X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2FCrossRef.java;h=4ba7e416e7d115d183d68ce50878e21f1377b892;hb=97f870ff4890ebfadcbb73e6793c83f856948658;hp=1ddeb20213479bbbf2a0ff0c7639b1307c2f3550;hpb=639ac6c56cbbd596611ba7a2152380d6c9474090;p=jalview.git diff --git a/src/jalview/analysis/CrossRef.java b/src/jalview/analysis/CrossRef.java index 1ddeb20..4ba7e41 100644 --- a/src/jalview/analysis/CrossRef.java +++ b/src/jalview/analysis/CrossRef.java @@ -24,6 +24,7 @@ import jalview.datamodel.AlignedCodonFrame; import jalview.datamodel.Alignment; import jalview.datamodel.AlignmentI; import jalview.datamodel.DBRefEntry; +import jalview.datamodel.DBRefSource; import jalview.datamodel.Mapping; import jalview.datamodel.Sequence; import jalview.datamodel.SequenceFeature; @@ -106,6 +107,16 @@ public class CrossRef findXrefSourcesForSequence(seq, dna, sources); } } + sources.remove(DBRefSource.EMBL); // hack to prevent EMBL xrefs resulting in + // redundant datasets + if (dna) + { + sources.remove(DBRefSource.ENSEMBL); // hack to prevent Ensembl and + // EnsemblGenomes xref option shown + // from cdna panel + sources.remove(DBRefSource.ENSEMBLGENOMES); + } + // redundant datasets return sources; } @@ -209,8 +220,7 @@ public class CrossRef rseqs = new ArrayList(); AlignedCodonFrame cf = new AlignedCodonFrame(); - matcher = new SequenceIdMatcher( - dataset.getSequences()); + matcher = new SequenceIdMatcher(dataset.getSequences()); for (SequenceI seq : fromSeqs) { @@ -421,19 +431,40 @@ public class CrossRef if (retrieved != null) { - updateDbrefMappings(seq, xrfs, retrieved, cf, fromDna); + boolean addedXref = false; + List newDsSeqs = new ArrayList(), doNotAdd = new ArrayList(); + for (SequenceI retrievedSequence : retrieved) { // dataset gets contaminated ccwith non-ds sequences. why ??! // try: Ensembl -> Nuc->Ensembl, Nuc->Uniprot-->Protein->EMBL-> SequenceI retrievedDss = retrievedSequence.getDatasetSequence() == null ? retrievedSequence : retrievedSequence.getDatasetSequence(); - importCrossRefSeq(cf, dss, retrievedDss); - rseqs.add(retrievedDss); - if (dataset.findIndex(retrievedDss) == -1) + addedXref |= importCrossRefSeq(cf, newDsSeqs, doNotAdd, dss, + retrievedDss); + } + if (!addedXref) + { + // try again, after looking for matching IDs + // shouldn't need to do this unless the dbref mechanism has broken. + updateDbrefMappings(seq, xrfs, retrieved, cf, fromDna); + for (SequenceI retrievedSequence : retrieved) { - dataset.addSequence(retrievedDss); - matcher.add(retrievedDss); + // dataset gets contaminated ccwith non-ds sequences. why ??! + // try: Ensembl -> Nuc->Ensembl, Nuc->Uniprot-->Protein->EMBL-> + SequenceI retrievedDss = retrievedSequence.getDatasetSequence() == null ? retrievedSequence + : retrievedSequence.getDatasetSequence(); + addedXref |= importCrossRefSeq(cf, newDsSeqs, doNotAdd, dss, + retrievedDss); + } + } + for (SequenceI newToSeq : newDsSeqs) + { + if (!doNotAdd.contains(newToSeq) + && dataset.findIndex(newToSeq) == -1) + { + dataset.addSequence(newToSeq); + matcher.add(newToSeq); } } } @@ -470,6 +501,7 @@ public class CrossRef } if (dupeFound) { + // rebuild the search array from the filtered sourceRefs list dbrSourceSet = sourceRefs.toArray(new DBRefEntry[0]); } } @@ -482,15 +514,28 @@ public class CrossRef * @param cf * @param sourceSequence * @param retrievedSequence + * @return true if retrieveSequence was imported */ - private void importCrossRefSeq(AlignedCodonFrame cf, + private boolean importCrossRefSeq(AlignedCodonFrame cf, + List newDsSeqs, List doNotAdd, SequenceI sourceSequence, SequenceI retrievedSequence) { + /** + * set when retrievedSequence has been verified as a crossreference for + * sourceSequence + */ + boolean imported = false; DBRefEntry[] dbr = retrievedSequence.getDBRefs(); if (dbr != null) { for (DBRefEntry dbref : dbr) { + SequenceI matched = findInDataset(dbref); + if (matched == sourceSequence) + { + // verified retrieved and source sequence cross-reference each other + imported = true; + } // find any entry where we should put in the sequence being // cross-referenced into the map Mapping map = dbref.getMap(); @@ -498,61 +543,76 @@ public class CrossRef { if (map.getTo() != null && map.getMap() != null) { - // TODO findInDataset requires exact sequence match but - // 'congruent' test is only for the mapped part - // maybe not a problem in practice since only ENA provide a - // mapping and it is to the full protein translation of CDS - SequenceI matched = findInDataset(dbref); - // matcher.findIdMatch(map.getTo()); - if (matched != null) + if (map.getTo() == sourceSequence) { - /* - * already got an xref to this sequence; update this - * map to point to the same sequence, and add - * any new dbrefs to it - */ - DBRefEntry[] toRefs = map.getTo().getDBRefs(); - if (toRefs != null) - { - for (DBRefEntry ref : toRefs) - { - matched.addDBRef(ref); // add or update mapping - } - } - map.setTo(matched); + // already called to import once, and most likely this sequence + // already imported ! + continue; } - else + if (matched == null) { - if (dataset.findIndex(map.getTo()) == -1) - { - dataset.addSequence(map.getTo()); - matcher.add(map.getTo()); - } + /* + * sequence is new to dataset, so save a reference so it can be added. + */ + newDsSeqs.add(map.getTo()); + continue; } + /* + * there was a matching sequence in dataset, so now, check to see if we can update the map.getTo() sequence to the existing one. + */ + try { // compare ms with dss and replace with dss in mapping // if map is congruent SequenceI ms = map.getTo(); + // TODO findInDataset requires exact sequence match but + // 'congruent' test is only for the mapped part + // maybe not a problem in practice since only ENA provide a + // mapping and it is to the full protein translation of CDS + // matcher.findIdMatch(map.getTo()); + // TODO addendum: if matched is shorter than getTo, this will fail + // - when it should really succeed. int sf = map.getMap().getToLowest(); int st = map.getMap().getToHighest(); SequenceI mappedrg = ms.getSubSequence(sf, st); if (mappedrg.getLength() > 0 && ms.getSequenceAsString().equals( - sourceSequence.getSequenceAsString())) + matched.getSequenceAsString())) { + /* + * sequences were a match, + */ String msg = "Mapping updated from " + ms.getName() + " to retrieved crossreference " - + sourceSequence.getName(); + + matched.getName(); System.out.println(msg); - map.setTo(sourceSequence); + + DBRefEntry[] toRefs = map.getTo().getDBRefs(); + if (toRefs != null) + { + /* + * transfer database refs + */ + for (DBRefEntry ref : toRefs) + { + if (dbref.getSrcAccString().equals( + ref.getSrcAccString())) + { + continue; // avoid overwriting the ref on source sequence + } + matched.addDBRef(ref); // add or update mapping + } + } + doNotAdd.add(map.getTo()); + map.setTo(matched); /* * give the reverse reference the inverse mapping * (if it doesn't have one already) */ - setReverseMapping(sourceSequence, dbref, cf); + setReverseMapping(matched, dbref, cf); /* * copy sequence features as well, avoiding @@ -569,8 +629,7 @@ public class CrossRef * attribute in equality test; this avoids creating many * otherwise duplicate exon features on genomic sequence */ - SequenceFeature newFeature = new SequenceFeature( - feat) + SequenceFeature newFeature = new SequenceFeature(feat) { @Override public boolean equals(Object o) @@ -578,9 +637,10 @@ public class CrossRef return super.equals(o, true); } }; - sourceSequence.addSequenceFeature(newFeature); + matched.addSequenceFeature(newFeature); } } + } cf.addMap(retrievedSequence, map.getTo(), map.getMap()); } catch (Exception e) @@ -593,8 +653,19 @@ public class CrossRef } } } - retrievedSequence.updatePDBIds(); + if (imported) + { + retrievedSequence.updatePDBIds(); + rseqs.add(retrievedSequence); + if (dataset.findIndex(retrievedSequence) == -1) + { + dataset.addSequence(retrievedSequence); + matcher.add(retrievedSequence); + } + } + return imported; } + /** * Sets the inverse sequence mapping in the corresponding dbref of the mapped * to sequence (if any). This is used after fetching a cross-referenced @@ -794,8 +865,8 @@ public class CrossRef MapList mapping = null; SequenceI dsmapFrom = mapFrom.getDatasetSequence() == null ? mapFrom : mapFrom.getDatasetSequence(); - SequenceI dsmapTo = mapTo.getDatasetSequence() == null ? mapTo - : mapTo.getDatasetSequence(); + SequenceI dsmapTo = mapTo.getDatasetSequence() == null ? mapTo : mapTo + .getDatasetSequence(); /* * look for a reverse mapping, if found make its inverse. * Note - we do this on dataset sequences only.