package jalview.ext.ensembl;
+import jalview.analysis.AlignmentUtils;
import jalview.datamodel.Alignment;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.DBRefEntry;
import jalview.exceptions.JalviewException;
import jalview.io.FastaFile;
import jalview.io.FileParse;
-import jalview.io.gff.SequenceOntology;
+import jalview.io.gff.SequenceOntologyFactory;
+import jalview.io.gff.SequenceOntologyI;
import jalview.util.DBRefUtils;
import jalview.util.MapList;
/**
* Base class for Ensembl sequence fetchers
*
+ * @see http://rest.ensembl.org/documentation/info/sequence_id
* @author gmcarstairs
*/
public abstract class EnsemblSeqProxy extends EnsemblRestClient
{
+ private static final List<String> CROSS_REFERENCES = Arrays
+ .asList(new String[] { "CCDS", "Uniprot/SWISSPROT",
+ "Uniprot/SPTREMBL" });
+
+ protected static final String CONSEQUENCE_TYPE = "consequence_type";
+
protected static final String PARENT = "Parent";
protected static final String ID = "ID";
+ protected static final String NAME = "Name";
+
+ protected static final String DESCRIPTION = "description";
+
+ /*
+ * enum for 'type' parameter to the /sequence REST service
+ */
public enum EnsemblSeqType
{
/**
- * type=genomic for the full dna including introns
+ * type=genomic to fetch full dna including introns
*/
GENOMIC("genomic"),
/**
- * type=cdna for transcribed dna including UTRs
+ * type=cdna to fetch dna including UTRs
*/
CDNA("cdna"),
/**
- * type=cds for coding dna excluding UTRs
+ * type=cds to fetch coding dna excluding UTRs
*/
CDS("cds"),
/**
- * type=protein for the peptide product sequence
+ * type=protein to fetch peptide product sequence
*/
PROTEIN("protein");
}
- /*
- * genomic sequence, with features retrieved from the REST overlap service
+ /**
+ * Default constructor (to use rest.ensembl.org)
*/
- private SequenceI genomicSequence;
+ public EnsemblSeqProxy()
+ {
+ super();
+ }
/**
- * Constructor
+ * Constructor given the target domain to fetch data from
*/
- public EnsemblSeqProxy()
+ public EnsemblSeqProxy(String d)
{
+ super(d);
}
/**
* Makes the sequence queries to Ensembl's REST service and returns an
- * alignment consisting of the returned sequences
+ * alignment consisting of the returned sequences.
*/
@Override
public AlignmentI getSequenceRecords(String query) throws Exception
{
- long now = System.currentTimeMillis();
// TODO use a String... query vararg instead?
// danger: accession separator used as a regex here, a string elsewhere
// in this case it is ok (it is just a space), but (e.g.) '\' would not be
- List<String> allIds = Arrays.asList(query.split(getAccessionSeparator()));
+ List<String> allIds = Arrays.asList(query
+ .split(getAccessionSeparator()));
AlignmentI alignment = null;
inProgress = true;
+ " chunks. Unexpected problem (" + r.getLocalizedMessage()
+ ")";
System.err.println(msg);
- if (alignment != null)
- {
- break; // return what we got
- }
- else
- {
- throw new JalviewException(msg, r);
- }
+ break;
}
}
+ if (alignment == null)
+ {
+ return null;
+ }
+
/*
- * fetch and transfer genomic sequence features
+ * fetch and transfer genomic sequence features,
+ * fetch protein product and add as cross-reference
*/
for (String accId : allIds)
{
addFeaturesAndProduct(accId, alignment);
}
- inProgress = false;
- System.out.println(getClass().getName() + " took "
- + (System.currentTimeMillis() - now) + "ms to fetch");
+ for (SequenceI seq : alignment.getSequences())
+ {
+ getCrossReferences(seq);
+ }
+
return alignment;
}
*/
protected void addFeaturesAndProduct(String accId, AlignmentI alignment)
{
+ if (alignment == null)
+ {
+ return;
+ }
+
try
{
/*
* get 'dummy' genomic sequence with exon, cds and variation features
*/
- if (genomicSequence == null)
+ SequenceI genomicSequence = null;
+ EnsemblFeatures gffFetcher = new EnsemblFeatures(getDomain());
+ EnsemblFeatureType[] features = getFeaturesToFetch();
+ AlignmentI geneFeatures = gffFetcher.getSequenceRecords(accId,
+ features);
+ if (geneFeatures.getHeight() > 0)
{
- EnsemblOverlap gffFetcher = new EnsemblOverlap();
- EnsemblFeatureType[] features = getFeaturesToFetch();
- AlignmentI geneFeatures = gffFetcher.getSequenceRecords(accId,
- features);
- if (geneFeatures.getHeight() > 0)
- {
- /*
- * transfer features to the query sequence
- */
- genomicSequence = geneFeatures.getSequenceAt(0);
- }
+ genomicSequence = geneFeatures.getSequenceAt(0);
}
if (genomicSequence != null)
{
+ /*
+ * transfer features to the query sequence
+ */
SequenceI querySeq = alignment.findName(accId);
if (transferFeatures(accId, genomicSequence, querySeq))
{
String accId = querySeq.getName();
try
{
- AlignmentI protein = new EnsemblProtein().getSequenceRecords(accId);
+ AlignmentI protein = new EnsemblProtein(getDomain())
+ .getSequenceRecords(accId);
if (protein == null || protein.getHeight() == 0)
{
- System.out.println("Failed to retrieve protein for " + accId);
+ System.out.println("No protein product found for " + accId);
return;
}
SequenceI proteinSeq = protein.getSequenceAt(0);
proteinSeq.createDatasetSequence();
querySeq.createDatasetSequence();
- MapList mapList = mapCdsToProtein(querySeq, proteinSeq);
+ MapList mapList = AlignmentUtils.mapCdsToProtein(querySeq, proteinSeq);
if (mapList != null)
{
- Mapping map = new Mapping(proteinSeq.getDatasetSequence(), mapList);
+ // clunky: ensure Uniprot xref if we have one is on mapped sequence
+ SequenceI ds = proteinSeq.getDatasetSequence();
+ ds.setSourceDBRef(proteinSeq.getSourceDBRef());
+ Mapping map = new Mapping(ds, mapList);
DBRefEntry dbr = new DBRefEntry(getDbSource(), getDbVersion(),
accId, map);
querySeq.getDatasetSequence().addDBRef(dbr);
+
+ /*
+ * 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);
}
} catch (Exception e)
{
}
/**
- * Returns a mapping from dna to protein by inspecting sequence features of
- * type "CDS" on the dna.
+ * Get database xrefs from Ensembl, and attach them to the sequence
*
- * @param dnaSeq
- * @param proteinSeq
- * @return
+ * @param seq
*/
- protected MapList mapCdsToProtein(SequenceI dnaSeq, SequenceI proteinSeq)
+ protected void getCrossReferences(SequenceI seq)
{
- SequenceFeature[] sfs = dnaSeq.getSequenceFeatures();
- if (sfs == null)
+ while (seq.getDatasetSequence() != null)
{
- return null;
+ seq = seq.getDatasetSequence();
}
- List<int[]> ranges = new ArrayList<int[]>(50);
- SequenceOntology so = SequenceOntology.getInstance();
-
- int mappedDnaLength = 0;
-
- /*
- * Map CDS columns of dna to peptide. No need to worry about reverse strand
- * dna here since the retrieved sequence is as transcribed (reverse
- * complement for reverse strand), i.e in the same sense as the peptide.
- */
- for (SequenceFeature sf : sfs)
+ EnsemblXref xrefFetcher = new EnsemblXref(getDomain());
+ List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName(),
+ getCrossReferenceDatabases());
+ for (DBRefEntry xref : xrefs)
{
+ seq.addDBRef(xref);
/*
- * process a CDS feature (or a sub-type of CDS)
+ * Save any Uniprot xref to be the reference for SIFTS mapping
*/
- if (so.isA(sf.getType(), SequenceOntology.CDS))
+ if (DBRefSource.UNIPROT.equals(xref.getSource()))
{
- ranges.add(new int[] { sf.getBegin(), sf.getEnd() });
- mappedDnaLength += Math.abs(sf.getEnd() - sf.getBegin()) + 1;
+ seq.setSourceDBRef(xref);
}
}
- int proteinLength = proteinSeq.getLength();
- List<int[]> proteinRange = new ArrayList<int[]>();
- proteinRange.add(new int[] { 1, proteinLength });
+ }
- /*
- * dna length should map to protein (or protein minus stop codon)
- */
- if (mappedDnaLength == 3 * proteinLength
- || mappedDnaLength == 3 * (proteinLength + 1))
- {
- return new MapList(ranges, proteinRange, 3, 1);
- }
- return null;
+ /**
+ * Returns a list of database names to be used when fetching cross-references.
+ * Specifically, the names are used to filter data returned by the Ensembl
+ * xrefs REST service on the value in field 'dbname'.
+ *
+ * @return
+ */
+ protected List<String> getCrossReferenceDatabases()
+ {
+ return CROSS_REFERENCES;
}
/**
"Only retrieved %d sequences for %d query strings", fr
.getSeqs().size(), ids.size()));
}
+
+ if (fr.getSeqs().size() == 1 && fr.getSeqs().get(0).getLength() == 0)
+ {
+ /*
+ * POST request has returned an empty FASTA file e.g. for invalid id
+ */
+ throw new IOException("No data returned for " + ids);
+ }
+
if (fr.getSeqs().size() > 0)
{
AlignmentI seqal = new Alignment(
@Override
protected URL getUrl(List<String> ids) throws MalformedURLException
{
- // ids are not used - they go in the POST body instead
+ /*
+ * a single id is included in the URL path
+ * multiple ids go in the POST body instead
+ */
StringBuffer urlstring = new StringBuffer(128);
- urlstring.append(SEQUENCE_ID_URL);
-
+ urlstring.append(getDomain() + "/sequence/id");
+ if (ids.size() == 1)
+ {
+ urlstring.append("/").append(ids.get(0));
+ }
// @see https://github.com/Ensembl/ensembl-rest/wiki/Output-formats
urlstring.append("?type=").append(getSourceEnsemblType().getType());
urlstring.append(("&Accept=text/x-fasta"));
}
@Override
- protected String getRequestMimeType()
+ protected String getRequestMimeType(boolean multipleIds)
{
- return "application/json";
+ return multipleIds ? "application/json" : "text/x-fasta";
}
@Override
* the start position of the sequence we are mapping to
* @return
*/
- protected MapList getGenomicRanges(SequenceI sourceSequence,
+ protected MapList getGenomicRangesFromFeatures(SequenceI sourceSequence,
String accId, int start)
{
SequenceFeature[] sfs = sourceSequence.getSequenceFeatures();
}
/*
- * generously size for initial number of cds regions
+ * generously initial size for number of cds regions
* (worst case titin Q8WZ42 has c. 313 exons)
*/
List<int[]> regions = new ArrayList<int[]>(100);
- int sourceLength = sourceSequence.getLength();
int mappedLength = 0;
int direction = 1; // forward
boolean directionSet = false;
*/
if (identifiesSequence(sf, accId))
{
- int strand = sf.getStrand();
-
- if (directionSet && strand != direction)
- {
- // abort - mix of forward and backward
+ int strand = sf.getStrand();
+ strand = strand == 0 ? 1 : strand; // treat unknown as forward
+
+ if (directionSet && strand != direction)
+ {
+ // abort - mix of forward and backward
System.err.println("Error: forward and backward strand for "
+ accId);
return null;
}
mappedLength += Math.abs(sf.getEnd() - sf.getBegin() + 1);
- if (mappedLength >= sourceLength)
+ if (!isSpliceable())
{
/*
- * break for the case of matching gene features v gene sequence
- * - only need to locate the 'gene' feature for accId
+ * 'gene' sequence is contiguous so we can stop as soon as its
+ * identifying feature has been found
*/
break;
}
*/
Collections.sort(regions, new RangeSorter(direction == 1));
- List<int[]> to = new ArrayList<int[]>();
- to.add(new int[] { start, start + mappedLength - 1 });
+ List<int[]> to = Arrays.asList(new int[] { start,
+ start + mappedLength - 1 });
return new MapList(regions, to, 1, 1);
}
/**
+ * Answers true if the sequence being retrieved may occupy discontiguous
+ * regions on the genomic sequence.
+ */
+ protected boolean isSpliceable()
+ {
+ return true;
+ }
+
+ /**
* Returns true if the sequence feature marks positions of the genomic
* sequence feature which are within the sequence being retrieved. For
* example, an 'exon' feature whose parent is the target transcript marks the
String accId);
/**
- * Transfers the sequence feature to the target sequence, adjusting its start
- * and end range based on the 'overlap' ranges. Features which do not overlap
- * the target sequence are ignored, as are features with a parent other than
- * the target sequence id.
+ * Transfers the sequence feature to the target sequence, locating its start
+ * and end range based on the mapping. Features which do not overlap the
+ * target sequence are ignored.
*
* @param sf
* @param targetSequence
- * @param overlap
+ * @param mapping
+ * mapping from the sequence feature's coordinates to the target
+ * sequence
*/
protected void transferFeature(SequenceFeature sf,
- SequenceI targetSequence, MapList overlap)
+ SequenceI targetSequence, MapList mapping)
{
int start = sf.getBegin();
int end = sf.getEnd();
- int[] mappedRange = overlap.locateInTo(start, end);
+ int[] mappedRange = mapping.locateInTo(start, end);
if (mappedRange != null)
{
/*
* for sequence_variant, make an additional feature with consequence
*/
- if (SequenceOntology.getInstance().isSequenceVariant(sf.getType()))
- {
- String consequence = (String) sf.getValue("consequence_type");
- if (consequence != null)
- {
- SequenceFeature sf2 = new SequenceFeature("consequence",
- consequence, copy.getBegin(), copy.getEnd(), 0f,
- null);
- targetSequence.addSequenceFeature(sf2);
- }
- }
+ // if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
+ // SequenceOntologyI.SEQUENCE_VARIANT))
+ // {
+ // String consequence = (String) sf.getValue(CONSEQUENCE_TYPE);
+ // if (consequence != null)
+ // {
+ // SequenceFeature sf2 = new SequenceFeature("consequence",
+ // consequence, copy.getBegin(), copy.getEnd(), 0f,
+ // null);
+ // targetSequence.addSequenceFeature(sf2);
+ // }
+ // }
}
-
}
/**
return false;
}
+ // long start = System.currentTimeMillis();
SequenceFeature[] sfs = sourceSequence.getSequenceFeatures();
- MapList overlap = getGenomicRanges(sourceSequence, accessionId,
+ MapList mapping = getGenomicRangesFromFeatures(sourceSequence, accessionId,
targetSequence.getStart());
- if (overlap == null)
+ if (mapping == null)
{
return false;
}
- final boolean forwardStrand = overlap.isFromForwardStrand();
+ boolean result = transferFeatures(sfs, targetSequence, mapping,
+ accessionId);
+ // System.out.println("transferFeatures (" + (sfs.length) + " --> "
+ // + targetSequence.getSequenceFeatures().length + ") to "
+ // + targetSequence.getName()
+ // + " took " + (System.currentTimeMillis() - start) + "ms");
+ return result;
+ }
+
+ /**
+ * Transfer features to the target sequence. The start/end positions are
+ * converted using the mapping. Features which do not overlap are ignored.
+ * Features whose parent is not the specified identifier are also ignored.
+ *
+ * @param features
+ * @param targetSequence
+ * @param mapping
+ * @param parentId
+ * @return
+ */
+ protected boolean transferFeatures(SequenceFeature[] features,
+ SequenceI targetSequence, MapList mapping, String parentId)
+ {
+ final boolean forwardStrand = mapping.isFromForwardStrand();
/*
* sort features by start position (descending if reverse strand)
* before transferring (in forwards order) to the target sequence
*/
- Arrays.sort(sfs, new Comparator<SequenceFeature>()
+ Arrays.sort(features, new Comparator<SequenceFeature>()
{
@Override
public int compare(SequenceFeature o1, SequenceFeature o2)
});
boolean transferred = false;
- for (SequenceFeature sf : sfs)
+ for (SequenceFeature sf : features)
{
- if (retainFeature(sf, accessionId))
+ if (retainFeature(sf, parentId))
{
- transferFeature(sf, targetSequence, overlap);
+ transferFeature(sf, targetSequence, mapping);
transferred = true;
}
}
}
/**
- * Answers true if the feature is one to attach to the retrieved sequence
+ * Answers true if the feature type is one we want to keep for the sequence.
+ * Some features are only retrieved in order to identify the sequence range,
+ * and may then be discarded as redundant information (e.g. "CDS" feature for
+ * a CDS sequence).
+ */
+ @SuppressWarnings("unused")
+ protected boolean retainFeature(SequenceFeature sf, String accessionId)
+ {
+ return true; // override as required
+ }
+
+ /**
+ * Answers true if the feature has a Parent which refers to the given
+ * accession id, or if the feature has no parent. Answers false if the
+ * feature's Parent is for a different accession id.
*
- * @param type
+ * @param sf
+ * @param identifier
* @return
*/
- protected boolean retainFeature(SequenceFeature sf, String accessionId)
+ protected boolean featureMayBelong(SequenceFeature sf, String identifier)
{
String parent = (String) sf.getValue(PARENT);
- if (parent != null && !parent.contains(accessionId))
+ // using contains to allow for prefix "gene:", "transcript:" etc
+ if (parent != null && !parent.contains(identifier))
{
// this genomic feature belongs to a different transcript
return false;
+ " sequence with variant features";
}
- public AlignmentI getSequenceRecords(String transcriptId,
- SequenceI geneSeq) throws Exception
+ /**
+ * Returns a (possibly empty) list of features on the sequence which have the
+ * specified sequence ontology type (or a sub-type of it), and the given
+ * identifier as parent
+ *
+ * @param sequence
+ * @param type
+ * @param parentId
+ * @return
+ */
+ protected List<SequenceFeature> findFeatures(SequenceI sequence,
+ String type, String parentId)
+ {
+ List<SequenceFeature> result = new ArrayList<SequenceFeature>();
+
+ SequenceFeature[] sfs = sequence.getSequenceFeatures();
+ if (sfs != null) {
+ SequenceOntologyI so = SequenceOntologyFactory.getInstance();
+ for (SequenceFeature sf :sfs) {
+ if (so.isA(sf.getType(), type))
+ {
+ String parent = (String) sf.getValue(PARENT);
+ if (parent.equals(parentId))
+ {
+ result.add(sf);
+ }
+ }
+ }
+ }
+ return result;
+ }
+
+ /**
+ * Answers true if the feature type is either 'NMD_transcript_variant' or
+ * 'transcript' or one of its sub-types in the Sequence Ontology. This is
+ * needed because NMD_transcript_variant behaves like 'transcript' in Ensembl
+ * although strictly speaking it is not (it is a sub-type of
+ * sequence_variant).
+ *
+ * @param featureType
+ * @return
+ */
+ public static boolean isTranscript(String featureType)
{
- this.genomicSequence = geneSeq;
- return getSequenceRecords(transcriptId);
+ return SequenceOntologyI.NMD_TRANSCRIPT_VARIANT.equals(featureType)
+ || SequenceOntologyFactory.getInstance().isA(featureType,
+ SequenceOntologyI.TRANSCRIPT);
}
}