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;
44 import java.io.BufferedReader;
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;
53 import org.json.simple.JSONObject;
54 import org.json.simple.parser.JSONParser;
55 import org.json.simple.parser.ParseException;
58 * Base class for Ensembl sequence fetchers
60 * @see http://rest.ensembl.org/documentation/info/sequence_id
63 public abstract class EnsemblSeqProxy extends EnsemblRestClient
65 protected static final String DESCRIPTION = "description";
68 * enum for 'type' parameter to the /sequence REST service
70 public enum EnsemblSeqType
73 * type=genomic to fetch full dna including introns
78 * type=cdna to fetch coding dna including UTRs
83 * type=cds to fetch coding dna excluding UTRs
88 * type=protein to fetch peptide product sequence
93 * the value of the 'type' parameter to fetch this version of
98 EnsemblSeqType(String t)
103 public String getType()
111 * Default constructor (to use rest.ensembl.org)
113 public EnsemblSeqProxy()
119 * Makes the sequence queries to Ensembl's REST service and returns an
120 * alignment consisting of the returned sequences.
123 public AlignmentI getSequenceRecords(String query) throws Exception
125 // TODO use a String... query vararg instead?
127 // danger: accession separator used as a regex here, a string elsewhere
128 // in this case it is ok (it is just a space), but (e.g.) '\' would not be
129 List<String> allIds = Arrays
130 .asList(query.split(getAccessionSeparator()));
131 AlignmentI alignment = null;
135 * execute queries, if necessary in batches of the
136 * maximum allowed number of ids
138 int maxQueryCount = getMaximumQueryCount();
139 for (int v = 0, vSize = allIds.size(); v < vSize; v += maxQueryCount)
141 int p = Math.min(vSize, v + maxQueryCount);
142 List<String> ids = allIds.subList(v, p);
145 alignment = fetchSequences(ids, alignment);
146 } catch (Throwable r)
149 String msg = "Aborting ID retrieval after " + v
150 + " chunks. Unexpected problem (" + r.getLocalizedMessage()
152 System.err.println(msg);
158 if (alignment == null)
164 * fetch and transfer genomic sequence features,
165 * fetch protein product and add as cross-reference
167 for (String accId : allIds)
169 addFeaturesAndProduct(accId, alignment);
172 for (SequenceI seq : alignment.getSequences())
174 getCrossReferences(seq);
181 * Fetches Ensembl features using the /overlap REST endpoint, and adds them to
182 * the sequence in the alignment. Also fetches the protein product, maps it
183 * from the CDS features of the sequence, and saves it as a cross-reference of
189 protected void addFeaturesAndProduct(String accId, AlignmentI alignment)
191 if (alignment == null)
199 * get 'dummy' genomic sequence with gene, transcript,
200 * exon, cds and variation features
202 SequenceI genomicSequence = null;
203 EnsemblFeatures gffFetcher = new EnsemblFeatures();
204 EnsemblFeatureType[] features = getFeaturesToFetch();
205 AlignmentI geneFeatures = gffFetcher.getSequenceRecords(accId,
207 if (geneFeatures != null && geneFeatures.getHeight() > 0)
209 genomicSequence = geneFeatures.getSequenceAt(0);
211 if (genomicSequence != null)
214 * transfer features to the query sequence
216 SequenceI querySeq = alignment.findName(accId, true);
217 if (transferFeatures(accId, genomicSequence, querySeq))
221 * fetch and map protein product, and add it as a cross-reference
222 * of the retrieved sequence
224 addProteinProduct(querySeq);
227 } catch (IOException e)
230 "Error transferring Ensembl features: " + e.getMessage());
235 * Returns those sequence feature types to fetch from Ensembl. We may want
236 * features either because they are of interest to the user, or as means to
237 * identify the locations of the sequence on the genomic sequence (CDS
238 * features identify CDS, exon features identify cDNA etc).
242 protected abstract EnsemblFeatureType[] getFeaturesToFetch();
245 * Fetches and maps the protein product, and adds it as a cross-reference of
246 * the retrieved sequence
248 protected void addProteinProduct(SequenceI querySeq)
250 String accId = querySeq.getName();
253 AlignmentI protein = new EnsemblProtein()
254 .getSequenceRecords(accId);
255 if (protein == null || protein.getHeight() == 0)
257 System.out.println("No protein product found for " + accId);
260 SequenceI proteinSeq = protein.getSequenceAt(0);
263 * need dataset sequences (to be the subject of mappings)
265 proteinSeq.createDatasetSequence();
266 querySeq.createDatasetSequence();
268 MapList mapList = AlignmentUtils.mapCdsToProtein(querySeq,
272 // clunky: ensure Uniprot xref if we have one is on mapped sequence
273 SequenceI ds = proteinSeq.getDatasetSequence();
274 // TODO: Verify ensp primary ref is on proteinSeq.getDatasetSequence()
275 Mapping map = new Mapping(ds, mapList);
276 DBRefEntry dbr = new DBRefEntry(getDbSource(),
277 getEnsemblDataVersion(), proteinSeq.getName(), map);
278 querySeq.getDatasetSequence().addDBRef(dbr);
279 DBRefEntry[] uprots = DBRefUtils.selectRefs(ds.getDBRefs(),
281 { DBRefSource.UNIPROT });
282 DBRefEntry[] upxrefs = DBRefUtils.selectRefs(querySeq.getDBRefs(),
284 { DBRefSource.UNIPROT });
287 for (DBRefEntry up : uprots)
289 // locate local uniprot ref and map
290 List<DBRefEntry> upx = DBRefUtils.searchRefs(upxrefs,
291 up.getAccessionId());
300 "Implementation issue - multiple uniprot acc on product sequence.");
305 upxref = new DBRefEntry(DBRefSource.UNIPROT,
306 getEnsemblDataVersion(), up.getAccessionId());
309 Mapping newMap = new Mapping(ds, mapList);
310 upxref.setVersion(getEnsemblDataVersion());
311 upxref.setMap(newMap);
314 // add the new uniprot ref
315 querySeq.getDatasetSequence().addDBRef(upxref);
322 * copy exon features to protein, compute peptide variants from dna
323 * variants and add as features on the protein sequence ta-da
325 AlignmentUtils.computeProteinFeatures(querySeq, proteinSeq,
328 } catch (Exception e)
331 .println(String.format("Error retrieving protein for %s: %s",
332 accId, e.getMessage()));
337 * Get database xrefs from Ensembl, and attach them to the sequence
341 protected void getCrossReferences(SequenceI seq)
343 while (seq.getDatasetSequence() != null)
345 seq = seq.getDatasetSequence();
348 EnsemblXref xrefFetcher = new EnsemblXref(getDbSource(),
349 getEnsemblDataVersion());
350 List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName());
351 for (DBRefEntry xref : xrefs)
357 * and add a reference to itself
359 DBRefEntry self = new DBRefEntry(getDbSource(), getEnsemblDataVersion(),
365 * Fetches sequences for the list of accession ids and adds them to the
366 * alignment. Returns the extended (or created) alignment.
371 * @throws JalviewException
372 * @throws IOException
374 protected AlignmentI fetchSequences(List<String> ids,
375 AlignmentI alignment) throws JalviewException, IOException
377 if (!isEnsemblAvailable())
380 throw new JalviewException("ENSEMBL Rest API not available.");
382 BufferedReader br = getSequenceReader(ids);
388 List<SequenceI> seqs = parseSequenceJson(br);
392 throw new IOException("No data returned for " + ids);
395 if (seqs.size() != ids.size())
397 System.out.println(String.format(
398 "Only retrieved %d sequences for %d query strings",
399 seqs.size(), ids.size()));
404 AlignmentI seqal = new Alignment(
405 seqs.toArray(new SequenceI[seqs.size()]));
406 for (SequenceI seq : seqs)
408 if (seq.getDescription() == null)
410 seq.setDescription(getDbName());
412 String name = seq.getName();
413 if (ids.contains(name)
414 || ids.contains(name.replace("ENSP", "ENST")))
416 // TODO JAL-3077 use true accession version in dbref
417 DBRefEntry dbref = DBRefUtils.parseToDbRef(seq, getDbSource(),
418 getEnsemblDataVersion(), name);
422 if (alignment == null)
428 alignment.append(seqal);
435 * Parses a JSON response for a single sequence ID query
438 * @return a single jalview.datamodel.Sequence
439 * @see http://rest.ensembl.org/documentation/info/sequence_id
441 protected List<SequenceI> parseSequenceJson(BufferedReader br)
443 JSONParser jp = new JSONParser();
444 List<SequenceI> result = new ArrayList<>();
448 * for now, assumes only one sequence returned; refactor if needed
449 * in future to handle a JSONArray with more than one
451 final JSONObject val = (JSONObject) jp.parse(br);
452 Object s = val.get("desc");
453 String desc = s == null ? null : s.toString();
455 String id = s == null ? null : s.toString();
457 String seq = s == null ? null : s.toString();
458 Sequence sequence = new Sequence(id, seq);
461 sequence.setDescription(desc);
463 // todo JAL-3077 make a DBRefEntry with true accession version
464 // s = val.get("version");
465 // String version = s == null ? "0" : s.toString();
466 // DBRefEntry dbref = new DBRefEntry(getDbSource(), version, id);
467 // sequence.addDBRef(dbref);
468 result.add(sequence);
469 } catch (ParseException | IOException e)
471 System.err.println("Error processing JSON response: " + e.toString());
478 * Returns the URL for the REST call
481 * @throws MalformedURLException
484 protected URL getUrl(List<String> ids) throws MalformedURLException
487 * a single id is included in the URL path
488 * multiple ids go in the POST body instead
490 StringBuffer urlstring = new StringBuffer(128);
491 urlstring.append(getDomain() + "/sequence/id");
494 urlstring.append("/").append(ids.get(0));
496 // @see https://github.com/Ensembl/ensembl-rest/wiki/Output-formats
497 urlstring.append("?type=").append(getSourceEnsemblType().getType());
498 urlstring.append(("&Accept=application/json"));
499 urlstring.append(("&Content-Type=application/json"));
501 String objectType = getObjectType();
502 if (objectType != null)
504 urlstring.append("&").append(OBJECT_TYPE).append("=")
508 URL url = new URL(urlstring.toString());
513 * Override this method to specify object_type request parameter
517 protected String getObjectType()
523 * A sequence/id POST request currently allows up to 50 queries
525 * @see http://rest.ensembl.org/documentation/info/sequence_id_post
528 public int getMaximumQueryCount()
534 protected boolean useGetRequest()
541 * @return the configured sequence return type for this source
543 protected abstract EnsemblSeqType getSourceEnsemblType();
546 * Returns a list of [start, end] genomic ranges corresponding to the sequence
549 * The correspondence between the frames of reference is made by locating
550 * those features on the genomic sequence which identify the retrieved
551 * sequence. Specifically
553 * <li>genomic sequence is identified by "transcript" features with
554 * ID=transcript:transcriptId</li>
555 * <li>cdna sequence is identified by "exon" features with
556 * Parent=transcript:transcriptId</li>
557 * <li>cds sequence is identified by "CDS" features with
558 * Parent=transcript:transcriptId</li>
561 * The returned ranges are sorted to run forwards (for positive strand) or
562 * backwards (for negative strand). Aborts and returns null if both positive
563 * and negative strand are found (this should not normally happen).
565 * @param sourceSequence
568 * the start position of the sequence we are mapping to
571 protected MapList getGenomicRangesFromFeatures(SequenceI sourceSequence,
572 String accId, int start)
574 List<SequenceFeature> sfs = getIdentifyingFeatures(sourceSequence,
582 * generously initial size for number of cds regions
583 * (worst case titin Q8WZ42 has c. 313 exons)
585 List<int[]> regions = new ArrayList<>(100);
586 int mappedLength = 0;
587 int direction = 1; // forward
588 boolean directionSet = false;
590 for (SequenceFeature sf : sfs)
592 int strand = sf.getStrand();
593 strand = strand == 0 ? 1 : strand; // treat unknown as forward
595 if (directionSet && strand != direction)
597 // abort - mix of forward and backward
599 .println("Error: forward and backward strand for " + accId);
606 * add to CDS ranges, semi-sorted forwards/backwards
610 regions.add(0, new int[] { sf.getEnd(), sf.getBegin() });
614 regions.add(new int[] { sf.getBegin(), sf.getEnd() });
616 mappedLength += Math.abs(sf.getEnd() - sf.getBegin() + 1);
619 if (regions.isEmpty())
621 System.out.println("Failed to identify target sequence for " + accId
622 + " from genomic features");
627 * a final sort is needed since Ensembl returns CDS sorted within source
628 * (havana / ensembl_havana)
630 Collections.sort(regions, direction == 1 ? IntRangeComparator.ASCENDING
631 : IntRangeComparator.DESCENDING);
633 List<int[]> to = Arrays
635 { start, start + mappedLength - 1 });
637 return new MapList(regions, to, 1, 1);
641 * Answers a list of sequence features that mark positions of the genomic
642 * sequence feature which are within the sequence being retrieved. For
643 * example, an 'exon' feature whose parent is the target transcript marks the
644 * cdna positions of the transcript. For a gene sequence, this is trivially
645 * just the 'gene' feature with matching gene id.
651 protected abstract List<SequenceFeature> getIdentifyingFeatures(
652 SequenceI seq, String accId);
655 * Transfers the sequence feature to the target sequence, locating its start
656 * and end range based on the mapping. Features which do not overlap the
657 * target sequence are ignored.
660 * @param targetSequence
662 * mapping from the sequence feature's coordinates to the target
664 * @param forwardStrand
666 protected void transferFeature(SequenceFeature sf,
667 SequenceI targetSequence, MapList mapping, boolean forwardStrand)
669 int start = sf.getBegin();
670 int end = sf.getEnd();
671 int[] mappedRange = mapping.locateInTo(start, end);
673 if (mappedRange != null)
675 String group = sf.getFeatureGroup();
676 if (".".equals(group))
678 group = getDbSource();
680 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
681 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
682 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
683 group, sf.getScore());
684 targetSequence.addSequenceFeature(copy);
687 * for sequence_variant on reverse strand, have to convert the allele
688 * values to their complements
690 if (!forwardStrand && SequenceOntologyFactory.getInstance()
691 .isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
693 reverseComplementAlleles(copy);
699 * Change the 'alleles' value of a feature by converting to complementary
700 * bases, and also update the feature description to match
704 static void reverseComplementAlleles(SequenceFeature sf)
706 final String alleles = (String) sf.getValue(Gff3Helper.ALLELES);
711 StringBuilder complement = new StringBuilder(alleles.length());
712 for (String allele : alleles.split(","))
714 reverseComplementAllele(complement, allele);
716 String comp = complement.toString();
717 sf.setValue(Gff3Helper.ALLELES, comp);
718 sf.setDescription(comp);
721 * replace value of "alleles=" in sf.ATTRIBUTES as well
722 * so 'output as GFF' shows reverse complement alleles
724 String atts = sf.getAttributes();
727 atts = atts.replace(Gff3Helper.ALLELES + "=" + alleles,
728 Gff3Helper.ALLELES + "=" + comp);
729 sf.setAttributes(atts);
734 * Makes the 'reverse complement' of the given allele and appends it to the
735 * buffer, after a comma separator if not the first
740 static void reverseComplementAllele(StringBuilder complement,
743 if (complement.length() > 0)
745 complement.append(",");
749 * some 'alleles' are actually descriptive terms
750 * e.g. HGMD_MUTATION, PhenCode_variation
751 * - we don't want to 'reverse complement' these
753 if (!Comparison.isNucleotideSequence(allele, true))
755 complement.append(allele);
759 for (int i = allele.length() - 1; i >= 0; i--)
761 complement.append(Dna.getComplement(allele.charAt(i)));
767 * Transfers features from sourceSequence to targetSequence
770 * @param sourceSequence
771 * @param targetSequence
772 * @return true if any features were transferred, else false
774 protected boolean transferFeatures(String accessionId,
775 SequenceI sourceSequence, SequenceI targetSequence)
777 if (sourceSequence == null || targetSequence == null)
782 // long start = System.currentTimeMillis();
783 List<SequenceFeature> sfs = sourceSequence.getFeatures()
784 .getPositionalFeatures();
785 MapList mapping = getGenomicRangesFromFeatures(sourceSequence,
786 accessionId, targetSequence.getStart());
792 boolean result = transferFeatures(sfs, targetSequence, mapping,
794 // System.out.println("transferFeatures (" + (sfs.size()) + " --> "
795 // + targetSequence.getFeatures().getFeatureCount(true) + ") to "
796 // + targetSequence.getName() + " took "
797 // + (System.currentTimeMillis() - start) + "ms");
802 * Transfer features to the target sequence. The start/end positions are
803 * converted using the mapping. Features which do not overlap are ignored.
804 * Features whose parent is not the specified identifier are also ignored.
807 * @param targetSequence
812 protected boolean transferFeatures(List<SequenceFeature> sfs,
813 SequenceI targetSequence, MapList mapping, String parentId)
815 final boolean forwardStrand = mapping.isFromForwardStrand();
818 * sort features by start position (which corresponds to end
819 * position descending if reverse strand) so as to add them in
820 * 'forwards' order to the target sequence
822 SequenceFeatures.sortFeatures(sfs, forwardStrand);
824 boolean transferred = false;
825 for (SequenceFeature sf : sfs)
827 if (retainFeature(sf, parentId))
829 transferFeature(sf, targetSequence, mapping, forwardStrand);
837 * Answers true if the feature type is one we want to keep for the sequence.
838 * Some features are only retrieved in order to identify the sequence range,
839 * and may then be discarded as redundant information (e.g. "CDS" feature for
842 @SuppressWarnings("unused")
843 protected boolean retainFeature(SequenceFeature sf, String accessionId)
845 return true; // override as required
849 * Answers true if the feature has a Parent which refers to the given
850 * accession id, or if the feature has no parent. Answers false if the
851 * feature's Parent is for a different accession id.
857 protected boolean featureMayBelong(SequenceFeature sf, String identifier)
859 String parent = (String) sf.getValue(PARENT);
861 && !parent.equalsIgnoreCase(identifier))
863 // this genomic feature belongs to a different transcript
870 * Answers a short description of the sequence fetcher
873 public String getDescription()
875 return "Ensembl " + getSourceEnsemblType().getType()
876 + " sequence with variant features";
880 * Returns a (possibly empty) list of features on the sequence which have the
881 * specified sequence ontology term (or a sub-type of it), and the given
882 * identifier as parent
889 protected List<SequenceFeature> findFeatures(SequenceI sequence,
890 String term, String parentId)
892 List<SequenceFeature> result = new ArrayList<>();
894 List<SequenceFeature> sfs = sequence.getFeatures()
895 .getFeaturesByOntology(term);
896 for (SequenceFeature sf : sfs)
898 String parent = (String) sf.getValue(PARENT);
899 if (parent != null && parent.equalsIgnoreCase(parentId))
909 * Answers true if the feature type is either 'NMD_transcript_variant' or
910 * 'transcript' (or one of its sub-types in the Sequence Ontology). This is
911 * because NMD_transcript_variant behaves like 'transcript' in Ensembl
912 * although strictly speaking it is not (it is a sub-type of
915 * (This test was needed when fetching transcript features as GFF. As we are
916 * now fetching as JSON, all features have type 'transcript' so the check for
917 * NMD_transcript_variant is redundant. Left in for any future case arising.)
922 public static boolean isTranscript(String featureType)
924 return SequenceOntologyI.NMD_TRANSCRIPT_VARIANT.equals(featureType)
925 || SequenceOntologyFactory.getInstance().isA(featureType,
926 SequenceOntologyI.TRANSCRIPT);