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.bin.Cache;
\r
23 import jalview.datamodel.AlignmentI;
\r
24 import jalview.datamodel.DBRefEntry;
\r
25 import jalview.datamodel.DBRefSource;
\r
26 import jalview.datamodel.Mapping;
\r
27 import jalview.datamodel.SequenceFeature;
\r
28 import jalview.datamodel.SequenceI;
\r
29 import jalview.gui.AlignFrame;
\r
30 import jalview.gui.CutAndPasteTransfer;
\r
31 import jalview.gui.Desktop;
\r
32 import jalview.gui.IProgressIndicator;
\r
33 import jalview.gui.OOMWarning;
\r
35 import java.lang.reflect.Array;
\r
36 import java.util.Enumeration;
\r
37 import java.util.Hashtable;
\r
38 import java.util.StringTokenizer;
\r
39 import java.util.Vector;
\r
41 import org.biojava.dasobert.dasregistry.DasSource;
\r
43 import uk.ac.ebi.picr.model.UPEntry;
\r
46 * Implements a runnable for validating a sequence against external databases
\r
47 * and then propagating references and features onto the sequence(s)
\r
50 * @version $Revision$
\r
52 public class DBRefFetcher implements Runnable
\r
54 SequenceI[] dataset;
\r
56 IProgressIndicator af;
\r
58 CutAndPasteTransfer output = new CutAndPasteTransfer();
\r
60 StringBuffer sbuffer = new StringBuffer();
\r
62 boolean running = false;
\r
65 * picr client instance
\r
67 uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperInterface picrClient = null;
\r
69 // /This will be a collection of Vectors of sequenceI refs.
\r
70 // The key will be the seq name or accession id of the seq
\r
75 SequenceFetcher sfetcher;
\r
77 private SequenceI[] alseqs;
\r
79 public DBRefFetcher()
\r
84 * Creates a new SequenceFeatureFetcher object and fetches from the currently
\r
85 * selected set of databases.
\r
88 * fetch references for these sequences
\r
90 * the parent alignframe for progress bar monitoring.
\r
92 public DBRefFetcher(SequenceI[] seqs, AlignFrame af)
\r
94 this(seqs, af, null);
\r
98 * Creates a new SequenceFeatureFetcher object and fetches from the currently
\r
99 * selected set of databases.
\r
102 * fetch references for these sequences
\r
104 * the parent alignframe for progress bar monitoring.
\r
106 * array of database source strings to query references from
\r
108 public DBRefFetcher(SequenceI[] seqs, AlignFrame af, String[] sources)
\r
111 alseqs = new SequenceI[seqs.length];
\r
112 SequenceI[] ds = new SequenceI[seqs.length];
\r
113 for (int i = 0; i < seqs.length; i++)
\r
115 alseqs[i] = seqs[i];
\r
116 if (seqs[i].getDatasetSequence() != null)
\r
117 ds[i] = seqs[i].getDatasetSequence();
\r
122 // TODO Jalview 2.5 lots of this code should be in the gui package!
\r
123 sfetcher = jalview.gui.SequenceFetcher.getSequenceFetcherSingleton(af);
\r
124 if (sources == null)
\r
126 // af.featureSettings_actionPerformed(null);
\r
127 String[] defdb = null, otherdb = sfetcher
\r
128 .getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
\r
129 Vector selsources = new Vector(), dasselsrc = (af.featureSettings != null) ? af.featureSettings
\r
130 .getSelectedSources()
\r
131 : new jalview.gui.DasSourceBrowser().getSelectedSources();
\r
132 Enumeration en = dasselsrc.elements();
\r
133 while (en.hasMoreElements())
\r
135 DasSource src = (DasSource) en.nextElement();
\r
136 selsources.addElement(src.getNickname());
\r
139 for (int o = 0; otherdb != null && o < otherdb.length; o++)
\r
141 if (!selsources.contains(otherdb[o]))
\r
150 // select appropriate databases based on alignFrame context.
\r
151 if (af.getViewport().getAlignment().isNucleotide())
\r
153 defdb = DBRefSource.DNACODINGDBS;
\r
157 defdb = DBRefSource.PROTEINDBS;
\r
159 // append the selected sequence sources to the default dbs
\r
160 dbSources = new String[defdb.length + osel];
\r
161 System.arraycopy(defdb, 0, dbSources, 0, defdb.length);
\r
162 for (int o = 0, op = defdb.length; otherdb != null
\r
163 && o < otherdb.length; o++)
\r
165 if (otherdb[o] != null)
\r
167 dbSources[op++] = otherdb[o];
\r
173 // we assume the caller knows what they're doing and ensured that all the
\r
174 // db source names are valid
\r
175 dbSources = sources;
\r
180 * retrieve all the das sequence sources and add them to the list of db
\r
181 * sources to retrieve from
\r
183 public void appendAllDasSources()
\r
185 if (dbSources == null)
\r
187 dbSources = new String[]
\r
190 // append additional sources
\r
191 String[] otherdb = sfetcher
\r
192 .getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
\r
193 if (otherdb != null && otherdb.length > 0)
\r
195 String[] newsrc = new String[dbSources.length + otherdb.length];
\r
196 System.arraycopy(dbSources, 0, newsrc, 0, dbSources.length);
\r
198 .arraycopy(otherdb, 0, newsrc, dbSources.length,
\r
200 dbSources = newsrc;
\r
205 * start the fetcher thread
\r
207 * @param waitTillFinished
\r
208 * true to block until the fetcher has finished
\r
210 public void fetchDBRefs(boolean waitTillFinished)
\r
212 Thread thread = new Thread(this);
\r
216 if (waitTillFinished)
\r
223 } catch (Exception ex)
\r
231 * The sequence will be added to a vector of sequences belonging to key which
\r
232 * could be either seq name or dbref id
\r
239 void addSeqId(SequenceI seq, String key)
\r
241 key = key.toUpperCase();
\r
244 if (seqRefs.containsKey(key))
\r
246 seqs = (Vector) seqRefs.get(key);
\r
248 if (seqs != null && !seqs.contains(seq))
\r
250 seqs.addElement(seq);
\r
252 else if (seqs == null)
\r
254 seqs = new Vector();
\r
255 seqs.addElement(seq);
\r
261 seqs = new Vector();
\r
262 seqs.addElement(seq);
\r
265 seqRefs.put(key, seqs);
\r
273 if (dbSources == null)
\r
275 throw new Error("Implementation error. Must initialise dbSources");
\r
278 long startTime = System.currentTimeMillis();
\r
279 af.setProgressBar("Fetching db refs", startTime);
\r
282 if (Cache.getDefault("DBREFFETCH_USEPICR", false))
\r
284 picrClient = new uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperServiceLocator()
\r
285 .getAccessionMapperPort();
\r
287 } catch (Exception e)
\r
289 System.err.println("Couldn't locate PICR service instance.\n");
\r
290 e.printStackTrace();
\r
293 Vector sdataset = new Vector();
\r
294 for (int s = 0; s < dataset.length; s++)
\r
296 sdataset.addElement(dataset[s]);
\r
298 while (sdataset.size() > 0 && db < dbSources.length)
\r
300 int maxqlen = 1; // default number of queries made to at one time
\r
301 System.err.println("Verifying against " + dbSources[db]);
\r
302 jalview.ws.seqfetcher.DbSourceProxy dbsource = sfetcher
\r
303 .getSourceProxy(dbSources[db]);
\r
304 if (dbsource == null)
\r
306 System.err.println("No proxy for " + dbSources[db]);
\r
310 if (dbsource.getDbSourceProperties()
\r
311 .containsKey(DBRefSource.MULTIACC))
\r
313 maxqlen = ((Integer) dbsource.getDbSourceProperties().get(
\r
314 DBRefSource.MULTIACC)).intValue();
\r
320 // iterate through db for each remaining un-verified sequence
\r
321 SequenceI[] currSeqs = new SequenceI[sdataset.size()];
\r
322 sdataset.copyInto(currSeqs);// seqs that are to be validated against
\r
324 Vector queries = new Vector(); // generated queries curSeq
\r
325 seqRefs = new Hashtable();
\r
329 while (queries.size() > 0 || seqIndex < currSeqs.length)
\r
331 if (queries.size() > 0)
\r
333 // Still queries to make for current seqIndex
\r
334 StringBuffer queryString = new StringBuffer("");
\r
335 int numq = 0, nqSize = (maxqlen > queries.size()) ? queries
\r
338 while (queries.size() > 0 && numq < nqSize)
\r
340 String query = (String) queries.elementAt(0);
\r
341 if (dbsource.isValidReference(query))
\r
343 queryString.append((numq == 0) ? "" : dbsource
\r
344 .getAccessionSeparator());
\r
345 queryString.append(query);
\r
348 // remove the extracted query string
\r
349 queries.removeElementAt(0);
\r
351 // make the queries and process the response
\r
352 AlignmentI retrieved = null;
\r
355 if (jalview.bin.Cache.log.isDebugEnabled())
\r
357 jalview.bin.Cache.log.debug("Querying "
\r
358 + dbsource.getDbName() + " with : '"
\r
359 + queryString.toString() + "'");
\r
361 retrieved = dbsource.getSequenceRecords(queryString.toString());
\r
362 } catch (Exception ex)
\r
364 ex.printStackTrace();
\r
365 } catch (OutOfMemoryError err)
\r
367 new OOMWarning("retrieving database references ("
\r
368 + queryString.toString() + ")", err);
\r
370 if (retrieved != null)
\r
372 transferReferences(sdataset, dbSources[db], retrieved);
\r
377 // make some more strings for use as queries
\r
378 for (int i = 0; (seqIndex < dataset.length) && (i < 50); seqIndex++, i++)
\r
380 SequenceI sequence = dataset[seqIndex];
\r
381 DBRefEntry[] uprefs = jalview.util.DBRefUtils.selectRefs(
\r
382 sequence.getDBRef(), new String[]
\r
383 { dbSources[db] }); // jalview.datamodel.DBRefSource.UNIPROT
\r
385 // check for existing dbrefs to use
\r
386 if (uprefs != null && uprefs.length > 0)
\r
388 for (int j = 0; j < uprefs.length; j++)
\r
390 addSeqId(sequence, uprefs[j].getAccessionId());
\r
392 .addElement(uprefs[j].getAccessionId()
\r
398 // generate queries from sequence ID string
\r
399 StringTokenizer st = new StringTokenizer(sequence.getName(),
\r
401 while (st.hasMoreTokens())
\r
403 String token = st.nextToken();
\r
404 UPEntry[] presp = null;
\r
405 if (picrClient != null)
\r
407 // resolve the string against PICR to recover valid IDs
\r
411 .getUPIForAccession(token, null, picrClient
\r
412 .getMappedDatabaseNames(), null, true);
\r
413 } catch (Exception e)
\r
415 System.err.println("Exception with Picr for '" + token
\r
417 e.printStackTrace();
\r
420 if (presp != null && presp.length > 0)
\r
422 for (int id = 0; id < presp.length; id++)
\r
424 // construct sequences from response if sequences are
\r
425 // present, and do a transferReferences
\r
426 // otherwise transfer non sequence x-references directly.
\r
429 .println("Validated ID against PICR... (for what its worth):"
\r
431 addSeqId(sequence, token);
\r
432 queries.addElement(token.toUpperCase());
\r
437 // System.out.println("Not querying source with token="+token+"\n");
\r
438 addSeqId(sequence, token);
\r
439 queries.addElement(token.toUpperCase());
\r
446 // advance to next database
\r
448 } // all databases have been queries.
\r
449 if (sbuffer.length() > 0)
\r
452 .setText("Your sequences have been verified against known sequence databases. Some of the ids have been\n"
\r
453 + "altered, most likely the start/end residue will have been updated.\n"
\r
454 + "Save your alignment to maintain the updated id.\n\n"
\r
455 + sbuffer.toString());
\r
456 Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);
\r
457 // The above is the dataset, we must now find out the index
\r
458 // of the viewed sequence
\r
462 af.setProgressBar("DBRef search completed", startTime);
\r
463 // promptBeforeBlast();
\r
470 * Verify local sequences in seqRefs against the retrieved sequence database
\r
474 void transferReferences(Vector sdataset, String dbSource,
\r
475 AlignmentI retrievedAl) // File
\r
478 if (retrievedAl == null || retrievedAl.getHeight() == 0)
\r
482 SequenceI[] retrieved = recoverDbSequences(retrievedAl
\r
483 .getSequencesArray());
\r
484 SequenceI sequence = null;
\r
485 boolean transferred = false;
\r
486 StringBuffer messages = new StringBuffer();
\r
488 // Vector entries = new Uniprot().getUniprotEntries(file);
\r
490 int i, iSize = retrieved.length; // entries == null ? 0 : entries.size();
\r
491 // UniprotEntry entry;
\r
492 for (i = 0; i < iSize; i++)
\r
494 SequenceI entry = retrieved[i]; // (UniprotEntry) entries.elementAt(i);
\r
496 // Work out which sequences this sequence matches,
\r
497 // taking into account all accessionIds and names in the file
\r
498 Vector sequenceMatches = new Vector();
\r
499 // look for corresponding accession ids
\r
500 DBRefEntry[] entryRefs = jalview.util.DBRefUtils.selectRefs(entry
\r
501 .getDBRef(), new String[]
\r
503 for (int j = 0; j < entryRefs.length; j++)
\r
505 String accessionId = entryRefs[j].getAccessionId(); // .getAccession().elementAt(j).toString();
\r
506 // match up on accessionId
\r
507 if (seqRefs.containsKey(accessionId.toUpperCase()))
\r
509 Vector seqs = (Vector) seqRefs.get(accessionId);
\r
510 for (int jj = 0; jj < seqs.size(); jj++)
\r
512 sequence = (SequenceI) seqs.elementAt(jj);
\r
513 if (!sequenceMatches.contains(sequence))
\r
515 sequenceMatches.addElement(sequence);
\r
520 if (sequenceMatches.size() == 0)
\r
522 // failed to match directly on accessionId==query so just compare all
\r
523 // sequences to entry
\r
524 Enumeration e = seqRefs.keys();
\r
525 while (e.hasMoreElements())
\r
527 Vector sqs = (Vector) seqRefs.get(e.nextElement());
\r
528 if (sqs != null && sqs.size() > 0)
\r
530 Enumeration sqe = sqs.elements();
\r
531 while (sqe.hasMoreElements())
\r
533 sequenceMatches.addElement(sqe.nextElement());
\r
538 // look for corresponding names
\r
539 // this is uniprot specific ?
\r
540 // could be useful to extend this so we try to find any 'significant'
\r
541 // information in common between two sequence objects.
\r
543 * DBRefEntry[] entryRefs =
\r
544 * jalview.util.DBRefUtils.selectRefs(entry.getDBRef(), new String[] {
\r
545 * dbSource }); for (int j = 0; j < entry.getName().size(); j++) { String
\r
546 * name = entry.getName().elementAt(j).toString(); if
\r
547 * (seqRefs.containsKey(name)) { Vector seqs = (Vector) seqRefs.get(name);
\r
548 * for (int jj = 0; jj < seqs.size(); jj++) { sequence = (SequenceI)
\r
549 * seqs.elementAt(jj); if (!sequenceMatches.contains(sequence)) {
\r
550 * sequenceMatches.addElement(sequence); } } } }
\r
552 // sequenceMatches now contains the set of all sequences associated with
\r
553 // the returned db record
\r
554 String entrySeq = entry.getSequenceAsString().toUpperCase();
\r
555 for (int m = 0; m < sequenceMatches.size(); m++)
\r
557 sequence = (SequenceI) sequenceMatches.elementAt(m);
\r
558 // only update start and end positions and shift features if there are
\r
559 // no existing references
\r
560 // TODO: test for legacy where uniprot or EMBL refs exist but no
\r
561 // mappings are made (but content matches retrieved set)
\r
562 boolean updateRefFrame = sequence.getDBRef() == null
\r
563 || sequence.getDBRef().length == 0;
\r
564 // verify sequence against the entry sequence
\r
566 String nonGapped = AlignSeq.extractGaps("-. ",
\r
567 sequence.getSequenceAsString()).toUpperCase();
\r
569 int absStart = entrySeq.indexOf(nonGapped);
\r
570 int mapStart = entry.getStart();
\r
571 jalview.datamodel.Mapping mp;
\r
573 if (absStart == -1)
\r
575 // Is local sequence contained in dataset sequence?
\r
576 absStart = nonGapped.indexOf(entrySeq);
\r
577 if (absStart == -1)
\r
578 { // verification failed.
\r
579 messages.append(sequence.getName()
\r
580 + " SEQUENCE NOT %100 MATCH \n");
\r
583 transferred = true;
\r
584 sbuffer.append(sequence.getName() + " HAS " + absStart
\r
585 + " PREFIXED RESIDUES COMPARED TO " + dbSource + "\n");
\r
587 // + " - ANY SEQUENCE FEATURES"
\r
588 // + " HAVE BEEN ADJUSTED ACCORDINGLY \n");
\r
590 // create valid mapping between matching region of local sequence and
\r
591 // the mapped sequence
\r
592 mp = new Mapping(null, new int[]
\r
593 { sequence.getStart() + absStart,
\r
594 sequence.getStart() + absStart + entrySeq.length() - 1 },
\r
596 { entry.getStart(),
\r
597 entry.getStart() + entrySeq.length() - 1 }, 1, 1);
\r
598 updateRefFrame = false; // mapping is based on current start/end so
\r
599 // don't modify start and end
\r
603 transferred = true;
\r
604 // update start and end of local sequence to place it in entry's
\r
605 // reference frame.
\r
606 // apply identity map map from whole of local sequence to matching
\r
607 // region of database
\r
609 mp = null; // Mapping.getIdentityMap();
\r
610 // new Mapping(null,
\r
611 // new int[] { absStart+sequence.getStart(),
\r
612 // absStart+sequence.getStart()+entrySeq.length()-1},
\r
613 // new int[] { entry.getStart(), entry.getEnd() }, 1, 1);
\r
614 // relocate local features for updated start
\r
615 if (updateRefFrame) {
\r
616 if (sequence.getSequenceFeatures() != null)
\r
618 SequenceFeature[] sf = sequence.getSequenceFeatures();
\r
619 int start = sequence.getStart();
\r
620 int end = sequence.getEnd();
\r
621 int startShift = 1 - absStart - start; // how much the features are
\r
622 // to be shifted by
\r
623 for (int sfi = 0; sfi < sf.length; sfi++)
\r
625 if (sf[sfi].getBegin() >= start && sf[sfi].getEnd() <= end)
\r
627 // shift feature along by absstart
\r
628 sf[sfi].setBegin(sf[sfi].getBegin() + startShift);
\r
629 sf[sfi].setEnd(sf[sfi].getEnd() + startShift);
\r
636 System.out.println("Adding dbrefs to " + sequence.getName()
\r
637 + " from " + dbSource + " sequence : " + entry.getName());
\r
638 sequence.transferAnnotation(entry, mp);
\r
639 // unknownSequences.remove(sequence);
\r
640 int absEnd = absStart + nonGapped.length();
\r
642 if (updateRefFrame)
\r
644 // finally, update local sequence reference frame if we're allowed
\r
645 sequence.setStart(absStart);
\r
646 sequence.setEnd(absEnd);
\r
647 // search for alignment sequences to update coordinate frame for
\r
648 for (int alsq = 0; alsq<alseqs.length; alsq++)
\r
650 if (alseqs[alsq].getDatasetSequence()==sequence)
\r
652 String ngAlsq = AlignSeq.extractGaps("-. ",
\r
653 alseqs[alsq].getSequenceAsString()).toUpperCase();
\r
654 int oldstrt = alseqs[alsq].getStart();
\r
655 alseqs[alsq].setStart(sequence.getSequenceAsString().toUpperCase().indexOf(ngAlsq)+sequence.getStart());
\r
656 if (oldstrt != alseqs[alsq].getStart()) {
\r
657 alseqs[alsq].setEnd(ngAlsq.length()+alseqs[alsq].getStart()-1);
\r
661 // TODO: search for all other references to this dataset sequence, and update start/end
\r
662 // TODO: update all AlCodonMappings which involve this alignment sequence (e.g. Q30167 cdna translation from exon2 product (vamsas demo)
\r
664 // and remove it from the rest
\r
665 // TODO: decide if we should remove annotated sequence from set
\r
666 sdataset.remove(sequence);
\r
667 // TODO: should we make a note of sequences that have received new DB
\r
668 // ids, so we can query all enabled DAS servers for them ?
\r
673 // report the ID/sequence mismatches
\r
674 sbuffer.append(messages);
\r
679 * loop thru and collect additional sequences in Map.
\r
681 * @param sequencesArray
\r
684 private SequenceI[] recoverDbSequences(SequenceI[] sequencesArray)
\r
686 Vector nseq = new Vector();
\r
687 for (int i = 0; sequencesArray != null && i < sequencesArray.length; i++)
\r
689 nseq.addElement(sequencesArray[i]);
\r
690 DBRefEntry dbr[] = sequencesArray[i].getDBRef();
\r
691 jalview.datamodel.Mapping map = null;
\r
692 for (int r = 0; (dbr != null) && r < dbr.length; r++)
\r
694 if ((map = dbr[r].getMap()) != null)
\r
696 if (map.getTo() != null && !nseq.contains(map.getTo()))
\r
698 nseq.addElement(map.getTo());
\r
703 if (nseq.size() > 0)
\r
705 sequencesArray = new SequenceI[nseq.size()];
\r
706 nseq.toArray(sequencesArray);
\r
708 return sequencesArray;
\r