2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.ext.ensembl;
23 import jalview.analysis.AlignmentUtils;
24 import jalview.analysis.Dna;
25 import jalview.bin.Cache;
26 import jalview.datamodel.Alignment;
27 import jalview.datamodel.AlignmentI;
28 import jalview.datamodel.DBRefEntry;
29 import jalview.datamodel.DBRefSource;
30 import jalview.datamodel.Mapping;
31 import jalview.datamodel.Sequence;
32 import jalview.datamodel.SequenceFeature;
33 import jalview.datamodel.SequenceI;
34 import jalview.datamodel.features.SequenceFeatures;
35 import jalview.exceptions.JalviewException;
36 import jalview.io.gff.Gff3Helper;
37 import jalview.io.gff.SequenceOntologyFactory;
38 import jalview.io.gff.SequenceOntologyI;
39 import jalview.util.Comparison;
40 import jalview.util.DBRefUtils;
41 import jalview.util.IntRangeComparator;
42 import jalview.util.MapList;
43 import jalview.util.Platform;
45 import java.io.IOException;
46 import java.net.MalformedURLException;
48 import java.util.ArrayList;
49 import java.util.Arrays;
50 import java.util.Collections;
51 import java.util.List;
54 import org.json.simple.parser.ParseException;
57 * Base class for Ensembl sequence fetchers
59 * @see http://rest.ensembl.org/documentation/info/sequence_id
62 public abstract class EnsemblSeqProxy extends EnsemblRestClient
64 protected static final String DESCRIPTION = "description";
67 * enum for 'type' parameter to the /sequence REST service
69 public enum EnsemblSeqType
72 * type=genomic to fetch full dna including introns
77 * type=cdna to fetch coding dna including UTRs
82 * type=cds to fetch coding dna excluding UTRs
87 * type=protein to fetch peptide product sequence
92 * the value of the 'type' parameter to fetch this version of
97 EnsemblSeqType(String t)
102 public String getType()
110 * Default constructor (to use rest.ensembl.org)
112 public EnsemblSeqProxy()
118 * Constructor given the target domain to fetch data from
120 public EnsemblSeqProxy(String d)
126 * Makes the sequence queries to Ensembl's REST service and returns an
127 * alignment consisting of the returned sequences.
130 public AlignmentI getSequenceRecords(String query) throws Exception
132 // TODO use a String... query vararg instead?
134 // danger: accession separator used as a regex here, a string elsewhere
135 // in this case it is ok (it is just a space), but (e.g.) '\' would not be
136 List<String> allIds = Arrays
137 .asList(query.split(getAccessionSeparator()));
138 AlignmentI alignment = null;
142 * execute queries, if necessary in batches of the
143 * maximum allowed number of ids
145 int maxQueryCount = getMaximumQueryCount();
146 for (int v = 0, vSize = allIds.size(); v < vSize; v += maxQueryCount)
148 int p = Math.min(vSize, v + maxQueryCount);
149 List<String> ids = allIds.subList(v, p);
152 alignment = fetchSequences(ids, alignment);
153 } catch (Throwable r)
156 String msg = "Aborting ID retrieval after " + v
157 + " chunks. Unexpected problem (" + r.getLocalizedMessage()
159 System.err.println(msg);
165 if (alignment == null)
171 * fetch and transfer genomic sequence features,
172 * fetch protein product and add as cross-reference
174 for (String accId : allIds)
176 addFeaturesAndProduct(accId, alignment);
179 for (SequenceI seq : alignment.getSequences())
181 getCrossReferences(seq);
188 * Fetches Ensembl features using the /overlap REST endpoint, and adds them to
189 * the sequence in the alignment. Also fetches the protein product, maps it
190 * from the CDS features of the sequence, and saves it as a cross-reference of
196 protected void addFeaturesAndProduct(String accId, AlignmentI alignment)
198 if (alignment == null)
206 * get 'dummy' genomic sequence with gene, transcript,
207 * exon, cds and variation features
209 SequenceI genomicSequence = null;
210 EnsemblFeatures gffFetcher = new EnsemblFeatures(getDomain());
211 EnsemblFeatureType[] features = getFeaturesToFetch();
212 AlignmentI geneFeatures = gffFetcher.getSequenceRecords(accId,
214 if (geneFeatures != null && geneFeatures.getHeight() > 0)
216 genomicSequence = geneFeatures.getSequenceAt(0);
218 if (genomicSequence != null)
221 * transfer features to the query sequence
223 SequenceI querySeq = alignment.findName(accId, true);
224 if (transferFeatures(accId, genomicSequence, querySeq))
228 * fetch and map protein product, and add it as a cross-reference
229 * of the retrieved sequence
231 addProteinProduct(querySeq);
234 } catch (IOException e)
237 "Error transferring Ensembl features: " + e.getMessage());
242 * Returns those sequence feature types to fetch from Ensembl. We may want
243 * features either because they are of interest to the user, or as means to
244 * identify the locations of the sequence on the genomic sequence (CDS
245 * features identify CDS, exon features identify cDNA etc).
249 protected abstract EnsemblFeatureType[] getFeaturesToFetch();
252 * Fetches and maps the protein product, and adds it as a cross-reference of
253 * the retrieved sequence
255 protected void addProteinProduct(SequenceI querySeq)
257 String accId = querySeq.getName();
260 System.out.println("Adding protein product for " + accId);
261 AlignmentI protein = new EnsemblProtein(getDomain())
262 .getSequenceRecords(accId);
263 if (protein == null || protein.getHeight() == 0)
265 System.out.println("No protein product found for " + accId);
268 SequenceI proteinSeq = protein.getSequenceAt(0);
271 * need dataset sequences (to be the subject of mappings)
273 proteinSeq.createDatasetSequence();
274 querySeq.createDatasetSequence();
276 MapList mapList = AlignmentUtils.mapCdsToProtein(querySeq,
280 // clunky: ensure Uniprot xref if we have one is on mapped sequence
281 SequenceI ds = proteinSeq.getDatasetSequence();
282 // TODO: Verify ensp primary ref is on proteinSeq.getDatasetSequence()
283 Mapping map = new Mapping(ds, mapList);
284 DBRefEntry dbr = new DBRefEntry(getDbSource(),
285 getEnsemblDataVersion(), proteinSeq.getName(), map);
286 querySeq.getDatasetSequence().addDBRef(dbr);
287 List<DBRefEntry> uprots = DBRefUtils.selectRefs(ds.getDBRefs(),
289 { DBRefSource.UNIPROT });
290 List<DBRefEntry> upxrefs = DBRefUtils.selectRefs(querySeq.getDBRefs(),
292 { DBRefSource.UNIPROT });
295 for (DBRefEntry up : uprots)
297 // locate local uniprot ref and map
298 List<DBRefEntry> upx = DBRefUtils.searchRefs(upxrefs,
299 up.getAccessionId());
308 "Implementation issue - multiple uniprot acc on product sequence.");
313 upxref = new DBRefEntry(DBRefSource.UNIPROT,
314 getEnsemblDataVersion(), up.getAccessionId());
317 Mapping newMap = new Mapping(ds, mapList);
318 upxref.setVersion(getEnsemblDataVersion());
319 upxref.setMap(newMap);
322 // add the new uniprot ref
323 querySeq.getDatasetSequence().addDBRef(upxref);
330 * copy exon features to protein, compute peptide variants from dna
331 * variants and add as features on the protein sequence ta-da
333 AlignmentUtils.computeProteinFeatures(querySeq, proteinSeq,
336 } catch (Exception e)
339 .println(String.format("Error retrieving protein for %s: %s",
340 accId, e.getMessage()));
345 * Get database xrefs from Ensembl, and attach them to the sequence
349 protected void getCrossReferences(SequenceI seq)
351 while (seq.getDatasetSequence() != null)
353 seq = seq.getDatasetSequence();
355 EnsemblXref xrefFetcher = new EnsemblXref(getDomain(), getDbSource(),
356 getEnsemblDataVersion());
357 List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName());
359 for (int i = 0, n = xrefs.size(); i < n; i++)
361 // BH 2019.01.25 this next method was taking 174 ms PER addition for a 266-reference example.
362 // DBRefUtils.ensurePrimaries(seq)
363 // was at the end of seq.addDBRef, so executed after ever addition!
364 // This method was moved to seq.getPrimaryDBRefs()
365 seq.addDBRef(xrefs.get(i));
368 * and add a reference to itself
370 DBRefEntry self = new DBRefEntry(getDbSource(), getEnsemblDataVersion(),
376 * Fetches sequences for the list of accession ids and adds them to the
377 * alignment. Returns the extended (or created) alignment.
382 * @throws JalviewException
383 * @throws IOException
385 protected AlignmentI fetchSequences(List<String> ids,
386 AlignmentI alignment) throws JalviewException, IOException
388 if (!isEnsemblAvailable())
391 throw new JalviewException("ENSEMBL Rest API not available.");
393 List<SequenceI> seqs = parseSequenceJson(ids);
399 throw new IOException("No data returned for " + ids);
402 if (seqs.size() != ids.size())
404 System.out.println(String.format(
405 "Only retrieved %d sequences for %d query strings",
406 seqs.size(), ids.size()));
411 AlignmentI seqal = new Alignment(
412 seqs.toArray(new SequenceI[seqs.size()]));
413 for (SequenceI seq : seqs)
415 if (seq.getDescription() == null)
417 seq.setDescription(getDbName());
419 String name = seq.getName();
420 if (ids.contains(name)
421 || ids.contains(name.replace("ENSP", "ENST")))
423 // TODO JAL-3077 use true accession version in dbref
424 DBRefEntry dbref = DBRefUtils.parseToDbRef(seq, getDbSource(),
425 getEnsemblDataVersion(), name);
429 if (alignment == null)
435 alignment.append(seqal);
442 * Parses a JSON response for a single sequence ID query
445 * @return a single jalview.datamodel.Sequence
446 * @see http://rest.ensembl.org/documentation/info/sequence_id
448 @SuppressWarnings("unchecked")
449 protected List<SequenceI> parseSequenceJson(List<String> ids)
451 List<SequenceI> result = new ArrayList<>();
455 * for now, assumes only one sequence returned; refactor if needed
456 * in future to handle a JSONArray with more than one
458 Map<String, Object> val = (Map<String, Object>) getJSON(null, ids, -1, MODE_MAP, null);
461 Object s = val.get("desc");
462 String desc = s == null ? null : s.toString();
464 String id = s == null ? null : s.toString();
466 String seq = s == null ? null : s.toString();
467 Sequence sequence = new Sequence(id, seq);
470 sequence.setDescription(desc);
472 // todo JAL-3077 make a DBRefEntry with true accession version
473 // s = val.get("version");
474 // String version = s == null ? "0" : s.toString();
475 // DBRefEntry dbref = new DBRefEntry(getDbSource(), version, id);
476 // sequence.addDBRef(dbref);
477 result.add(sequence);
478 } catch (ParseException | IOException e)
480 System.err.println("Error processing JSON response: " + e.toString());
487 * Returns the URL for the REST call
490 * @throws MalformedURLException
493 protected URL getUrl(List<String> ids) throws MalformedURLException
496 * a single id is included in the URL path
497 * multiple ids go in the POST body instead
499 StringBuffer urlstring = new StringBuffer(128);
500 urlstring.append(getDomain() + "/sequence/id");
503 urlstring.append("/").append(ids.get(0));
505 // @see https://github.com/Ensembl/ensembl-rest/wiki/Output-formats
506 urlstring.append("?type=").append(getSourceEnsemblType().getType());
507 urlstring.append(("&Accept=application/json"));
508 urlstring.append(("&content-type=application/json"));
510 String objectType = getObjectType();
511 if (objectType != null)
513 urlstring.append("&").append(OBJECT_TYPE).append("=")
517 URL url = new URL(urlstring.toString());
522 * Override this method to specify object_type request parameter
526 protected String getObjectType()
532 * A sequence/id POST request currently allows up to 50 queries
534 * @see http://rest.ensembl.org/documentation/info/sequence_id_post
537 public int getMaximumQueryCount()
543 protected boolean useGetRequest()
550 * @return the configured sequence return type for this source
552 protected abstract EnsemblSeqType getSourceEnsemblType();
555 * Returns a list of [start, end] genomic ranges corresponding to the sequence
558 * The correspondence between the frames of reference is made by locating
559 * those features on the genomic sequence which identify the retrieved
560 * sequence. Specifically
562 * <li>genomic sequence is identified by "transcript" features with
563 * ID=transcript:transcriptId</li>
564 * <li>cdna sequence is identified by "exon" features with
565 * Parent=transcript:transcriptId</li>
566 * <li>cds sequence is identified by "CDS" features with
567 * Parent=transcript:transcriptId</li>
570 * The returned ranges are sorted to run forwards (for positive strand) or
571 * backwards (for negative strand). Aborts and returns null if both positive
572 * and negative strand are found (this should not normally happen).
574 * @param sourceSequence
577 * the start position of the sequence we are mapping to
580 protected MapList getGenomicRangesFromFeatures(SequenceI sourceSequence,
581 String accId, int start)
583 List<SequenceFeature> sfs = getIdentifyingFeatures(sourceSequence,
591 * generously initial size for number of cds regions
592 * (worst case titin Q8WZ42 has c. 313 exons)
594 List<int[]> regions = new ArrayList<>(100);
595 int mappedLength = 0;
596 int direction = 1; // forward
597 boolean directionSet = false;
599 for (SequenceFeature sf : sfs)
601 int strand = sf.getStrand();
602 strand = strand == 0 ? 1 : strand; // treat unknown as forward
604 if (directionSet && strand != direction)
606 // abort - mix of forward and backward
608 .println("Error: forward and backward strand for " + accId);
615 * add to CDS ranges, semi-sorted forwards/backwards
619 regions.add(0, new int[] { sf.getEnd(), sf.getBegin() });
623 regions.add(new int[] { sf.getBegin(), sf.getEnd() });
625 mappedLength += Math.abs(sf.getEnd() - sf.getBegin() + 1);
628 if (regions.isEmpty())
630 System.out.println("Failed to identify target sequence for " + accId
631 + " from genomic features");
636 * a final sort is needed since Ensembl returns CDS sorted within source
637 * (havana / ensembl_havana)
639 Collections.sort(regions, direction == 1 ? IntRangeComparator.ASCENDING
640 : IntRangeComparator.DESCENDING);
642 List<int[]> to = Arrays
644 { start, start + mappedLength - 1 });
646 return new MapList(regions, to, 1, 1);
650 * Answers a list of sequence features that mark positions of the genomic
651 * sequence feature which are within the sequence being retrieved. For
652 * example, an 'exon' feature whose parent is the target transcript marks the
653 * cdna positions of the transcript. For a gene sequence, this is trivially
654 * just the 'gene' feature with matching gene id.
660 protected abstract List<SequenceFeature> getIdentifyingFeatures(
661 SequenceI seq, String accId);
664 * Transfers the sequence feature to the target sequence, locating its start
665 * and end range based on the mapping. Features which do not overlap the
666 * target sequence are ignored.
669 * @param targetSequence
671 * mapping from the sequence feature's coordinates to the target
673 * @param forwardStrand
675 protected void transferFeature(SequenceFeature sf,
676 SequenceI targetSequence, MapList mapping, boolean forwardStrand)
678 int start = sf.getBegin();
679 int end = sf.getEnd();
680 int[] mappedRange = mapping.locateInTo(start, end);
682 if (mappedRange != null)
684 String group = sf.getFeatureGroup();
685 if (".".equals(group))
687 group = getDbSource();
689 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
690 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
691 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
692 group, sf.getScore());
693 targetSequence.addSequenceFeature(copy);
696 * for sequence_variant on reverse strand, have to convert the allele
697 * values to their complements
699 if (!forwardStrand && SequenceOntologyFactory.getInstance()
700 .isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
702 reverseComplementAlleles(copy);
708 * Change the 'alleles' value of a feature by converting to complementary
709 * bases, and also update the feature description to match
713 static void reverseComplementAlleles(SequenceFeature sf)
715 final String alleles = (String) sf.getValue(Gff3Helper.ALLELES);
720 StringBuilder complement = new StringBuilder(alleles.length());
721 for (String allele : alleles.split(","))
723 reverseComplementAllele(complement, allele);
725 String comp = complement.toString();
726 sf.setValue(Gff3Helper.ALLELES, comp);
727 sf.setDescription(comp);
730 * replace value of "alleles=" in sf.ATTRIBUTES as well
731 * so 'output as GFF' shows reverse complement alleles
733 String atts = sf.getAttributes();
736 atts = atts.replace(Gff3Helper.ALLELES + "=" + alleles,
737 Gff3Helper.ALLELES + "=" + comp);
738 sf.setAttributes(atts);
743 * Makes the 'reverse complement' of the given allele and appends it to the
744 * buffer, after a comma separator if not the first
749 static void reverseComplementAllele(StringBuilder complement,
752 if (complement.length() > 0)
754 complement.append(",");
758 * some 'alleles' are actually descriptive terms
759 * e.g. HGMD_MUTATION, PhenCode_variation
760 * - we don't want to 'reverse complement' these
762 if (!Comparison.isNucleotideSequence(allele, true))
764 complement.append(allele);
768 for (int i = allele.length() - 1; i >= 0; i--)
770 complement.append(Dna.getComplement(allele.charAt(i)));
776 * Transfers features from sourceSequence to targetSequence
779 * @param sourceSequence
780 * @param targetSequence
781 * @return true if any features were transferred, else false
783 protected boolean transferFeatures(String accessionId,
784 SequenceI sourceSequence, SequenceI targetSequence)
786 if (sourceSequence == null || targetSequence == null)
791 // long start = System.currentTimeMillis();
792 List<SequenceFeature> sfs = sourceSequence.getFeatures()
793 .getPositionalFeatures();
794 MapList mapping = getGenomicRangesFromFeatures(sourceSequence,
795 accessionId, targetSequence.getStart());
801 boolean result = transferFeatures(sfs, targetSequence, mapping,
803 // System.out.println("transferFeatures (" + (sfs.size()) + " --> "
804 // + targetSequence.getFeatures().getFeatureCount(true) + ") to "
805 // + targetSequence.getName() + " took "
806 // + (System.currentTimeMillis() - start) + "ms");
811 * Transfer features to the target sequence. The start/end positions are
812 * converted using the mapping. Features which do not overlap are ignored.
813 * Features whose parent is not the specified identifier are also ignored.
816 * @param targetSequence
821 protected boolean transferFeatures(List<SequenceFeature> sfs,
822 SequenceI targetSequence, MapList mapping, String parentId)
824 final boolean forwardStrand = mapping.isFromForwardStrand();
827 * sort features by start position (which corresponds to end
828 * position descending if reverse strand) so as to add them in
829 * 'forwards' order to the target sequence
831 SequenceFeatures.sortFeatures(sfs, forwardStrand);
833 boolean transferred = false;
834 for (SequenceFeature sf : sfs)
836 if (retainFeature(sf, parentId))
838 transferFeature(sf, targetSequence, mapping, forwardStrand);
846 * Answers true if the feature type is one we want to keep for the sequence.
847 * Some features are only retrieved in order to identify the sequence range,
848 * and may then be discarded as redundant information (e.g. "CDS" feature for
851 @SuppressWarnings("unused")
852 protected boolean retainFeature(SequenceFeature sf, String accessionId)
854 return true; // override as required
858 * Answers true if the feature has a Parent which refers to the given
859 * accession id, or if the feature has no parent. Answers false if the
860 * feature's Parent is for a different accession id.
866 protected boolean featureMayBelong(SequenceFeature sf, String identifier)
868 String parent = (String) sf.getValue(PARENT);
870 && !parent.equalsIgnoreCase(identifier))
872 // this genomic feature belongs to a different transcript
879 * Answers a short description of the sequence fetcher
882 public String getDescription()
884 return "Ensembl " + getSourceEnsemblType().getType()
885 + " sequence with variant features";
889 * Returns a (possibly empty) list of features on the sequence which have the
890 * specified sequence ontology term (or a sub-type of it), and the given
891 * identifier as parent
898 protected List<SequenceFeature> findFeatures(SequenceI sequence,
899 String term, String parentId)
901 List<SequenceFeature> result = new ArrayList<>();
903 List<SequenceFeature> sfs = sequence.getFeatures()
904 .getFeaturesByOntology(term);
905 for (SequenceFeature sf : sfs)
907 String parent = (String) sf.getValue(PARENT);
908 if (parent != null && parent.equalsIgnoreCase(parentId))
918 * Answers true if the feature type is either 'NMD_transcript_variant' or
919 * 'transcript' (or one of its sub-types in the Sequence Ontology). This is
920 * because NMD_transcript_variant behaves like 'transcript' in Ensembl
921 * although strictly speaking it is not (it is a sub-type of
924 * (This test was needed when fetching transcript features as GFF. As we are
925 * now fetching as JSON, all features have type 'transcript' so the check for
926 * NMD_transcript_variant is redundant. Left in for any future case arising.)
931 public static boolean isTranscript(String featureType)
933 return SequenceOntologyI.NMD_TRANSCRIPT_VARIANT.equals(featureType)
934 || SequenceOntologyFactory.getInstance().isA(featureType,
935 SequenceOntologyI.TRANSCRIPT);