2 * Jalview - A Sequence Alignment Editor and Viewer (Version 2.5)
\r
3 * Copyright (C) 2010 J Procter, AM Waterhouse, G Barton, M Clamp, S Searle
\r
5 * This file is part of Jalview.
\r
7 * Jalview is free software: you can redistribute it and/or
\r
8 * modify it under the terms of the GNU General Public License
\r
9 * as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
\r
11 * Jalview is distributed in the hope that it will be useful, but
\r
12 * WITHOUT ANY WARRANTY; without even the implied warranty
\r
13 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
\r
14 * PURPOSE. See the GNU General Public License for more details.
\r
16 * You should have received a copy of the GNU General Public License along with Jalview. If not, see <http://www.gnu.org/licenses/>.
\r
20 import jalview.analysis.AlignSeq;
\r
21 import jalview.bin.Cache;
\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
64 * picr client instance
\r
66 uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperInterface picrClient = null;
\r
68 // /This will be a collection of Vectors of sequenceI refs.
\r
69 // The key will be the seq name or accession id of the seq
\r
74 SequenceFetcher sfetcher;
\r
76 private SequenceI[] alseqs;
\r
78 public DBRefFetcher()
\r
83 * Creates a new SequenceFeatureFetcher object and fetches from the currently
\r
84 * selected set of databases.
\r
87 * fetch references for these sequences
\r
89 * the parent alignframe for progress bar monitoring.
\r
91 public DBRefFetcher(SequenceI[] seqs, AlignFrame af)
\r
93 this(seqs, af, null);
\r
97 * Creates a new SequenceFeatureFetcher object and fetches from the currently
\r
98 * selected set of databases.
\r
101 * fetch references for these sequences
\r
103 * the parent alignframe for progress bar monitoring.
\r
105 * array of database source strings to query references from
\r
107 public DBRefFetcher(SequenceI[] seqs, AlignFrame af, String[] sources)
\r
110 alseqs = new SequenceI[seqs.length];
\r
111 SequenceI[] ds = new SequenceI[seqs.length];
\r
112 for (int i = 0; i < seqs.length; i++)
\r
114 alseqs[i] = seqs[i];
\r
115 if (seqs[i].getDatasetSequence() != null)
\r
116 ds[i] = seqs[i].getDatasetSequence();
\r
121 // TODO Jalview 2.5 lots of this code should be in the gui package!
\r
122 sfetcher = jalview.gui.SequenceFetcher.getSequenceFetcherSingleton(af);
\r
123 if (sources == null)
\r
125 // af.featureSettings_actionPerformed(null);
\r
126 String[] defdb = null, otherdb = sfetcher
\r
127 .getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
\r
128 Vector selsources = new Vector(), dasselsrc = (af.featureSettings != null) ? af.featureSettings
\r
129 .getSelectedSources()
\r
130 : new jalview.gui.DasSourceBrowser().getSelectedSources();
\r
131 Enumeration en = dasselsrc.elements();
\r
132 while (en.hasMoreElements())
\r
134 DasSource src = (DasSource) en.nextElement();
\r
135 selsources.addElement(src.getNickname());
\r
138 for (int o = 0; otherdb != null && o < otherdb.length; o++)
\r
140 if (!selsources.contains(otherdb[o]))
\r
149 // select appropriate databases based on alignFrame context.
\r
150 if (af.getViewport().getAlignment().isNucleotide())
\r
152 defdb = DBRefSource.DNACODINGDBS;
\r
156 defdb = DBRefSource.PROTEINDBS;
\r
158 // append the selected sequence sources to the default dbs
\r
159 dbSources = new String[defdb.length + osel];
\r
160 System.arraycopy(defdb, 0, dbSources, 0, defdb.length);
\r
161 for (int o = 0, op = defdb.length; otherdb != null
\r
162 && o < otherdb.length; o++)
\r
164 if (otherdb[o] != null)
\r
166 dbSources[op++] = otherdb[o];
\r
172 // we assume the caller knows what they're doing and ensured that all the
\r
173 // db source names are valid
\r
174 dbSources = sources;
\r
179 * retrieve all the das sequence sources and add them to the list of db
\r
180 * sources to retrieve from
\r
182 public void appendAllDasSources()
\r
184 if (dbSources == null)
\r
186 dbSources = new String[]
\r
189 // append additional sources
\r
190 String[] otherdb = sfetcher
\r
191 .getDbInstances(jalview.ws.dbsources.DasSequenceSource.class);
\r
192 if (otherdb != null && otherdb.length > 0)
\r
194 String[] newsrc = new String[dbSources.length + otherdb.length];
\r
195 System.arraycopy(dbSources, 0, newsrc, 0, dbSources.length);
\r
197 .arraycopy(otherdb, 0, newsrc, dbSources.length,
\r
199 dbSources = newsrc;
\r
204 * start the fetcher thread
\r
206 * @param waitTillFinished
\r
207 * true to block until the fetcher has finished
\r
209 public void fetchDBRefs(boolean waitTillFinished)
\r
211 Thread thread = new Thread(this);
\r
215 if (waitTillFinished)
\r
222 } catch (Exception ex)
\r
230 * The sequence will be added to a vector of sequences belonging to key which
\r
231 * could be either seq name or dbref id
\r
238 void addSeqId(SequenceI seq, String key)
\r
240 key = key.toUpperCase();
\r
243 if (seqRefs.containsKey(key))
\r
245 seqs = (Vector) seqRefs.get(key);
\r
247 if (seqs != null && !seqs.contains(seq))
\r
249 seqs.addElement(seq);
\r
251 else if (seqs == null)
\r
253 seqs = new Vector();
\r
254 seqs.addElement(seq);
\r
260 seqs = new Vector();
\r
261 seqs.addElement(seq);
\r
264 seqRefs.put(key, seqs);
\r
272 if (dbSources == null)
\r
274 throw new Error("Implementation error. Must initialise dbSources");
\r
277 long startTime = System.currentTimeMillis();
\r
278 af.setProgressBar("Fetching db refs", startTime);
\r
281 if (Cache.getDefault("DBREFFETCH_USEPICR", false))
\r
283 picrClient = new uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperServiceLocator()
\r
284 .getAccessionMapperPort();
\r
286 } catch (Exception e)
\r
288 System.err.println("Couldn't locate PICR service instance.\n");
\r
289 e.printStackTrace();
\r
292 Vector sdataset = new Vector();
\r
293 for (int s = 0; s < dataset.length; s++)
\r
295 sdataset.addElement(dataset[s]);
\r
297 while (sdataset.size() > 0 && db < dbSources.length)
\r
299 int maxqlen = 1; // default number of queries made to at one time
\r
300 System.err.println("Verifying against " + dbSources[db]);
\r
301 jalview.ws.seqfetcher.DbSourceProxy dbsource = sfetcher
\r
302 .getSourceProxy(dbSources[db]);
\r
303 if (dbsource == null)
\r
305 System.err.println("No proxy for " + dbSources[db]);
\r
309 if (dbsource.getDbSourceProperties()
\r
310 .containsKey(DBRefSource.MULTIACC))
\r
312 maxqlen = ((Integer) dbsource.getDbSourceProperties().get(
\r
313 DBRefSource.MULTIACC)).intValue();
\r
319 // iterate through db for each remaining un-verified sequence
\r
320 SequenceI[] currSeqs = new SequenceI[sdataset.size()];
\r
321 sdataset.copyInto(currSeqs);// seqs that are to be validated against
\r
323 Vector queries = new Vector(); // generated queries curSeq
\r
324 seqRefs = new Hashtable();
\r
328 while (queries.size() > 0 || seqIndex < currSeqs.length)
\r
330 if (queries.size() > 0)
\r
332 // Still queries to make for current seqIndex
\r
333 StringBuffer queryString = new StringBuffer("");
\r
334 int numq = 0, nqSize = (maxqlen > queries.size()) ? queries
\r
337 while (queries.size() > 0 && numq < nqSize)
\r
339 String query = (String) queries.elementAt(0);
\r
340 if (dbsource.isValidReference(query))
\r
342 queryString.append((numq == 0) ? "" : dbsource
\r
343 .getAccessionSeparator());
\r
344 queryString.append(query);
\r
347 // remove the extracted query string
\r
348 queries.removeElementAt(0);
\r
350 // make the queries and process the response
\r
351 AlignmentI retrieved = null;
\r
354 if (jalview.bin.Cache.log.isDebugEnabled())
\r
356 jalview.bin.Cache.log.debug("Querying "
\r
357 + dbsource.getDbName() + " with : '"
\r
358 + queryString.toString() + "'");
\r
360 retrieved = dbsource.getSequenceRecords(queryString.toString());
\r
361 } catch (Exception ex)
\r
363 ex.printStackTrace();
\r
364 } catch (OutOfMemoryError err)
\r
366 new OOMWarning("retrieving database references ("
\r
367 + queryString.toString() + ")", err);
\r
369 if (retrieved != null)
\r
371 transferReferences(sdataset, dbSources[db], retrieved);
\r
376 // make some more strings for use as queries
\r
377 for (int i = 0; (seqIndex < dataset.length) && (i < 50); seqIndex++, i++)
\r
379 SequenceI sequence = dataset[seqIndex];
\r
380 DBRefEntry[] uprefs = jalview.util.DBRefUtils.selectRefs(
\r
381 sequence.getDBRef(), new String[]
\r
382 { dbSources[db] }); // jalview.datamodel.DBRefSource.UNIPROT
\r
384 // check for existing dbrefs to use
\r
385 if (uprefs != null && uprefs.length > 0)
\r
387 for (int j = 0; j < uprefs.length; j++)
\r
389 addSeqId(sequence, uprefs[j].getAccessionId());
\r
391 .addElement(uprefs[j].getAccessionId()
\r
397 // generate queries from sequence ID string
\r
398 StringTokenizer st = new StringTokenizer(sequence.getName(),
\r
400 while (st.hasMoreTokens())
\r
402 String token = st.nextToken();
\r
403 UPEntry[] presp = null;
\r
404 if (picrClient != null)
\r
406 // resolve the string against PICR to recover valid IDs
\r
410 .getUPIForAccession(token, null, picrClient
\r
411 .getMappedDatabaseNames(), null, true);
\r
412 } catch (Exception e)
\r
414 System.err.println("Exception with Picr for '" + token
\r
416 e.printStackTrace();
\r
419 if (presp != null && presp.length > 0)
\r
421 for (int id = 0; id < presp.length; id++)
\r
423 // construct sequences from response if sequences are
\r
424 // present, and do a transferReferences
\r
425 // otherwise transfer non sequence x-references directly.
\r
428 .println("Validated ID against PICR... (for what its worth):"
\r
430 addSeqId(sequence, token);
\r
431 queries.addElement(token.toUpperCase());
\r
436 // System.out.println("Not querying source with token="+token+"\n");
\r
437 addSeqId(sequence, token);
\r
438 queries.addElement(token.toUpperCase());
\r
445 // advance to next database
\r
447 } // all databases have been queries.
\r
448 if (sbuffer.length() > 0)
\r
451 .setText("Your sequences have been verified against known sequence databases. Some of the ids have been\n"
\r
452 + "altered, most likely the start/end residue will have been updated.\n"
\r
453 + "Save your alignment to maintain the updated id.\n\n"
\r
454 + sbuffer.toString());
\r
455 Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);
\r
456 // The above is the dataset, we must now find out the index
\r
457 // of the viewed sequence
\r
461 af.setProgressBar("DBRef search completed", startTime);
\r
462 // promptBeforeBlast();
\r
469 * Verify local sequences in seqRefs against the retrieved sequence database
\r
473 void transferReferences(Vector sdataset, String dbSource,
\r
474 AlignmentI retrievedAl) // File
\r
477 if (retrievedAl == null || retrievedAl.getHeight() == 0)
\r
481 SequenceI[] retrieved = recoverDbSequences(retrievedAl
\r
482 .getSequencesArray());
\r
483 SequenceI sequence = null;
\r
484 boolean transferred = false;
\r
485 StringBuffer messages = new StringBuffer();
\r
487 // Vector entries = new Uniprot().getUniprotEntries(file);
\r
489 int i, iSize = retrieved.length; // entries == null ? 0 : entries.size();
\r
490 // UniprotEntry entry;
\r
491 for (i = 0; i < iSize; i++)
\r
493 SequenceI entry = retrieved[i]; // (UniprotEntry) entries.elementAt(i);
\r
495 // Work out which sequences this sequence matches,
\r
496 // taking into account all accessionIds and names in the file
\r
497 Vector sequenceMatches = new Vector();
\r
498 // look for corresponding accession ids
\r
499 DBRefEntry[] entryRefs = jalview.util.DBRefUtils.selectRefs(entry
\r
500 .getDBRef(), new String[]
\r
502 if (entryRefs == null)
\r
505 .println("Dud dbSource string ? no entryrefs selected for "
\r
506 + dbSource + " on " + entry.getName());
\r
509 for (int j = 0; j < entryRefs.length; j++)
\r
511 String accessionId = entryRefs[j].getAccessionId(); // .getAccession().elementAt(j).toString();
\r
512 // match up on accessionId
\r
513 if (seqRefs.containsKey(accessionId.toUpperCase()))
\r
515 Vector seqs = (Vector) seqRefs.get(accessionId);
\r
516 for (int jj = 0; jj < seqs.size(); jj++)
\r
518 sequence = (SequenceI) seqs.elementAt(jj);
\r
519 if (!sequenceMatches.contains(sequence))
\r
521 sequenceMatches.addElement(sequence);
\r
526 if (sequenceMatches.size() == 0)
\r
528 // failed to match directly on accessionId==query so just compare all
\r
529 // sequences to entry
\r
530 Enumeration e = seqRefs.keys();
\r
531 while (e.hasMoreElements())
\r
533 Vector sqs = (Vector) seqRefs.get(e.nextElement());
\r
534 if (sqs != null && sqs.size() > 0)
\r
536 Enumeration sqe = sqs.elements();
\r
537 while (sqe.hasMoreElements())
\r
539 sequenceMatches.addElement(sqe.nextElement());
\r
544 // look for corresponding names
\r
545 // this is uniprot specific ?
\r
546 // could be useful to extend this so we try to find any 'significant'
\r
547 // information in common between two sequence objects.
\r
549 * DBRefEntry[] entryRefs =
\r
550 * jalview.util.DBRefUtils.selectRefs(entry.getDBRef(), new String[] {
\r
551 * dbSource }); for (int j = 0; j < entry.getName().size(); j++) { String
\r
552 * name = entry.getName().elementAt(j).toString(); if
\r
553 * (seqRefs.containsKey(name)) { Vector seqs = (Vector) seqRefs.get(name);
\r
554 * for (int jj = 0; jj < seqs.size(); jj++) { sequence = (SequenceI)
\r
555 * seqs.elementAt(jj); if (!sequenceMatches.contains(sequence)) {
\r
556 * sequenceMatches.addElement(sequence); } } } }
\r
558 // sequenceMatches now contains the set of all sequences associated with
\r
559 // the returned db record
\r
560 String entrySeq = entry.getSequenceAsString().toUpperCase();
\r
561 for (int m = 0; m < sequenceMatches.size(); m++)
\r
563 sequence = (SequenceI) sequenceMatches.elementAt(m);
\r
564 // only update start and end positions and shift features if there are
\r
565 // no existing references
\r
566 // TODO: test for legacy where uniprot or EMBL refs exist but no
\r
567 // mappings are made (but content matches retrieved set)
\r
568 boolean updateRefFrame = sequence.getDBRef() == null
\r
569 || sequence.getDBRef().length == 0;
\r
570 // verify sequence against the entry sequence
\r
572 String nonGapped = AlignSeq.extractGaps("-. ",
\r
573 sequence.getSequenceAsString()).toUpperCase();
\r
575 int absStart = entrySeq.indexOf(nonGapped);
\r
576 int mapStart = entry.getStart();
\r
577 jalview.datamodel.Mapping mp;
\r
579 if (absStart == -1)
\r
581 // Is local sequence contained in dataset sequence?
\r
582 absStart = nonGapped.indexOf(entrySeq);
\r
583 if (absStart == -1)
\r
584 { // verification failed.
\r
585 messages.append(sequence.getName()
\r
586 + " SEQUENCE NOT %100 MATCH \n");
\r
589 transferred = true;
\r
590 sbuffer.append(sequence.getName() + " HAS " + absStart
\r
591 + " PREFIXED RESIDUES COMPARED TO " + dbSource + "\n");
\r
593 // + " - ANY SEQUENCE FEATURES"
\r
594 // + " HAVE BEEN ADJUSTED ACCORDINGLY \n");
\r
596 // create valid mapping between matching region of local sequence and
\r
597 // the mapped sequence
\r
598 mp = new Mapping(null, new int[]
\r
599 { sequence.getStart() + absStart,
\r
600 sequence.getStart() + absStart + entrySeq.length() - 1 },
\r
602 { entry.getStart(),
\r
603 entry.getStart() + entrySeq.length() - 1 }, 1, 1);
\r
604 updateRefFrame = false; // mapping is based on current start/end so
\r
605 // don't modify start and end
\r
609 transferred = true;
\r
610 // update start and end of local sequence to place it in entry's
\r
611 // reference frame.
\r
612 // apply identity map map from whole of local sequence to matching
\r
613 // region of database
\r
615 mp = null; // Mapping.getIdentityMap();
\r
616 // new Mapping(null,
\r
617 // new int[] { absStart+sequence.getStart(),
\r
618 // absStart+sequence.getStart()+entrySeq.length()-1},
\r
619 // new int[] { entry.getStart(), entry.getEnd() }, 1, 1);
\r
620 // relocate local features for updated start
\r
621 if (updateRefFrame)
\r
623 if (sequence.getSequenceFeatures() != null)
\r
625 SequenceFeature[] sf = sequence.getSequenceFeatures();
\r
626 int start = sequence.getStart();
\r
627 int end = sequence.getEnd();
\r
628 int startShift = 1 - absStart - start; // how much the features
\r
630 // to be shifted by
\r
631 for (int sfi = 0; sfi < sf.length; sfi++)
\r
633 if (sf[sfi].getBegin() >= start && sf[sfi].getEnd() <= end)
\r
635 // shift feature along by absstart
\r
636 sf[sfi].setBegin(sf[sfi].getBegin() + startShift);
\r
637 sf[sfi].setEnd(sf[sfi].getEnd() + startShift);
\r
644 System.out.println("Adding dbrefs to " + sequence.getName()
\r
645 + " from " + dbSource + " sequence : " + entry.getName());
\r
646 sequence.transferAnnotation(entry, mp);
\r
647 // unknownSequences.remove(sequence);
\r
648 int absEnd = absStart + nonGapped.length();
\r
650 if (updateRefFrame)
\r
652 // finally, update local sequence reference frame if we're allowed
\r
653 sequence.setStart(absStart);
\r
654 sequence.setEnd(absEnd);
\r
655 // search for alignment sequences to update coordinate frame for
\r
656 for (int alsq = 0; alsq < alseqs.length; alsq++)
\r
658 if (alseqs[alsq].getDatasetSequence() == sequence)
\r
660 String ngAlsq = AlignSeq.extractGaps("-. ",
\r
661 alseqs[alsq].getSequenceAsString()).toUpperCase();
\r
662 int oldstrt = alseqs[alsq].getStart();
\r
663 alseqs[alsq].setStart(sequence.getSequenceAsString()
\r
664 .toUpperCase().indexOf(ngAlsq)
\r
665 + sequence.getStart());
\r
666 if (oldstrt != alseqs[alsq].getStart())
\r
668 alseqs[alsq].setEnd(ngAlsq.length()
\r
669 + alseqs[alsq].getStart() - 1);
\r
673 // TODO: search for all other references to this dataset sequence, and
\r
674 // update start/end
\r
675 // TODO: update all AlCodonMappings which involve this alignment
\r
676 // sequence (e.g. Q30167 cdna translation from exon2 product (vamsas
\r
679 // and remove it from the rest
\r
680 // TODO: decide if we should remove annotated sequence from set
\r
681 sdataset.remove(sequence);
\r
682 // TODO: should we make a note of sequences that have received new DB
\r
683 // ids, so we can query all enabled DAS servers for them ?
\r
688 // report the ID/sequence mismatches
\r
689 sbuffer.append(messages);
\r
694 * loop thru and collect additional sequences in Map.
\r
696 * @param sequencesArray
\r
699 private SequenceI[] recoverDbSequences(SequenceI[] sequencesArray)
\r
701 Vector nseq = new Vector();
\r
702 for (int i = 0; sequencesArray != null && i < sequencesArray.length; i++)
\r
704 nseq.addElement(sequencesArray[i]);
\r
705 DBRefEntry dbr[] = sequencesArray[i].getDBRef();
\r
706 jalview.datamodel.Mapping map = null;
\r
707 for (int r = 0; (dbr != null) && r < dbr.length; r++)
\r
709 if ((map = dbr[r].getMap()) != null)
\r
711 if (map.getTo() != null && !nseq.contains(map.getTo()))
\r
713 nseq.addElement(map.getTo());
\r
718 if (nseq.size() > 0)
\r
720 sequencesArray = new SequenceI[nseq.size()];
\r
721 nseq.toArray(sequencesArray);
\r
723 return sequencesArray;
\r