2 * Jalview - A Sequence Alignment Editor and Viewer (Version 2.7)
\r
3 * Copyright (C) 2011 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() : new jalview.gui.DasSourceBrowser()
\r
130 .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
196 System.arraycopy(otherdb, 0, newsrc, dbSources.length, otherdb.length);
\r
197 dbSources = newsrc;
\r
202 * start the fetcher thread
\r
204 * @param waitTillFinished
\r
205 * true to block until the fetcher has finished
\r
207 public void fetchDBRefs(boolean waitTillFinished)
\r
209 Thread thread = new Thread(this);
\r
213 if (waitTillFinished)
\r
220 } catch (Exception ex)
\r
228 * The sequence will be added to a vector of sequences belonging to key which
\r
229 * could be either seq name or dbref id
\r
236 void addSeqId(SequenceI seq, String key)
\r
238 key = key.toUpperCase();
\r
241 if (seqRefs.containsKey(key))
\r
243 seqs = (Vector) seqRefs.get(key);
\r
245 if (seqs != null && !seqs.contains(seq))
\r
247 seqs.addElement(seq);
\r
249 else if (seqs == null)
\r
251 seqs = new Vector();
\r
252 seqs.addElement(seq);
\r
258 seqs = new Vector();
\r
259 seqs.addElement(seq);
\r
262 seqRefs.put(key, seqs);
\r
270 if (dbSources == null)
\r
272 throw new Error("Implementation error. Must initialise dbSources");
\r
275 long startTime = System.currentTimeMillis();
\r
276 af.setProgressBar("Fetching db refs", startTime);
\r
279 if (Cache.getDefault("DBREFFETCH_USEPICR", false))
\r
281 picrClient = new uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperServiceLocator()
\r
282 .getAccessionMapperPort();
\r
284 } catch (Exception e)
\r
286 System.err.println("Couldn't locate PICR service instance.\n");
\r
287 e.printStackTrace();
\r
290 Vector sdataset = new Vector();
\r
291 for (int s = 0; s < dataset.length; s++)
\r
293 sdataset.addElement(dataset[s]);
\r
295 while (sdataset.size() > 0 && db < dbSources.length)
\r
297 int maxqlen = 1; // default number of queries made to at one time
\r
298 System.err.println("Verifying against " + dbSources[db]);
\r
299 jalview.ws.seqfetcher.DbSourceProxy dbsource = sfetcher
\r
300 .getSourceProxy(dbSources[db]);
\r
301 if (dbsource == null)
\r
303 System.err.println("No proxy for " + dbSources[db]);
\r
307 if (dbsource.getDbSourceProperties()
\r
308 .containsKey(DBRefSource.MULTIACC))
\r
310 maxqlen = ((Integer) dbsource.getDbSourceProperties().get(
\r
311 DBRefSource.MULTIACC)).intValue();
\r
317 // iterate through db for each remaining un-verified sequence
\r
318 SequenceI[] currSeqs = new SequenceI[sdataset.size()];
\r
319 sdataset.copyInto(currSeqs);// seqs that are to be validated against
\r
321 Vector queries = new Vector(); // generated queries curSeq
\r
322 seqRefs = new Hashtable();
\r
326 while (queries.size() > 0 || seqIndex < currSeqs.length)
\r
328 if (queries.size() > 0)
\r
330 // Still queries to make for current seqIndex
\r
331 StringBuffer queryString = new StringBuffer("");
\r
332 int numq = 0, nqSize = (maxqlen > queries.size()) ? queries
\r
335 while (queries.size() > 0 && numq < nqSize)
\r
337 String query = (String) queries.elementAt(0);
\r
338 if (dbsource.isValidReference(query))
\r
340 queryString.append((numq == 0) ? "" : dbsource
\r
341 .getAccessionSeparator());
\r
342 queryString.append(query);
\r
345 // remove the extracted query string
\r
346 queries.removeElementAt(0);
\r
348 // make the queries and process the response
\r
349 AlignmentI retrieved = null;
\r
352 if (jalview.bin.Cache.log.isDebugEnabled())
\r
354 jalview.bin.Cache.log.debug("Querying "
\r
355 + dbsource.getDbName() + " with : '"
\r
356 + queryString.toString() + "'");
\r
358 retrieved = dbsource.getSequenceRecords(queryString.toString());
\r
359 } catch (Exception ex)
\r
361 ex.printStackTrace();
\r
362 } catch (OutOfMemoryError err)
\r
364 new OOMWarning("retrieving database references ("
\r
365 + queryString.toString() + ")", err);
\r
367 if (retrieved != null)
\r
369 transferReferences(sdataset, dbSources[db], retrieved);
\r
374 // make some more strings for use as queries
\r
375 for (int i = 0; (seqIndex < dataset.length) && (i < 50); seqIndex++, i++)
\r
377 SequenceI sequence = dataset[seqIndex];
\r
378 DBRefEntry[] uprefs = jalview.util.DBRefUtils.selectRefs(
\r
379 sequence.getDBRef(), new String[]
\r
380 { dbSources[db] }); // jalview.datamodel.DBRefSource.UNIPROT
\r
382 // check for existing dbrefs to use
\r
383 if (uprefs != null && uprefs.length > 0)
\r
385 for (int j = 0; j < uprefs.length; j++)
\r
387 addSeqId(sequence, uprefs[j].getAccessionId());
\r
388 queries.addElement(uprefs[j].getAccessionId().toUpperCase());
\r
393 // generate queries from sequence ID string
\r
394 StringTokenizer st = new StringTokenizer(sequence.getName(),
\r
396 while (st.hasMoreTokens())
\r
398 String token = st.nextToken();
\r
399 UPEntry[] presp = null;
\r
400 if (picrClient != null)
\r
402 // resolve the string against PICR to recover valid IDs
\r
406 .getUPIForAccession(token, null,
\r
407 picrClient.getMappedDatabaseNames(),
\r
409 } catch (Exception e)
\r
411 System.err.println("Exception with Picr for '" + token
\r
413 e.printStackTrace();
\r
416 if (presp != null && presp.length > 0)
\r
418 for (int id = 0; id < presp.length; id++)
\r
420 // construct sequences from response if sequences are
\r
421 // present, and do a transferReferences
\r
422 // otherwise transfer non sequence x-references directly.
\r
425 .println("Validated ID against PICR... (for what its worth):"
\r
427 addSeqId(sequence, token);
\r
428 queries.addElement(token.toUpperCase());
\r
433 // System.out.println("Not querying source with token="+token+"\n");
\r
434 addSeqId(sequence, token);
\r
435 queries.addElement(token.toUpperCase());
\r
442 // advance to next database
\r
444 } // all databases have been queries.
\r
445 if (sbuffer.length() > 0)
\r
447 output.setText("Your sequences have been verified against known sequence databases. Some of the ids have been\n"
\r
448 + "altered, most likely the start/end residue will have been updated.\n"
\r
449 + "Save your alignment to maintain the updated id.\n\n"
\r
450 + sbuffer.toString());
\r
451 Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);
\r
452 // The above is the dataset, we must now find out the index
\r
453 // of the viewed sequence
\r
457 af.setProgressBar("DBRef search completed", startTime);
\r
458 // promptBeforeBlast();
\r
465 * Verify local sequences in seqRefs against the retrieved sequence database
\r
469 void transferReferences(Vector sdataset, String dbSource,
\r
470 AlignmentI retrievedAl) // File
\r
473 if (retrievedAl == null || retrievedAl.getHeight() == 0)
\r
477 SequenceI[] retrieved = recoverDbSequences(retrievedAl
\r
478 .getSequencesArray());
\r
479 SequenceI sequence = null;
\r
480 boolean transferred = false;
\r
481 StringBuffer messages = new StringBuffer();
\r
483 // Vector entries = new Uniprot().getUniprotEntries(file);
\r
485 int i, iSize = retrieved.length; // entries == null ? 0 : entries.size();
\r
486 // UniprotEntry entry;
\r
487 for (i = 0; i < iSize; i++)
\r
489 SequenceI entry = retrieved[i]; // (UniprotEntry) entries.elementAt(i);
\r
491 // Work out which sequences this sequence matches,
\r
492 // taking into account all accessionIds and names in the file
\r
493 Vector sequenceMatches = new Vector();
\r
494 // look for corresponding accession ids
\r
495 DBRefEntry[] entryRefs = jalview.util.DBRefUtils.selectRefs(
\r
496 entry.getDBRef(), new String[]
\r
498 if (entryRefs == null)
\r
501 .println("Dud dbSource string ? no entryrefs selected for "
\r
502 + dbSource + " on " + entry.getName());
\r
505 for (int j = 0; j < entryRefs.length; j++)
\r
507 String accessionId = entryRefs[j].getAccessionId(); // .getAccession().elementAt(j).toString();
\r
508 // match up on accessionId
\r
509 if (seqRefs.containsKey(accessionId.toUpperCase()))
\r
511 Vector seqs = (Vector) seqRefs.get(accessionId);
\r
512 for (int jj = 0; jj < seqs.size(); jj++)
\r
514 sequence = (SequenceI) seqs.elementAt(jj);
\r
515 if (!sequenceMatches.contains(sequence))
\r
517 sequenceMatches.addElement(sequence);
\r
522 if (sequenceMatches.size() == 0)
\r
524 // failed to match directly on accessionId==query so just compare all
\r
525 // sequences to entry
\r
526 Enumeration e = seqRefs.keys();
\r
527 while (e.hasMoreElements())
\r
529 Vector sqs = (Vector) seqRefs.get(e.nextElement());
\r
530 if (sqs != null && sqs.size() > 0)
\r
532 Enumeration sqe = sqs.elements();
\r
533 while (sqe.hasMoreElements())
\r
535 sequenceMatches.addElement(sqe.nextElement());
\r
540 // look for corresponding names
\r
541 // this is uniprot specific ?
\r
542 // could be useful to extend this so we try to find any 'significant'
\r
543 // information in common between two sequence objects.
\r
545 * DBRefEntry[] entryRefs =
\r
546 * jalview.util.DBRefUtils.selectRefs(entry.getDBRef(), new String[] {
\r
547 * dbSource }); for (int j = 0; j < entry.getName().size(); j++) { String
\r
548 * name = entry.getName().elementAt(j).toString(); if
\r
549 * (seqRefs.containsKey(name)) { Vector seqs = (Vector) seqRefs.get(name);
\r
550 * for (int jj = 0; jj < seqs.size(); jj++) { sequence = (SequenceI)
\r
551 * seqs.elementAt(jj); if (!sequenceMatches.contains(sequence)) {
\r
552 * sequenceMatches.addElement(sequence); } } } }
\r
554 // sequenceMatches now contains the set of all sequences associated with
\r
555 // the returned db record
\r
556 String entrySeq = entry.getSequenceAsString().toUpperCase();
\r
557 for (int m = 0; m < sequenceMatches.size(); m++)
\r
559 sequence = (SequenceI) sequenceMatches.elementAt(m);
\r
560 // only update start and end positions and shift features if there are
\r
561 // no existing references
\r
562 // TODO: test for legacy where uniprot or EMBL refs exist but no
\r
563 // mappings are made (but content matches retrieved set)
\r
564 boolean updateRefFrame = sequence.getDBRef() == null
\r
565 || sequence.getDBRef().length == 0;
\r
566 // verify sequence against the entry sequence
\r
568 String nonGapped = AlignSeq.extractGaps("-. ",
\r
569 sequence.getSequenceAsString()).toUpperCase();
\r
571 int absStart = entrySeq.indexOf(nonGapped);
\r
572 int mapStart = entry.getStart();
\r
573 jalview.datamodel.Mapping mp;
\r
575 if (absStart == -1)
\r
577 // Is local sequence contained in dataset sequence?
\r
578 absStart = nonGapped.indexOf(entrySeq);
\r
579 if (absStart == -1)
\r
580 { // verification failed.
\r
581 messages.append(sequence.getName()
\r
582 + " SEQUENCE NOT %100 MATCH \n");
\r
585 transferred = true;
\r
586 sbuffer.append(sequence.getName() + " HAS " + absStart
\r
587 + " PREFIXED RESIDUES COMPARED TO " + dbSource + "\n");
\r
589 // + " - ANY SEQUENCE FEATURES"
\r
590 // + " HAVE BEEN ADJUSTED ACCORDINGLY \n");
\r
592 // create valid mapping between matching region of local sequence and
\r
593 // the mapped sequence
\r
594 mp = new Mapping(null, new int[]
\r
595 { sequence.getStart() + absStart,
\r
596 sequence.getStart() + absStart + entrySeq.length() - 1 },
\r
598 { entry.getStart(),
\r
599 entry.getStart() + entrySeq.length() - 1 }, 1, 1);
\r
600 updateRefFrame = false; // mapping is based on current start/end so
\r
601 // don't modify start and end
\r
605 transferred = true;
\r
606 // update start and end of local sequence to place it in entry's
\r
607 // reference frame.
\r
608 // apply identity map map from whole of local sequence to matching
\r
609 // region of database
\r
611 mp = null; // Mapping.getIdentityMap();
\r
612 // new Mapping(null,
\r
613 // new int[] { absStart+sequence.getStart(),
\r
614 // absStart+sequence.getStart()+entrySeq.length()-1},
\r
615 // new int[] { entry.getStart(), entry.getEnd() }, 1, 1);
\r
616 // relocate local features for updated start
\r
617 if (updateRefFrame)
\r
619 if (sequence.getSequenceFeatures() != null)
\r
621 SequenceFeature[] sf = sequence.getSequenceFeatures();
\r
622 int start = sequence.getStart();
\r
623 int end = sequence.getEnd();
\r
624 int startShift = 1 - absStart - start; // how much the features
\r
626 // to be shifted by
\r
627 for (int sfi = 0; sfi < sf.length; sfi++)
\r
629 if (sf[sfi].getBegin() >= start && sf[sfi].getEnd() <= end)
\r
631 // shift feature along by absstart
\r
632 sf[sfi].setBegin(sf[sfi].getBegin() + startShift);
\r
633 sf[sfi].setEnd(sf[sfi].getEnd() + startShift);
\r
640 System.out.println("Adding dbrefs to " + sequence.getName()
\r
641 + " from " + dbSource + " sequence : " + entry.getName());
\r
642 sequence.transferAnnotation(entry, mp);
\r
643 // unknownSequences.remove(sequence);
\r
644 int absEnd = absStart + nonGapped.length();
\r
646 if (updateRefFrame)
\r
648 // finally, update local sequence reference frame if we're allowed
\r
649 sequence.setStart(absStart);
\r
650 sequence.setEnd(absEnd);
\r
651 // search for alignment sequences to update coordinate frame for
\r
652 for (int alsq = 0; alsq < alseqs.length; alsq++)
\r
654 if (alseqs[alsq].getDatasetSequence() == sequence)
\r
656 String ngAlsq = AlignSeq.extractGaps("-. ",
\r
657 alseqs[alsq].getSequenceAsString()).toUpperCase();
\r
658 int oldstrt = alseqs[alsq].getStart();
\r
659 alseqs[alsq].setStart(sequence.getSequenceAsString()
\r
660 .toUpperCase().indexOf(ngAlsq)
\r
661 + sequence.getStart());
\r
662 if (oldstrt != alseqs[alsq].getStart())
\r
664 alseqs[alsq].setEnd(ngAlsq.length()
\r
665 + alseqs[alsq].getStart() - 1);
\r
669 // TODO: search for all other references to this dataset sequence, and
\r
670 // update start/end
\r
671 // TODO: update all AlCodonMappings which involve this alignment
\r
672 // sequence (e.g. Q30167 cdna translation from exon2 product (vamsas
\r
675 // and remove it from the rest
\r
676 // TODO: decide if we should remove annotated sequence from set
\r
677 sdataset.remove(sequence);
\r
678 // TODO: should we make a note of sequences that have received new DB
\r
679 // ids, so we can query all enabled DAS servers for them ?
\r
684 // report the ID/sequence mismatches
\r
685 sbuffer.append(messages);
\r
690 * loop thru and collect additional sequences in Map.
\r
692 * @param sequencesArray
\r
695 private SequenceI[] recoverDbSequences(SequenceI[] sequencesArray)
\r
697 Vector nseq = new Vector();
\r
698 for (int i = 0; sequencesArray != null && i < sequencesArray.length; i++)
\r
700 nseq.addElement(sequencesArray[i]);
\r
701 DBRefEntry dbr[] = sequencesArray[i].getDBRef();
\r
702 jalview.datamodel.Mapping map = null;
\r
703 for (int r = 0; (dbr != null) && r < dbr.length; r++)
\r
705 if ((map = dbr[r].getMap()) != null)
\r
707 if (map.getTo() != null && !nseq.contains(map.getTo()))
\r
709 nseq.addElement(map.getTo());
\r
714 if (nseq.size() > 0)
\r
716 sequencesArray = new SequenceI[nseq.size()];
\r
717 nseq.toArray(sequencesArray);
\r
719 return sequencesArray;
\r