2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.analysis;
23 import jalview.datamodel.AlignedCodonFrame;
24 import jalview.datamodel.Alignment;
25 import jalview.datamodel.AlignmentI;
26 import jalview.datamodel.DBRefEntry;
27 import jalview.datamodel.DBRefSource;
28 import jalview.datamodel.Mapping;
29 import jalview.datamodel.Sequence;
30 import jalview.datamodel.SequenceFeature;
31 import jalview.datamodel.SequenceI;
32 import jalview.util.DBRefUtils;
33 import jalview.util.MapList;
34 import jalview.ws.SequenceFetcher;
36 import java.util.ArrayList;
37 import java.util.Iterator;
38 import java.util.List;
40 import intervalstore.api.IntervalI;
43 * Functions for cross-referencing sequence databases.
51 * the dataset of the alignment for which we are searching for
52 * cross-references; in some cases we may resolve xrefs by
53 * searching in the dataset
55 private AlignmentI dataset;
58 * the sequences for which we are seeking cross-references
60 private SequenceI[] fromSeqs;
63 * matcher built from dataset
65 SequenceIdMatcher matcher;
68 * sequences found by cross-ref searches to fromSeqs
70 List<SequenceI> rseqs;
76 * the sequences for which we are seeking cross-references
78 * the containing alignment dataset (may be searched to resolve
81 public CrossRef(SequenceI[] seqs, AlignmentI ds)
84 dataset = ds.getDataset() == null ? ds : ds.getDataset();
88 * Returns a list of distinct database sources for which sequences have either
90 * <li>a (dna-to-protein or protein-to-dna) cross-reference</li>
91 * <li>an indirect cross-reference - a (dna-to-protein or protein-to-dna)
92 * reference from another sequence in the dataset which has a cross-reference
93 * to a direct DBRefEntry on the given sequence</li>
97 * - when true, cross-references *from* dna returned. When false,
98 * cross-references *from* protein are returned
101 public List<String> findXrefSourcesForSequences(boolean dna)
103 List<String> sources = new ArrayList<>();
104 for (SequenceI seq : fromSeqs)
108 findXrefSourcesForSequence(seq, dna, sources);
111 sources.remove(DBRefSource.EMBL); // hack to prevent EMBL xrefs resulting in
112 // redundant datasets
115 sources.remove(DBRefSource.ENSEMBL); // hack to prevent Ensembl and
116 // EnsemblGenomes xref option shown
118 sources.remove(DBRefSource.ENSEMBLGENOMES);
120 // redundant datasets
125 * Returns a list of distinct database sources for which a sequence has either
127 * <li>a (dna-to-protein or protein-to-dna) cross-reference</li>
128 * <li>an indirect cross-reference - a (dna-to-protein or protein-to-dna)
129 * reference from another sequence in the dataset which has a cross-reference
130 * to a direct DBRefEntry on the given sequence</li>
134 * the sequence whose dbrefs we are searching against
136 * when true, context is DNA - so sources identifying protein
137 * products will be returned.
139 * a list of sources to add matches to
141 void findXrefSourcesForSequence(SequenceI seq, boolean fromDna,
142 List<String> sources)
145 * first find seq's xrefs (dna-to-peptide or peptide-to-dna)
147 List<DBRefEntry> rfs = DBRefUtils.selectDbRefs(!fromDna, seq.getDBRefs());
148 addXrefsToSources(rfs, sources);
152 * find sequence's direct (dna-to-dna, peptide-to-peptide) xrefs
154 List<DBRefEntry> lrfs = DBRefUtils.selectDbRefs(fromDna, seq.getDBRefs());
155 List<SequenceI> foundSeqs = new ArrayList<>();
158 * find sequences in the alignment which xref one of these DBRefs
159 * i.e. is xref-ed to a common sequence identifier
161 searchDatasetXrefs(fromDna, seq, lrfs, foundSeqs, null);
164 * add those sequences' (dna-to-peptide or peptide-to-dna) dbref sources
166 for (SequenceI rs : foundSeqs)
168 List<DBRefEntry> xrs = DBRefUtils.selectDbRefs(!fromDna,
170 addXrefsToSources(xrs, sources);
176 * Helper method that adds the source identifiers of some cross-references to
177 * a (non-redundant) list of database sources
182 void addXrefsToSources(List<DBRefEntry> xrefs, List<String> sources)
186 for (DBRefEntry ref : xrefs)
189 * avoid duplication e.g. ENSEMBL and Ensembl
191 String source = DBRefUtils.getCanonicalName(ref.getSource());
192 if (!sources.contains(source))
201 * Attempts to find cross-references from the sequences provided in the
202 * constructor to the given source database. Cross-references may be found
204 * <li>in dbrefs on the sequence which hold a mapping to a sequence
206 * <li>provided with a fetched sequence (e.g. ENA translation), or</li>
207 * <li>populated previously after getting cross-references</li>
209 * <li>as other sequences in the alignment which share a dbref identifier with
211 * <li>by fetching from the remote database</li>
213 * The cross-referenced sequences, and mappings to them, are added to the
217 * @return cross-referenced sequences (as dataset sequences)
219 public Alignment findXrefSequences(String source, boolean fromDna)
222 rseqs = new ArrayList<>();
223 AlignedCodonFrame cf = new AlignedCodonFrame();
224 matcher = new SequenceIdMatcher(dataset.getSequences());
226 for (SequenceI seq : fromSeqs)
229 while (dss.getDatasetSequence() != null)
231 dss = dss.getDatasetSequence();
233 boolean found = false;
234 List<DBRefEntry> xrfs = DBRefUtils.selectDbRefs(!fromDna,
236 // ENST & ENSP comes in to both Protein and nucleotide, so we need to
239 if ((xrfs == null || xrfs.size() == 0) && dataset != null)
242 * found no suitable dbrefs on sequence - look for sequences in the
243 * alignment which share a dbref with this one
245 List<DBRefEntry> lrfs = DBRefUtils.selectDbRefs(fromDna,
249 * find sequences (except this one!), of complementary type,
250 * which have a dbref to an accession id for this sequence,
251 * and add them to the results
253 found = searchDatasetXrefs(fromDna, dss, lrfs, rseqs, cf);
255 if (xrfs == null && !found)
258 * no dbref to source on this sequence or matched
259 * complementary sequence in the dataset
263 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefsForSource(xrfs,
265 Iterator<DBRefEntry> refIterator = sourceRefs.iterator();
266 // At this point, if we are retrieving Ensembl, we still don't filter out
267 // ENST when looking for protein crossrefs.
268 while (refIterator.hasNext())
270 DBRefEntry xref = refIterator.next();
272 // we're only interested in coding cross-references, not
274 if (xref.hasMap() && xref.getMap().getMap().isTripletMap())
276 SequenceI mappedTo = xref.getMap().getTo();
277 if (mappedTo != null)
280 * dbref contains the sequence it maps to; add it to the
281 * results unless we have done so already (could happen if
282 * fetching xrefs for sequences which have xrefs in common)
283 * for example: UNIPROT {P0CE19, P0CE20} -> EMBL {J03321, X06707}
287 * problem: matcher.findIdMatch() is lenient - returns a sequence
288 * with a dbref to the search arg e.g. ENST for ENSP - wrong
289 * but findInDataset() matches ENSP when looking for Uniprot...
291 SequenceI matchInDataset = findInDataset(xref);
292 if (matchInDataset != null && xref.getMap().getTo() != null
293 && matchInDataset != xref.getMap().getTo())
296 "Implementation problem (reopen JAL-2154): CrossRef.findInDataset seems to have recovered a different sequence than the one explicitly mapped for xref."
297 + "Found:" + matchInDataset + "\nExpected:"
298 + xref.getMap().getTo() + "\nFor xref:"
301 /*matcher.findIdMatch(mappedTo);*/
302 if (matchInDataset != null)
304 if (!rseqs.contains(matchInDataset))
306 rseqs.add(matchInDataset);
308 // even if rseqs contained matchInDataset - check mappings between
309 // these seqs are added
310 // need to try harder to only add unique mappings
311 if (xref.getMap().getMap().isTripletMap()
312 && dataset.getMapping(seq, matchInDataset) == null
313 && cf.getMappingBetween(seq, matchInDataset) == null)
315 // materialise a mapping for highlighting between these
319 cf.addMap(dss, matchInDataset, xref.getMap().getMap(),
320 xref.getMap().getMappedFromId());
324 cf.addMap(matchInDataset, dss,
325 xref.getMap().getMap().getInverse(),
326 xref.getMap().getMappedFromId());
330 refIterator.remove();
333 // TODO: need to determine if this should be a deriveSequence
334 SequenceI rsq = new Sequence(mappedTo);
336 if (xref.getMap().getMap().isTripletMap())
338 // get sense of map correct for adding to product alignment.
341 // map is from dna seq to a protein product
342 cf.addMap(dss, rsq, xref.getMap().getMap(),
343 xref.getMap().getMappedFromId());
347 // map should be from protein seq to its coding dna
348 cf.addMap(rsq, dss, xref.getMap().getMap().getInverse(),
349 xref.getMap().getMappedFromId());
357 SequenceI matchedSeq = matcher.findIdMatch(
358 xref.getSource() + "|" + xref.getAccessionId());
359 // if there was a match, check it's at least the right type of
361 if (matchedSeq != null && matchedSeq.isProtein() == fromDna)
363 if (constructMapping(seq, matchedSeq, xref, cf, fromDna))
372 // do a bit more work - search for sequences with references matching
373 // xrefs on this sequence.
374 found = searchDataset(fromDna, dss, xref, rseqs, cf, false, DBRefUtils.SEARCH_MODE_FULL);
378 refIterator.remove();
383 * fetch from source database any dbrefs we haven't resolved up to here
385 if (!sourceRefs.isEmpty())
387 retrieveCrossRef(sourceRefs, seq, xrfs, fromDna, cf);
391 Alignment ral = null;
392 if (rseqs.size() > 0)
394 ral = new Alignment(rseqs.toArray(new SequenceI[rseqs.size()]));
397 dataset.addCodonFrame(cf);
403 private void retrieveCrossRef(List<DBRefEntry> sourceRefs, SequenceI seq,
404 List<DBRefEntry> xrfs, boolean fromDna, AlignedCodonFrame cf)
406 SequenceI[] retrieved = null;
407 SequenceI dss = seq.getDatasetSequence() == null ? seq
408 : seq.getDatasetSequence();
409 // first filter in case we are retrieving crossrefs that have already been
410 // retrieved. this happens for cases where a database record doesn't yield
411 // protein products for CDS
412 removeAlreadyRetrievedSeqs(sourceRefs, fromDna);
413 if (sourceRefs.size() == 0)
415 // no more work to do! We already had all requested sequence records in
421 retrieved = SequenceFetcher.getInstance()
422 .getSequences(sourceRefs, !fromDna);
423 } catch (Exception e)
426 "Problem whilst retrieving cross references for Sequence : "
431 if (retrieved != null)
433 boolean addedXref = false;
434 List<SequenceI> newDsSeqs = new ArrayList<>(),
435 doNotAdd = new ArrayList<>();
437 for (SequenceI retrievedSequence : retrieved)
439 // dataset gets contaminated ccwith non-ds sequences. why ??!
440 // try: Ensembl -> Nuc->Ensembl, Nuc->Uniprot-->Protein->EMBL->
441 SequenceI retrievedDss = retrievedSequence
442 .getDatasetSequence() == null ? retrievedSequence
443 : retrievedSequence.getDatasetSequence();
444 addedXref |= importCrossRefSeq(cf, newDsSeqs, doNotAdd, dss,
449 // try again, after looking for matching IDs
450 // shouldn't need to do this unless the dbref mechanism has broken.
451 updateDbrefMappings(seq, xrfs, retrieved, cf, fromDna);
452 for (SequenceI retrievedSequence : retrieved)
454 // dataset gets contaminated ccwith non-ds sequences. why ??!
455 // try: Ensembl -> Nuc->Ensembl, Nuc->Uniprot-->Protein->EMBL->
456 SequenceI retrievedDss = retrievedSequence
457 .getDatasetSequence() == null ? retrievedSequence
458 : retrievedSequence.getDatasetSequence();
459 addedXref |= importCrossRefSeq(cf, newDsSeqs, doNotAdd, dss,
463 for (SequenceI newToSeq : newDsSeqs)
465 if (!doNotAdd.contains(newToSeq)
466 && dataset.findIndex(newToSeq) == -1)
468 dataset.addSequence(newToSeq);
469 matcher.add(newToSeq);
476 * Search dataset for sequences with a primary reference contained in
480 * - list of references to filter.
482 * - type of sequence to search for matching primary reference.
484 private void removeAlreadyRetrievedSeqs(List<DBRefEntry> sourceRefs,
487 List<DBRefEntry> dbrSourceSet = new ArrayList<>(sourceRefs);
488 List<SequenceI> dsSeqs = dataset.getSequences();
489 for (int ids = 0, nds = dsSeqs.size(); ids < nds; ids++)
491 SequenceI sq = dsSeqs.get(ids);
492 boolean dupeFound = false;
493 // !fromDna means we are looking only for nucleotide sequences, not
495 if (sq.isProtein() == fromDna)
497 List<DBRefEntry> sqdbrefs = sq.getPrimaryDBRefs();
498 for (int idb = 0, ndb = sqdbrefs.size(); idb < ndb; idb++)
500 DBRefEntry dbr = sqdbrefs.get(idb);
501 List<DBRefEntry> searchrefs = DBRefUtils.searchRefs(dbrSourceSet, dbr, DBRefUtils.SEARCH_MODE_FULL);
502 for (int isr = 0, nsr = searchrefs.size(); isr < nsr; isr++)
504 sourceRefs.remove(searchrefs.get(isr));
511 // rebuild the search array from the filtered sourceRefs list
512 dbrSourceSet.clear();
513 dbrSourceSet.addAll(sourceRefs);
519 * process sequence retrieved via a dbref on source sequence to resolve and
523 * @param sourceSequence
524 * @param retrievedSequence
525 * @return true if retrieveSequence was imported
527 private boolean importCrossRefSeq(AlignedCodonFrame cf,
528 List<SequenceI> newDsSeqs, List<SequenceI> doNotAdd,
529 SequenceI sourceSequence, SequenceI retrievedSequence)
532 * set when retrievedSequence has been verified as a crossreference for
535 boolean imported = false;
536 List<DBRefEntry> dbr = retrievedSequence.getDBRefs();
539 for (int ib = 0, nb = dbr.size(); ib < nb; ib++)
542 DBRefEntry dbref = dbr.get(ib);
543 SequenceI matched = findInDataset(dbref);
544 if (matched == sourceSequence)
546 // verified retrieved and source sequence cross-reference each other
549 // find any entry where we should put in the sequence being
550 // cross-referenced into the map
551 Mapping map = dbref.getMap();
554 SequenceI ms = map.getTo();
555 if (ms != null && map.getMap() != null)
557 if (ms == sourceSequence)
559 // already called to import once, and most likely this sequence
560 // already imported !
566 * sequence is new to dataset, so save a reference so it can be added.
573 * there was a matching sequence in dataset, so now, check to see if we can update the map.getTo() sequence to the existing one.
578 // compare ms with dss and replace with dss in mapping
579 // if map is congruent
580 // TODO findInDataset requires exact sequence match but
581 // 'congruent' test is only for the mapped part
582 // maybe not a problem in practice since only ENA provide a
583 // mapping and it is to the full protein translation of CDS
584 // matcher.findIdMatch(map.getTo());
585 // TODO addendum: if matched is shorter than getTo, this will fail
586 // - when it should really succeed.
587 int sf = map.getMap().getToLowest();
588 int st = map.getMap().getToHighest();
589 SequenceI mappedrg = ms.getSubSequence(sf, st);
590 if (mappedrg.getLength() > 0 && ms.getSequenceAsString()
591 .equals(matched.getSequenceAsString()))
594 * sequences were a match,
596 String msg = "Mapping updated from " + ms.getName()
597 + " to retrieved crossreference "
599 System.out.println(msg);
601 List<DBRefEntry> toRefs = map.getTo().getDBRefs();
605 * transfer database refs
607 for (DBRefEntry ref : toRefs)
609 if (dbref.getSrcAccString()
610 .equals(ref.getSrcAccString()))
612 continue; // avoid overwriting the ref on source sequence
614 matched.addDBRef(ref); // add or update mapping
617 doNotAdd.add(map.getTo());
621 * give the reverse reference the inverse mapping
622 * (if it doesn't have one already)
624 setReverseMapping(matched, dbref, cf);
627 * copy sequence features as well, avoiding
628 * duplication (e.g. same variation from two
631 List<SequenceFeature> sfs = ms.getFeatures()
633 for (SequenceFeature feat : sfs)
636 * make a flyweight feature object which ignores Parent
637 * attribute in equality test; this avoids creating many
638 * otherwise duplicate exon features on genomic sequence
640 SequenceFeature newFeature = new SequenceFeature(feat)
642 // BH 2019.08.15 We must override equalsInterval, not
643 // equals, because that is part of the IntervalI interface,
644 // and IntervalStore may need that for proper, faster
647 // public boolean equals(Object o)
649 // return super.equals(o, true);
653 public boolean equalsInterval(IntervalI sf)
655 return equals((SequenceFeature) sf, true);
659 matched.addSequenceFeature(newFeature);
662 cf.addMap(retrievedSequence, map.getTo(), map.getMap());
663 } catch (Exception e)
666 "Exception when consolidating Mapped sequence set...");
667 e.printStackTrace(System.err);
675 retrievedSequence.updatePDBIds();
676 rseqs.add(retrievedSequence);
677 if (dataset.findIndex(retrievedSequence) == -1)
679 dataset.addSequence(retrievedSequence);
680 matcher.add(retrievedSequence);
687 * Sets the inverse sequence mapping in the corresponding dbref of the mapped
688 * to sequence (if any). This is used after fetching a cross-referenced
689 * sequence, if the fetched sequence has a mapping to the original sequence,
690 * to set the mapping in the original sequence's dbref.
693 * the sequence mapped from
697 void setReverseMapping(SequenceI mapFrom, DBRefEntry dbref,
698 AlignedCodonFrame mappings)
700 SequenceI mapTo = dbref.getMap().getTo();
705 List<DBRefEntry> dbrefs = mapTo.getDBRefs();
710 for (DBRefEntry toRef : dbrefs)
712 if (toRef.hasMap() && mapFrom == toRef.getMap().getTo())
715 * found the reverse dbref; update its mapping if null
717 if (toRef.getMap().getMap() == null)
719 MapList inverse = dbref.getMap().getMap().getInverse();
720 toRef.getMap().setMap(inverse);
721 mappings.addMap(mapTo, mapFrom, inverse);
728 * Returns null or the first sequence in the dataset which is identical to
729 * xref.mapTo, and has a) a primary dbref matching xref, or if none found, the
730 * first one with an ID source|xrefacc
733 * with map and mapped-to sequence
736 SequenceI findInDataset(DBRefEntry xref)
738 if (xref == null || !xref.hasMap() || xref.getMap().getTo() == null)
742 SequenceI mapsTo = xref.getMap().getTo();
743 String name = xref.getAccessionId();
744 String name2 = xref.getSource() + "|" + name;
745 SequenceI dss = mapsTo.getDatasetSequence() == null ? mapsTo
746 : mapsTo.getDatasetSequence();
747 // first check ds if ds is directly referenced
748 if (dataset.findIndex(dss) > -1)
752 DBRefEntry template = new DBRefEntry(xref.getSource(), null,
753 xref.getAccessionId());
755 * remember the first ID match - in case we don't find a match to template
757 SequenceI firstIdMatch = null;
758 for (SequenceI seq : dataset.getSequences())
760 // first check primary refs.
761 List<DBRefEntry> match = DBRefUtils.searchRefs(
762 seq.getPrimaryDBRefs(), template, DBRefUtils.SEARCH_MODE_FULL);
763 if (match != null && match.size() == 1 && sameSequence(seq, dss))
768 * clumsy alternative to using SequenceIdMatcher which currently
769 * returns sequences with a dbref to the matched accession id
770 * which we don't want
772 if (firstIdMatch == null && (name.equals(seq.getName())
773 || seq.getName().startsWith(name2)))
775 if (sameSequence(seq, dss))
785 * Answers true if seq1 and seq2 contain exactly the same characters (ignoring
786 * case), else false. This method compares the lengths, then each character in
787 * turn, in order to 'fail fast'. For case-sensitive comparison, it would be
788 * possible to use Arrays.equals(seq1.getSequence(), seq2.getSequence()).
794 // TODO move to Sequence / SequenceI
795 static boolean sameSequence(SequenceI seq1, SequenceI seq2)
801 if (seq1 == null || seq2 == null)
806 if (seq1.getLength() != seq2.getLength())
810 int length = seq1.getLength();
811 for (int i = 0; i < length; i++)
813 int diff = seq1.getCharAt(i) - seq2.getCharAt(i);
815 * same char or differ in case only ('a'-'A' == 32)
817 if (diff != 0 && diff != 32 && diff != -32)
826 * Updates any empty mappings in the cross-references with one to a compatible
827 * retrieved sequence if found, and adds any new mappings to the
835 void updateDbrefMappings(SequenceI mapFrom, List<DBRefEntry> xrefs,
836 SequenceI[] retrieved, AlignedCodonFrame acf, boolean fromDna)
838 SequenceIdMatcher idMatcher = new SequenceIdMatcher(retrieved);
839 for (DBRefEntry xref : xrefs)
843 String targetSeqName = xref.getSource() + "|"
844 + xref.getAccessionId();
845 SequenceI[] matches = idMatcher.findAllIdMatches(targetSeqName);
850 for (SequenceI seq : matches)
852 constructMapping(mapFrom, seq, xref, acf, fromDna);
859 * Tries to make a mapping between sequences. If successful, adds the mapping
860 * to the dbref and the mappings collection and answers true, otherwise
861 * answers false. The following methods of making are mapping are tried in
864 * <li>if 'mapTo' holds a mapping to 'mapFrom', take the inverse; this is, for
865 * example, the case after fetching EMBL cross-references for a Uniprot
867 * <li>else check if the dna translates exactly to the protein (give or take
868 * start and stop codons></li>
869 * <li>else try to map based on CDS features on the dna sequence</li>
878 boolean constructMapping(SequenceI mapFrom, SequenceI mapTo,
879 DBRefEntry xref, AlignedCodonFrame mappings, boolean fromDna)
881 MapList mapping = null;
882 SequenceI dsmapFrom = mapFrom.getDatasetSequence() == null ? mapFrom
883 : mapFrom.getDatasetSequence();
884 SequenceI dsmapTo = mapTo.getDatasetSequence() == null ? mapTo
885 : mapTo.getDatasetSequence();
887 * look for a reverse mapping, if found make its inverse.
888 * Note - we do this on dataset sequences only.
890 if (dsmapTo.getDBRefs() != null)
892 for (DBRefEntry dbref : dsmapTo.getDBRefs())
894 String name = dbref.getSource() + "|" + dbref.getAccessionId();
895 if (dbref.hasMap() && dsmapFrom.getName().startsWith(name))
898 * looks like we've found a map from 'mapTo' to 'mapFrom'
899 * - invert it to make the mapping the other way
901 MapList reverse = dbref.getMap().getMap().getInverse();
902 xref.setMap(new Mapping(dsmapTo, reverse));
903 mappings.addMap(mapFrom, dsmapTo, reverse);
911 mapping = AlignmentUtils.mapCdnaToProtein(mapTo, mapFrom);
915 mapping = AlignmentUtils.mapCdnaToProtein(mapFrom, mapTo);
918 mapping = mapping.getInverse();
925 xref.setMap(new Mapping(mapTo, mapping));
928 * and add a reverse DbRef with the inverse mapping
930 if (mapFrom.getDatasetSequence() != null && false)
931 // && mapFrom.getDatasetSequence().getSourceDBRef() != null)
933 // possible need to search primary references... except, why doesn't xref
934 // == getSourceDBRef ??
935 // DBRefEntry dbref = new DBRefEntry(mapFrom.getDatasetSequence()
936 // .getSourceDBRef());
937 // dbref.setMap(new Mapping(mapFrom.getDatasetSequence(), mapping
939 // mapTo.addDBRef(dbref);
944 AlignmentUtils.computeProteinFeatures(mapFrom, mapTo, mapping);
945 mappings.addMap(mapFrom, mapTo, mapping);
949 mappings.addMap(mapTo, mapFrom, mapping.getInverse());
956 * find references to lrfs in the cross-reference set of each sequence in
957 * dataset (that is not equal to sequenceI) Identifies matching DBRefEntry
958 * based on source and accession string only - Map and Version are nulled.
961 * - true if context was searching from Dna sequences, false if
962 * context was searching from Protein sequences
966 * @return true if matches were found.
968 private boolean searchDatasetXrefs(boolean fromDna, SequenceI sequenceI,
969 List<DBRefEntry> lrfs, List<SequenceI> foundSeqs,
970 AlignedCodonFrame cf)
972 boolean found = false;
977 for (int i = 0, n = lrfs.size(); i < n; i++)
979 // DBRefEntry xref = new DBRefEntry(lrfs.get(i));
980 // // add in wildcards
981 // xref.setVersion(null);
982 // xref.setMap(null);
983 found |= searchDataset(fromDna, sequenceI, lrfs.get(i), foundSeqs, cf,
984 false, DBRefUtils.SEARCH_MODE_NO_MAP_NO_VERSION);
990 * Searches dataset for DBRefEntrys matching the given one (xrf) and adds the
991 * associated sequence to rseqs
994 * true if context was searching for refs *from* dna sequence, false
995 * if context was searching for refs *from* protein sequence
997 * a sequence to ignore (start point of search)
999 * a cross-reference to try to match
1001 * result list to add to
1003 * a set of sequence mappings to add to
1005 * - indicates the type of relationship between returned sequences,
1006 * xrf, and sequenceI that is required.
1008 * <li>direct implies xrf is a primary reference for sequenceI AND
1009 * the sequences to be located (eg a uniprot ID for a protein
1010 * sequence, and a uniprot ref on a transcript sequence).</li>
1011 * <li>indirect means xrf is a cross reference with respect to
1012 * sequenceI or all the returned sequences (eg a genomic reference
1013 * associated with a locus and one or more transcripts)</li>
1015 * @param mode SEARCH_MODE_FULL for all; SEARCH_MODE_NO_MAP_NO_VERSION optional
1016 * @return true if relationship found and sequence added.
1018 boolean searchDataset(boolean fromDna, SequenceI fromSeq, DBRefEntry xrf,
1019 List<SequenceI> foundSeqs, AlignedCodonFrame mappings,
1020 boolean direct, int mode)
1022 boolean found = false;
1023 if (dataset == null)
1027 if (dataset.getSequences() == null)
1029 System.err.println("Empty dataset sequence set - NO VECTOR");
1032 List<SequenceI> ds = dataset.getSequences();
1035 for (SequenceI nxt : ds)
1039 if (nxt.getDatasetSequence() != null)
1042 "Implementation warning: CrossRef initialised with a dataset alignment with non-dataset sequences in it! ("
1043 + nxt.getDisplayId(true) + " has ds reference "
1044 + nxt.getDatasetSequence().getDisplayId(true)
1047 if (nxt == fromSeq || nxt == fromSeq.getDatasetSequence())
1052 * only look at same molecule type if 'direct', or
1053 * complementary type if !direct
1056 boolean isDna = !nxt.isProtein();
1057 if (direct ? (isDna != fromDna) : (isDna == fromDna))
1059 // skip this sequence because it is wrong molecule type
1064 // look for direct or indirect references in common
1065 List<DBRefEntry> poss = nxt.getDBRefs();
1066 List<DBRefEntry> cands = null;
1068 // todo: indirect specifies we select either direct references to nxt
1069 // that match xrf which is indirect to sequenceI, or indirect
1070 // references to nxt that match xrf which is direct to sequenceI
1071 cands = DBRefUtils.searchRefs(poss, xrf, mode);
1074 // poss = DBRefUtils.selectDbRefs(nxt.isProtein()!fromDna, poss);
1075 // cands = DBRefUtils.searchRefs(poss, xrf);
1077 if (!cands.isEmpty())
1079 if (foundSeqs.contains(nxt))
1085 if (mappings != null && !direct)
1088 * if the matched sequence has mapped dbrefs to
1089 * protein product / cdna, add equivalent mappings to
1090 * our source sequence
1092 for (DBRefEntry candidate : cands)
1094 Mapping mapping = candidate.getMap();
1095 if (mapping != null)
1097 MapList map = mapping.getMap();
1098 if (mapping.getTo() != null
1099 && map.getFromRatio() != map.getToRatio())
1102 * add a mapping, as from dna to peptide sequence
1104 if (map.getFromRatio() == 3)
1106 mappings.addMap(nxt, fromSeq, map);
1110 mappings.addMap(nxt, fromSeq, map.getInverse());