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 SequenceFeatureFetcher implements Runnable
\r
48 ArrayList unknownSequences;
\r
49 CutAndPasteTransfer output = new CutAndPasteTransfer();
\r
50 StringBuffer sbuffer = new StringBuffer();
\r
51 boolean uniprotFlag = false;
\r
53 public SequenceFeatureFetcher()
\r
56 public Vector getUniprotEntries(File file)
\r
59 UniprotFile uni = new UniprotFile();
\r
62 // 1. Load the mapping information from the file
\r
63 Mapping map = new Mapping(uni.getClass().getClassLoader());
\r
64 java.net.URL url = getClass().getResource("/uniprot_mapping.xml");
\r
65 map.loadMapping(url);
\r
67 // 2. Unmarshal the data
\r
68 Unmarshaller unmar = new Unmarshaller(uni);
\r
69 unmar.setIgnoreExtraElements(true);
\r
70 unmar.setMapping(map);
\r
71 // unmar.setDebug(true);
\r
73 uni = (UniprotFile) unmar.unmarshal(new FileReader(file));
\r
77 System.out.println("Error getUniprotEntries() "+e);
\r
81 return uni.getUniprotEntries();
\r
85 * Creates a new SequenceFeatureFetcher object.
\r
87 * @param align DOCUMENT ME!
\r
88 * @param ap DOCUMENT ME!
\r
90 public SequenceFeatureFetcher(AlignmentI align, AlignmentPanel ap)
\r
92 unknownSequences = new ArrayList();
\r
94 this.dataset = align.getDataset();
\r
97 Thread thread = new Thread(this);
\r
109 Vector sequences = dataset.getSequences();
\r
111 while (seqIndex < sequences.size())
\r
113 Vector ids = new Vector();
\r
115 for (int i = 0; (seqIndex < sequences.size()) && (i < 50);
\r
118 Sequence sequence = (Sequence) sequences.get(seqIndex);
\r
119 Vector uprefs = jalview.util.DBRefUtils.selectRefs(sequence.getDBRef(), new String[] { "Uniprot"});
\r
122 // we know the id for this entry, so don't note its ID in the unknownSequences list
\r
123 for (int j=0,k=uprefs.size(); j<k; j++)
\r
124 ids.add(((DBRefEntry) uprefs.get(j)).getAccessionId());
\r
125 unknownSequences.add(sequence);
\r
127 if (!ids.contains(sequence.getName()))
\r
129 ids.add(sequence.getName());
\r
130 unknownSequences.add(sequence);
\r
135 ///////////////////////////////////
\r
137 if (ids.size() > 0)
\r
139 StringBuffer remainingIds = new StringBuffer("uniprot:");
\r
140 for (int i = 0; i < ids.size(); i++)
\r
142 if(ids.get(i).toString().indexOf("|")>-1)
\r
144 remainingIds.append(ids.get(i).toString().substring(
\r
145 ids.get(i).toString().lastIndexOf("|") + 1));
\r
146 uniprotFlag = true;
\r
148 remainingIds.append(ids.get(i) + ";");
\r
150 EBIFetchClient ebi = new EBIFetchClient();
\r
151 File file = ebi.fetchDataAsFile(remainingIds.toString(),
\r
158 ReadUniprotFile(file, ids);
\r
163 catch (Exception ex)
\r
165 ex.printStackTrace();
\r
168 if (sbuffer.length() > 0)
\r
171 "Your sequences have been matched to Uniprot. Some of the ids have been\n" +
\r
172 "altered, most likely the start/end residue will have been updated.\n" +
\r
173 "Save your alignment to maintain the updated id.\n\n" +
\r
174 sbuffer.toString());
\r
175 Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);
\r
176 // The above is the dataset, we must now find out the index
\r
177 // of the viewed sequence
\r
181 promptBeforeBlast();
\r
186 void promptBeforeBlast()
\r
188 // This must be outside the run() body as java 1.5
\r
189 // will not return any value from the OptionPane to the expired thread.
\r
190 if (unknownSequences.size() > 0)
\r
192 // int reply = javax.swing.JOptionPane.showConfirmDialog(
\r
193 // Desktop.desktop, "Couldn't find a match for "+unknownSequences.size()+" sequences."
\r
194 // +"\nPerform blast for unknown sequences?",
\r
195 // "Blast for Unidentified Sequences",
\r
196 // javax.swing.JOptionPane.YES_NO_OPTION, javax.swing.JOptionPane.QUESTION_MESSAGE);
\r
197 javax.swing.JOptionPane.showMessageDialog(
\r
198 Desktop.desktop, "Couldn't find a match for "+unknownSequences.size()+" sequences.",
\r
199 "Unidentified Sequences",
\r
200 javax.swing.JOptionPane.WARNING_MESSAGE);
\r
203 // if(reply == javax.swing.JOptionPane.YES_OPTION)
\r
204 // new WSWUBlastClient(ap, align, unknownSequences);
\r
214 * @param result DOCUMENT ME!
\r
215 * @param out DOCUMENT ME!
\r
216 * @param align DOCUMENT ME!
\r
218 void ReadUniprotFile(File file, Vector ids)
\r
223 SequenceI sequence = null;
\r
225 Vector entries = getUniprotEntries(file);
\r
227 int i, iSize = entries==null?0:entries.size();
\r
228 UniprotEntry entry;
\r
229 for (i = 0; i < iSize; i++)
\r
231 entry = (UniprotEntry) entries.elementAt(i);
\r
232 String idmatch = entry.getAccession().elementAt(0).toString();
\r
233 sequence = dataset.findName(idmatch);
\r
235 if (sequence == null)
\r
237 //Sequence maybe Name, not Accession
\r
238 idmatch = entry.getName().elementAt(0).toString();
\r
239 sequence = dataset.findName(idmatch);
\r
243 ids.remove(sequence.getName());
\r
245 else if (sequence == null && uniprotFlag)
\r
247 sequence = dataset.findName("UniProt/Swiss-Prot|"+entry.getAccession().elementAt(0)+"|"+idmatch);
\r
248 ids.remove(idmatch);
\r
251 if(sequence ==null)
\r
253 System.out.println(idmatch+" not found");
\r
258 String nonGapped = AlignSeq.extractGaps("-. ", sequence.getSequence());
\r
260 int absStart = entry.getUniprotSequence().getContent().indexOf(
\r
261 nonGapped.toString());
\r
263 if (absStart == -1)
\r
265 // Is UniprotSequence contained in dataset sequence?
\r
266 absStart = nonGapped.toString().indexOf(entry.getUniprotSequence().getContent());
\r
269 sbuffer.append(sequence.getName() +
\r
270 " SEQUENCE NOT %100 MATCH \n");
\r
277 if(entry.getFeature()!=null)
\r
279 Enumeration e = entry.getFeature().elements();
\r
280 while (e.hasMoreElements())
\r
282 SequenceFeature sf = (SequenceFeature) e.nextElement();
\r
283 sf.setBegin(sf.getBegin() + absStart + 1);
\r
284 sf.setEnd(sf.getEnd() + absStart + 1);
\r
288 sbuffer.append(sequence.getName() +
\r
289 " HAS "+absStart+" PREFIXED RESIDUES COMPARED TO UNIPROT - ANY SEQUENCE FEATURES"
\r
290 +" HAVE BEEN ADJUSTED ACCORDINGLY \n");
\r
296 unknownSequences.remove(sequence);
\r
298 int absEnd = absStart + nonGapped.toString().length();
\r
301 Enumeration e = entry.getDbReference().elements();
\r
302 Vector onlyPdbEntries = new Vector();
\r
303 while(e.hasMoreElements())
\r
305 PDBEntry pdb = (PDBEntry)e.nextElement();
\r
306 if(!pdb.getType().equals("PDB"))
\r
309 onlyPdbEntries.addElement(pdb);
\r
312 sequence.setPDBId(onlyPdbEntries);
\r
313 if (entry.getFeature()!=null) {
\r
314 e = entry.getFeature().elements();
\r
315 while (e.hasMoreElements())
\r
317 SequenceFeature sf = (SequenceFeature) e.nextElement();
\r
318 sf.setFeatureGroup("Uniprot");
\r
319 sequence.addSequenceFeature( sf );
\r
322 sequence.setStart(absStart);
\r
323 sequence.setEnd(absEnd);
\r
328 while (n < align.getHeight())
\r
330 //This loop enables multiple sequences with the same
\r
331 //id to have features added and seq limits updated
\r
332 seq2 = align.getSequenceAt(n);
\r
333 if (seq2.getName().equals(idmatch))
\r
336 nonGapped = AlignSeq.extractGaps("-. ", seq2.getSequence());
\r
338 absStart = sequence.getSequence().indexOf(nonGapped);
\r
339 absEnd = absStart + nonGapped.toString().length() - 1;
\r
341 // This is the Viewd alignment sequences
\r
342 // No need to tell the user of the dataset updates
\r
343 if ( (seq2.getStart() != absStart+sequence.getStart())
\r
344 || (seq2.getEnd() != absEnd+sequence.getStart()))
\r
346 sbuffer.append("Updated: " + seq2.getName() + " " +
\r
347 seq2.getStart() + "/" + seq2.getEnd() +
\r
348 " to " + (absStart + sequence.getStart()) + "/" +
\r
349 (absEnd + sequence.getStart()) + "\n");
\r
351 seq2.setStart(absStart + sequence.getStart());
\r
352 seq2.setEnd(absEnd + sequence.getStart());
\r