X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fext%2Fensembl%2FEnsemblGene.java;h=84b5dcfd2063f6cba073264b7fbd0112d7659ae4;hb=6d9835352394d78d9e114b317a3fe2545e27940d;hp=dc287965158d740ab4bb21c1ea094df4cdd5f012;hpb=46b2bc114452c0e30463214b3e82eb0a5c98d23f;p=jalview.git diff --git a/src/jalview/ext/ensembl/EnsemblGene.java b/src/jalview/ext/ensembl/EnsemblGene.java index dc28796..84b5dcf 100644 --- a/src/jalview/ext/ensembl/EnsemblGene.java +++ b/src/jalview/ext/ensembl/EnsemblGene.java @@ -1,13 +1,21 @@ 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.FeatureColourAdapter; +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; @@ -21,19 +29,41 @@ import com.stevesoft.pat.Regex; */ public class EnsemblGene extends EnsemblSeqProxy { - // TODO modify to accept other species e.g. ENSMUSGnnn - private static final Regex ACCESSION_REGEX = new Regex( - "((ENSG)[0-9]{11})"); + private static final String GENE_PREFIX = "gene:"; + + /* + * accepts anything as we will attempt lookup of gene or + * transcript id or gene name + */ + private static final Regex ACCESSION_REGEX = new Regex(".*"); private static final EnsemblFeatureType[] FEATURES_TO_FETCH = { EnsemblFeatureType.gene, EnsemblFeatureType.transcript, 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 @@ -49,8 +79,15 @@ public class EnsemblGene extends EnsemblSeqProxy } /** - * Builds an alignment of all transcripts for the requested gene: + * Returns an alignment containing the gene(s) for the given gene or + * transcript identifier, or external identifier (e.g. Uniprot id). If given a + * gene name or external identifier, returns any related gene sequences found + * for model organisms. If only a single gene is queried for, then its + * transcripts are also retrieved and added to the alignment.
+ * Method: * + * + * @param query + * 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 ? if an ENST identifier is supplied, convert to ENSG? + /* + * if given a transcript id, look up its gene parent + */ + if (isTranscriptIdentifier(query)) + { + query = new EnsemblLookup(getDomain()).getParent(query); + if (query == null) + { + return null; + } + } + + /* + * if given a gene or other external name, lookup and fetch + * the corresponding gene for all model organisms + */ + if (!isGeneIdentifier(query)) + { + List geneIds = new EnsemblSymbol(getDomain()).getIds(query); + if (geneIds.isEmpty()) + { + return null; + } + String theIds = StringUtils.listToDelimitedString(geneIds, + getAccessionSeparator()); + return getSequenceRecords(theIds); + } + + /* + * fetch the gene sequence(s) with features and xrefs + */ AlignmentI al = super.getSequenceRecords(query); - if (al.getHeight() > 0) + + /* + * if we retrieved a single gene, get its transcripts as well + */ + if (al.getHeight() == 1) { getTranscripts(al, query); } @@ -77,6 +152,29 @@ public class EnsemblGene extends EnsemblSeqProxy } /** + * Attempts to get Ensembl stable identifiers for model organisms for a gene + * name by calling the xrefs symbol REST service to resolve the gene name. + * + * @param query + * @return + */ + protected String getGeneIdentifiersForName(String query) + { + List ids = new EnsemblSymbol(getDomain()).getIds(query); + if (ids != null) + { + for (String id : ids) + { + if (isGeneIdentifier(id)) + { + return id; + } + } + } + return null; + } + + /** * Constructs all transcripts for the gene, as identified by "transcript" * features whose Parent is the requested gene. The coding transcript * sequences (i.e. with introns omitted) are added to the alignment. @@ -96,6 +194,36 @@ 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()])); + } } /** @@ -114,7 +242,7 @@ public class EnsemblGene extends EnsemblSeqProxy SequenceI makeTranscript(SequenceFeature transcriptFeature, AlignmentI al, SequenceI gene) { - String accId = (String) transcriptFeature.getValue("transcript_id"); + String accId = getTranscriptId(transcriptFeature); if (accId == null) { return null; @@ -161,10 +289,25 @@ 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) + { + description = (String) transcriptFeature.getValue(DESCRIPTION); + } + if (description != null) { - transcript.setDescription(geneName); + try + { + transcript.setDescription(URLDecoder.decode(description, "UTF-8")); + } catch (UnsupportedEncodingException e) + { + e.printStackTrace(); // as if + } } transcript.createDatasetSequence(); @@ -177,18 +320,35 @@ 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 + */ + cdna.getCrossReferences(transcript); + + /* * and finally fetch the protein product and save as a cross-reference */ - new EnsemblCdna().addProteinProduct(transcript); + cdna.addProteinProduct(transcript); return transcript; } /** + * Returns the 'transcript_id' property of the sequence feature (or null) + * + * @param feature + * @return + */ + protected String getTranscriptId(SequenceFeature feature) + { + return (String) feature.getValue("transcript_id"); + } + + /** * Returns a list of the transcript features on the sequence whose Parent is * the gene for the accession id. * @@ -201,7 +361,7 @@ public class EnsemblGene extends EnsemblSeqProxy { List transcriptFeatures = new ArrayList(); - String parentIdentifier = "gene:" + accId; + String parentIdentifier = GENE_PREFIX + accId; SequenceFeature[] sfs = geneSequence.getSequenceFeatures(); if (sfs != null) @@ -225,11 +385,11 @@ public class EnsemblGene extends EnsemblSeqProxy @Override public String getDescription() { - return "Fetches all transcripts and variant features for a gene"; + return "Fetches all transcripts and variant features for a gene or transcript"; } /** - * Default test query is a transcript + * Default test query is a gene id (can also enter a transcript id) */ @Override public String getTestQuery() @@ -251,7 +411,7 @@ public class EnsemblGene extends EnsemblSeqProxy SequenceOntologyI.GENE)) { String id = (String) sf.getValue(ID); - if (("gene:" + accId).equals(id)) + if ((GENE_PREFIX + accId).equals(id)) { return true; } @@ -269,16 +429,16 @@ 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:" + accessionId).equals(parent)) + if (!(GENE_PREFIX + accessionId).equals(parent)) { return false; } @@ -297,15 +457,6 @@ 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 @@ -321,4 +472,89 @@ 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 FeatureColourAdapter() + { + @Override + public boolean isColourByLabel() + { + return true; + } + }; + } + if (so.isA(type, SequenceOntologyI.SEQUENCE_VARIANT)) + { + return new FeatureColourAdapter() + { + + @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; + } + }; + } + + @Override + public int getMaximumQueryCount() + { + return 1; + } + }