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.SequenceFetcherFactory;
35 import jalview.ws.seqfetcher.ASequenceFetcher;
37 import java.util.ArrayList;
38 import java.util.Iterator;
39 import java.util.List;
42 * Functions for cross-referencing sequence databases.
50 * the dataset of the alignment for which we are searching for
51 * cross-references; in some cases we may resolve xrefs by
52 * searching in the dataset
54 private AlignmentI dataset;
57 * the sequences for which we are seeking cross-references
59 private SequenceI[] fromSeqs;
62 * matcher built from dataset
64 SequenceIdMatcher matcher;
67 * sequences found by cross-ref searches to fromSeqs
69 List<SequenceI> rseqs;
75 * the sequences for which we are seeking cross-references
77 * the containing alignment dataset (may be searched to resolve
80 public CrossRef(SequenceI[] seqs, AlignmentI ds)
83 dataset = ds.getDataset() == null ? ds : ds.getDataset();
87 * Returns a list of distinct database sources for which sequences have either
89 * <li>a (dna-to-protein or protein-to-dna) cross-reference</li>
90 * <li>an indirect cross-reference - a (dna-to-protein or protein-to-dna)
91 * reference from another sequence in the dataset which has a cross-reference
92 * to a direct DBRefEntry on the given sequence</li>
96 * - when true, cross-references *from* dna returned. When false,
97 * cross-references *from* protein are returned
100 public List<String> findXrefSourcesForSequences(boolean dna)
102 List<String> sources = new ArrayList<String>();
103 for (SequenceI seq : fromSeqs)
107 findXrefSourcesForSequence(seq, dna, sources);
110 sources.remove(DBRefSource.EMBL); // hack to prevent EMBL xrefs resulting in
111 // redundant datasets
114 sources.remove(DBRefSource.ENSEMBL); // hack to prevent Ensembl and
115 // EnsemblGenomes xref option shown
117 sources.remove(DBRefSource.ENSEMBLGENOMES);
119 // redundant datasets
124 * Returns a list of distinct database sources for which a sequence has either
126 * <li>a (dna-to-protein or protein-to-dna) cross-reference</li>
127 * <li>an indirect cross-reference - a (dna-to-protein or protein-to-dna)
128 * reference from another sequence in the dataset which has a cross-reference
129 * to a direct DBRefEntry on the given sequence</li>
133 * the sequence whose dbrefs we are searching against
135 * when true, context is DNA - so sources identifying protein
136 * products will be returned.
138 * a list of sources to add matches to
140 void findXrefSourcesForSequence(SequenceI seq, boolean fromDna,
141 List<String> sources)
144 * first find seq's xrefs (dna-to-peptide or peptide-to-dna)
146 DBRefEntry[] rfs = DBRefUtils.selectDbRefs(!fromDna, seq.getDBRefs());
147 addXrefsToSources(rfs, sources);
151 * find sequence's direct (dna-to-dna, peptide-to-peptide) xrefs
153 DBRefEntry[] lrfs = DBRefUtils.selectDbRefs(fromDna, seq.getDBRefs());
154 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
157 * find sequences in the alignment which xref one of these DBRefs
158 * i.e. is xref-ed to a common sequence identifier
160 searchDatasetXrefs(fromDna, seq, lrfs, foundSeqs, null);
163 * add those sequences' (dna-to-peptide or peptide-to-dna) dbref sources
165 for (SequenceI rs : foundSeqs)
167 DBRefEntry[] xrs = DBRefUtils
168 .selectDbRefs(!fromDna, rs.getDBRefs());
169 addXrefsToSources(xrs, sources);
175 * Helper method that adds the source identifiers of some cross-references to
176 * a (non-redundant) list of database sources
181 void addXrefsToSources(DBRefEntry[] xrefs, List<String> sources)
185 for (DBRefEntry ref : xrefs)
188 * avoid duplication e.g. ENSEMBL and Ensembl
190 String source = DBRefUtils.getCanonicalName(ref.getSource());
191 if (!sources.contains(source))
200 * Attempts to find cross-references from the sequences provided in the
201 * constructor to the given source database. Cross-references may be found
203 * <li>in dbrefs on the sequence which hold a mapping to a sequence
205 * <li>provided with a fetched sequence (e.g. ENA translation), or</li>
206 * <li>populated previously after getting cross-references</li>
208 * <li>as other sequences in the alignment which share a dbref identifier with
210 * <li>by fetching from the remote database</li>
212 * The cross-referenced sequences, and mappings to them, are added to the
216 * @return cross-referenced sequences (as dataset sequences)
218 public Alignment findXrefSequences(String source, boolean fromDna)
221 rseqs = new ArrayList<SequenceI>();
222 AlignedCodonFrame cf = new AlignedCodonFrame();
223 matcher = new SequenceIdMatcher(
224 dataset.getSequences());
226 for (SequenceI seq : fromSeqs)
229 while (dss.getDatasetSequence() != null)
231 dss = dss.getDatasetSequence();
233 boolean found = false;
234 DBRefEntry[] xrfs = DBRefUtils
235 .selectDbRefs(!fromDna, dss.getDBRefs());
236 // ENST & ENSP comes in to both Protein and nucleotide, so we need to
239 if ((xrfs == null || xrfs.length == 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 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 .println("Implementation problem (reopen JAL-2154): CrossRef.findInDataset seems to have recovered a different sequence than the one explicitly mapped for xref."
300 + xref.getMap().getTo()
304 /*matcher.findIdMatch(mappedTo);*/
305 if (matchInDataset != null)
307 if (!rseqs.contains(matchInDataset))
309 rseqs.add(matchInDataset);
311 // even if rseqs contained matchInDataset - check mappings between
312 // these seqs are added
313 // need to try harder to only add unique mappings
314 if (xref.getMap().getMap().isTripletMap()
315 && dataset.getMapping(seq, matchInDataset) == null
316 && cf.getMappingBetween(seq, matchInDataset) == null)
318 // materialise a mapping for highlighting between these
322 cf.addMap(dss, matchInDataset, xref.getMap().getMap(),
323 xref.getMap().getMappedFromId());
327 cf.addMap(matchInDataset, dss, xref.getMap().getMap()
328 .getInverse(), xref.getMap().getMappedFromId());
332 refIterator.remove();
335 // TODO: need to determine if this should be a deriveSequence
336 SequenceI rsq = new Sequence(mappedTo);
338 if (xref.getMap().getMap().isTripletMap())
340 // get sense of map correct for adding to product alignment.
343 // map is from dna seq to a protein product
344 cf.addMap(dss, rsq, xref.getMap().getMap(), xref.getMap()
349 // map should be from protein seq to its coding dna
350 cf.addMap(rsq, dss, xref.getMap().getMap().getInverse(),
351 xref.getMap().getMappedFromId());
359 SequenceI matchedSeq = matcher.findIdMatch(xref.getSource() + "|"
360 + xref.getAccessionId());
361 // if there was a match, check it's at least the right type of
363 if (matchedSeq != null && matchedSeq.isProtein() == fromDna)
365 if (constructMapping(seq, matchedSeq, xref, cf, fromDna))
374 // do a bit more work - search for sequences with references matching
375 // xrefs on this sequence.
376 found = searchDataset(fromDna, dss, xref, rseqs, cf, false);
380 refIterator.remove();
385 * fetch from source database any dbrefs we haven't resolved up to here
387 if (!sourceRefs.isEmpty())
389 retrieveCrossRef(sourceRefs, seq, xrfs, fromDna, cf);
393 Alignment ral = null;
394 if (rseqs.size() > 0)
396 ral = new Alignment(rseqs.toArray(new SequenceI[rseqs.size()]));
399 dataset.addCodonFrame(cf);
405 private void retrieveCrossRef(List<DBRefEntry> sourceRefs, SequenceI seq,
406 DBRefEntry[] xrfs, boolean fromDna, AlignedCodonFrame cf)
408 ASequenceFetcher sftch = SequenceFetcherFactory.getSequenceFetcher();
409 SequenceI[] retrieved = null;
410 SequenceI dss = seq.getDatasetSequence() == null ? seq : seq
411 .getDatasetSequence();
412 // first filter in case we are retrieving crossrefs that have already been
413 // retrieved. this happens for cases where a database record doesn't yield
414 // protein products for CDS
415 removeAlreadyRetrievedSeqs(sourceRefs, fromDna);
416 if (sourceRefs.size() == 0)
418 // no more work to do! We already had all requested sequence records in
424 retrieved = sftch.getSequences(sourceRefs, !fromDna);
425 } catch (Exception e)
428 .println("Problem whilst retrieving cross references for Sequence : "
433 if (retrieved != null)
435 boolean addedXref = false;
436 List<SequenceI> newDsSeqs = new ArrayList<SequenceI>(), doNotAdd = new ArrayList<SequenceI>();
438 for (SequenceI retrievedSequence : retrieved)
440 // dataset gets contaminated ccwith non-ds sequences. why ??!
441 // try: Ensembl -> Nuc->Ensembl, Nuc->Uniprot-->Protein->EMBL->
442 SequenceI retrievedDss = retrievedSequence.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.getDatasetSequence() == null ? retrievedSequence
457 : retrievedSequence.getDatasetSequence();
458 addedXref |= importCrossRefSeq(cf, newDsSeqs, doNotAdd, dss,
462 for (SequenceI newToSeq : newDsSeqs)
464 if (!doNotAdd.contains(newToSeq)
465 && dataset.findIndex(newToSeq) == -1)
467 dataset.addSequence(newToSeq);
468 matcher.add(newToSeq);
475 * Search dataset for sequences with a primary reference contained in
479 * - list of references to filter.
481 * - type of sequence to search for matching primary reference.
483 private void removeAlreadyRetrievedSeqs(List<DBRefEntry> sourceRefs,
486 DBRefEntry[] dbrSourceSet = sourceRefs.toArray(new DBRefEntry[0]);
487 for (SequenceI sq : dataset.getSequences())
489 boolean dupeFound = false;
490 // !fromDna means we are looking only for nucleotide sequences, not
492 if (sq.isProtein() == fromDna)
494 for (DBRefEntry dbr : sq.getPrimaryDBRefs())
496 for (DBRefEntry found : DBRefUtils.searchRefs(dbrSourceSet, dbr))
498 sourceRefs.remove(found);
505 // rebuild the search array from the filtered sourceRefs list
506 dbrSourceSet = sourceRefs.toArray(new DBRefEntry[0]);
512 * process sequence retrieved via a dbref on source sequence to resolve and
516 * @param sourceSequence
517 * @param retrievedSequence
518 * @return true if retrieveSequence was imported
520 private boolean importCrossRefSeq(AlignedCodonFrame cf,
521 List<SequenceI> newDsSeqs, List<SequenceI> doNotAdd,
522 SequenceI sourceSequence, SequenceI retrievedSequence)
525 * set when retrievedSequence has been verified as a crossreference for
528 boolean imported = false;
529 DBRefEntry[] dbr = retrievedSequence.getDBRefs();
532 for (DBRefEntry dbref : dbr)
534 SequenceI matched = findInDataset(dbref);
535 if (matched == sourceSequence)
537 // verified retrieved and source sequence cross-reference each other
540 // find any entry where we should put in the sequence being
541 // cross-referenced into the map
542 Mapping map = dbref.getMap();
545 if (map.getTo() != null && map.getMap() != null)
547 if (map.getTo() == sourceSequence)
549 // already called to import once, and most likely this sequence
550 // already imported !
556 * sequence is new to dataset, so save a reference so it can be added.
558 newDsSeqs.add(map.getTo());
563 * there was a matching sequence in dataset, so now, check to see if we can update the map.getTo() sequence to the existing one.
568 // compare ms with dss and replace with dss in mapping
569 // if map is congruent
570 SequenceI ms = map.getTo();
571 // TODO findInDataset requires exact sequence match but
572 // 'congruent' test is only for the mapped part
573 // maybe not a problem in practice since only ENA provide a
574 // mapping and it is to the full protein translation of CDS
575 // matcher.findIdMatch(map.getTo());
576 // TODO addendum: if matched is shorter than getTo, this will fail
577 // - when it should really succeed.
578 int sf = map.getMap().getToLowest();
579 int st = map.getMap().getToHighest();
580 SequenceI mappedrg = ms.getSubSequence(sf, st);
581 if (mappedrg.getLength() > 0
582 && ms.getSequenceAsString().equals(
583 matched.getSequenceAsString()))
586 * sequences were a match,
588 String msg = "Mapping updated from " + ms.getName()
589 + " to retrieved crossreference "
591 System.out.println(msg);
593 DBRefEntry[] toRefs = map.getTo().getDBRefs();
597 * transfer database refs
599 for (DBRefEntry ref : toRefs)
601 if (dbref.getSrcAccString().equals(
602 ref.getSrcAccString()))
604 continue; // avoid overwriting the ref on source sequence
606 matched.addDBRef(ref); // add or update mapping
609 doNotAdd.add(map.getTo());
613 * give the reverse reference the inverse mapping
614 * (if it doesn't have one already)
616 setReverseMapping(matched, dbref, cf);
619 * copy sequence features as well, avoiding
620 * duplication (e.g. same variation from two
623 SequenceFeature[] sfs = ms.getSequenceFeatures();
626 for (SequenceFeature feat : sfs)
629 * make a flyweight feature object which ignores Parent
630 * attribute in equality test; this avoids creating many
631 * otherwise duplicate exon features on genomic sequence
633 SequenceFeature newFeature = new SequenceFeature(
637 public boolean equals(Object o)
639 return super.equals(o, true);
642 matched.addSequenceFeature(newFeature);
647 cf.addMap(retrievedSequence, map.getTo(), map.getMap());
648 } catch (Exception e)
651 .println("Exception when consolidating Mapped sequence set...");
652 e.printStackTrace(System.err);
660 retrievedSequence.updatePDBIds();
661 rseqs.add(retrievedSequence);
662 if (dataset.findIndex(retrievedSequence) == -1)
664 dataset.addSequence(retrievedSequence);
665 matcher.add(retrievedSequence);
671 * Sets the inverse sequence mapping in the corresponding dbref of the mapped
672 * to sequence (if any). This is used after fetching a cross-referenced
673 * sequence, if the fetched sequence has a mapping to the original sequence,
674 * to set the mapping in the original sequence's dbref.
677 * the sequence mapped from
681 void setReverseMapping(SequenceI mapFrom, DBRefEntry dbref,
682 AlignedCodonFrame mappings)
684 SequenceI mapTo = dbref.getMap().getTo();
689 DBRefEntry[] dbrefs = mapTo.getDBRefs();
694 for (DBRefEntry toRef : dbrefs)
696 if (toRef.hasMap() && mapFrom == toRef.getMap().getTo())
699 * found the reverse dbref; update its mapping if null
701 if (toRef.getMap().getMap() == null)
703 MapList inverse = dbref.getMap().getMap().getInverse();
704 toRef.getMap().setMap(inverse);
705 mappings.addMap(mapTo, mapFrom, inverse);
712 * Returns null or the first sequence in the dataset which is identical to
713 * xref.mapTo, and has a) a primary dbref matching xref, or if none found, the
714 * first one with an ID source|xrefacc
717 * with map and mapped-to sequence
720 SequenceI findInDataset(DBRefEntry xref)
722 if (xref == null || !xref.hasMap() || xref.getMap().getTo() == null)
726 SequenceI mapsTo = xref.getMap().getTo();
727 String name = xref.getAccessionId();
728 String name2 = xref.getSource() + "|" + name;
729 SequenceI dss = mapsTo.getDatasetSequence() == null ? mapsTo : mapsTo
730 .getDatasetSequence();
731 // first check ds if ds is directly referenced
732 if (dataset.findIndex(dss) > -1)
736 DBRefEntry template = new DBRefEntry(xref.getSource(), null,
737 xref.getAccessionId());
739 * remember the first ID match - in case we don't find a match to template
741 SequenceI firstIdMatch = null;
742 for (SequenceI seq : dataset.getSequences())
744 // first check primary refs.
745 List<DBRefEntry> match = DBRefUtils.searchRefs(seq.getPrimaryDBRefs()
746 .toArray(new DBRefEntry[0]), template);
747 if (match != null && match.size() == 1 && sameSequence(seq, dss))
752 * clumsy alternative to using SequenceIdMatcher which currently
753 * returns sequences with a dbref to the matched accession id
754 * which we don't want
756 if (firstIdMatch == null
757 && (name.equals(seq.getName()) || seq.getName().startsWith(
760 if (sameSequence(seq, dss))
770 * Answers true if seq1 and seq2 contain exactly the same characters (ignoring
771 * case), else false. This method compares the lengths, then each character in
772 * turn, in order to 'fail fast'. For case-sensitive comparison, it would be
773 * possible to use Arrays.equals(seq1.getSequence(), seq2.getSequence()).
779 // TODO move to Sequence / SequenceI
780 static boolean sameSequence(SequenceI seq1, SequenceI seq2)
786 if (seq1 == null || seq2 == null)
790 char[] c1 = seq1.getSequence();
791 char[] c2 = seq2.getSequence();
792 if (c1.length != c2.length)
796 for (int i = 0; i < c1.length; i++)
798 int diff = c1[i] - c2[i];
800 * same char or differ in case only ('a'-'A' == 32)
802 if (diff != 0 && diff != 32 && diff != -32)
811 * Updates any empty mappings in the cross-references with one to a compatible
812 * retrieved sequence if found, and adds any new mappings to the
820 void updateDbrefMappings(SequenceI mapFrom, DBRefEntry[] xrefs,
821 SequenceI[] retrieved, AlignedCodonFrame acf, boolean fromDna)
823 SequenceIdMatcher idMatcher = new SequenceIdMatcher(retrieved);
824 for (DBRefEntry xref : xrefs)
828 String targetSeqName = xref.getSource() + "|"
829 + xref.getAccessionId();
830 SequenceI[] matches = idMatcher.findAllIdMatches(targetSeqName);
835 for (SequenceI seq : matches)
837 constructMapping(mapFrom, seq, xref, acf, fromDna);
844 * Tries to make a mapping between sequences. If successful, adds the mapping
845 * to the dbref and the mappings collection and answers true, otherwise
846 * answers false. The following methods of making are mapping are tried in
849 * <li>if 'mapTo' holds a mapping to 'mapFrom', take the inverse; this is, for
850 * example, the case after fetching EMBL cross-references for a Uniprot
852 * <li>else check if the dna translates exactly to the protein (give or take
853 * start and stop codons></li>
854 * <li>else try to map based on CDS features on the dna sequence</li>
863 boolean constructMapping(SequenceI mapFrom, SequenceI mapTo,
864 DBRefEntry xref, AlignedCodonFrame mappings, boolean fromDna)
866 MapList mapping = null;
867 SequenceI dsmapFrom = mapFrom.getDatasetSequence() == null ? mapFrom
868 : mapFrom.getDatasetSequence();
869 SequenceI dsmapTo = mapTo.getDatasetSequence() == null ? mapTo
870 : mapTo.getDatasetSequence();
872 * look for a reverse mapping, if found make its inverse.
873 * Note - we do this on dataset sequences only.
875 if (dsmapTo.getDBRefs() != null)
877 for (DBRefEntry dbref : dsmapTo.getDBRefs())
879 String name = dbref.getSource() + "|" + dbref.getAccessionId();
880 if (dbref.hasMap() && dsmapFrom.getName().startsWith(name))
883 * looks like we've found a map from 'mapTo' to 'mapFrom'
884 * - invert it to make the mapping the other way
886 MapList reverse = dbref.getMap().getMap().getInverse();
887 xref.setMap(new Mapping(dsmapTo, reverse));
888 mappings.addMap(mapFrom, dsmapTo, reverse);
896 mapping = AlignmentUtils.mapCdnaToProtein(mapTo, mapFrom);
900 mapping = AlignmentUtils.mapCdnaToProtein(mapFrom, mapTo);
903 mapping = mapping.getInverse();
910 xref.setMap(new Mapping(mapTo, mapping));
913 * and add a reverse DbRef with the inverse mapping
915 if (mapFrom.getDatasetSequence() != null && false)
916 // && mapFrom.getDatasetSequence().getSourceDBRef() != null)
918 // possible need to search primary references... except, why doesn't xref
919 // == getSourceDBRef ??
920 // DBRefEntry dbref = new DBRefEntry(mapFrom.getDatasetSequence()
921 // .getSourceDBRef());
922 // dbref.setMap(new Mapping(mapFrom.getDatasetSequence(), mapping
924 // mapTo.addDBRef(dbref);
929 AlignmentUtils.computeProteinFeatures(mapFrom, mapTo, mapping);
930 mappings.addMap(mapFrom, mapTo, mapping);
934 mappings.addMap(mapTo, mapFrom, mapping.getInverse());
941 * find references to lrfs in the cross-reference set of each sequence in
942 * dataset (that is not equal to sequenceI) Identifies matching DBRefEntry
943 * based on source and accession string only - Map and Version are nulled.
946 * - true if context was searching from Dna sequences, false if
947 * context was searching from Protein sequences
951 * @return true if matches were found.
953 private boolean searchDatasetXrefs(boolean fromDna, SequenceI sequenceI,
954 DBRefEntry[] lrfs, List<SequenceI> foundSeqs, AlignedCodonFrame cf)
956 boolean found = false;
961 for (int i = 0; i < lrfs.length; i++)
963 DBRefEntry xref = new DBRefEntry(lrfs[i]);
965 xref.setVersion(null);
967 found |= searchDataset(fromDna, sequenceI, xref, foundSeqs, cf, false);
973 * Searches dataset for DBRefEntrys matching the given one (xrf) and adds the
974 * associated sequence to rseqs
977 * true if context was searching for refs *from* dna sequence, false
978 * if context was searching for refs *from* protein sequence
980 * a sequence to ignore (start point of search)
982 * a cross-reference to try to match
984 * result list to add to
986 * a set of sequence mappings to add to
988 * - indicates the type of relationship between returned sequences,
989 * xrf, and sequenceI that is required.
991 * <li>direct implies xrf is a primary reference for sequenceI AND
992 * the sequences to be located (eg a uniprot ID for a protein
993 * sequence, and a uniprot ref on a transcript sequence).</li>
994 * <li>indirect means xrf is a cross reference with respect to
995 * sequenceI or all the returned sequences (eg a genomic reference
996 * associated with a locus and one or more transcripts)</li>
998 * @return true if relationship found and sequence added.
1000 boolean searchDataset(boolean fromDna, SequenceI fromSeq, DBRefEntry xrf,
1001 List<SequenceI> foundSeqs, AlignedCodonFrame mappings,
1004 boolean found = false;
1005 if (dataset == null)
1009 if (dataset.getSequences() == null)
1011 System.err.println("Empty dataset sequence set - NO VECTOR");
1015 synchronized (ds = dataset.getSequences())
1017 for (SequenceI nxt : ds)
1021 if (nxt.getDatasetSequence() != null)
1024 .println("Implementation warning: CrossRef initialised with a dataset alignment with non-dataset sequences in it! ("
1025 + nxt.getDisplayId(true)
1026 + " has ds reference "
1027 + nxt.getDatasetSequence().getDisplayId(true)
1030 if (nxt == fromSeq || nxt == fromSeq.getDatasetSequence())
1035 * only look at same molecule type if 'direct', or
1036 * complementary type if !direct
1039 boolean isDna = !nxt.isProtein();
1040 if (direct ? (isDna != fromDna) : (isDna == fromDna))
1042 // skip this sequence because it is wrong molecule type
1047 // look for direct or indirect references in common
1048 DBRefEntry[] poss = nxt.getDBRefs();
1049 List<DBRefEntry> cands = null;
1051 // todo: indirect specifies we select either direct references to nxt
1052 // that match xrf which is indirect to sequenceI, or indirect
1053 // references to nxt that match xrf which is direct to sequenceI
1054 cands = DBRefUtils.searchRefs(poss, xrf);
1057 // poss = DBRefUtils.selectDbRefs(nxt.isProtein()!fromDna, poss);
1058 // cands = DBRefUtils.searchRefs(poss, xrf);
1060 if (!cands.isEmpty())
1062 if (foundSeqs.contains(nxt))
1068 if (mappings != null && !direct)
1071 * if the matched sequence has mapped dbrefs to
1072 * protein product / cdna, add equivalent mappings to
1073 * our source sequence
1075 for (DBRefEntry candidate : cands)
1077 Mapping mapping = candidate.getMap();
1078 if (mapping != null)
1080 MapList map = mapping.getMap();
1081 if (mapping.getTo() != null
1082 && map.getFromRatio() != map.getToRatio())
1085 * add a mapping, as from dna to peptide sequence
1087 if (map.getFromRatio() == 3)
1089 mappings.addMap(nxt, fromSeq, map);
1093 mappings.addMap(nxt, fromSeq, map.getInverse());