X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2FCrossRef.java;h=40401fb69564874b0c72d374d1c9f8a77551394c;hb=9b85e1552fa57f02cf6cd312cfbd7efdfd079ea3;hp=05814c2784e5a0da1be2c4a21907c41de8e9f269;hpb=a7591fdadf3edeb556d1417967787a5bfdb1da32;p=jalview.git diff --git a/src/jalview/analysis/CrossRef.java b/src/jalview/analysis/CrossRef.java index 05814c2..40401fb 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; @@ -98,7 +99,7 @@ public class CrossRef */ public List findXrefSourcesForSequences(boolean dna) { - List sources = new ArrayList(); + List sources = new ArrayList<>(); for (SequenceI seq : fromSeqs) { if (seq != null) @@ -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; } @@ -140,7 +151,7 @@ public class CrossRef * find sequence's direct (dna-to-dna, peptide-to-peptide) xrefs */ DBRefEntry[] lrfs = DBRefUtils.selectDbRefs(fromDna, seq.getDBRefs()); - List foundSeqs = new ArrayList(); + List foundSeqs = new ArrayList<>(); /* * find sequences in the alignment which xref one of these DBRefs @@ -153,8 +164,8 @@ public class CrossRef */ for (SequenceI rs : foundSeqs) { - DBRefEntry[] xrs = DBRefUtils - .selectDbRefs(!fromDna, rs.getDBRefs()); + DBRefEntry[] xrs = DBRefUtils.selectDbRefs(!fromDna, + rs.getDBRefs()); addXrefsToSources(xrs, sources); } } @@ -207,10 +218,9 @@ public class CrossRef public Alignment findXrefSequences(String source, boolean fromDna) { - rseqs = new ArrayList(); + rseqs = new ArrayList<>(); AlignedCodonFrame cf = new AlignedCodonFrame(); - matcher = new SequenceIdMatcher( - dataset.getSequences()); + matcher = new SequenceIdMatcher(dataset.getSequences()); for (SequenceI seq : fromSeqs) { @@ -220,8 +230,8 @@ public class CrossRef dss = dss.getDatasetSequence(); } boolean found = false; - DBRefEntry[] xrfs = DBRefUtils - .selectDbRefs(!fromDna, dss.getDBRefs()); + DBRefEntry[] xrfs = DBRefUtils.selectDbRefs(!fromDna, + dss.getDBRefs()); // ENST & ENSP comes in to both Protein and nucleotide, so we need to // filter them // out later. @@ -281,13 +291,10 @@ public class CrossRef if (matchInDataset != null && xref.getMap().getTo() != null && matchInDataset != xref.getMap().getTo()) { - System.err - .println("Implementation problem (reopen JAL-2154): CrossRef.findInDataset seems to have recovered a different sequence than the one explicitly mapped for xref." - + "Found:" - + matchInDataset - + "\nExpected:" - + xref.getMap().getTo() - + "\nFor xref:" + System.err.println( + "Implementation problem (reopen JAL-2154): CrossRef.findInDataset seems to have recovered a different sequence than the one explicitly mapped for xref." + + "Found:" + matchInDataset + "\nExpected:" + + xref.getMap().getTo() + "\nFor xref:" + xref); } /*matcher.findIdMatch(mappedTo);*/ @@ -313,8 +320,9 @@ public class CrossRef } else { - cf.addMap(matchInDataset, dss, xref.getMap().getMap() - .getInverse(), xref.getMap().getMappedFromId()); + cf.addMap(matchInDataset, dss, + xref.getMap().getMap().getInverse(), + xref.getMap().getMappedFromId()); } } @@ -330,8 +338,8 @@ public class CrossRef if (fromDna) { // map is from dna seq to a protein product - cf.addMap(dss, rsq, xref.getMap().getMap(), xref.getMap() - .getMappedFromId()); + cf.addMap(dss, rsq, xref.getMap().getMap(), + xref.getMap().getMappedFromId()); } else { @@ -345,8 +353,8 @@ public class CrossRef if (!found) { - SequenceI matchedSeq = matcher.findIdMatch(xref.getSource() + "|" - + xref.getAccessionId()); + SequenceI matchedSeq = matcher.findIdMatch( + xref.getSource() + "|" + xref.getAccessionId()); // if there was a match, check it's at least the right type of // molecule! if (matchedSeq != null && matchedSeq.isProtein() == fromDna) @@ -396,8 +404,8 @@ public class CrossRef { ASequenceFetcher sftch = SequenceFetcherFactory.getSequenceFetcher(); SequenceI[] retrieved = null; - SequenceI dss = seq.getDatasetSequence() == null ? seq : seq - .getDatasetSequence(); + SequenceI dss = seq.getDatasetSequence() == null ? seq + : seq.getDatasetSequence(); // first filter in case we are retrieving crossrefs that have already been // retrieved. this happens for cases where a database record doesn't yield // protein products for CDS @@ -413,23 +421,55 @@ public class CrossRef retrieved = sftch.getSequences(sourceRefs, !fromDna); } catch (Exception e) { - System.err - .println("Problem whilst retrieving cross references for Sequence : " + System.err.println( + "Problem whilst retrieving cross references for Sequence : " + seq.getName()); e.printStackTrace(); } 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(); - DBRefEntry[] dbr = retrievedSequence.getDBRefs(); - if (dbr != null) + SequenceI retrievedDss = retrievedSequence + .getDatasetSequence() == null ? retrievedSequence + : retrievedSequence.getDatasetSequence(); + 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 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); + } + } + } + } /** * Search dataset for sequences with a primary reference contained in @@ -462,128 +502,167 @@ public class CrossRef } if (dupeFound) { + // rebuild the search array from the filtered sourceRefs list dbrSourceSet = sourceRefs.toArray(new DBRefEntry[0]); } } } + /** + * process sequence retrieved via a dbref on source sequence to resolve and + * transfer data + * + * @param cf + * @param sourceSequence + * @param retrievedSequence + * @return true if retrieveSequence was imported + */ + 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(); + if (map != null) { - for (DBRefEntry dbref : dbr) + if (map.getTo() != null && map.getMap() != null) { - // find any entry where we should put in the sequence being - // cross-referenced into the map - Mapping map = dbref.getMap(); - if (map != null) + if (map.getTo() == sourceSequence) { - if (map.getTo() != null && map.getMap() != null) + // already called to import once, and most likely this sequence + // already imported ! + continue; + } + if (matched == null) + { + /* + * 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(matched.getSequenceAsString())) { - // 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) + /* + * sequences were a match, + */ + String msg = "Mapping updated from " + ms.getName() + + " to retrieved crossreference " + + matched.getName(); + System.out.println(msg); + + DBRefEntry[] toRefs = map.getTo().getDBRefs(); + if (toRefs != null) { /* - * already got an xref to this sequence; update this - * map to point to the same sequence, and add - * any new dbrefs to it + * transfer database refs */ - DBRefEntry[] toRefs = map.getTo().getDBRefs(); - if (toRefs != null) + for (DBRefEntry ref : toRefs) { - for (DBRefEntry ref : toRefs) + if (dbref.getSrcAccString() + .equals(ref.getSrcAccString())) { - matched.addDBRef(ref); // add or update mapping + continue; // avoid overwriting the ref on source sequence } - } - map.setTo(matched); - } - else - { - if (dataset.findIndex(map.getTo()) == -1) - { - dataset.addSequence(map.getTo()); - matcher.add(map.getTo()); + matched.addDBRef(ref); // add or update mapping } } - try - { - // compare ms with dss and replace with dss in mapping - // if map is congruent - SequenceI ms = map.getTo(); - int sf = map.getMap().getToLowest(); - int st = map.getMap().getToHighest(); - SequenceI mappedrg = ms.getSubSequence(sf, st); - // SequenceI loc = dss.getSubSequence(sf, st); - if (mappedrg.getLength() > 0 - && ms.getSequenceAsString().equals( - dss.getSequenceAsString())) - // && mappedrg.getSequenceAsString().equals( - // loc.getSequenceAsString())) - { - String msg = "Mapping updated from " + ms.getName() - + " to retrieved crossreference " - + dss.getName(); - System.out.println(msg); - map.setTo(dss); + doNotAdd.add(map.getTo()); + map.setTo(matched); - /* - * give the reverse reference the inverse mapping - * (if it doesn't have one already) - */ - setReverseMapping(dss, dbref, cf); + /* + * give the reverse reference the inverse mapping + * (if it doesn't have one already) + */ + setReverseMapping(matched, dbref, cf); - /* - * copy sequence features as well, avoiding - * duplication (e.g. same variation from two - * transcripts) - */ - SequenceFeature[] sfs = ms.getSequenceFeatures(); - if (sfs != null) + /* + * copy sequence features as well, avoiding + * duplication (e.g. same variation from two + * transcripts) + */ + List sfs = ms.getFeatures() + .getAllFeatures(); + for (SequenceFeature feat : sfs) + { + /* + * make a flyweight feature object which ignores Parent + * attribute in equality test; this avoids creating many + * otherwise duplicate exon features on genomic sequence + */ + SequenceFeature newFeature = new SequenceFeature(feat) + { + @Override + public boolean equals(Object o) { - for (SequenceFeature feat : sfs) - { - /* - * make a flyweight feature object which ignores Parent - * attribute in equality test; this avoids creating many - * otherwise duplicate exon features on genomic sequence - */ - SequenceFeature newFeature = new SequenceFeature( - feat) - { - @Override - public boolean equals(Object o) - { - return super.equals(o, true); - } - }; - dss.addSequenceFeature(newFeature); - } + return super.equals(o, true); } - } - cf.addMap(retrievedDss, map.getTo(), map.getMap()); - } catch (Exception e) - { - System.err - .println("Exception when consolidating Mapped sequence set..."); - e.printStackTrace(System.err); + }; + matched.addSequenceFeature(newFeature); } } + cf.addMap(retrievedSequence, map.getTo(), map.getMap()); + } catch (Exception e) + { + System.err.println( + "Exception when consolidating Mapped sequence set..."); + e.printStackTrace(System.err); } } } - retrievedSequence.updatePDBIds(); - rseqs.add(retrievedDss); - if (dataset.findIndex(retrievedDss) == -1) - { - dataset.addSequence(retrievedDss); - matcher.add(retrievedDss); - } } } + 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 @@ -626,9 +705,12 @@ public class CrossRef } /** - * Returns the first identical sequence in the dataset if any, else null + * Returns null or the first sequence in the dataset which is identical to + * xref.mapTo, and has a) a primary dbref matching xref, or if none found, the + * first one with an ID source|xrefacc * * @param xref + * with map and mapped-to sequence * @return */ SequenceI findInDataset(DBRefEntry xref) @@ -640,30 +722,43 @@ public class CrossRef SequenceI mapsTo = xref.getMap().getTo(); String name = xref.getAccessionId(); String name2 = xref.getSource() + "|" + name; - SequenceI dss = mapsTo.getDatasetSequence() == null ? mapsTo : mapsTo - .getDatasetSequence(); + SequenceI dss = mapsTo.getDatasetSequence() == null ? mapsTo + : mapsTo.getDatasetSequence(); // first check ds if ds is directly referenced if (dataset.findIndex(dss) > -1) { return dss; } - ; + DBRefEntry template = new DBRefEntry(xref.getSource(), null, + xref.getAccessionId()); + /** + * remember the first ID match - in case we don't find a match to template + */ + SequenceI firstIdMatch = null; for (SequenceI seq : dataset.getSequences()) { + // first check primary refs. + List match = DBRefUtils.searchRefs( + seq.getPrimaryDBRefs().toArray(new DBRefEntry[0]), template); + if (match != null && match.size() == 1 && sameSequence(seq, dss)) + { + return seq; + } /* * clumsy alternative to using SequenceIdMatcher which currently * returns sequences with a dbref to the matched accession id * which we don't want */ - if (name.equals(seq.getName()) || seq.getName().startsWith(name2)) + if (firstIdMatch == null && (name.equals(seq.getName()) + || seq.getName().startsWith(name2))) { if (sameSequence(seq, dss)) { - return seq; + firstIdMatch = seq; } } } - return null; + return firstIdMatch; } /** @@ -687,15 +782,15 @@ public class CrossRef { return false; } - char[] c1 = seq1.getSequence(); - char[] c2 = seq2.getSequence(); - if (c1.length != c2.length) + + if (seq1.getLength() != seq2.getLength()) { return false; } - for (int i = 0; i < c1.length; i++) + int length = seq1.getLength(); + for (int i = 0; i < length; i++) { - int diff = c1[i] - c2[i]; + int diff = seq1.getCharAt(i) - seq2.getCharAt(i); /* * same char or differ in case only ('a'-'A' == 32) */ @@ -826,7 +921,7 @@ public class CrossRef if (fromDna) { - AlignmentUtils.computeProteinFeatures(mapFrom, mapTo, mapping); + // AlignmentUtils.computeProteinFeatures(mapFrom, mapTo, mapping); mappings.addMap(mapFrom, mapTo, mapping); } else @@ -851,7 +946,8 @@ public class CrossRef * @return true if matches were found. */ private boolean searchDatasetXrefs(boolean fromDna, SequenceI sequenceI, - DBRefEntry[] lrfs, List foundSeqs, AlignedCodonFrame cf) + DBRefEntry[] lrfs, List foundSeqs, + AlignedCodonFrame cf) { boolean found = false; if (lrfs == null) @@ -864,7 +960,8 @@ public class CrossRef // add in wildcards xref.setVersion(null); xref.setMap(null); - found |= searchDataset(fromDna, sequenceI, xref, foundSeqs, cf, false); + found |= searchDataset(fromDna, sequenceI, xref, foundSeqs, cf, + false); } return found; } @@ -920,10 +1017,9 @@ public class CrossRef { if (nxt.getDatasetSequence() != null) { - System.err - .println("Implementation warning: CrossRef initialised with a dataset alignment with non-dataset sequences in it! (" - + nxt.getDisplayId(true) - + " has ds reference " + System.err.println( + "Implementation warning: CrossRef initialised with a dataset alignment with non-dataset sequences in it! (" + + nxt.getDisplayId(true) + " has ds reference " + nxt.getDatasetSequence().getDisplayId(true) + ")"); }