2 * Jalview - A Sequence Alignment Editor and Viewer (Development Version 2.4.1)
\r
3 * Copyright (C) 2009 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.analysis.AlignSeq;
\r
22 import jalview.datamodel.AlignmentI;
\r
23 import jalview.datamodel.DBRefEntry;
\r
24 import jalview.datamodel.DBRefSource;
\r
25 import jalview.datamodel.Mapping;
\r
26 import jalview.datamodel.SequenceFeature;
\r
27 import jalview.datamodel.SequenceI;
\r
28 import jalview.gui.AlignFrame;
\r
29 import jalview.gui.CutAndPasteTransfer;
\r
30 import jalview.gui.Desktop;
\r
31 import jalview.gui.IProgressIndicator;
\r
32 import jalview.gui.OOMWarning;
\r
34 import java.lang.reflect.Array;
\r
35 import java.util.Enumeration;
\r
36 import java.util.Hashtable;
\r
37 import java.util.StringTokenizer;
\r
38 import java.util.Vector;
\r
40 import org.biojava.dasobert.dasregistry.DasSource;
\r
42 import uk.ac.ebi.picr.model.UPEntry;
\r
45 * Implements a runnable for validating a sequence against external databases
\r
46 * and then propagating references and features onto the sequence(s)
\r
49 * @version $Revision$
\r
51 public class DBRefFetcher implements Runnable
\r
53 SequenceI[] dataset;
\r
55 IProgressIndicator af;
\r
57 CutAndPasteTransfer output = new CutAndPasteTransfer();
\r
59 StringBuffer sbuffer = new StringBuffer();
\r
61 boolean running = false;
\r
63 * picr client instance
\r
65 uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperInterface picrClient = null;
\r
67 // /This will be a collection of Vectors of sequenceI refs.
\r
68 // The key will be the seq name or accession id of the seq
\r
73 SequenceFetcher sfetcher;
\r
75 public DBRefFetcher()
\r
80 * Creates a new SequenceFeatureFetcher object and fetches from the
\r
81 * currently selected set of databases.
\r
84 * fetch references for these sequences
\r
86 * the parent alignframe for progress bar monitoring.
\r
88 public DBRefFetcher(SequenceI[] seqs, AlignFrame af)
\r
90 this(seqs, af, null);
\r
93 * Creates a new SequenceFeatureFetcher object and fetches from the
\r
94 * currently selected set of databases.
\r
97 * fetch references for these sequences
\r
99 * the parent alignframe for progress bar monitoring.
\r
100 * @param sources array of database source strings to query references from
\r
102 public DBRefFetcher(SequenceI[] seqs, AlignFrame af, String[] sources)
\r
105 SequenceI[] ds = new SequenceI[seqs.length];
\r
106 for (int i = 0; i < seqs.length; i++)
\r
108 if (seqs[i].getDatasetSequence() != null)
\r
109 ds[i] = seqs[i].getDatasetSequence();
\r
114 // TODO Jalview 2.5 lots of this code should be in the gui package!
\r
115 sfetcher = jalview.gui.SequenceFetcher.getSequenceFetcherSingleton(af);
\r
118 // af.featureSettings_actionPerformed(null);
\r
119 String[] defdb=null,otherdb = sfetcher.getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
\r
120 Vector selsources = new Vector(), dasselsrc= (af.featureSettings!=null) ? af.featureSettings.getSelectedSources()
\r
121 : new jalview.gui.DasSourceBrowser().getSelectedSources();
\r
122 Enumeration en = dasselsrc.elements();
\r
123 while (en.hasMoreElements())
\r
125 DasSource src = (DasSource) en.nextElement();
\r
126 selsources.addElement(src.getNickname());
\r
129 for (int o=0;otherdb!=null && o<otherdb.length;o++)
\r
131 if (!selsources.contains(otherdb[o]))
\r
138 // select appropriate databases based on alignFrame context.
\r
139 if (af.getViewport().getAlignment().isNucleotide())
\r
141 defdb = DBRefSource.DNACODINGDBS;
\r
145 defdb = DBRefSource.PROTEINDBS;
\r
147 // append the selected sequence sources to the default dbs
\r
148 dbSources = new String[defdb.length+osel];
\r
149 System.arraycopy(defdb, 0, dbSources, 0, defdb.length);
\r
150 for (int o=0,op=defdb.length; otherdb!=null && o<otherdb.length; o++)
\r
152 if (otherdb[o]!=null)
\r
154 dbSources[op++] = otherdb[o];
\r
158 // we assume the caller knows what they're doing and ensured that all the db source names are valid
\r
159 dbSources = sources;
\r
163 * retrieve all the das sequence sources and add them to the list of db sources to retrieve from
\r
165 public void appendAllDasSources()
\r
167 if (dbSources == null)
\r
169 dbSources = new String[] {};
\r
171 // append additional sources
\r
172 String[] otherdb = sfetcher.getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
\r
173 if (otherdb!=null && otherdb.length>0)
\r
175 String[] newsrc = new String[dbSources.length+otherdb.length];
\r
176 System.arraycopy(dbSources, 0, newsrc,0,dbSources.length);
\r
177 System.arraycopy(otherdb, 0, newsrc,dbSources.length, otherdb.length);
\r
178 dbSources = newsrc;
\r
182 * start the fetcher thread
\r
184 * @param waitTillFinished
\r
185 * true to block until the fetcher has finished
\r
187 public void fetchDBRefs(boolean waitTillFinished)
\r
189 Thread thread = new Thread(this);
\r
193 if (waitTillFinished)
\r
200 } catch (Exception ex)
\r
208 * The sequence will be added to a vector of sequences belonging to key which
\r
209 * could be either seq name or dbref id
\r
216 void addSeqId(SequenceI seq, String key)
\r
218 key = key.toUpperCase();
\r
221 if (seqRefs.containsKey(key))
\r
223 seqs = (Vector) seqRefs.get(key);
\r
225 if (seqs != null && !seqs.contains(seq))
\r
227 seqs.addElement(seq);
\r
229 else if (seqs == null)
\r
231 seqs = new Vector();
\r
232 seqs.addElement(seq);
\r
238 seqs = new Vector();
\r
239 seqs.addElement(seq);
\r
242 seqRefs.put(key, seqs);
\r
250 if (dbSources == null)
\r
252 throw new Error("Implementation error. Must initialise dbSources");
\r
255 long startTime = System.currentTimeMillis();
\r
256 af.setProgressBar("Fetching db refs", startTime);
\r
258 picrClient = new uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperServiceLocator().getAccessionMapperPort();
\r
259 } catch (Exception e)
\r
261 System.err.println("Couldn't locate PICR service instance.\n");
\r
262 e.printStackTrace();
\r
265 Vector sdataset = new Vector();
\r
266 for (int s = 0; s < dataset.length; s++)
\r
268 sdataset.addElement(dataset[s]);
\r
270 while (sdataset.size() > 0 && db < dbSources.length)
\r
272 int maxqlen = 1; // default number of queries made to at one time
\r
273 System.err.println("Verifying against " + dbSources[db]);
\r
274 jalview.ws.seqfetcher.DbSourceProxy dbsource = sfetcher
\r
275 .getSourceProxy(dbSources[db]);
\r
276 if (dbsource == null)
\r
278 System.err.println("No proxy for " + dbSources[db]);
\r
282 if (dbsource.getDbSourceProperties()
\r
283 .containsKey(DBRefSource.MULTIACC))
\r
285 maxqlen = ((Integer) dbsource.getDbSourceProperties().get(
\r
286 DBRefSource.MULTIACC)).intValue();
\r
292 // iterate through db for each remaining un-verified sequence
\r
293 SequenceI[] currSeqs = new SequenceI[sdataset.size()];
\r
294 sdataset.copyInto(currSeqs);// seqs that are to be validated against
\r
296 Vector queries = new Vector(); // generated queries curSeq
\r
297 seqRefs = new Hashtable();
\r
301 while (queries.size() > 0 || seqIndex < currSeqs.length)
\r
303 if (queries.size() > 0)
\r
305 // Still queries to make for current seqIndex
\r
306 StringBuffer queryString = new StringBuffer("");
\r
307 int numq=0,nqSize = (maxqlen > queries.size()) ? queries.size()
\r
310 while (queries.size()>0 && numq < nqSize)
\r
312 String query = (String) queries.elementAt(0);
\r
313 if (dbsource.isValidReference(query))
\r
315 queryString.append((numq == 0) ? "" : dbsource
\r
316 .getAccessionSeparator());
\r
317 queryString.append(query);
\r
320 // remove the extracted query string
\r
321 queries.removeElementAt(0);
\r
323 // make the queries and process the response
\r
324 AlignmentI retrieved = null;
\r
327 if (jalview.bin.Cache.log.isDebugEnabled())
\r
329 jalview.bin.Cache.log.debug("Querying "+dbsource.getDbName()+" with : '"+queryString.toString()+"'");
\r
331 retrieved = dbsource.getSequenceRecords(queryString.toString());
\r
332 } catch (Exception ex)
\r
334 ex.printStackTrace();
\r
335 } catch (OutOfMemoryError err)
\r
337 new OOMWarning("retrieving database references ("
\r
338 + queryString.toString() + ")", err);
\r
340 if (retrieved != null)
\r
342 transferReferences(sdataset, dbSources[db], retrieved);
\r
347 // make some more strings for use as queries
\r
348 for (int i = 0; (seqIndex < dataset.length) && (i < 50); seqIndex++, i++)
\r
350 SequenceI sequence = dataset[seqIndex];
\r
351 DBRefEntry[] uprefs = jalview.util.DBRefUtils.selectRefs(
\r
352 sequence.getDBRef(), new String[]
\r
353 { dbSources[db] }); // jalview.datamodel.DBRefSource.UNIPROT
\r
355 // check for existing dbrefs to use
\r
356 if (uprefs != null && uprefs.length>0)
\r
358 for (int j = 0; j < uprefs.length; j++)
\r
360 addSeqId(sequence, uprefs[j].getAccessionId());
\r
362 .addElement(uprefs[j].getAccessionId()
\r
368 // generate queries from sequence ID string
\r
369 StringTokenizer st = new StringTokenizer(sequence.getName(),
\r
371 while (st.hasMoreTokens())
\r
373 String token = st.nextToken();
\r
374 UPEntry[] presp = null;
\r
376 presp=picrClient.getUPIForAccession(token, null, picrClient.getMappedDatabaseNames(), null, true);
\r
377 } catch (Exception e) {
\r
378 System.err.println("Exception with Picr for '"+token+"'\n");
\r
379 e.printStackTrace();
\r
381 if (presp!=null && presp.length>0)
\r
383 for (int id=0;id<presp.length; id++)
\r
385 // construct sequences from response if sequences are present, and do a transferReferences
\r
386 // otherwise transfer non sequence x-references directly.
\r
388 System.out.println("Validated ID against PICR... (for what its worth):"+token);
\r
389 addSeqId(sequence, token);
\r
390 queries.addElement(token.toUpperCase());
\r
393 //System.out.println("Not querying source with token="+token+"\n");
\r
394 addSeqId(sequence, token);
\r
395 queries.addElement(token.toUpperCase());
\r
402 // advance to next database
\r
404 } // all databases have been queries.
\r
405 if (sbuffer.length() > 0)
\r
408 .setText("Your sequences have been verified against known sequence databases. Some of the ids have been\n"
\r
409 + "altered, most likely the start/end residue will have been updated.\n"
\r
410 + "Save your alignment to maintain the updated id.\n\n"
\r
411 + sbuffer.toString());
\r
412 Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);
\r
413 // The above is the dataset, we must now find out the index
\r
414 // of the viewed sequence
\r
418 af.setProgressBar("DBRef search completed", startTime);
\r
419 // promptBeforeBlast();
\r
426 * Verify local sequences in seqRefs against the retrieved sequence database
\r
430 void transferReferences(Vector sdataset, String dbSource,
\r
431 AlignmentI retrievedAl) // File
\r
435 if (retrievedAl == null || retrievedAl.getHeight() == 0)
\r
439 SequenceI[] retrieved = recoverDbSequences(retrievedAl.getSequencesArray());
\r
440 SequenceI sequence = null;
\r
442 // Vector entries = new Uniprot().getUniprotEntries(file);
\r
444 int i, iSize = retrieved.length; // entries == null ? 0 : entries.size();
\r
445 // UniprotEntry entry;
\r
446 for (i = 0; i < iSize; i++)
\r
448 SequenceI entry = retrieved[i]; // (UniprotEntry) entries.elementAt(i);
\r
450 // Work out which sequences this sequence matches,
\r
451 // taking into account all accessionIds and names in the file
\r
452 Vector sequenceMatches = new Vector();
\r
453 // look for corresponding accession ids
\r
454 DBRefEntry[] entryRefs = jalview.util.DBRefUtils.selectRefs(entry
\r
455 .getDBRef(), new String[]
\r
457 for (int j = 0; j < entryRefs.length; j++)
\r
459 String accessionId = entryRefs[j].getAccessionId(); // .getAccession().elementAt(j).toString();
\r
460 // match up on accessionId
\r
461 if (seqRefs.containsKey(accessionId.toUpperCase()))
\r
463 Vector seqs = (Vector) seqRefs.get(accessionId);
\r
464 for (int jj = 0; jj < seqs.size(); jj++)
\r
466 sequence = (SequenceI) seqs.elementAt(jj);
\r
467 if (!sequenceMatches.contains(sequence))
\r
469 sequenceMatches.addElement(sequence);
\r
474 if (sequenceMatches.size() == 0)
\r
476 // failed to match directly on accessionId==query so just compare all
\r
477 // sequences to entry
\r
478 Enumeration e = seqRefs.keys();
\r
479 while (e.hasMoreElements())
\r
481 Vector sqs = (Vector) seqRefs.get(e.nextElement());
\r
482 if (sqs != null && sqs.size() > 0)
\r
484 Enumeration sqe = sqs.elements();
\r
485 while (sqe.hasMoreElements())
\r
487 sequenceMatches.addElement(sqe.nextElement());
\r
492 // look for corresponding names
\r
493 // this is uniprot specific ?
\r
494 // could be useful to extend this so we try to find any 'significant'
\r
495 // information in common between two sequence objects.
\r
497 * DBRefEntry[] entryRefs =
\r
498 * jalview.util.DBRefUtils.selectRefs(entry.getDBRef(), new String[] {
\r
499 * dbSource }); for (int j = 0; j < entry.getName().size(); j++) { String
\r
500 * name = entry.getName().elementAt(j).toString(); if
\r
501 * (seqRefs.containsKey(name)) { Vector seqs = (Vector) seqRefs.get(name);
\r
502 * for (int jj = 0; jj < seqs.size(); jj++) { sequence = (SequenceI)
\r
503 * seqs.elementAt(jj); if (!sequenceMatches.contains(sequence)) {
\r
504 * sequenceMatches.addElement(sequence); } } } }
\r
506 // sequenceMatches now contains the set of all sequences associated with
\r
507 // the returned db record
\r
508 String entrySeq = entry.getSequenceAsString().toUpperCase();
\r
509 for (int m = 0; m < sequenceMatches.size(); m++)
\r
511 sequence = (SequenceI) sequenceMatches.elementAt(m);
\r
512 // only update start and end positions and shift features if there are
\r
513 // no existing references
\r
514 // TODO: test for legacy where uniprot or EMBL refs exist but no
\r
515 // mappings are made (but content matches retrieved set)
\r
516 boolean updateRefFrame = sequence.getDBRef() == null
\r
517 || sequence.getDBRef().length == 0;
\r
518 // verify sequence against the entry sequence
\r
520 String nonGapped = AlignSeq.extractGaps("-. ",
\r
521 sequence.getSequenceAsString()).toUpperCase();
\r
523 int absStart = entrySeq.indexOf(nonGapped);
\r
524 int mapStart = entry.getStart();
\r
525 jalview.datamodel.Mapping mp;
\r
527 if (absStart == -1)
\r
529 // Is local sequence contained in dataset sequence?
\r
530 absStart = nonGapped.indexOf(entrySeq);
\r
531 if (absStart == -1)
\r
532 { // verification failed.
\r
533 sbuffer.append(sequence.getName()
\r
534 + " SEQUENCE NOT %100 MATCH \n");
\r
538 sbuffer.append(sequence.getName() + " HAS " + absStart
\r
539 + " PREFIXED RESIDUES COMPARED TO " + dbSource + "\n");
\r
541 // + " - ANY SEQUENCE FEATURES"
\r
542 // + " HAVE BEEN ADJUSTED ACCORDINGLY \n");
\r
544 // create valid mapping between matching region of local sequence and
\r
545 // the mapped sequence
\r
546 mp = new Mapping(null, new int[]
\r
547 { sequence.getStart() + absStart,
\r
548 sequence.getStart() + absStart + entrySeq.length() - 1 },
\r
550 { entry.getStart(),
\r
551 entry.getStart() + entrySeq.length() - 1 }, 1, 1);
\r
552 updateRefFrame = false; // mapping is based on current start/end so
\r
553 // don't modify start and end
\r
557 // update start and end of local sequence to place it in entry's
\r
558 // reference frame.
\r
559 // apply identity map map from whole of local sequence to matching
\r
560 // region of database
\r
562 mp = null; // Mapping.getIdentityMap();
\r
563 // new Mapping(null,
\r
564 // new int[] { absStart+sequence.getStart(),
\r
565 // absStart+sequence.getStart()+entrySeq.length()-1},
\r
566 // new int[] { entry.getStart(), entry.getEnd() }, 1, 1);
\r
567 // relocate local features for updated start
\r
568 if (updateRefFrame && sequence.getSequenceFeatures() != null)
\r
570 SequenceFeature[] sf = sequence.getSequenceFeatures();
\r
571 int start = sequence.getStart();
\r
572 int end = sequence.getEnd();
\r
573 int startShift = 1 - absStart - start; // how much the features are
\r
574 // to be shifted by
\r
575 for (int sfi = 0; sfi < sf.length; sfi++)
\r
577 if (sf[sfi].getBegin() >= start && sf[sfi].getEnd() <= end)
\r
579 // shift feature along by absstart
\r
580 sf[sfi].setBegin(sf[sfi].getBegin() + startShift);
\r
581 sf[sfi].setEnd(sf[sfi].getEnd() + startShift);
\r
587 System.out.println("Adding dbrefs to " + sequence.getName()
\r
588 + " from " + dbSource + " sequence : " + entry.getName());
\r
589 sequence.transferAnnotation(entry, mp);
\r
590 // unknownSequences.remove(sequence);
\r
591 int absEnd = absStart + nonGapped.length();
\r
593 if (updateRefFrame)
\r
595 // finally, update local sequence reference frame if we're allowed
\r
596 sequence.setStart(absStart);
\r
597 sequence.setEnd(absEnd);
\r
599 // and remove it from the rest
\r
600 // TODO: decide if we should remove annotated sequence from set
\r
601 sdataset.remove(sequence);
\r
602 // TODO: should we make a note of sequences that have received new DB ids, so we can query all enabled DAS servers for them ?
\r
608 * loop thru and collect additional sequences in Map.
\r
609 * @param sequencesArray
\r
612 private SequenceI[] recoverDbSequences(SequenceI[] sequencesArray)
\r
614 Vector nseq = new Vector();
\r
615 for (int i=0;sequencesArray!=null && i<sequencesArray.length;i++)
\r
617 nseq.addElement(sequencesArray[i]);
\r
618 DBRefEntry dbr[] = sequencesArray[i].getDBRef();
\r
619 jalview.datamodel.Mapping map = null;
\r
620 for (int r=0;(dbr!=null) && r<dbr.length; r++) {
\r
621 if ((map=dbr[r].getMap())!=null)
\r
623 if (!nseq.contains(map.getTo()))
\r
625 nseq.addElement(map.getTo());
\r
630 if (nseq.size()>0) {
\r
631 sequencesArray = new SequenceI[nseq.size()];
\r
632 nseq.toArray(sequencesArray);
\r
634 return sequencesArray;
\r