import jalview.analysis.AlignmentUtils;
import jalview.analysis.Dna;
+import jalview.bin.Cache;
import jalview.datamodel.Alignment;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.DBRefEntry;
import jalview.io.FileParse;
import jalview.io.gff.SequenceOntologyFactory;
import jalview.io.gff.SequenceOntologyI;
+import jalview.util.Comparison;
import jalview.util.DBRefUtils;
import jalview.util.MapList;
+ " chunks. Unexpected problem (" + r.getLocalizedMessage()
+ ")";
System.err.println(msg);
+ r.printStackTrace();
break;
}
}
proteinSeq.createDatasetSequence();
querySeq.createDatasetSequence();
- MapList mapList = AlignmentUtils.mapCdsToProtein(querySeq, proteinSeq);
+ MapList mapList = AlignmentUtils
+ .mapCdsToProtein(querySeq, proteinSeq);
if (mapList != null)
{
// clunky: ensure Uniprot xref if we have one is on mapped sequence
SequenceI ds = proteinSeq.getDatasetSequence();
- ds.setSourceDBRef(proteinSeq.getSourceDBRef());
-
+ // TODO: Verify ensp primary ref is on proteinSeq.getDatasetSequence()
Mapping map = new Mapping(ds, mapList);
- DBRefEntry dbr = new DBRefEntry(getDbSource(), getDbVersion(),
- proteinSeq.getName(), map);
+ DBRefEntry dbr = new DBRefEntry(getDbSource(),
+ getEnsemblDataVersion(), proteinSeq.getName(), map);
querySeq.getDatasetSequence().addDBRef(dbr);
-
+ DBRefEntry[] uprots = DBRefUtils.selectRefs(ds.getDBRefs(),
+ new String[] { DBRefSource.UNIPROT });
+ DBRefEntry[] upxrefs = DBRefUtils.selectRefs(querySeq.getDBRefs(),
+ new String[] { DBRefSource.UNIPROT });
+ if (uprots != null)
+ {
+ for (DBRefEntry up : uprots)
+ {
+ // locate local uniprot ref and map
+ List<DBRefEntry> upx = DBRefUtils.searchRefs(upxrefs,
+ up.getAccessionId());
+ DBRefEntry upxref;
+ if (upx.size() != 0)
+ {
+ upxref = upx.get(0);
+
+ if (upx.size() > 1)
+ {
+ Cache.log
+ .warn("Implementation issue - multiple uniprot acc on product sequence.");
+ }
+ }
+ else
+ {
+ upxref = new DBRefEntry(DBRefSource.UNIPROT,
+ getEnsemblDataVersion(), up.getAccessionId());
+ }
+
+ Mapping newMap = new Mapping(ds, mapList);
+ upxref.setVersion(getEnsemblDataVersion());
+ upxref.setMap(newMap);
+ if (upx.size() == 0)
+ {
+ // add the new uniprot ref
+ querySeq.getDatasetSequence().addDBRef(upxref);
+ }
+
+ }
+ }
+
/*
* copy exon features to protein, compute peptide variants from dna
* variants and add as features on the protein sequence ta-da
*/
- AlignmentUtils.computeProteinFeatures(querySeq, proteinSeq, mapList);
+ AlignmentUtils
+ .computeProteinFeatures(querySeq, proteinSeq, mapList);
}
} catch (Exception e)
{
seq = seq.getDatasetSequence();
}
- EnsemblXref xrefFetcher = new EnsemblXref(getDomain());
+ EnsemblXref xrefFetcher = new EnsemblXref(getDomain(), getDbSource(),
+ getEnsemblDataVersion());
List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName());
for (DBRefEntry xref : xrefs)
{
seq.addDBRef(xref);
- /*
- * Save any Uniprot xref to be the reference for SIFTS mapping
- */
- if (DBRefSource.UNIPROT.equals(xref.getSource()))
- {
- seq.setSourceDBRef(xref);
- }
}
/*
* and add a reference to itself
*/
- DBRefEntry self = new DBRefEntry(getDbSource(), "0", seq.getName());
+ DBRefEntry self = new DBRefEntry(getDbSource(),
+ getEnsemblDataVersion(), seq.getName());
seq.addDBRef(self);
}
throw new JalviewException("ENSEMBL Rest API not available.");
}
FileParse fp = getSequenceReader(ids);
+ if (fp == null)
+ {
+ return alignment;
+ }
+
FastaFile fr = new FastaFile(fp);
if (fr.hasWarningMessage())
{
if (fr.getSeqs().size() > 0)
{
- AlignmentI seqal = new Alignment(
- fr.getSeqsAsArray());
- for (SequenceI sq:seqal.getSequences())
+ AlignmentI seqal = new Alignment(fr.getSeqsAsArray());
+ for (SequenceI sq : seqal.getSequences())
{
if (sq.getDescription() == null)
{
if (ids.contains(name)
|| ids.contains(name.replace("ENSP", "ENST")))
{
- DBRefUtils.parseToDbRef(sq, DBRefSource.ENSEMBL, "0", name);
+ DBRefEntry dbref = DBRefUtils.parseToDbRef(sq, getDbSource(),
+ getEnsemblDataVersion(), name);
+ sq.addDBRef(dbref);
}
}
if (alignment == null)
int mappedLength = 0;
int direction = 1; // forward
boolean directionSet = false;
-
+
for (SequenceFeature sf : sfs)
{
/*
// abort - mix of forward and backward
System.err.println("Error: forward and backward strand for "
+ accId);
- return null;
- }
- direction = strand;
- directionSet = true;
-
- /*
- * add to CDS ranges, semi-sorted forwards/backwards
- */
- if (strand < 0)
- {
- regions.add(0, new int[] { sf.getEnd(), sf.getBegin() });
- }
- else
- {
+ return null;
+ }
+ direction = strand;
+ directionSet = true;
+
+ /*
+ * add to CDS ranges, semi-sorted forwards/backwards
+ */
+ if (strand < 0)
+ {
+ regions.add(0, new int[] { sf.getEnd(), sf.getBegin() });
+ }
+ else
+ {
regions.add(new int[] { sf.getBegin(), sf.getEnd() });
}
mappedLength += Math.abs(sf.getEnd() - sf.getBegin() + 1);
}
}
}
-
+
if (regions.isEmpty())
{
System.out.println("Failed to identify target sequence for " + accId
* (havana / ensembl_havana)
*/
Collections.sort(regions, new RangeSorter(direction == 1));
-
+
List<int[]> to = Arrays.asList(new int[] { start,
start + mappedLength - 1 });
-
+
return new MapList(regions, to, 1, 1);
}
int start = sf.getBegin();
int end = sf.getEnd();
int[] mappedRange = mapping.locateInTo(start, end);
-
+
if (mappedRange != null)
{
SequenceFeature copy = new SequenceFeature(sf);
copy.setBegin(Math.min(mappedRange[0], mappedRange[1]));
copy.setEnd(Math.max(mappedRange[0], mappedRange[1]));
+ if (".".equals(copy.getFeatureGroup()))
+ {
+ copy.setFeatureGroup(getDbSource());
+ }
targetSequence.addSequenceFeature(copy);
/*
{
complement.append(",");
}
- if ("HGMD_MUTATION".equalsIgnoreCase(allele))
+
+ /*
+ * some 'alleles' are actually descriptive terms
+ * e.g. HGMD_MUTATION, PhenCode_variation
+ * - we don't want to 'reverse complement' these
+ */
+ if (!Comparison.isNucleotideSequence(allele, true))
{
complement.append(allele);
}
else
{
- char[] alleles = allele.toCharArray();
- for (int i = alleles.length - 1; i >= 0; i--)
+ for (int i = allele.length() - 1; i >= 0; i--)
{
- complement.append(Dna.getComplement(alleles[i]));
+ complement.append(Dna.getComplement(allele.charAt(i)));
}
}
}
// long start = System.currentTimeMillis();
SequenceFeature[] sfs = sourceSequence.getSequenceFeatures();
- MapList mapping = getGenomicRangesFromFeatures(sourceSequence, accessionId,
- targetSequence.getStart());
+ MapList mapping = getGenomicRangesFromFeatures(sourceSequence,
+ accessionId, targetSequence.getStart());
if (mapping == null)
{
return false;
String type, String parentId)
{
List<SequenceFeature> result = new ArrayList<SequenceFeature>();
-
+
SequenceFeature[] sfs = sequence.getSequenceFeatures();
- if (sfs != null) {
+ if (sfs != null)
+ {
SequenceOntologyI so = SequenceOntologyFactory.getInstance();
- for (SequenceFeature sf :sfs) {
+ for (SequenceFeature sf : sfs)
+ {
if (so.isA(sf.getType(), type))
{
String parent = (String) sf.getValue(PARENT);
git://source.jalview.org / jalview.git / blobdiff