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 java.io.IOException;
24 import java.net.MalformedURLException;
26 import java.util.ArrayList;
27 import java.util.Arrays;
28 import java.util.Collections;
29 import java.util.List;
32 import org.json.simple.parser.ParseException;
34 import jalview.analysis.AlignmentUtils;
35 import jalview.analysis.Dna;
36 import jalview.bin.Console;
37 import jalview.datamodel.Alignment;
38 import jalview.datamodel.AlignmentI;
39 import jalview.datamodel.DBRefEntry;
40 import jalview.datamodel.DBRefSource;
41 import jalview.datamodel.Mapping;
42 import jalview.datamodel.Sequence;
43 import jalview.datamodel.SequenceFeature;
44 import jalview.datamodel.SequenceI;
45 import jalview.datamodel.features.SequenceFeatures;
46 import jalview.exceptions.JalviewException;
47 import jalview.io.gff.Gff3Helper;
48 import jalview.io.gff.SequenceOntologyFactory;
49 import jalview.io.gff.SequenceOntologyI;
50 import jalview.util.Comparison;
51 import jalview.util.DBRefUtils;
52 import jalview.util.IntRangeComparator;
53 import jalview.util.MapList;
56 * Base class for Ensembl sequence fetchers
58 * @see http://rest.ensembl.org/documentation/info/sequence_id
61 public abstract class EnsemblSeqProxy extends EnsemblRestClient
63 protected static final String DESCRIPTION = "description";
66 * enum for 'type' parameter to the /sequence REST service
68 public enum EnsemblSeqType
71 * type=genomic to fetch full dna including introns
76 * type=cdna to fetch coding dna including UTRs
81 * type=cds to fetch coding dna excluding UTRs
86 * type=protein to fetch peptide product sequence
91 * the value of the 'type' parameter to fetch this version of
96 EnsemblSeqType(String t)
101 public String getType()
109 * Default constructor (to use rest.ensembl.org)
111 public EnsemblSeqProxy()
117 * Constructor given the target domain to fetch data from
119 public EnsemblSeqProxy(String d)
125 * Makes the sequence queries to Ensembl's REST service and returns an
126 * alignment consisting of the returned sequences.
129 public AlignmentI getSequenceRecords(String query) throws Exception
131 // TODO use a String... query vararg instead?
133 // danger: accession separator used as a regex here, a string elsewhere
134 // in this case it is ok (it is just a space), but (e.g.) '\' would not be
135 List<String> allIds = Arrays
136 .asList(query.split(getAccessionSeparator()));
137 AlignmentI alignment = null;
141 * execute queries, if necessary in batches of the
142 * maximum allowed number of ids
144 int maxQueryCount = getMaximumQueryCount();
145 for (int v = 0, vSize = allIds.size(); v < vSize; v += maxQueryCount)
147 int p = Math.min(vSize, v + maxQueryCount);
148 List<String> ids = allIds.subList(v, p);
151 alignment = fetchSequences(ids, alignment);
152 } catch (Throwable r)
155 String msg = "Aborting ID retrieval after " + v
156 + " chunks. Unexpected problem (" + r.getLocalizedMessage()
158 jalview.bin.Console.errPrintln(msg);
164 if (alignment == null)
170 * fetch and transfer genomic sequence features,
171 * fetch protein product and add as cross-reference
173 for (int i = 0, n = allIds.size(); i < n; i++)
175 addFeaturesAndProduct(allIds.get(i), alignment);
178 List<SequenceI> seqs = alignment.getSequences();
179 for (int i = 0, n = seqs.size(); i < n; i++)
181 getCrossReferences(seqs.get(i));
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();
213 // Platform.timeCheck("ESP.getsequencerec1", Platform.TIME_MARK);
215 AlignmentI geneFeatures = gffFetcher.getSequenceRecords(accId,
217 if (geneFeatures != null && geneFeatures.getHeight() > 0)
219 genomicSequence = geneFeatures.getSequenceAt(0);
222 // Platform.timeCheck("ESP.getsequencerec2", Platform.TIME_MARK);
224 if (genomicSequence != null)
227 * transfer features to the query sequence
229 SequenceI querySeq = alignment.findName(accId, true);
230 if (transferFeatures(accId, genomicSequence, querySeq))
234 * fetch and map protein product, and add it as a cross-reference
235 * of the retrieved sequence
237 // Platform.timeCheck("ESP.transferFeatures", Platform.TIME_MARK);
238 addProteinProduct(querySeq);
241 } catch (IOException e)
243 jalview.bin.Console.errPrintln(
244 "Error transferring Ensembl features: " + e.getMessage());
246 // Platform.timeCheck("ESP.addfeat done", Platform.TIME_MARK);
250 * Returns those sequence feature types to fetch from Ensembl. We may want
251 * features either because they are of interest to the user, or as means to
252 * identify the locations of the sequence on the genomic sequence (CDS
253 * features identify CDS, exon features identify cDNA etc).
257 protected abstract EnsemblFeatureType[] getFeaturesToFetch();
260 * Fetches and maps the protein product, and adds it as a cross-reference of
261 * the retrieved sequence
263 protected void addProteinProduct(SequenceI querySeq)
265 String accId = querySeq.getName();
268 jalview.bin.Console.outPrintln("Adding protein product for " + accId);
269 AlignmentI protein = new EnsemblProtein(getDomain())
270 .getSequenceRecords(accId);
271 if (protein == null || protein.getHeight() == 0)
274 .outPrintln("No protein product found for " + accId);
277 SequenceI proteinSeq = protein.getSequenceAt(0);
280 * need dataset sequences (to be the subject of mappings)
282 proteinSeq.createDatasetSequence();
283 querySeq.createDatasetSequence();
285 MapList mapList = AlignmentUtils.mapCdsToProtein(querySeq,
289 // clunky: ensure Uniprot xref if we have one is on mapped sequence
290 SequenceI ds = proteinSeq.getDatasetSequence();
291 // TODO: Verify ensp primary ref is on proteinSeq.getDatasetSequence()
292 Mapping map = new Mapping(ds, mapList);
293 DBRefEntry dbr = new DBRefEntry(getDbSource(),
294 getEnsemblDataVersion(), proteinSeq.getName(), map);
295 querySeq.getDatasetSequence().addDBRef(dbr);
296 List<DBRefEntry> uprots = DBRefUtils.selectRefs(ds.getDBRefs(),
298 { DBRefSource.UNIPROT });
299 List<DBRefEntry> upxrefs = DBRefUtils
300 .selectRefs(querySeq.getDBRefs(), new String[]
301 { DBRefSource.UNIPROT });
304 for (DBRefEntry up : uprots)
306 // locate local uniprot ref and map
307 List<DBRefEntry> upx = DBRefUtils.searchRefs(upxrefs,
308 up.getAccessionId());
317 "Implementation issue - multiple uniprot acc on product sequence.");
322 upxref = new DBRefEntry(DBRefSource.UNIPROT,
323 getEnsemblDataVersion(), up.getAccessionId());
326 Mapping newMap = new Mapping(ds, mapList);
327 upxref.setVersion(getEnsemblDataVersion());
328 upxref.setMap(newMap);
331 // add the new uniprot ref
332 querySeq.getDatasetSequence().addDBRef(upxref);
339 * copy exon features to protein, compute peptide variants from dna
340 * variants and add as features on the protein sequence ta-da
342 // JAL-3187 render on the fly instead
343 // AlignmentUtils.computeProteinFeatures(querySeq, proteinSeq, mapList);
345 } catch (Exception e)
348 .println(String.format("Error retrieving protein for %s: %s",
349 accId, e.getMessage()));
354 * Get database xrefs from Ensembl, and attach them to the sequence
358 protected void getCrossReferences(SequenceI seq)
361 // Platform.timeCheck("ESP. getdataseq ", Platform.TIME_MARK);
363 while (seq.getDatasetSequence() != null)
365 seq = seq.getDatasetSequence();
368 // Platform.timeCheck("ESP. getxref ", Platform.TIME_MARK);
370 EnsemblXref xrefFetcher = new EnsemblXref(getDomain(), getDbSource(),
371 getEnsemblDataVersion());
372 List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName());
374 for (int i = 0, n = xrefs.size(); i < n; i++)
376 // Platform.timeCheck("ESP. getxref + " + (i) + "/" + n,
377 // Platform.TIME_MARK);
378 // BH 2019.01.25 this next method was taking 174 ms PER addition for a
379 // 266-reference example.
380 // DBRefUtils.ensurePrimaries(seq)
381 // was at the end of seq.addDBRef, so executed after ever addition!
382 // This method was moved to seq.getPrimaryDBRefs()
383 seq.addDBRef(xrefs.get(i));
386 // jalview.bin.Console.outPrintln("primaries are " +
387 // seq.getPrimaryDBRefs().toString());
389 * and add a reference to itself
392 // Platform.timeCheck("ESP. getxref self ", Platform.TIME_MARK);
394 DBRefEntry self = new DBRefEntry(getDbSource(), getEnsemblDataVersion(),
397 // Platform.timeCheck("ESP. getxref self add ", Platform.TIME_MARK);
401 // Platform.timeCheck("ESP. seqprox done ", Platform.TIME_MARK);
405 * Fetches sequences for the list of accession ids and adds them to the
406 * alignment. Returns the extended (or created) alignment.
411 * @throws JalviewException
412 * @throws IOException
414 protected AlignmentI fetchSequences(List<String> ids,
415 AlignmentI alignment) throws JalviewException, IOException
417 if (!isEnsemblAvailable())
420 throw new JalviewException("ENSEMBL Rest API not available.");
422 // Platform.timeCheck("EnsemblSeqProx.fetchSeq ", Platform.TIME_MARK);
424 List<SequenceI> seqs = parseSequenceJson(ids);
430 throw new IOException("No data returned for " + ids);
433 if (seqs.size() != ids.size())
435 jalview.bin.Console.outPrintln(String.format(
436 "Only retrieved %d sequences for %d query strings",
437 seqs.size(), ids.size()));
442 AlignmentI seqal = new Alignment(
443 seqs.toArray(new SequenceI[seqs.size()]));
444 for (SequenceI seq : seqs)
446 if (seq.getDescription() == null)
448 seq.setDescription(getDbName());
450 String name = seq.getName();
451 if (ids.contains(name)
452 || ids.contains(name.replace("ENSP", "ENST")))
454 // TODO JAL-3077 use true accession version in dbref
455 DBRefEntry dbref = DBRefUtils.parseToDbRef(seq, getDbSource(),
456 getEnsemblDataVersion(), name);
460 if (alignment == null)
466 alignment.append(seqal);
473 * Parses a JSON response for a single sequence ID query
476 * @return a single jalview.datamodel.Sequence
477 * @see http://rest.ensembl.org/documentation/info/sequence_id
479 @SuppressWarnings("unchecked")
480 protected List<SequenceI> parseSequenceJson(List<String> ids)
482 List<SequenceI> result = new ArrayList<>();
486 * for now, assumes only one sequence returned; refactor if needed
487 * in future to handle a JSONArray with more than one
489 // Platform.timeCheck("ENS seqproxy", Platform.TIME_MARK);
490 Map<String, Object> val = (Map<String, Object>) getJSON(null, ids, -1,
494 Object s = val.get("desc");
495 String desc = s == null ? null : s.toString();
497 String id = s == null ? null : s.toString();
499 String seq = s == null ? null : s.toString();
500 Sequence sequence = new Sequence(id, seq);
503 sequence.setDescription(desc);
505 // todo JAL-3077 make a DBRefEntry with true accession version
506 // s = val.get("version");
507 // String version = s == null ? "0" : s.toString();
508 // DBRefEntry dbref = new DBRefEntry(getDbSource(), version, id);
509 // sequence.addDBRef(dbref);
510 result.add(sequence);
511 } catch (ParseException | IOException e)
513 jalview.bin.Console.errPrintln(
514 "Error processing JSON response: " + e.toString());
517 // Platform.timeCheck("ENS seqproxy2", Platform.TIME_MARK);
522 * Returns the URL for the REST call
525 * @throws MalformedURLException
528 protected URL getUrl(List<String> ids) throws MalformedURLException
531 * a single id is included in the URL path
532 * multiple ids go in the POST body instead
534 StringBuffer urlstring = new StringBuffer(128);
535 urlstring.append(getDomain() + "/sequence/id");
538 urlstring.append("/").append(ids.get(0));
540 // @see https://github.com/Ensembl/ensembl-rest/wiki/Output-formats
541 urlstring.append("?type=").append(getSourceEnsemblType().getType());
542 urlstring.append(("&Accept=application/json"));
543 urlstring.append(("&content-type=application/json"));
545 String objectType = getObjectType();
546 if (objectType != null)
548 urlstring.append("&").append(OBJECT_TYPE).append("=")
552 URL url = new URL(urlstring.toString());
557 * Override this method to specify object_type request parameter
561 protected String getObjectType()
567 * A sequence/id POST request currently allows up to 50 queries
569 * @see http://rest.ensembl.org/documentation/info/sequence_id_post
572 public int getMaximumQueryCount()
578 protected boolean useGetRequest()
585 * @return the configured sequence return type for this source
587 protected abstract EnsemblSeqType getSourceEnsemblType();
590 * Returns a list of [start, end] genomic ranges corresponding to the sequence
593 * The correspondence between the frames of reference is made by locating
594 * those features on the genomic sequence which identify the retrieved
595 * sequence. Specifically
597 * <li>genomic sequence is identified by "transcript" features with
598 * ID=transcript:transcriptId</li>
599 * <li>cdna sequence is identified by "exon" features with
600 * Parent=transcript:transcriptId</li>
601 * <li>cds sequence is identified by "CDS" features with
602 * Parent=transcript:transcriptId</li>
605 * The returned ranges are sorted to run forwards (for positive strand) or
606 * backwards (for negative strand). Aborts and returns null if both positive
607 * and negative strand are found (this should not normally happen).
609 * @param sourceSequence
612 * the start position of the sequence we are mapping to
615 protected MapList getGenomicRangesFromFeatures(SequenceI sourceSequence,
616 String accId, int start)
618 List<SequenceFeature> sfs = getIdentifyingFeatures(sourceSequence,
626 * generously initial size for number of cds regions
627 * (worst case titin Q8WZ42 has c. 313 exons)
629 List<int[]> regions = new ArrayList<>(100);
630 int mappedLength = 0;
631 int direction = 1; // forward
632 boolean directionSet = false;
634 for (SequenceFeature sf : sfs)
636 int strand = sf.getStrand();
637 strand = strand == 0 ? 1 : strand; // treat unknown as forward
639 if (directionSet && strand != direction)
641 // abort - mix of forward and backward
643 .println("Error: forward and backward strand for " + accId);
650 * add to CDS ranges, semi-sorted forwards/backwards
654 regions.add(0, new int[] { sf.getEnd(), sf.getBegin() });
658 regions.add(new int[] { sf.getBegin(), sf.getEnd() });
660 mappedLength += Math.abs(sf.getEnd() - sf.getBegin() + 1);
663 if (regions.isEmpty())
666 .outPrintln("Failed to identify target sequence for " + accId
667 + " from genomic features");
672 * a final sort is needed since Ensembl returns CDS sorted within source
673 * (havana / ensembl_havana)
675 Collections.sort(regions, direction == 1 ? IntRangeComparator.ASCENDING
676 : IntRangeComparator.DESCENDING);
678 List<int[]> to = Arrays
680 { start, start + mappedLength - 1 });
682 return new MapList(regions, to, 1, 1);
686 * Answers a list of sequence features that mark positions of the genomic
687 * sequence feature which are within the sequence being retrieved. For
688 * example, an 'exon' feature whose parent is the target transcript marks the
689 * cdna positions of the transcript. For a gene sequence, this is trivially
690 * just the 'gene' feature with matching gene id.
696 protected abstract List<SequenceFeature> getIdentifyingFeatures(
697 SequenceI seq, String accId);
702 * Transfers the sequence feature to the target sequence, locating its start
703 * and end range based on the mapping. Features which do not overlap the
704 * target sequence are ignored.
707 * @param targetSequence
709 * mapping from the sequence feature's coordinates to the target
711 * @param forwardStrand
713 protected void transferFeature(SequenceFeature sf,
714 SequenceI targetSequence, MapList mapping, boolean forwardStrand)
716 int start = sf.getBegin();
717 int end = sf.getEnd();
718 int[] mappedRange = mapping.locateInTo(start, end);
720 if (mappedRange != null)
722 // Platform.timeCheck(null, Platform.TIME_SET);
723 String group = sf.getFeatureGroup();
724 if (".".equals(group))
726 group = getDbSource();
728 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
729 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
730 // Platform.timeCheck(null, Platform.TIME_MARK);
733 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
734 group, sf.getScore());
736 targetSequence.addSequenceFeature(copy);
739 * for sequence_variant on reverse strand, have to convert the allele
740 * values to their complements
742 if (!forwardStrand && SequenceOntologyFactory.getInstance()
743 .isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
745 reverseComplementAlleles(copy);
751 * Change the 'alleles' value of a feature by converting to complementary
752 * bases, and also update the feature description to match
756 static void reverseComplementAlleles(SequenceFeature sf)
758 final String alleles = (String) sf.getValue(Gff3Helper.ALLELES);
763 StringBuilder complement = new StringBuilder(alleles.length());
764 for (String allele : alleles.split(","))
766 reverseComplementAllele(complement, allele);
768 String comp = complement.toString();
769 sf.setValue(Gff3Helper.ALLELES, comp);
770 sf.setDescription(comp);
774 * Makes the 'reverse complement' of the given allele and appends it to the
775 * buffer, after a comma separator if not the first
780 static void reverseComplementAllele(StringBuilder complement,
783 if (complement.length() > 0)
785 complement.append(",");
789 * some 'alleles' are actually descriptive terms
790 * e.g. HGMD_MUTATION, PhenCode_variation
791 * - we don't want to 'reverse complement' these
793 if (!Comparison.isNucleotideSequence(allele, true))
795 complement.append(allele);
799 for (int i = allele.length() - 1; i >= 0; i--)
801 complement.append(Dna.getComplement(allele.charAt(i)));
807 * Transfers features from sourceSequence to targetSequence
810 * @param sourceSequence
811 * @param targetSequence
812 * @return true if any features were transferred, else false
814 protected boolean transferFeatures(String accessionId,
815 SequenceI sourceSequence, SequenceI targetSequence)
817 if (sourceSequence == null || targetSequence == null)
822 // long start = System.currentTimeMillis();
823 List<SequenceFeature> sfs = sourceSequence.getFeatures()
824 .getPositionalFeatures();
825 MapList mapping = getGenomicRangesFromFeatures(sourceSequence,
826 accessionId, targetSequence.getStart());
832 // Platform.timeCheck("ESP. xfer " + sfs.size(), Platform.TIME_MARK);
834 boolean result = transferFeatures(sfs, targetSequence, mapping,
836 // jalview.bin.Console.outPrintln("transferFeatures (" + (sfs.size()) + "
838 // + targetSequence.getFeatures().getFeatureCount(true) + ") to "
839 // + targetSequence.getName() + " took "
840 // + (System.currentTimeMillis() - start) + "ms");
845 * Transfer features to the target sequence. The start/end positions are
846 * converted using the mapping. Features which do not overlap are ignored.
847 * Features whose parent is not the specified identifier are also ignored.
850 * @param targetSequence
855 protected boolean transferFeatures(List<SequenceFeature> sfs,
856 SequenceI targetSequence, MapList mapping, String parentId)
858 final boolean forwardStrand = mapping.isFromForwardStrand();
861 * sort features by start position (which corresponds to end
862 * position descending if reverse strand) so as to add them in
863 * 'forwards' order to the target sequence
865 SequenceFeatures.sortFeatures(sfs, forwardStrand);
867 boolean transferred = false;
869 for (int i = 0, n = sfs.size(); i < n; i++)
872 // if ((i%1000) == 0) {
873 //// Platform.timeCheck("Feature " + bhtest, Platform.TIME_GET);
874 // Platform.timeCheck("ESP. xferFeature + " + (i) + "/" + n,
875 // Platform.TIME_MARK);
878 SequenceFeature sf = sfs.get(i);
879 if (retainFeature(sf, parentId))
881 transferFeature(sf, targetSequence, mapping, forwardStrand);
890 * Answers true if the feature type is one we want to keep for the sequence.
891 * Some features are only retrieved in order to identify the sequence range,
892 * and may then be discarded as redundant information (e.g. "CDS" feature for
895 @SuppressWarnings("unused")
896 protected boolean retainFeature(SequenceFeature sf, String accessionId)
898 return true; // override as required
902 * Answers true if the feature has a Parent which refers to the given
903 * accession id, or if the feature has no parent. Answers false if the
904 * feature's Parent is for a different accession id.
910 protected boolean featureMayBelong(SequenceFeature sf, String identifier)
912 String parent = (String) sf.getValue(PARENT);
913 if (parent != null && !parent.equalsIgnoreCase(identifier))
915 // this genomic feature belongs to a different transcript
922 * Answers a short description of the sequence fetcher
925 public String getDescription()
927 return "Ensembl " + getSourceEnsemblType().getType()
928 + " sequence with variant features";
932 * Returns a (possibly empty) list of features on the sequence which have the
933 * specified sequence ontology term (or a sub-type of it), and the given
934 * identifier as parent
941 protected List<SequenceFeature> findFeatures(SequenceI sequence,
942 String term, String parentId)
944 List<SequenceFeature> result = new ArrayList<>();
946 List<SequenceFeature> sfs = sequence.getFeatures()
947 .getFeaturesByOntology(term);
948 for (SequenceFeature sf : sfs)
950 String parent = (String) sf.getValue(PARENT);
951 if (parent != null && parent.equalsIgnoreCase(parentId))
961 * Answers true if the feature type is either 'NMD_transcript_variant' or
962 * 'transcript' (or one of its sub-types in the Sequence Ontology). This is
963 * because NMD_transcript_variant behaves like 'transcript' in Ensembl
964 * although strictly speaking it is not (it is a sub-type of
967 * (This test was needed when fetching transcript features as GFF. As we are
968 * now fetching as JSON, all features have type 'transcript' so the check for
969 * NMD_transcript_variant is redundant. Left in for any future case arising.)
974 public static boolean isTranscript(String featureType)
976 return SequenceOntologyI.NMD_TRANSCRIPT_VARIANT.equals(featureType)
977 || SequenceOntologyFactory.getInstance().isA(featureType,
978 SequenceOntologyI.TRANSCRIPT);