X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fext%2Fensembl%2FEnsemblGene.java;h=0c20e123f26c45ce4289a6b8d6dc210f5557d6ab;hb=fa2b80f5f3252f1f4801d2048b80a8745b399616;hp=10841bd8cd26278b01ef4172ecbd77c967bb1e5a;hpb=550c391f0c113658e540783dc89034a34280ef18;p=jalview.git diff --git a/src/jalview/ext/ensembl/EnsemblGene.java b/src/jalview/ext/ensembl/EnsemblGene.java index 10841bd..0c20e12 100644 --- a/src/jalview/ext/ensembl/EnsemblGene.java +++ b/src/jalview/ext/ensembl/EnsemblGene.java @@ -1,14 +1,20 @@ package jalview.ext.ensembl; +import jalview.api.FeatureColourI; +import jalview.api.FeatureSettingsModelI; import jalview.datamodel.AlignmentI; import jalview.datamodel.Sequence; import jalview.datamodel.SequenceFeature; import jalview.datamodel.SequenceI; import jalview.io.gff.SequenceOntologyFactory; import jalview.io.gff.SequenceOntologyI; +import jalview.schemes.FeatureColour; +import jalview.schemes.FeatureSettingsAdapter; import jalview.util.MapList; -import jalview.util.StringUtils; +import java.awt.Color; +import java.io.UnsupportedEncodingException; +import java.net.URLDecoder; import java.util.ArrayList; import java.util.Arrays; import java.util.List; @@ -35,10 +41,28 @@ public class EnsemblGene extends EnsemblSeqProxy EnsemblFeatureType.exon, EnsemblFeatureType.cds, EnsemblFeatureType.variation }; + /** + * Default constructor (to use rest.ensembl.org) + */ + public EnsemblGene() + { + super(); + } + + /** + * Constructor given the target domain to fetch data from + * + * @param d + */ + public EnsemblGene(String d) + { + super(d); + } + @Override public String getDbName() { - return "ENSEMBL (GENE)"; + return "ENSEMBL"; } @Override @@ -76,56 +100,95 @@ public class EnsemblGene extends EnsemblSeqProxy * * * @param query - * one or more identifiers separated by a space - * @return an alignment containing one or more genes, and possibly - * transcripts, or null + * a single gene or transcript identifier or gene name + * @return an alignment containing a gene, and possibly transcripts, or null */ @Override public AlignmentI getSequenceRecords(String query) throws Exception { - // todo: tidy up handling of one or multiple accession ids - String[] queries = query.split(getAccessionSeparator()); - /* - * if given a transcript id, look up its gene parent + * convert to a non-duplicated list of gene identifiers */ - if (isTranscriptIdentifier(query)) + List geneIds = getGeneIds(query); + + AlignmentI al = null; + for (String geneId : geneIds) { - // we are assuming all transcripts have the same gene parent here - query = new EnsemblLookup().getParent(queries[0]); - if (query == null) + /* + * fetch the gene sequence(s) with features and xrefs + */ + AlignmentI geneAlignment = super.getSequenceRecords(geneId); + if (geneAlignment == null) { - return null; + continue; + } + if (geneAlignment.getHeight() == 1) + { + getTranscripts(geneAlignment, geneId); + } + if (al == null) + { + al = geneAlignment; + } + else + { + al.append(geneAlignment); } } + return al; + } - /* - * if given a gene or other external name, lookup and fetch - * the corresponding gene for all model organisms - */ - if (!isGeneIdentifier(query)) + /** + * Converts a query, which may contain one or more gene or transcript + * identifiers, into a non-redundant list of gene identifiers. + * + * @param accessions + * @return + */ + List getGeneIds(String accessions) + { + List geneIds = new ArrayList(); + + for (String acc : accessions.split(getAccessionSeparator())) { - List geneIds = new EnsemblSymbol().getIds(query); - if (geneIds.isEmpty()) + if (isGeneIdentifier(acc)) { - return null; + if (!geneIds.contains(acc)) + { + geneIds.add(acc); + } } - String theIds = StringUtils.listToDelimitedString(geneIds, - getAccessionSeparator()); - return getSequenceRecords(theIds); - } - AlignmentI al = super.getSequenceRecords(query); + /* + * if given a transcript id, look up its gene parent + */ + else if (isTranscriptIdentifier(acc)) + { + String geneId = new EnsemblLookup(getDomain()).getParent(acc); + if (geneId != null && !geneIds.contains(geneId)) + { + geneIds.add(geneId); + } + } - /* - * if we retrieved a single gene, get its transcripts as well - */ - if (al.getHeight() == 1) - { - getTranscripts(al, query); + /* + * if given a gene or other external name, lookup and fetch + * the corresponding gene for all model organisms + */ + else + { + List ids = new EnsemblSymbol(getDomain(), getDbSource(), + getDbVersion()).getIds(acc); + for (String geneId : ids) + { + if (!geneIds.contains(geneId)) + { + geneIds.add(geneId); + } + } + } } - - return al; + return geneIds; } /** @@ -137,7 +200,8 @@ public class EnsemblGene extends EnsemblSeqProxy */ protected String getGeneIdentifiersForName(String query) { - List ids = new EnsemblSymbol().getIds(query); + List ids = new EnsemblSymbol(getDomain(), getDbSource(), + getDbVersion()).getIds(query); if (ids != null) { for (String id : ids) @@ -171,6 +235,36 @@ public class EnsemblGene extends EnsemblSeqProxy { makeTranscript(transcriptFeature, al, gene); } + + clearGeneFeatures(gene); + } + + /** + * Remove unwanted features (transcript, exon, CDS) from the gene sequence + * after we have used them to derive transcripts and transfer features + * + * @param gene + */ + protected void clearGeneFeatures(SequenceI gene) + { + SequenceFeature[] sfs = gene.getSequenceFeatures(); + if (sfs != null) + { + SequenceOntologyI so = SequenceOntologyFactory.getInstance(); + List filtered = new ArrayList(); + for (SequenceFeature sf : sfs) + { + String type = sf.getType(); + if (!isTranscript(type) && !so.isA(type, SequenceOntologyI.EXON) + && !so.isA(type, SequenceOntologyI.CDS)) + { + filtered.add(sf); + } + } + gene.setSequenceFeatures(filtered + .toArray(new SequenceFeature[filtered + .size()])); + } } /** @@ -236,10 +330,25 @@ public class EnsemblGene extends EnsemblSeqProxy } Sequence transcript = new Sequence(accId, seqChars, 1, transcriptLength); - String geneName = (String) transcriptFeature.getValue(NAME); - if (geneName != null) + + /* + * Ensembl has gene name as transcript Name + * EnsemblGenomes doesn't, but has a url-encoded description field + */ + String description = (String) transcriptFeature.getValue(NAME); + if (description == null) { - transcript.setDescription(geneName); + description = (String) transcriptFeature.getValue(DESCRIPTION); + } + if (description != null) + { + try + { + transcript.setDescription(URLDecoder.decode(description, "UTF-8")); + } catch (UnsupportedEncodingException e) + { + e.printStackTrace(); // as if + } } transcript.createDatasetSequence(); @@ -252,18 +361,19 @@ public class EnsemblGene extends EnsemblSeqProxy List mapTo = new ArrayList(); mapTo.add(new int[] { 1, transcriptLength }); MapList mapping = new MapList(mappedFrom, mapTo, 1, 1); - new EnsemblCdna().transferFeatures(gene.getSequenceFeatures(), + EnsemblCdna cdna = new EnsemblCdna(getDomain()); + cdna.transferFeatures(gene.getSequenceFeatures(), transcript.getDatasetSequence(), mapping, parentId); /* * fetch and save cross-references */ - super.getCrossReferences(transcript); + cdna.getCrossReferences(transcript); /* * and finally fetch the protein product and save as a cross-reference */ - new EnsemblCdna().addProteinProduct(transcript); + cdna.addProteinProduct(transcript); return transcript; } @@ -360,13 +470,13 @@ public class EnsemblGene extends EnsemblSeqProxy @Override protected boolean retainFeature(SequenceFeature sf, String accessionId) { - if (SequenceOntologyFactory.getInstance().isA(sf.getType(), - SequenceOntologyI.GENE)) + SequenceOntologyI so = SequenceOntologyFactory.getInstance(); + String type = sf.getType(); + if (so.isA(type, SequenceOntologyI.GENE)) { return false; } - - if (isTranscript(sf.getType())) + if (isTranscript(type)) { String parent = (String) sf.getValue(PARENT); if (!(GENE_PREFIX + accessionId).equals(parent)) @@ -388,15 +498,6 @@ public class EnsemblGene extends EnsemblSeqProxy return false; } - @Override - protected List getCrossReferenceDatabases() - { - // found these for ENSG00000157764 on 30/01/2016: - // return new String[] {"Vega_gene", "OTTG", "ENS_LRG_gene", "ArrayExpress", - // "EntrezGene", "HGNC", "MIM_GENE", "MIM_MORBID", "WikiGene"}; - return super.getCrossReferenceDatabases(); - } - /** * Override to do nothing as Ensembl doesn't return a protein sequence for a * gene identifier @@ -412,4 +513,83 @@ public class EnsemblGene extends EnsemblSeqProxy return ACCESSION_REGEX; } + /** + * Returns a descriptor for suitable feature display settings with + *
    + *
  • only exon or sequence_variant features (or their subtypes in the + * Sequence Ontology) visible
    • + *
  • variant features coloured red
    • + *
  • exon features coloured by label (exon name)
    • + *
  • variants displayed above (on top of) exons
    • + *
+ */ + @Override + public FeatureSettingsModelI getFeatureColourScheme() + { + return new FeatureSettingsAdapter() + { + SequenceOntologyI so = SequenceOntologyFactory.getInstance(); + @Override + public boolean isFeatureDisplayed(String type) + { + return (so.isA(type, SequenceOntologyI.EXON) || so.isA(type, + SequenceOntologyI.SEQUENCE_VARIANT)); + } + + @Override + public FeatureColourI getFeatureColour(String type) + { + if (so.isA(type, SequenceOntologyI.EXON)) + { + return new FeatureColour() + { + @Override + public boolean isColourByLabel() + { + return true; + } + }; + } + if (so.isA(type, SequenceOntologyI.SEQUENCE_VARIANT)) + { + return new FeatureColour() + { + + @Override + public Color getColour() + { + return Color.RED; + } + }; + } + return null; + } + + /** + * order to render sequence_variant after exon after the rest + */ + @Override + public int compare(String feature1, String feature2) + { + if (so.isA(feature1, SequenceOntologyI.SEQUENCE_VARIANT)) + { + return +1; + } + if (so.isA(feature2, SequenceOntologyI.SEQUENCE_VARIANT)) + { + return -1; + } + if (so.isA(feature1, SequenceOntologyI.EXON)) + { + return +1; + } + if (so.isA(feature2, SequenceOntologyI.EXON)) + { + return -1; + } + return 0; + } + }; + } + }