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.getFeatures().getPositionalFeatures(),
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.
+ *
+ * Transcript features are those of type "transcript", or any of its sub-types
+ * in the Sequence Ontology e.g. "mRNA", "processed_transcript". We also
+ * include "NMD_transcript_variant", because this type behaves like a
+ * transcript identifier in Ensembl, although strictly speaking it is not in
+ * the SO.
*
* @param accId
* @param geneSequence
@@ -193,23 +401,21 @@ public class EnsemblGene extends EnsemblSeqProxy
protected List getTranscriptFeatures(String accId,
SequenceI geneSequence)
{
- List transcriptFeatures = new ArrayList();
+ List transcriptFeatures = new ArrayList<>();
+
+ String parentIdentifier = GENE_PREFIX + accId;
- String parentIdentifier = "gene:" + accId;
- SequenceFeature[] sfs = geneSequence.getSequenceFeatures();
+ List sfs = geneSequence.getFeatures()
+ .getFeaturesByOntology(SequenceOntologyI.TRANSCRIPT);
+ sfs.addAll(geneSequence.getFeatures().getPositionalFeatures(
+ SequenceOntologyI.NMD_TRANSCRIPT_VARIANT));
- if (sfs != null)
+ for (SequenceFeature sf : sfs)
{
- for (SequenceFeature sf : sfs)
+ String parent = (String) sf.getValue(PARENT);
+ if (parentIdentifier.equals(parent))
{
- if (isTranscript(sf.getType()))
- {
- String parent = (String) sf.getValue(PARENT);
- if (parentIdentifier.equals(parent))
- {
- transcriptFeatures.add(sf);
- }
- }
+ transcriptFeatures.add(sf);
}
}
@@ -219,11 +425,11 @@ 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()
@@ -244,8 +450,9 @@ public class EnsemblGene extends EnsemblSeqProxy
if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
SequenceOntologyI.GENE))
{
- String id = (String) sf.getValue(ID);
- if (("gene:" + accId).equals(id))
+ // NB features as gff use 'ID'; rest services return as 'id'
+ String id = (String) sf.getValue("ID");
+ if ((GENE_PREFIX + accId).equals(id))
{
return true;
}
@@ -263,16 +470,16 @@ 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;
}
@@ -291,22 +498,99 @@ public class EnsemblGene extends EnsemblSeqProxy
return false;
}
+ /**
+ * Override to do nothing as Ensembl doesn't return a protein sequence for a
+ * gene identifier
+ */
@Override
- protected List getCrossReferenceDatabases()
+ protected void addProteinProduct(SequenceI querySeq)
{
- // 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
+ public Regex getAccessionValidator()
+ {
+ return ACCESSION_REGEX;
}
/**
- * Override to do nothing as Ensembl doesn't return a protein sequence for a
- * gene identifier
+ * 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
- protected void addProteinProduct(SequenceI querySeq)
+ 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;
+ }
+ };
}
}