceFeature) e.nextElement();
\r
334 // sf.setFeatureGroup("Uniprot");
\r
335 // sequence.addSequenceFeature( sf );
\r
338 sequence.setStart(absStart);
\r
339 sequence.setEnd(absEnd);
\r
344 while (n < align.getHeight())
\r
346 //This loop enables multiple sequences with the same
\r
347 //id to have features added and seq limits updated
\r
348 seq2 = align.getSequenceAt(n);
\r
349 if (seq2.getName().equals(idmatch))
\r
351 nonGapped = AlignSeq.extractGaps("-. ", seq2.getSequence());
\r
353 absStart = sequence.getSequence().indexOf(nonGapped);
\r
354 absEnd = absStart + nonGapped.toString().length() - 1;
\r
356 // This is the Viewd alignment sequences
\r
357 // No need to tell the user of the dataset updates
\r
358 if ( (seq2.getStart() != absStart+sequence.getStart())
\r
359 || (seq2.getEnd() != absEnd+sequence.getStart()))
\r
361 sbuffer.append("Updated: " + seq2.getName() + " " +
\r
362 seq2.getStart() + "/" + seq2.getEnd() +
\r
363 " to " + (absStart + sequence.getStart()) + "/" +
\r
364 (absEnd + sequence.getStart()) + "\n");
\r
366 seq2.setStart(absStart + sequence.getStart());
\r
367 seq2.setEnd(absEnd + sequence.getStart());
\r