2 * Jalview - A Sequence Alignment Editor and Viewer (Version 2.7)
\r
3 * Copyright (C) 2011 J Procter, AM Waterhouse, J Engelhardt, LM Lui, 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
33 import jalview.ws.dbsources.das.api.jalviewSourceI;
\r
34 import jalview.ws.seqfetcher.DbSourceProxy;
\r
36 import java.lang.reflect.Array;
\r
37 import java.util.ArrayList;
\r
38 import java.util.Enumeration;
\r
39 import java.util.Hashtable;
\r
40 import java.util.List;
\r
41 import java.util.StringTokenizer;
\r
42 import java.util.Vector;
\r
44 import uk.ac.ebi.picr.model.UPEntry;
\r
47 * Implements a runnable for validating a sequence against external databases
\r
48 * and then propagating references and features onto the sequence(s)
\r
51 * @version $Revision$
\r
53 public class DBRefFetcher implements Runnable
\r
55 SequenceI[] dataset;
\r
57 IProgressIndicator af;
\r
59 CutAndPasteTransfer output = new CutAndPasteTransfer();
\r
61 StringBuffer sbuffer = new StringBuffer();
\r
63 boolean running = false;
\r
66 * picr client instance
\r
68 uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperInterface picrClient = null;
\r
70 // /This will be a collection of Vectors of sequenceI refs.
\r
71 // The key will be the seq name or accession id of the seq
\r
74 DbSourceProxy[] dbSources;
\r
76 SequenceFetcher sfetcher;
\r
78 private SequenceI[] alseqs;
\r
80 public DBRefFetcher()
\r
85 * Creates a new SequenceFeatureFetcher object and fetches from the currently
\r
86 * selected set of databases.
\r
89 * fetch references for these sequences
\r
91 * the parent alignframe for progress bar monitoring.
\r
93 public DBRefFetcher(SequenceI[] seqs, AlignFrame af)
\r
95 this(seqs, af, null);
\r
99 * Creates a new SequenceFeatureFetcher object and fetches from the currently
\r
100 * selected set of databases.
\r
103 * fetch references for these sequences
\r
105 * the parent alignframe for progress bar monitoring.
\r
107 * array of database source strings to query references from
\r
109 public DBRefFetcher(SequenceI[] seqs, AlignFrame af, DbSourceProxy[] sources)
\r
112 alseqs = new SequenceI[seqs.length];
\r
113 SequenceI[] ds = new SequenceI[seqs.length];
\r
114 for (int i = 0; i < seqs.length; i++)
\r
116 alseqs[i] = seqs[i];
\r
117 if (seqs[i].getDatasetSequence() != null)
\r
118 ds[i] = seqs[i].getDatasetSequence();
\r
123 // TODO Jalview 2.5 lots of this code should be in the gui package!
\r
124 sfetcher = jalview.gui.SequenceFetcher.getSequenceFetcherSingleton(af);
\r
125 if (sources == null)
\r
127 // af.featureSettings_actionPerformed(null);
\r
128 String[] defdb = null, otherdb = sfetcher
\r
129 .getDbInstances(jalview.ws.dbsources.das.datamodel.DasSequenceSource.class);
\r
130 List<DbSourceProxy> selsources = new ArrayList<DbSourceProxy>();
\r
131 Vector dasselsrc = (af.featureSettings != null) ? af.featureSettings
\r
132 .getSelectedSources() : new jalview.gui.DasSourceBrowser()
\r
133 .getSelectedSources();
\r
134 Enumeration<jalviewSourceI> en = dasselsrc.elements();
\r
135 while (en.hasMoreElements())
\r
137 jalviewSourceI src = en.nextElement();
\r
138 List<DbSourceProxy> sp=src.getSequenceSourceProxies();
\r
139 selsources.addAll(sp);
\r
142 Cache.log.debug("Added many Db Sources for :"+src.getTitle());
\r
145 // select appropriate databases based on alignFrame context.
\r
146 if (af.getViewport().getAlignment().isNucleotide())
\r
148 defdb = DBRefSource.DNACODINGDBS;
\r
152 defdb = DBRefSource.PROTEINDBS;
\r
154 List<DbSourceProxy> srces=new ArrayList<DbSourceProxy>();
\r
155 for (String ddb:defdb) {
\r
156 List<DbSourceProxy> srcesfordb=sfetcher.getSourceProxy(ddb);
\r
157 if (srcesfordb!=null) {
\r
158 srces.addAll(srcesfordb);
\r
162 // append the selected sequence sources to the default dbs
\r
163 srces.addAll(selsources);
\r
164 dbSources = srces.toArray(new DbSourceProxy[0]);
\r
168 // we assume the caller knows what they're doing and ensured that all the
\r
169 // db source names are valid
\r
170 dbSources = sources;
\r
175 * retrieve all the das sequence sources and add them to the list of db
\r
176 * sources to retrieve from
\r
178 public void appendAllDasSources()
\r
180 if (dbSources == null)
\r
182 dbSources = new DbSourceProxy[0];
\r
184 // append additional sources
\r
185 DbSourceProxy[] otherdb=sfetcher
\r
186 .getDbSourceProxyInstances(jalview.ws.dbsources.das.datamodel.DasSequenceSource.class);
\r
187 if (otherdb != null && otherdb.length > 0)
\r
189 DbSourceProxy[] newsrc = new DbSourceProxy[dbSources.length + otherdb.length];
\r
190 System.arraycopy(dbSources, 0, newsrc, 0, dbSources.length);
\r
191 System.arraycopy(otherdb, 0, newsrc, dbSources.length, otherdb.length);
\r
192 dbSources = newsrc;
\r
197 * start the fetcher thread
\r
199 * @param waitTillFinished
\r
200 * true to block until the fetcher has finished
\r
202 public void fetchDBRefs(boolean waitTillFinished)
\r
204 Thread thread = new Thread(this);
\r
208 if (waitTillFinished)
\r
215 } catch (Exception ex)
\r
223 * The sequence will be added to a vector of sequences belonging to key which
\r
224 * could be either seq name or dbref id
\r
231 void addSeqId(SequenceI seq, String key)
\r
233 key = key.toUpperCase();
\r
236 if (seqRefs.containsKey(key))
\r
238 seqs = (Vector) seqRefs.get(key);
\r
240 if (seqs != null && !seqs.contains(seq))
\r
242 seqs.addElement(seq);
\r
244 else if (seqs == null)
\r
246 seqs = new Vector();
\r
247 seqs.addElement(seq);
\r
253 seqs = new Vector();
\r
254 seqs.addElement(seq);
\r
257 seqRefs.put(key, seqs);
\r
265 if (dbSources == null)
\r
267 throw new Error("Implementation error. Must initialise dbSources");
\r
270 long startTime = System.currentTimeMillis();
\r
271 af.setProgressBar("Fetching db refs", startTime);
\r
274 if (Cache.getDefault("DBREFFETCH_USEPICR", false))
\r
276 picrClient = new uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperServiceLocator()
\r
277 .getAccessionMapperPort();
\r
279 } catch (Exception e)
\r
281 System.err.println("Couldn't locate PICR service instance.\n");
\r
282 e.printStackTrace();
\r
285 Vector sdataset = new Vector();
\r
286 for (int s = 0; s < dataset.length; s++)
\r
288 sdataset.addElement(dataset[s]);
\r
290 while (sdataset.size() > 0 && db < dbSources.length)
\r
292 int maxqlen = 1; // default number of queries made to at one time
\r
293 System.err.println("Verifying against " + dbSources[db].getDbName());
\r
294 boolean dn = false;
\r
296 // iterate through db for each remaining un-verified sequence
\r
297 SequenceI[] currSeqs = new SequenceI[sdataset.size()];
\r
298 sdataset.copyInto(currSeqs);// seqs that are to be validated against
\r
300 Vector queries = new Vector(); // generated queries curSeq
\r
301 seqRefs = new Hashtable();
\r
305 jalview.ws.seqfetcher.DbSourceProxy dbsource = dbSources[db];
\r
307 // for moment, we dumbly iterate over all retrieval sources for a particular database
\r
308 // TODO: introduce multithread multisource queries and logic to remove a query from other sources if any source for a database returns a record
\r
309 if (dbsource.getDbSourceProperties().containsKey(
\r
310 DBRefSource.MULTIACC))
\r
312 maxqlen = ((Integer) dbsource.getDbSourceProperties().get(
\r
313 DBRefSource.MULTIACC)).intValue();
\r
319 while (queries.size() > 0 || seqIndex < currSeqs.length)
\r
321 if (queries.size() > 0)
\r
323 // Still queries to make for current seqIndex
\r
324 StringBuffer queryString = new StringBuffer("");
\r
325 int numq = 0, nqSize = (maxqlen > queries.size()) ? queries
\r
328 while (queries.size() > 0 && numq < nqSize)
\r
330 String query = (String) queries.elementAt(0);
\r
331 if (dbsource.isValidReference(query))
\r
333 queryString.append((numq == 0) ? "" : dbsource
\r
334 .getAccessionSeparator());
\r
335 queryString.append(query);
\r
338 // remove the extracted query string
\r
339 queries.removeElementAt(0);
\r
341 // make the queries and process the response
\r
342 AlignmentI retrieved = null;
\r
345 if (jalview.bin.Cache.log.isDebugEnabled())
\r
347 jalview.bin.Cache.log.debug("Querying "
\r
348 + dbsource.getDbName() + " with : '"
\r
349 + queryString.toString() + "'");
\r
351 retrieved = dbsource.getSequenceRecords(queryString
\r
353 } catch (Exception ex)
\r
355 ex.printStackTrace();
\r
356 } catch (OutOfMemoryError err)
\r
358 new OOMWarning("retrieving database references ("
\r
359 + queryString.toString() + ")", err);
\r
361 if (retrieved != null)
\r
363 transferReferences(sdataset, dbsource.getDbSource(), retrieved);
\r
368 // make some more strings for use as queries
\r
369 for (int i = 0; (seqIndex < dataset.length) && (i < 50); seqIndex++, i++)
\r
371 SequenceI sequence = dataset[seqIndex];
\r
372 DBRefEntry[] uprefs = jalview.util.DBRefUtils.selectRefs(
\r
373 sequence.getDBRef(), new String[]
\r
374 { dbsource.getDbSource() }); // jalview.datamodel.DBRefSource.UNIPROT
\r
376 // check for existing dbrefs to use
\r
377 if (uprefs != null && uprefs.length > 0)
\r
379 for (int j = 0; j < uprefs.length; j++)
\r
381 addSeqId(sequence, uprefs[j].getAccessionId());
\r
382 queries.addElement(uprefs[j].getAccessionId()
\r
388 // generate queries from sequence ID string
\r
389 StringTokenizer st = new StringTokenizer(
\r
390 sequence.getName(), "|");
\r
391 while (st.hasMoreTokens())
\r
393 String token = st.nextToken();
\r
394 UPEntry[] presp = null;
\r
395 if (picrClient != null)
\r
397 // resolve the string against PICR to recover valid IDs
\r
400 presp = picrClient.getUPIForAccession(token, null,
\r
401 picrClient.getMappedDatabaseNames(), null,
\r
403 } catch (Exception e)
\r
405 System.err.println("Exception with Picr for '"
\r
407 e.printStackTrace();
\r
410 if (presp != null && presp.length > 0)
\r
412 for (int id = 0; id < presp.length; id++)
\r
414 // construct sequences from response if sequences are
\r
415 // present, and do a transferReferences
\r
416 // otherwise transfer non sequence x-references directly.
\r
419 .println("Validated ID against PICR... (for what its worth):"
\r
421 addSeqId(sequence, token);
\r
422 queries.addElement(token.toUpperCase());
\r
427 // System.out.println("Not querying source with token="+token+"\n");
\r
428 addSeqId(sequence, token);
\r
429 queries.addElement(token.toUpperCase());
\r
437 // advance to next database
\r
439 } // all databases have been queries.
\r
440 if (sbuffer.length() > 0)
\r
442 output.setText("Your sequences have been verified against known sequence databases. Some of the ids have been\n"
\r
443 + "altered, most likely the start/end residue will have been updated.\n"
\r
444 + "Save your alignment to maintain the updated id.\n\n"
\r
445 + sbuffer.toString());
\r
446 Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);
\r
447 // The above is the dataset, we must now find out the index
\r
448 // of the viewed sequence
\r
452 af.setProgressBar("DBRef search completed", startTime);
\r
453 // promptBeforeBlast();
\r
460 * Verify local sequences in seqRefs against the retrieved sequence database
\r
464 void transferReferences(Vector sdataset, String dbSource,
\r
465 AlignmentI retrievedAl) // File
\r
468 if (retrievedAl == null || retrievedAl.getHeight() == 0)
\r
472 SequenceI[] retrieved = recoverDbSequences(retrievedAl
\r
473 .getSequencesArray());
\r
474 SequenceI sequence = null;
\r
475 boolean transferred = false;
\r
476 StringBuffer messages = new StringBuffer();
\r
478 // Vector entries = new Uniprot().getUniprotEntries(file);
\r
480 int i, iSize = retrieved.length; // entries == null ? 0 : entries.size();
\r
481 // UniprotEntry entry;
\r
482 for (i = 0; i < iSize; i++)
\r
484 SequenceI entry = retrieved[i]; // (UniprotEntry) entries.elementAt(i);
\r
486 // Work out which sequences this sequence matches,
\r
487 // taking into account all accessionIds and names in the file
\r
488 Vector sequenceMatches = new Vector();
\r
489 // look for corresponding accession ids
\r
490 DBRefEntry[] entryRefs = jalview.util.DBRefUtils.selectRefs(
\r
491 entry.getDBRef(), new String[]
\r
493 if (entryRefs == null)
\r
496 .println("Dud dbSource string ? no entryrefs selected for "
\r
497 + dbSource + " on " + entry.getName());
\r
500 for (int j = 0; j < entryRefs.length; j++)
\r
502 String accessionId = entryRefs[j].getAccessionId(); // .getAccession().elementAt(j).toString();
\r
503 // match up on accessionId
\r
504 if (seqRefs.containsKey(accessionId.toUpperCase()))
\r
506 Vector seqs = (Vector) seqRefs.get(accessionId);
\r
507 for (int jj = 0; jj < seqs.size(); jj++)
\r
509 sequence = (SequenceI) seqs.elementAt(jj);
\r
510 if (!sequenceMatches.contains(sequence))
\r
512 sequenceMatches.addElement(sequence);
\r
517 if (sequenceMatches.size() == 0)
\r
519 // failed to match directly on accessionId==query so just compare all
\r
520 // sequences to entry
\r
521 Enumeration e = seqRefs.keys();
\r
522 while (e.hasMoreElements())
\r
524 Vector sqs = (Vector) seqRefs.get(e.nextElement());
\r
525 if (sqs != null && sqs.size() > 0)
\r
527 Enumeration sqe = sqs.elements();
\r
528 while (sqe.hasMoreElements())
\r
530 sequenceMatches.addElement(sqe.nextElement());
\r
535 // look for corresponding names
\r
536 // this is uniprot specific ?
\r
537 // could be useful to extend this so we try to find any 'significant'
\r
538 // information in common between two sequence objects.
\r
540 * DBRefEntry[] entryRefs =
\r
541 * jalview.util.DBRefUtils.selectRefs(entry.getDBRef(), new String[] {
\r
542 * dbSource }); for (int j = 0; j < entry.getName().size(); j++) { String
\r
543 * name = entry.getName().elementAt(j).toString(); if
\r
544 * (seqRefs.containsKey(name)) { Vector seqs = (Vector) seqRefs.get(name);
\r
545 * for (int jj = 0; jj < seqs.size(); jj++) { sequence = (SequenceI)
\r
546 * seqs.elementAt(jj); if (!sequenceMatches.contains(sequence)) {
\r
547 * sequenceMatches.addElement(sequence); } } } }
\r
549 // sequenceMatches now contains the set of all sequences associated with
\r
550 // the returned db record
\r
551 String entrySeq = entry.getSequenceAsString().toUpperCase();
\r
552 for (int m = 0; m < sequenceMatches.size(); m++)
\r
554 sequence = (SequenceI) sequenceMatches.elementAt(m);
\r
555 // only update start and end positions and shift features if there are
\r
556 // no existing references
\r
557 // TODO: test for legacy where uniprot or EMBL refs exist but no
\r
558 // mappings are made (but content matches retrieved set)
\r
559 boolean updateRefFrame = sequence.getDBRef() == null
\r
560 || sequence.getDBRef().length == 0;
\r
561 // verify sequence against the entry sequence
\r
563 String nonGapped = AlignSeq.extractGaps("-. ",
\r
564 sequence.getSequenceAsString()).toUpperCase();
\r
566 int absStart = entrySeq.indexOf(nonGapped);
\r
567 int mapStart = entry.getStart();
\r
568 jalview.datamodel.Mapping mp;
\r
570 if (absStart == -1)
\r
572 // Is local sequence contained in dataset sequence?
\r
573 absStart = nonGapped.indexOf(entrySeq);
\r
574 if (absStart == -1)
\r
575 { // verification failed.
\r
576 messages.append(sequence.getName()
\r
577 + " SEQUENCE NOT %100 MATCH \n");
\r
580 transferred = true;
\r
581 sbuffer.append(sequence.getName() + " HAS " + absStart
\r
582 + " PREFIXED RESIDUES COMPARED TO " + dbSource + "\n");
\r
584 // + " - ANY SEQUENCE FEATURES"
\r
585 // + " HAVE BEEN ADJUSTED ACCORDINGLY \n");
\r
587 // create valid mapping between matching region of local sequence and
\r
588 // the mapped sequence
\r
589 mp = new Mapping(null, new int[]
\r
590 { sequence.getStart() + absStart,
\r
591 sequence.getStart() + absStart + entrySeq.length() - 1 },
\r
593 { entry.getStart(),
\r
594 entry.getStart() + entrySeq.length() - 1 }, 1, 1);
\r
595 updateRefFrame = false; // mapping is based on current start/end so
\r
596 // don't modify start and end
\r
600 transferred = true;
\r
601 // update start and end of local sequence to place it in entry's
\r
602 // reference frame.
\r
603 // apply identity map map from whole of local sequence to matching
\r
604 // region of database
\r
606 mp = null; // Mapping.getIdentityMap();
\r
607 // new Mapping(null,
\r
608 // new int[] { absStart+sequence.getStart(),
\r
609 // absStart+sequence.getStart()+entrySeq.length()-1},
\r
610 // new int[] { entry.getStart(), entry.getEnd() }, 1, 1);
\r
611 // relocate local features for updated start
\r
612 if (updateRefFrame)
\r
614 if (sequence.getSequenceFeatures() != null)
\r
616 SequenceFeature[] sf = sequence.getSequenceFeatures();
\r
617 int start = sequence.getStart();
\r
618 int end = sequence.getEnd();
\r
619 int startShift = 1 - absStart - start; // how much the features
\r
621 // to be shifted by
\r
622 for (int sfi = 0; sfi < sf.length; sfi++)
\r
624 if (sf[sfi].getBegin() >= start && sf[sfi].getEnd() <= end)
\r
626 // shift feature along by absstart
\r
627 sf[sfi].setBegin(sf[sfi].getBegin() + startShift);
\r
628 sf[sfi].setEnd(sf[sfi].getEnd() + startShift);
\r
635 System.out.println("Adding dbrefs to " + sequence.getName()
\r
636 + " from " + dbSource + " sequence : " + entry.getName());
\r
637 sequence.transferAnnotation(entry, mp);
\r
638 // unknownSequences.remove(sequence);
\r
639 int absEnd = absStart + nonGapped.length();
\r
641 if (updateRefFrame)
\r
643 // finally, update local sequence reference frame if we're allowed
\r
644 sequence.setStart(absStart);
\r
645 sequence.setEnd(absEnd);
\r
646 // search for alignment sequences to update coordinate frame for
\r
647 for (int alsq = 0; alsq < alseqs.length; alsq++)
\r
649 if (alseqs[alsq].getDatasetSequence() == sequence)
\r
651 String ngAlsq = AlignSeq.extractGaps("-. ",
\r
652 alseqs[alsq].getSequenceAsString()).toUpperCase();
\r
653 int oldstrt = alseqs[alsq].getStart();
\r
654 alseqs[alsq].setStart(sequence.getSequenceAsString()
\r
655 .toUpperCase().indexOf(ngAlsq)
\r
656 + sequence.getStart());
\r
657 if (oldstrt != alseqs[alsq].getStart())
\r
659 alseqs[alsq].setEnd(ngAlsq.length()
\r
660 + alseqs[alsq].getStart() - 1);
\r
664 // TODO: search for all other references to this dataset sequence, and
\r
665 // update start/end
\r
666 // TODO: update all AlCodonMappings which involve this alignment
\r
667 // sequence (e.g. Q30167 cdna translation from exon2 product (vamsas
\r
670 // and remove it from the rest
\r
671 // TODO: decide if we should remove annotated sequence from set
\r
672 sdataset.remove(sequence);
\r
673 // TODO: should we make a note of sequences that have received new DB
\r
674 // ids, so we can query all enabled DAS servers for them ?
\r
679 // report the ID/sequence mismatches
\r
680 sbuffer.append(messages);
\r
685 * loop thru and collect additional sequences in Map.
\r
687 * @param sequencesArray
\r
690 private SequenceI[] recoverDbSequences(SequenceI[] sequencesArray)
\r
692 Vector nseq = new Vector();
\r
693 for (int i = 0; sequencesArray != null && i < sequencesArray.length; i++)
\r
695 nseq.addElement(sequencesArray[i]);
\r
696 DBRefEntry dbr[] = sequencesArray[i].getDBRef();
\r
697 jalview.datamodel.Mapping map = null;
\r
698 for (int r = 0; (dbr != null) && r < dbr.length; r++)
\r
700 if ((map = dbr[r].getMap()) != null)
\r
702 if (map.getTo() != null && !nseq.contains(map.getTo()))
\r
704 nseq.addElement(map.getTo());
\r
709 if (nseq.size() > 0)
\r
711 sequencesArray = new SequenceI[nseq.size()];
\r
712 nseq.toArray(sequencesArray);
\r
714 return sequencesArray;
\r