package jalview.ext.ensembl;
+import jalview.api.FeatureColourI;
+import jalview.api.FeatureSettingsI;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.Sequence;
import jalview.datamodel.SequenceFeature;
import jalview.datamodel.SequenceI;
-import jalview.io.gff.SequenceOntology;
+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.util.ArrayList;
import java.util.Arrays;
import java.util.List;
+import com.stevesoft.pat.Regex;
+
/**
* A class that fetches genomic sequence and all transcripts for an Ensembl gene
*
*/
public class EnsemblGene extends EnsemblSeqProxy
{
+ 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,
}
/**
- * 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. <br>
+ * Method:
* <ul>
+ * <li>resolves a transcript identifier by looking up its parent gene id</li>
+ * <li>resolves an external identifier by looking up xref-ed gene ids</li>
* <li>fetches the gene sequence</li>
* <li>fetches features on the sequence</li>
* <li>identifies "transcript" features whose Parent is the requested gene</li>
* <li>aligns each transcript against the gene sequence based on the position
* mappings</li>
* </ul>
+ *
+ * @param query
+ * one or more identifiers separated by a space
+ * @return an alignment containing one or more genes, and possibly
+ * transcripts, or null
*/
@Override
public AlignmentI getSequenceRecords(String query) throws Exception
{
- // TODO ? if an ENST identifier is supplied, convert to ENSG?
+ // 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
+ */
+ if (isTranscriptIdentifier(query))
+ {
+ // we are assuming all transcripts have the same gene parent here
+ query = new EnsemblLookup().getParent(queries[0]);
+ 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<String> geneIds = new EnsemblSymbol().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);
}
}
/**
+ * 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<String> ids = new EnsemblSymbol().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.
{
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<SequenceFeature> filtered = new ArrayList<SequenceFeature>();
+ 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()]));
+ }
}
/**
SequenceI makeTranscript(SequenceFeature transcriptFeature,
AlignmentI al, SequenceI gene)
{
- String accId = (String) transcriptFeature.getValue("transcript_id");
+ String accId = getTranscriptId(transcriptFeature);
if (accId == null)
{
return null;
*/
String parentId = "transcript:" + accId;
List<SequenceFeature> splices = findFeatures(gene,
- SequenceOntology.EXON, parentId);
+ SequenceOntologyI.EXON, parentId);
if (splices.isEmpty())
{
- splices = findFeatures(gene, SequenceOntology.CDS, parentId);
+ splices = findFeatures(gene, SequenceOntologyI.CDS, parentId);
}
int transcriptLength = 0;
}
Sequence transcript = new Sequence(accId, seqChars, 1, transcriptLength);
+ String geneName = (String) transcriptFeature.getValue(NAME);
+ if (geneName != null)
+ {
+ transcript.setDescription(geneName);
+ }
transcript.createDatasetSequence();
al.addSequence(transcript);
transcript.getDatasetSequence(), mapping, parentId);
/*
+ * fetch and save cross-references
+ */
+ super.getCrossReferences(transcript);
+
+ /*
* and finally fetch the protein product and save as a cross-reference
*/
- addProteinProduct(transcript);
+ new EnsemblCdna().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. Also removes all transcript features from
- * the gene sequence, as we have no further need for them and they obscure
- * more useful features on the display.
+ * the gene for the accession id.
*
* @param accId
* @param geneSequence
{
List<SequenceFeature> transcriptFeatures = new ArrayList<SequenceFeature>();
- List<SequenceFeature> keptFeatures = new ArrayList<SequenceFeature>();
- String parentIdentifier = "gene:" + accId;
+ String parentIdentifier = GENE_PREFIX + accId;
SequenceFeature[] sfs = geneSequence.getSequenceFeatures();
if (sfs != null)
transcriptFeatures.add(sf);
}
}
- else
- {
- keptFeatures.add(sf);
- }
}
}
- SequenceFeature[] featuresRetained = keptFeatures.toArray(new SequenceFeature[keptFeatures.size()]);
- geneSequence.getDatasetSequence().setSequenceFeatures(featuresRetained);
return transcriptFeatures;
}
@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()
@Override
protected boolean identifiesSequence(SequenceFeature sf, String accId)
{
- if (SequenceOntology.getInstance().isA(sf.getType(),
- SequenceOntology.GENE))
+ if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
+ SequenceOntologyI.GENE))
{
String id = (String) sf.getValue(ID);
- if (("gene:" + accId).equals(id))
+ if ((GENE_PREFIX + accId).equals(id))
{
return true;
}
@Override
protected boolean retainFeature(SequenceFeature sf, String accessionId)
{
- if (SequenceOntology.getInstance().isA(sf.getType(),
- SequenceOntology.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;
}
return false;
}
+ @Override
+ protected List<String> 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
+ */
+ @Override
+ protected void addProteinProduct(SequenceI querySeq)
+ {
+ }
+
+ @Override
+ public Regex getAccessionValidator()
+ {
+ return ACCESSION_REGEX;
+ }
+
+ @Override
+ public FeatureSettingsI 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;
+ }
+ };
+ }
+
}
git://source.jalview.org / jalview.git / blobdiff