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
24 import org.exolab.castor.mapping.*;
\r
25 import org.exolab.castor.xml.*;
\r
26 import jalview.analysis.*;
\r
27 import jalview.datamodel.*;
\r
28 import jalview.gui.*;
\r
34 * @version $Revision$
\r
36 public class DBRefFetcher
\r
39 SequenceI [] dataset;
\r
41 CutAndPasteTransfer output = new CutAndPasteTransfer();
\r
42 StringBuffer sbuffer = new StringBuffer();
\r
43 boolean running = false;
\r
45 ///This will be a collection of Vectors of sequenceI refs.
\r
46 //The key will be the seq name or accession id of the seq
\r
49 public DBRefFetcher()
\r
52 public Vector getUniprotEntries(File file)
\r
54 UniprotFile uni = new UniprotFile();
\r
57 // 1. Load the mapping information from the file
\r
58 org.exolab.castor.mapping.Mapping map = new org.exolab.castor.mapping.Mapping(uni.getClass().getClassLoader());
\r
59 java.net.URL url = getClass().getResource("/uniprot_mapping.xml");
\r
60 map.loadMapping(url);
\r
62 // 2. Unmarshal the data
\r
63 Unmarshaller unmar = new Unmarshaller(uni);
\r
64 unmar.setIgnoreExtraElements(true);
\r
65 unmar.setMapping(map);
\r
67 uni = (UniprotFile) unmar.unmarshal(new FileReader(file));
\r
71 System.out.println("Error getUniprotEntries() " + e);
\r
74 return uni.getUniprotEntries();
\r
78 * Creates a new SequenceFeatureFetcher object.
\r
80 * @param align DOCUMENT ME!
\r
81 * @param ap DOCUMENT ME!
\r
83 public DBRefFetcher(SequenceI [] seqs, AlignFrame af)
\r
86 SequenceI [] ds = new SequenceI[seqs.length];
\r
87 for (int i = 0; i < seqs.length; i++)
\r
89 if(seqs[i].getDatasetSequence()!=null)
\r
90 ds[i] = seqs[i].getDatasetSequence();
\r
97 public boolean fetchDBRefs(boolean waitTillFinished)
\r
99 Thread thread = new Thread(this);
\r
103 if (waitTillFinished)
\r
111 catch (Exception ex)
\r
120 * The sequence will be added to a vector of sequences
\r
121 * belonging to key which could be either seq name or dbref id
\r
122 * @param seq SequenceI
\r
123 * @param key String
\r
125 void addSeqId(SequenceI seq, String key)
\r
127 key = key.toUpperCase();
\r
130 if (seqRefs.containsKey(key))
\r
132 seqs = (Vector) seqRefs.get(key);
\r
134 if (seqs != null && !seqs.contains(seq))
\r
136 seqs.addElement(seq);
\r
138 else if (seqs == null)
\r
140 seqs = new Vector();
\r
141 seqs.addElement(seq);
\r
147 seqs = new Vector();
\r
148 seqs.addElement(seq);
\r
151 seqRefs.put(key, seqs);
\r
159 long startTime = System.currentTimeMillis();
\r
160 af.setProgressBar("Fetching db refs", startTime);
\r
163 seqRefs = new Hashtable();
\r
169 while (seqIndex < dataset.length)
\r
171 StringBuffer queryString = new StringBuffer("uniprot:");
\r
173 for (int i = 0; (seqIndex < dataset.length) && (i < 50);
\r
176 SequenceI sequence = dataset[seqIndex];
\r
177 DBRefEntry[] uprefs = jalview.util.DBRefUtils.selectRefs(sequence.
\r
178 getDBRef(), new String[]
\r
180 jalview.datamodel.DBRefSource.UNIPROT});
\r
181 if (uprefs != null)
\r
183 if (uprefs.length + i > 50)
\r
188 for (int j = 0; j < uprefs.length; j++)
\r
190 addSeqId(sequence, uprefs[j].getAccessionId());
\r
191 queryString.append(uprefs[j].getAccessionId() + ";");
\r
196 StringTokenizer st = new StringTokenizer(sequence.getName(), "|");
\r
197 if (st.countTokens() + i > 50)
\r
199 //Dont send more than 50 id strings to dbFetch!!
\r
204 while (st.hasMoreTokens())
\r
206 String token = st.nextToken();
\r
207 addSeqId(sequence, token);
\r
208 queryString.append(token + ";");
\r
214 ///////////////////////////////////
\r
216 EBIFetchClient ebi = new EBIFetchClient();
\r
217 File file = ebi.fetchDataAsFile(queryString.toString(), "xml", "raw");
\r
220 ReadUniprotFile(file);
\r
224 catch (Exception ex)
\r
226 ex.printStackTrace();
\r
229 if (sbuffer.length() > 0)
\r
232 "Your sequences have been matched to Uniprot. Some of the ids have been\n" +
\r
233 "altered, most likely the start/end residue will have been updated.\n" +
\r
234 "Save your alignment to maintain the updated id.\n\n" +
\r
235 sbuffer.toString());
\r
236 Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);
\r
237 // The above is the dataset, we must now find out the index
\r
238 // of the viewed sequence
\r
242 af.setProgressBar("DBRef search completed", startTime);
\r
243 // promptBeforeBlast();
\r
252 * @param result DOCUMENT ME!
\r
253 * @param out DOCUMENT ME!
\r
254 * @param align DOCUMENT ME!
\r
256 void ReadUniprotFile(File file)
\r
258 if (!file.exists())
\r
263 SequenceI sequence = null;
\r
265 Vector entries = getUniprotEntries(file);
\r
267 int i, iSize = entries == null ? 0 : entries.size();
\r
268 UniprotEntry entry;
\r
269 for (i = 0; i < iSize; i++)
\r
271 entry = (UniprotEntry) entries.elementAt(i);
\r
273 //Work out which sequences this Uniprot file has matches to,
\r
274 //taking into account all accessionIds and names in the file
\r
275 Vector sequenceMatches = new Vector();
\r
276 for (int j = 0; j < entry.getAccession().size(); j++)
\r
278 String accessionId = entry.getAccession().elementAt(j).toString();
\r
279 if (seqRefs.containsKey(accessionId))
\r
281 Vector seqs = (Vector) seqRefs.get(accessionId);
\r
282 for (int jj = 0; jj < seqs.size(); jj++)
\r
284 sequence = (SequenceI) seqs.elementAt(jj);
\r
285 if (!sequenceMatches.contains(sequence))
\r
287 sequenceMatches.addElement(sequence);
\r
292 for (int j = 0; j < entry.getName().size(); j++)
\r
294 String name = entry.getName().elementAt(j).toString();
\r
295 if (seqRefs.containsKey(name))
\r
297 Vector seqs = (Vector) seqRefs.get(name);
\r
298 for (int jj = 0; jj < seqs.size(); jj++)
\r
300 sequence = (SequenceI) seqs.elementAt(jj);
\r
301 if (!sequenceMatches.contains(sequence))
\r
303 sequenceMatches.addElement(sequence);
\r
309 for (int m = 0; m < sequenceMatches.size(); m++)
\r
311 sequence = (SequenceI) sequenceMatches.elementAt(m);
\r
312 sequence.addDBRef(new DBRefEntry(DBRefSource.UNIPROT,
\r
313 "0", // TODO: VERSION FROM UNIPROT
\r
314 entry.getAccession().elementAt(0).
\r
317 System.out.println("Adding dbref to " + sequence.getName() + " : " +
\r
318 entry.getAccession().elementAt(0).toString());
\r
320 String nonGapped = AlignSeq.extractGaps("-. ",
\r
321 sequence.getSequenceAsString()).
\r
324 int absStart = entry.getUniprotSequence().getContent().indexOf(
\r
325 nonGapped.toString());
\r
327 if (absStart == -1)
\r
329 // Is UniprotSequence contained in dataset sequence?
\r
330 absStart = nonGapped.toString().indexOf(entry.getUniprotSequence().
\r
332 if (absStart == -1)
\r
334 sbuffer.append(sequence.getName() + " SEQUENCE NOT %100 MATCH \n");
\r
338 if (entry.getFeature() != null)
\r
340 Enumeration e = entry.getFeature().elements();
\r
341 while (e.hasMoreElements())
\r
343 SequenceFeature sf = (SequenceFeature) e.nextElement();
\r
344 sf.setBegin(sf.getBegin() + absStart + 1);
\r
345 sf.setEnd(sf.getEnd() + absStart + 1);
\r
348 sbuffer.append(sequence.getName() +
\r
349 " HAS " + absStart +
\r
350 " PREFIXED RESIDUES COMPARED TO UNIPROT - ANY SEQUENCE FEATURES"
\r
351 + " HAVE BEEN ADJUSTED ACCORDINGLY \n");
\r
357 //unknownSequences.remove(sequence);
\r
359 int absEnd = absStart + nonGapped.toString().length();
\r
362 Enumeration e = entry.getDbReference().elements();
\r
363 Vector onlyPdbEntries = new Vector();
\r
364 while (e.hasMoreElements())
\r
366 PDBEntry pdb = (PDBEntry) e.nextElement();
\r
367 if (!pdb.getType().equals(DBRefSource.PDB))
\r
369 DBRefEntry xref = new DBRefEntry(pdb.getType(), DBRefSource.UNIPROT, pdb.getId());
\r
370 sequence.addDBRef(xref);
\r
374 sequence.addDBRef(new DBRefEntry(DBRefSource.PDB,
\r
378 onlyPdbEntries.addElement(pdb);
\r
381 sequence.setPDBId(onlyPdbEntries);
\r
383 sequence.setStart(absStart);
\r
384 sequence.setEnd(absEnd);
\r