import jalview.datamodel.Sequence;
import jalview.datamodel.SequenceFeature;
import jalview.datamodel.SequenceI;
-import jalview.util.Comparison;
import jalview.util.DBRefUtils;
import jalview.util.MapList;
import jalview.ws.SequenceFetcherFactory;
private SequenceI[] fromSeqs;
/**
+ * matcher built from dataset
+ */
+ SequenceIdMatcher matcher;
+
+ /**
+ * sequences found by cross-ref searches to fromSeqs
+ */
+ List<SequenceI> rseqs;
+
+ /**
* Constructor
*
* @param seqs
*
* @param seq
* the sequence whose dbrefs we are searching against
+ * @param fromDna
+ * when true, context is DNA - so sources identifying protein
+ * products will be returned.
* @param sources
* a list of sources to add matches to
*/
* find sequence's direct (dna-to-dna, peptide-to-peptide) xrefs
*/
DBRefEntry[] lrfs = DBRefUtils.selectDbRefs(fromDna, seq.getDBRefs());
- List<SequenceI> rseqs = new ArrayList<SequenceI>();
+ List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
/*
* find sequences in the alignment which xref one of these DBRefs
* i.e. is xref-ed to a common sequence identifier
*/
- searchDatasetXrefs(fromDna, seq, lrfs, rseqs, null);
+ searchDatasetXrefs(fromDna, seq, lrfs, foundSeqs, null);
/*
* add those sequences' (dna-to-peptide or peptide-to-dna) dbref sources
*/
- for (SequenceI rs : rseqs)
+ for (SequenceI rs : foundSeqs)
{
DBRefEntry[] xrs = DBRefUtils
.selectDbRefs(!fromDna, rs.getDBRefs());
public Alignment findXrefSequences(String source, boolean fromDna)
{
- List<SequenceI> rseqs = new ArrayList<SequenceI>();
+ rseqs = new ArrayList<SequenceI>();
AlignedCodonFrame cf = new AlignedCodonFrame();
- SequenceIdMatcher matcher = new SequenceIdMatcher(
+ matcher = new SequenceIdMatcher(
dataset.getSequences());
for (SequenceI seq : fromSeqs)
if (fromDna)
{
// map is from dna seq to a protein product
- cf.addMap(dss, rsq, xref.getMap().getMap());
+ cf.addMap(dss, rsq, xref.getMap().getMap(), xref.getMap()
+ .getMappedFromId());
}
else
{
// map should be from protein seq to its coding dna
- cf.addMap(rsq, dss, xref.getMap().getMap().getInverse());
+ cf.addMap(rsq, dss, xref.getMap().getMap().getInverse(),
+ xref.getMap().getMappedFromId());
}
}
}
*/
if (!sourceRefs.isEmpty())
{
- ASequenceFetcher sftch = SequenceFetcherFactory
- .getSequenceFetcher();
- SequenceI[] retrieved = null;
- try
- {
- retrieved = sftch.getSequences(sourceRefs, !fromDna);
- } catch (Exception e)
- {
- System.err
- .println("Problem whilst retrieving cross references for Sequence : "
- + seq.getName());
- e.printStackTrace();
- }
+ retrieveCrossRef(sourceRefs, seq, xrfs, fromDna, cf);
+ }
+ }
+
+ Alignment ral = null;
+ if (rseqs.size() > 0)
+ {
+ ral = new Alignment(rseqs.toArray(new SequenceI[rseqs.size()]));
+ if (!cf.isEmpty())
+ {
+ dataset.addCodonFrame(cf);
+ }
+ }
+ return ral;
+ }
+
+ private void retrieveCrossRef(List<DBRefEntry> sourceRefs, SequenceI seq,
+ DBRefEntry[] xrfs, boolean fromDna, AlignedCodonFrame cf)
+ {
+ ASequenceFetcher sftch = SequenceFetcherFactory.getSequenceFetcher();
+ SequenceI[] retrieved = null;
+ SequenceI dss = seq.getDatasetSequence() == null ? seq : seq
+ .getDatasetSequence();
+ try
+ {
+ retrieved = sftch.getSequences(sourceRefs, !fromDna);
+ } catch (Exception e)
+ {
+ System.err
+ .println("Problem whilst retrieving cross references for Sequence : "
+ + seq.getName());
+ e.printStackTrace();
+ }
- if (retrieved != null)
+ if (retrieved != null)
+ {
+ 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();
+ DBRefEntry[] dbr = retrievedSequence.getDBRefs();
+ if (dbr != null)
{
- updateDbrefMappings(seq, xrfs, retrieved, cf, fromDna);
- for (SequenceI retrievedSequence : retrieved)
+ for (DBRefEntry dbref : dbr)
{
- SequenceI retrievedDss = retrievedSequence.getDatasetSequence() == null ? retrievedSequence
- : retrievedSequence.getDatasetSequence();
- DBRefEntry[] dbr = retrievedSequence.getDBRefs();
- if (dbr != null)
+ // 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)
+ // 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() != null && map.getMap() != null)
+ /*
+ * 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)
{
- // 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)
- {
- /*
- * 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);
- }
- else
+ for (DBRefEntry ref : toRefs)
{
- matcher.add(map.getTo());
+ matched.addDBRef(ref); // add or update mapping
}
- try
+ }
+ map.setTo(matched);
+ }
+ else
+ {
+ matcher.add(map.getTo());
+ }
+ 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);
+
+ /*
+ * give the reverse reference the inverse mapping
+ * (if it doesn't have one already)
+ */
+ setReverseMapping(dss, dbref, cf);
+
+ /*
+ * copy sequence features as well, avoiding
+ * duplication (e.g. same variation from two
+ * transcripts)
+ */
+ SequenceFeature[] sfs = ms.getSequenceFeatures();
+ if (sfs != null)
{
- // 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()))
+ for (SequenceFeature feat : sfs)
{
- String msg = "Mapping updated from " + ms.getName()
- + " to retrieved crossreference "
- + dss.getName();
- System.out.println(msg);
- map.setTo(dss);
-
/*
- * give the reverse reference the inverse mapping
- * (if it doesn't have one already)
+ * make a flyweight feature object which ignores Parent
+ * attribute in equality test; this avoids creating many
+ * otherwise duplicate exon features on genomic sequence
*/
- setReverseMapping(dss, dbref, cf);
-
- /*
- * copy sequence features as well, avoiding
- * duplication (e.g. same variation from two
- * transcripts)
- */
- SequenceFeature[] sfs = ms.getSequenceFeatures();
- if (sfs != null)
+ SequenceFeature newFeature = new SequenceFeature(
+ feat)
{
- for (SequenceFeature feat : sfs)
+ @Override
+ public boolean equals(Object o)
{
- /*
- * 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);
}
- }
+ };
+ dss.addSequenceFeature(newFeature);
}
- cf.addMap(retrievedDss, map.getTo(), map.getMap());
- } catch (Exception e)
- {
- System.err
- .println("Exception when consolidating Mapped sequence set...");
- e.printStackTrace(System.err);
}
}
+ cf.addMap(retrievedDss, 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);
- dataset.addSequence(retrievedDss);
- matcher.add(retrievedDss);
}
}
+ retrievedSequence.updatePDBIds();
+ rseqs.add(retrievedDss);
+ dataset.addSequence(retrievedDss);
+ matcher.add(retrievedDss);
}
}
-
- Alignment ral = null;
- if (rseqs.size() > 0)
- {
- ral = new Alignment(rseqs.toArray(new SequenceI[rseqs.size()]));
- if (!cf.isEmpty())
- {
- dataset.addCodonFrame(cf);
- }
- }
- return ral;
}
-
/**
* Sets the inverse sequence mapping in the corresponding dbref of the mapped
* to sequence (if any). This is used after fetching a cross-referenced
void updateDbrefMappings(SequenceI mapFrom, DBRefEntry[] xrefs,
SequenceI[] retrieved, AlignedCodonFrame acf, boolean fromDna)
{
- SequenceIdMatcher matcher = new SequenceIdMatcher(retrieved);
+ SequenceIdMatcher idMatcher = new SequenceIdMatcher(retrieved);
for (DBRefEntry xref : xrefs)
{
if (!xref.hasMap())
{
String targetSeqName = xref.getSource() + "|"
+ xref.getAccessionId();
- SequenceI[] matches = matcher.findAllIdMatches(targetSeqName);
+ SequenceI[] matches = idMatcher.findAllIdMatches(targetSeqName);
if (matches == null)
{
return;
return false;
}
xref.setMap(new Mapping(mapTo, mapping));
+
+ /*
+ * and add a reverse DbRef with the inverse mapping
+ */
+ if (mapFrom.getDatasetSequence() != null
+ && mapFrom.getDatasetSequence().getSourceDBRef() != null)
+ {
+ DBRefEntry dbref = new DBRefEntry(mapFrom.getDatasetSequence()
+ .getSourceDBRef());
+ dbref.setMap(new Mapping(mapFrom.getDatasetSequence(), mapping
+ .getInverse()));
+ mapTo.addDBRef(dbref);
+ }
+
if (fromDna)
{
AlignmentUtils.computeProteinFeatures(mapFrom, mapTo, mapping);
* context was searching from Protein sequences
* @param sequenceI
* @param lrfs
- * @param rseqs
+ * @param foundSeqs
* @return true if matches were found.
*/
private boolean searchDatasetXrefs(boolean fromDna, SequenceI sequenceI,
- DBRefEntry[] lrfs, List<SequenceI> rseqs, AlignedCodonFrame cf)
+ DBRefEntry[] lrfs, List<SequenceI> foundSeqs, AlignedCodonFrame cf)
{
boolean found = false;
if (lrfs == null)
// add in wildcards
xref.setVersion(null);
xref.setMap(null);
- found |= searchDataset(fromDna, sequenceI, xref, rseqs, cf, false);
+ found |= searchDataset(fromDna, sequenceI, xref, foundSeqs, cf, false);
}
return found;
}
* @param fromDna
* true if context was searching for refs *from* dna sequence, false
* if context was searching for refs *from* protein sequence
- * @param sequenceI
+ * @param fromSeq
* a sequence to ignore (start point of search)
* @param xrf
* a cross-reference to try to match
- * @param rseqs
+ * @param foundSeqs
* result list to add to
- * @param cf
+ * @param mappings
* a set of sequence mappings to add to
* @param direct
- * - search all references or only subset
+ * - indicates the type of relationship between returned sequences,
+ * xrf, and sequenceI that is required.
+ * <ul>
+ * <li>direct implies xrf is a primary reference for sequenceI AND
+ * the sequences to be located (eg a uniprot ID for a protein
+ * sequence, and a uniprot ref on a transcript sequence).</li>
+ * <li>indirect means xrf is a cross reference with respect to
+ * sequenceI or all the returned sequences (eg a genomic reference
+ * associated with a locus and one or more transcripts)</li>
+ * </ul>
* @return true if relationship found and sequence added.
*/
- boolean searchDataset(boolean fromDna, SequenceI sequenceI,
- DBRefEntry xrf, List<SequenceI> rseqs, AlignedCodonFrame cf,
+ boolean searchDataset(boolean fromDna, SequenceI fromSeq,
+ DBRefEntry xrf, List<SequenceI> foundSeqs, AlignedCodonFrame mappings,
boolean direct)
{
boolean found = false;
if (nxt.getDatasetSequence() != null)
{
System.err
- .println("Implementation warning: getProducts passed a dataset alignment without dataset sequences in it!");
+ .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)
+ + ")");
}
- if (nxt == sequenceI || nxt == sequenceI.getDatasetSequence())
+ if (nxt == fromSeq || nxt == fromSeq.getDatasetSequence())
{
continue;
}
* complementary type if !direct
*/
{
- boolean isDna = Comparison
- .isNucleotide(new SequenceI[] { nxt });
+ boolean isDna = !nxt.isProtein();
if (direct ? (isDna != fromDna) : (isDna == fromDna))
{
// skip this sequence because it is wrong molecule type
// look for direct or indirect references in common
DBRefEntry[] poss = nxt.getDBRefs();
List<DBRefEntry> cands = null;
- /*
- * TODO does this make any sense?
- * if 'direct', search the dbrefs for xrf
- * else, filter the dbrefs by type and then search for xrf
- * - the result is the same isn't it?
- */
- if (direct)
- {
- cands = DBRefUtils.searchRefs(poss, xrf);
- }
- else
- {
- poss = DBRefUtils.selectDbRefs(!fromDna, poss);
- cands = DBRefUtils.searchRefs(poss, xrf);
- }
+
+ // 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);
+ // else
+ // {
+ // poss = DBRefUtils.selectDbRefs(nxt.isProtein()!fromDna, poss);
+ // cands = DBRefUtils.searchRefs(poss, xrf);
+ // }
if (!cands.isEmpty())
{
- if (!rseqs.contains(nxt))
+ if (!foundSeqs.contains(nxt))
{
found = true;
- rseqs.add(nxt);
- if (cf != null)
+ foundSeqs.add(nxt);
+ if (mappings != null && !direct)
{
- // don't search if we aren't given a codon map object
+ /*
+ * if the matched sequence has mapped dbrefs to
+ * protein product / cdna, add equivalent mappings to
+ * our source sequence
+ */
for (DBRefEntry candidate : cands)
{
Mapping mapping = candidate.getMap();
if (mapping.getTo() != null
&& map.getFromRatio() != map.getToRatio())
{
- // get sense of map correct for adding to product
- // alignment.
- if (fromDna)
+ /*
+ * add a mapping, as from dna to peptide sequence
+ */
+ if (map.getFromRatio() == 3)
{
- // map is from dna seq to a protein product
- cf.addMap(sequenceI, nxt, map);
+ mappings.addMap(nxt, fromSeq, map);
}
else
{
- // map should be from protein seq to its coding dna
- cf.addMap(nxt, sequenceI, map.getInverse());
+ mappings.addMap(nxt, fromSeq, map.getInverse());
}
}
}
}
}
- // TODO: add mapping between sequences if necessary
}
}
}
git://source.jalview.org / jalview.git / blobdiff