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.SequenceFetcher;
import jalview.ws.seqfetcher.ASequenceFetcher;
*/
public class CrossRef
{
+ /*
+ * A sub-class that ignores Parent attribute when comparing sequence
+ * features. This avoids 'duplicate' CDS features that only
+ * differ in their parent Transcript ids.
+ */
+ class MySequenceFeature extends SequenceFeature
+ {
+ private SequenceFeature feat;
+
+ MySequenceFeature(SequenceFeature sf)
+ {
+ this.feat = sf;
+ }
+
+ @Override
+ public boolean equals(Object o)
+ {
+ return feat.equals(o, true);
+ }
+ }
+
/**
* Select just the DNA or protein references for a protein or dna sequence
*
{
String[] dbrefs = null;
List<String> refs = new ArrayList<String>();
- for (int s = 0; s < seqs.length; s++)
+ for (SequenceI seq : seqs)
{
- if (seqs[s] != null)
+ if (seq != null)
{
- SequenceI dss = seqs[s];
+ SequenceI dss = seq;
while (dss.getDatasetSequence() != null)
{
dss = dss.getDatasetSequence();
}
DBRefEntry[] rfs = findXDbRefs(dna, dss.getDBRefs());
- for (int r = 0; rfs != null && r < rfs.length; r++)
+ if (rfs != null)
{
- if (!refs.contains(rfs[r].getSource()))
+ for (DBRefEntry ref : rfs)
{
- refs.add(rfs[r].getSource());
+ if (!refs.contains(ref.getSource()))
+ {
+ refs.add(ref.getSource());
+ }
}
}
if (dataset != null)
{
// search for references to this sequence's direct references.
- DBRefEntry[] lrfs = CrossRef
- .findXDbRefs(!dna, seqs[s].getDBRefs());
+ DBRefEntry[] lrfs = CrossRef.findXDbRefs(!dna, seq.getDBRefs());
List<SequenceI> rseqs = new ArrayList<SequenceI>();
- CrossRef.searchDatasetXrefs(seqs[s], !dna, lrfs, dataset, rseqs,
+ CrossRef.searchDatasetXrefs(seq, !dna, lrfs, dataset, rseqs,
null); // don't need to specify codon frame for mapping here
for (SequenceI rs : rseqs)
{
- DBRefEntry[] xrs = findXDbRefs(dna, rs.getDBRefs()); // not used??
- for (int r = 0; rfs != null && r < rfs.length; r++)
+ DBRefEntry[] xrs = findXDbRefs(dna, rs.getDBRefs());
+ if (xrs != null)
{
- if (!refs.contains(rfs[r].getSource()))
+ for (DBRefEntry ref : xrs)
{
- refs.add(rfs[r].getSource());
+ if (!refs.contains(ref.getSource()))
+ {
+ refs.add(ref.getSource());
+ }
}
}
+ // looks like copy and paste - change rfs to xrs?
+ // for (int r = 0; rfs != null && r < rfs.length; r++)
+ // {
+ // if (!refs.contains(rfs[r].getSource()))
+ // {
+ // refs.add(rfs[r].getSource());
+ // }
+ // }
}
}
}
return dbrefs;
}
- /*
- * if (dna) { if (rfs[r].hasMap()) { // most likely this is a protein cross
- * reference if (!refs.contains(rfs[r].getSource())) {
- * refs.addElement(rfs[r].getSource()); } } }
- */
public static boolean hasCdnaMap(SequenceI[] seqs)
{
+ // TODO unused - remove?
String[] reftypes = findSequenceXrefTypes(false, seqs);
for (int s = 0; s < reftypes.length; s++)
{
public static SequenceI[] getCdnaMap(SequenceI[] seqs)
{
+ // TODO unused - remove?
Vector cseqs = new Vector();
for (int s = 0; s < seqs.length; s++)
{
/**
*
- * @param dna
- * @param seqs
- * @return
- */
- public static Alignment findXrefSequences(SequenceI[] seqs, boolean dna,
- String source)
- {
- return findXrefSequences(seqs, dna, source, null);
- }
-
- /**
- *
* @param seqs
+ * sequences whose xrefs are being retrieved
* @param dna
+ * true if sequences are nucleotide
* @param source
- * @param dataset
- * alignment to search for product sequences.
+ * @param al
+ * alignment to search for cross-referenced sequences (and possibly
+ * add to)
* @return products (as dataset sequences)
*/
- public static Alignment findXrefSequences(SequenceI[] seqs, boolean dna,
- String source, AlignmentI dataset)
+ public static Alignment findXrefSequences(SequenceI[] seqs,
+ final boolean dna, final String source, AlignmentI al)
{
+ AlignmentI dataset = al.getDataset() == null ? al : al.getDataset();
List<SequenceI> rseqs = new ArrayList<SequenceI>();
- Alignment ral = null;
- AlignedCodonFrame cf = new AlignedCodonFrame(); // nominal width
- for (int s = 0; s < seqs.length; s++)
+ AlignedCodonFrame cf = new AlignedCodonFrame();
+ for (SequenceI seq : seqs)
{
- SequenceI dss = seqs[s];
+ SequenceI dss = seq;
while (dss.getDatasetSequence() != null)
{
dss = dss.getDatasetSequence();
if ((xrfs == null || xrfs.length == 0) && dataset != null)
{
System.out.println("Attempting to find ds Xrefs refs.");
- DBRefEntry[] lrfs = CrossRef.findXDbRefs(!dna, seqs[s].getDBRefs());
+ // FIXME should be dss not seq here?
+ DBRefEntry[] lrfs = CrossRef.findXDbRefs(!dna, seq.getDBRefs());
// less ambiguous would be a 'find primary dbRefEntry' method.
// filter for desired source xref here
found = CrossRef.searchDatasetXrefs(dss, !dna, lrfs, dataset,
}
for (int r = 0; xrfs != null && r < xrfs.length; r++)
{
- if (source != null && !source.equals(xrfs[r].getSource()))
+ DBRefEntry xref = xrfs[r];
+ if (source != null && !source.equals(xref.getSource()))
{
continue;
}
- if (xrfs[r].hasMap())
+ if (xref.hasMap())
{
- if (xrfs[r].getMap().getTo() != null)
+ if (xref.getMap().getTo() != null)
{
- SequenceI rsq = new Sequence(xrfs[r].getMap().getTo());
+ SequenceI rsq = new Sequence(xref.getMap().getTo());
rseqs.add(rsq);
- if (xrfs[r].getMap().getMap().getFromRatio() != xrfs[r]
+ if (xref.getMap().getMap().getFromRatio() != xref
.getMap().getMap().getToRatio())
{
// get sense of map correct for adding to product alignment.
if (dna)
{
// map is from dna seq to a protein product
- cf.addMap(dss, rsq, xrfs[r].getMap().getMap());
+ cf.addMap(dss, rsq, xref.getMap().getMap());
}
else
{
// map should be from protein seq to its coding dna
- cf.addMap(rsq, dss, xrfs[r].getMap().getMap().getInverse());
+ cf.addMap(rsq, dss, xref.getMap().getMap().getInverse());
}
}
found = true;
// xrefs on this sequence.
if (dataset != null)
{
- found |= searchDataset(dss, xrfs[r], dataset, rseqs, cf); // ,false,!dna);
+ found |= searchDataset(dss, xref, dataset, rseqs, cf, false,
+ !dna);
if (found)
{
xrfs[r] = null; // we've recovered seqs for this one.
}
if (l > 0)
{
- System.out
- .println("Attempting to retrieve cross referenced sequences.");
+ // System.out
+ // .println("Attempting to retrieve cross referenced sequences.");
DBRefEntry[] t = new DBRefEntry[l];
l = 0;
for (int r = 0; r < xrfs.length; r++)
xrfs = t;
try
{
- retrieved = sftch.getSequences(xrfs); // problem here is we don't
- // know which of xrfs
- // resulted in which
+ retrieved = sftch.getSequences(xrfs, !dna);
+ // problem here is we don't know which of xrfs resulted in which
// retrieved element
} catch (Exception e)
{
System.err
.println("Problem whilst retrieving cross references for Sequence : "
- + seqs[s].getName());
+ + seq.getName());
e.printStackTrace();
}
+
if (retrieved != null)
{
+ updateDbrefMappings(dna, seq, xrfs, retrieved, cf);
+
+ SequenceIdMatcher matcher = new SequenceIdMatcher(
+ dataset.getSequences());
+ List<SequenceFeature> copiedFeatures = new ArrayList<SequenceFeature>();
+ CrossRef me = new CrossRef();
for (int rs = 0; rs < retrieved.length; rs++)
{
// TODO: examine each sequence for 'redundancy'
- jalview.datamodel.DBRefEntry[] dbr = retrieved[rs]
- .getDBRefs();
+ DBRefEntry[] dbr = retrieved[rs].getDBRefs();
if (dbr != null && dbr.length > 0)
{
for (int di = 0; di < dbr.length; di++)
{
// find any entry where we should put in the sequence being
// cross-referenced into the map
- jalview.datamodel.Mapping map = dbr[di].getMap();
+ Mapping map = dbr[di].getMap();
if (map != null)
{
if (map.getTo() != null && map.getMap() != null)
{
- // should search the local dataset to find any existing
- // candidates for To !
+ SequenceI matched = 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
+ */
+ for (DBRefEntry ref : map.getTo().getDBRefs())
+ {
+ matched.addDBRef(ref); // add or update mapping
+ }
+ map.setTo(matched);
+ }
+ else
+ {
+ matcher.add(map.getTo());
+ }
try
{
// compare ms with dss and replace with dss in mapping
int sf = map.getMap().getToLowest();
int st = map.getMap().getToHighest();
SequenceI mappedrg = ms.getSubSequence(sf, st);
- SequenceI loc = dss.getSubSequence(sf, st);
+ // SequenceI loc = dss.getSubSequence(sf, st);
if (mappedrg.getLength() > 0
- && mappedrg.getSequenceAsString().equals(
- loc.getSequenceAsString()))
+ && ms.getSequenceAsString().equals(
+ dss.getSequenceAsString()))
+ // && mappedrg.getSequenceAsString().equals(
+ // loc.getSequenceAsString()))
{
- System.err
- .println("Mapping updated for retrieved crossreference");
+ String msg = "Mapping updated from "
+ + ms.getName()
+ + " to retrieved crossreference "
+ + dss.getName();
+ System.out.println(msg);
// method to update all refs of existing To on
// retrieved sequence with dss and merge any props
// on To onto dss.
map.setTo(dss);
+ /*
+ * copy sequence features as well, avoiding
+ * duplication (e.g. same variation from 2
+ * transcripts)
+ */
+ SequenceFeature[] sfs = ms
+ .getSequenceFeatures();
+ if (sfs != null)
+ {
+ for (SequenceFeature feat : sfs)
+ {
+ /*
+ * we override SequenceFeature.equals here (but
+ * not elsewhere) to ignore Parent attribute
+ * TODO not quite working yet!
+ */
+ if (!copiedFeatures
+ .contains(me.new MySequenceFeature(
+ feat)))
+ {
+ dss.addSequenceFeature(feat);
+ copiedFeatures.add(feat);
+ }
+ }
+ }
+ cf.addMap(retrieved[rs].getDatasetSequence(),
+ dss, map.getMap());
+ }
+ else
+ {
+ cf.addMap(retrieved[rs].getDatasetSequence(),
+ map.getTo(), map.getMap());
}
} catch (Exception e)
{
}
}
}
+
+ Alignment ral = null;
if (rseqs.size() > 0)
{
- SequenceI[] rsqs = new SequenceI[rseqs.size()];
- rseqs.toArray(rsqs);
- ral = new Alignment(rsqs);
- if (cf != null && cf.getProtMappings() != null)
+ ral = new Alignment(rseqs.toArray(new SequenceI[rseqs.size()]));
+ if (cf != null && !cf.isEmpty())
{
ral.addCodonFrame(cf);
}
}
/**
+ * Updates any empty mappings in the cross-references with one to a compatible
+ * retrieved sequence if found, and adds any new mappings to the
+ * AlignedCodonFrame
+ *
+ * @param dna
+ * @param mapFrom
+ * @param xrefs
+ * @param retrieved
+ * @param acf
+ */
+ static void updateDbrefMappings(boolean dna, SequenceI mapFrom,
+ DBRefEntry[] xrefs, SequenceI[] retrieved, AlignedCodonFrame acf)
+ {
+ SequenceIdMatcher matcher = new SequenceIdMatcher(retrieved);
+ for (DBRefEntry xref : xrefs)
+ {
+ if (!xref.hasMap())
+ {
+ String targetSeqName = xref.getSource() + "|"
+ + xref.getAccessionId();
+ SequenceI[] matches = matcher.findAllIdMatches(targetSeqName);
+ if (matches == null)
+ {
+ return;
+ }
+ for (SequenceI seq : matches)
+ {
+ MapList mapping = null;
+ if (dna)
+ {
+ mapping = AlignmentUtils.mapCdnaToProtein(seq, mapFrom);
+ }
+ else
+ {
+ mapping = AlignmentUtils.mapCdnaToProtein(mapFrom, seq);
+ if (mapping != null)
+ {
+ mapping = mapping.getInverse();
+ }
+ }
+ if (mapping != null)
+ {
+ xref.setMap(new Mapping(seq, mapping));
+ if (dna)
+ {
+ AlignmentUtils.computeProteinFeatures(mapFrom, seq, mapping);
+ }
+ if (dna)
+ {
+ acf.addMap(mapFrom, seq, mapping);
+ }
+ else
+ {
+ acf.addMap(seq, mapFrom, mapping.getInverse());
+ }
+ continue;
+ }
+ }
+ }
+ }
+ }
+
+ /**
* find references to lrfs in the cross-reference set of each sequence in
* dataset (that is not equal to sequenceI) Identifies matching DBRefEntry
* based on source and accession string only - Map and Version are nulled.