X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2FCrossRef.java;h=c69858f01d080be00983a90aaf5cf3e863e3592f;hb=da1ce6ca071deb35b983a41f6eeed43f30b3c3b2;hp=46e211995160b7ba78e6db73038185ad1e8eb57a;hpb=6e07596ded06cdd6fb75978a4ee9debb1294a17c;p=jalview.git diff --git a/src/jalview/analysis/CrossRef.java b/src/jalview/analysis/CrossRef.java index 46e2119..c69858f 100644 --- a/src/jalview/analysis/CrossRef.java +++ b/src/jalview/analysis/CrossRef.java @@ -24,14 +24,14 @@ 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; import jalview.datamodel.SequenceI; import jalview.util.DBRefUtils; import jalview.util.MapList; -import jalview.ws.SequenceFetcherFactory; -import jalview.ws.seqfetcher.ASequenceFetcher; +import jalview.ws.SequenceFetcher; import java.util.ArrayList; import java.util.Iterator; @@ -98,7 +98,7 @@ 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 +106,16 @@ 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; } @@ -132,15 +142,15 @@ public class CrossRef /* * first find seq's xrefs (dna-to-peptide or peptide-to-dna) */ - DBRefEntry[] rfs = DBRefUtils.selectDbRefs(!fromDna, seq.getDBRefs()); + List rfs = DBRefUtils.selectDbRefs(!fromDna, seq.getDBRefs()); addXrefsToSources(rfs, sources); if (dataset != null) { /* * find sequence's direct (dna-to-dna, peptide-to-peptide) xrefs */ - DBRefEntry[] lrfs = DBRefUtils.selectDbRefs(fromDna, seq.getDBRefs()); - List foundSeqs = new ArrayList(); + List lrfs = DBRefUtils.selectDbRefs(fromDna, seq.getDBRefs()); + List foundSeqs = new ArrayList<>(); /* * find sequences in the alignment which xref one of these DBRefs @@ -153,8 +163,8 @@ public class CrossRef */ for (SequenceI rs : foundSeqs) { - DBRefEntry[] xrs = DBRefUtils - .selectDbRefs(!fromDna, rs.getDBRefs()); + List xrs = DBRefUtils.selectDbRefs(!fromDna, + rs.getDBRefs()); addXrefsToSources(xrs, sources); } } @@ -167,7 +177,7 @@ public class CrossRef * @param xrefs * @param sources */ - void addXrefsToSources(DBRefEntry[] xrefs, List sources) + void addXrefsToSources(List xrefs, List sources) { if (xrefs != null) { @@ -207,10 +217,9 @@ 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) { @@ -220,18 +229,18 @@ public class CrossRef dss = dss.getDatasetSequence(); } boolean found = false; - DBRefEntry[] xrfs = DBRefUtils - .selectDbRefs(!fromDna, dss.getDBRefs()); + List xrfs = DBRefUtils.selectDbRefs(!fromDna, + dss.getDBRefs()); // ENST & ENSP comes in to both Protein and nucleotide, so we need to // filter them // out later. - if ((xrfs == null || xrfs.length == 0) && dataset != null) + if ((xrfs == null || xrfs.size() == 0) && dataset != null) { /* * found no suitable dbrefs on sequence - look for sequences in the * alignment which share a dbref with this one */ - DBRefEntry[] lrfs = DBRefUtils.selectDbRefs(fromDna, + List lrfs = DBRefUtils.selectDbRefs(fromDna, seq.getDBRefs()); /* @@ -281,13 +290,10 @@ 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 +319,9 @@ 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 +337,8 @@ 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 +352,8 @@ 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) @@ -362,7 +369,7 @@ public class CrossRef { // do a bit more work - search for sequences with references matching // xrefs on this sequence. - found = searchDataset(fromDna, dss, xref, rseqs, cf, false); + found = searchDataset(fromDna, dss, xref, rseqs, cf, false, DBRefUtils.SEARCH_MODE_FULL); } if (found) { @@ -392,12 +399,11 @@ public class CrossRef } private void retrieveCrossRef(List sourceRefs, SequenceI seq, - DBRefEntry[] xrfs, boolean fromDna, AlignedCodonFrame cf) + List xrfs, boolean fromDna, AlignedCodonFrame cf) { - 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 @@ -410,30 +416,55 @@ public class CrossRef } try { - retrieved = sftch.getSequences(sourceRefs, !fromDna); + retrieved = SequenceFetcher.getInstance() + .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(); - importCrossRefSeq(cf, dss, retrievedDss); - rseqs.add(retrievedDss); - if (dataset.findIndex(retrievedDss) == -1) + 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.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); } } } @@ -451,19 +482,24 @@ public class CrossRef private void removeAlreadyRetrievedSeqs(List sourceRefs, boolean fromDna) { - DBRefEntry[] dbrSourceSet = sourceRefs.toArray(new DBRefEntry[0]); - for (SequenceI sq : dataset.getSequences()) + List dbrSourceSet = new ArrayList<>(sourceRefs); + List dsSeqs = dataset.getSequences(); + for (int ids = 0, nds = dsSeqs.size(); ids < nds; ids++) { + SequenceI sq = dsSeqs.get(ids); boolean dupeFound = false; // !fromDna means we are looking only for nucleotide sequences, not // protein if (sq.isProtein() == fromDna) { - for (DBRefEntry dbr : sq.getPrimaryDBRefs()) + List sqdbrefs = sq.getPrimaryDBRefs(); + for (int idb = 0, ndb = sqdbrefs.size(); idb < ndb; idb++) { - for (DBRefEntry found : DBRefUtils.searchRefs(dbrSourceSet, dbr)) + DBRefEntry dbr = sqdbrefs.get(idb); + List searchrefs = DBRefUtils.searchRefs(dbrSourceSet, dbr, DBRefUtils.SEARCH_MODE_FULL); + for (int isr = 0, nsr = searchrefs.size(); isr < nsr; isr++) { - sourceRefs.remove(found); + sourceRefs.remove(searchrefs.get(isr)); dupeFound = true; } } @@ -471,7 +507,8 @@ public class CrossRef if (dupeFound) { // rebuild the search array from the filtered sourceRefs list - dbrSourceSet = sourceRefs.toArray(new DBRefEntry[0]); + dbrSourceSet.clear(); + dbrSourceSet.addAll(sourceRefs); } } } @@ -483,119 +520,156 @@ 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) { - DBRefEntry[] dbr = retrievedSequence.getDBRefs(); + /** + * set when retrievedSequence has been verified as a crossreference for + * sourceSequence + */ + boolean imported = false; + List dbr = retrievedSequence.getDBRefs(); if (dbr != null) { - for (DBRefEntry dbref : dbr) + for (int ib = 0, nb = dbr.size(); ib < nb; ib++) { + + DBRefEntry dbref = dbr.get(ib); + 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) { - if (map.getTo() != null && map.getMap() != null) + SequenceI ms = map.getTo(); + if (ms != 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 (ms == 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(ms); + 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())) + if (mappedrg.getLength() > 0 && ms.getSequenceAsString() + .equals(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); + + List 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 * duplication (e.g. same variation from two * transcripts) */ - SequenceFeature[] sfs = ms.getSequenceFeatures(); - if (sfs != null) + List sfs = ms.getFeatures() + .getAllFeatures(); + for (SequenceFeature feat : sfs) { - 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) { - /* - * 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) { - @Override - public boolean equals(Object o) - { - return super.equals(o, true); - } - }; - sourceSequence.addSequenceFeature(newFeature); - } + return super.equals(o, true); + } + }; + matched.addSequenceFeature(newFeature); } } cf.addMap(retrievedSequence, map.getTo(), map.getMap()); } catch (Exception e) { - System.err - .println("Exception when consolidating Mapped sequence set..."); + System.err.println( + "Exception when consolidating Mapped sequence set..."); e.printStackTrace(System.err); } } } } } - 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 @@ -615,7 +689,7 @@ public class CrossRef { return; } - DBRefEntry[] dbrefs = mapTo.getDBRefs(); + List dbrefs = mapTo.getDBRefs(); if (dbrefs == null) { return; @@ -655,8 +729,8 @@ 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) { @@ -671,8 +745,8 @@ public class CrossRef for (SequenceI seq : dataset.getSequences()) { // first check primary refs. - List match = DBRefUtils.searchRefs(seq.getPrimaryDBRefs() - .toArray(new DBRefEntry[0]), template); + List match = DBRefUtils.searchRefs( + seq.getPrimaryDBRefs(), template, DBRefUtils.SEARCH_MODE_FULL); if (match != null && match.size() == 1 && sameSequence(seq, dss)) { return seq; @@ -682,9 +756,8 @@ public class CrossRef * returns sequences with a dbref to the matched accession id * which we don't want */ - if (firstIdMatch == null - && (name.equals(seq.getName()) || seq.getName().startsWith( - name2))) + if (firstIdMatch == null && (name.equals(seq.getName()) + || seq.getName().startsWith(name2))) { if (sameSequence(seq, dss)) { @@ -716,15 +789,15 @@ 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) */ @@ -746,7 +819,7 @@ public class CrossRef * @param retrieved * @param acf */ - void updateDbrefMappings(SequenceI mapFrom, DBRefEntry[] xrefs, + void updateDbrefMappings(SequenceI mapFrom, List xrefs, SequenceI[] retrieved, AlignedCodonFrame acf, boolean fromDna) { SequenceIdMatcher idMatcher = new SequenceIdMatcher(retrieved); @@ -880,20 +953,22 @@ public class CrossRef * @return true if matches were found. */ private boolean searchDatasetXrefs(boolean fromDna, SequenceI sequenceI, - DBRefEntry[] lrfs, List foundSeqs, AlignedCodonFrame cf) + List lrfs, List foundSeqs, + AlignedCodonFrame cf) { boolean found = false; if (lrfs == null) { return false; } - for (int i = 0; i < lrfs.length; i++) + for (int i = 0, n = lrfs.size(); i < n; i++) { - DBRefEntry xref = new DBRefEntry(lrfs[i]); - // add in wildcards - xref.setVersion(null); - xref.setMap(null); - found |= searchDataset(fromDna, sequenceI, xref, foundSeqs, cf, false); +// DBRefEntry xref = new DBRefEntry(lrfs.get(i)); +// // add in wildcards +// xref.setVersion(null); +// xref.setMap(null); + found |= searchDataset(fromDna, sequenceI, lrfs.get(i), foundSeqs, cf, + false, DBRefUtils.SEARCH_MODE_NO_MAP_NO_VERSION); } return found; } @@ -924,11 +999,12 @@ public class CrossRef * sequenceI or all the returned sequences (eg a genomic reference * associated with a locus and one or more transcripts) * + * @param mode SEARCH_MODE_FULL for all; SEARCH_MODE_NO_MAP_NO_VERSION optional * @return true if relationship found and sequence added. */ boolean searchDataset(boolean fromDna, SequenceI fromSeq, DBRefEntry xrf, List foundSeqs, AlignedCodonFrame mappings, - boolean direct) + boolean direct, int mode) { boolean found = false; if (dataset == null) @@ -940,8 +1016,8 @@ public class CrossRef System.err.println("Empty dataset sequence set - NO VECTOR"); return false; } - List ds; - synchronized (ds = dataset.getSequences()) + List ds = dataset.getSequences(); + synchronized (ds) { for (SequenceI nxt : ds) { @@ -949,10 +1025,9 @@ 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) + ")"); } @@ -974,13 +1049,13 @@ public class CrossRef } // look for direct or indirect references in common - DBRefEntry[] poss = nxt.getDBRefs(); + List poss = nxt.getDBRefs(); List cands = null; // todo: indirect specifies we select either direct references to nxt // that match xrf which is indirect to sequenceI, or indirect // references to nxt that match xrf which is direct to sequenceI - cands = DBRefUtils.searchRefs(poss, xrf); + cands = DBRefUtils.searchRefs(poss, xrf, mode); // else // { // poss = DBRefUtils.selectDbRefs(nxt.isProtein()!fromDna, poss);