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 NAME = "Name";
67 protected static final String DESCRIPTION = "description";
70 * enum for 'type' parameter to the /sequence REST service
72 public enum EnsemblSeqType
75 * type=genomic to fetch full dna including introns
80 * type=cdna to fetch coding dna including UTRs
85 * type=cds to fetch coding dna excluding UTRs
90 * type=protein to fetch peptide product sequence
95 * the value of the 'type' parameter to fetch this version of
100 EnsemblSeqType(String t)
105 public String getType()
113 * Default constructor (to use rest.ensembl.org)
115 public EnsemblSeqProxy()
121 * Constructor given the target domain to fetch data from
123 public EnsemblSeqProxy(String d)
129 * Makes the sequence queries to Ensembl's REST service and returns an
130 * alignment consisting of the returned sequences.
133 public AlignmentI getSequenceRecords(String query) throws Exception
135 // TODO use a String... query vararg instead?
137 // danger: accession separator used as a regex here, a string elsewhere
138 // in this case it is ok (it is just a space), but (e.g.) '\' would not be
139 List<String> allIds = Arrays
140 .asList(query.split(getAccessionSeparator()));
141 AlignmentI alignment = null;
145 * execute queries, if necessary in batches of the
146 * maximum allowed number of ids
148 int maxQueryCount = getMaximumQueryCount();
149 for (int v = 0, vSize = allIds.size(); v < vSize; v += maxQueryCount)
151 int p = Math.min(vSize, v + maxQueryCount);
152 List<String> ids = allIds.subList(v, p);
155 alignment = fetchSequences(ids, alignment);
156 } catch (Throwable r)
159 String msg = "Aborting ID retrieval after " + v
160 + " chunks. Unexpected problem (" + r.getLocalizedMessage()
162 System.err.println(msg);
168 if (alignment == null)
174 * fetch and transfer genomic sequence features,
175 * fetch protein product and add as cross-reference
177 for (String accId : allIds)
179 addFeaturesAndProduct(accId, alignment);
182 for (SequenceI seq : alignment.getSequences())
184 getCrossReferences(seq);
191 * Fetches Ensembl features using the /overlap REST endpoint, and adds them to
192 * the sequence in the alignment. Also fetches the protein product, maps it
193 * from the CDS features of the sequence, and saves it as a cross-reference of
199 protected void addFeaturesAndProduct(String accId, AlignmentI alignment)
201 if (alignment == null)
209 * get 'dummy' genomic sequence with gene, transcript,
210 * exon, cds and variation features
212 SequenceI genomicSequence = null;
213 EnsemblFeatures gffFetcher = new EnsemblFeatures(getDomain());
214 EnsemblFeatureType[] features = getFeaturesToFetch();
215 AlignmentI geneFeatures = gffFetcher.getSequenceRecords(accId,
217 if (geneFeatures != null && geneFeatures.getHeight() > 0)
219 genomicSequence = geneFeatures.getSequenceAt(0);
221 if (genomicSequence != null)
224 * transfer features to the query sequence
226 SequenceI querySeq = alignment.findName(accId, true);
227 if (transferFeatures(accId, genomicSequence, querySeq))
231 * fetch and map protein product, and add it as a cross-reference
232 * of the retrieved sequence
234 addProteinProduct(querySeq);
237 } catch (IOException e)
240 "Error transferring Ensembl features: " + e.getMessage());
245 * Returns those sequence feature types to fetch from Ensembl. We may want
246 * features either because they are of interest to the user, or as means to
247 * identify the locations of the sequence on the genomic sequence (CDS
248 * features identify CDS, exon features identify cDNA etc).
252 protected abstract EnsemblFeatureType[] getFeaturesToFetch();
255 * Fetches and maps the protein product, and adds it as a cross-reference of
256 * the retrieved sequence
258 protected void addProteinProduct(SequenceI querySeq)
260 String accId = querySeq.getName();
263 AlignmentI protein = new EnsemblProtein(getDomain())
264 .getSequenceRecords(accId);
265 if (protein == null || protein.getHeight() == 0)
267 System.out.println("No protein product found for " + accId);
270 SequenceI proteinSeq = protein.getSequenceAt(0);
273 * need dataset sequences (to be the subject of mappings)
275 proteinSeq.createDatasetSequence();
276 querySeq.createDatasetSequence();
278 MapList mapList = AlignmentUtils.mapCdsToProtein(querySeq,
282 // clunky: ensure Uniprot xref if we have one is on mapped sequence
283 SequenceI ds = proteinSeq.getDatasetSequence();
284 // TODO: Verify ensp primary ref is on proteinSeq.getDatasetSequence()
285 Mapping map = new Mapping(ds, mapList);
286 DBRefEntry dbr = new DBRefEntry(getDbSource(),
287 getEnsemblDataVersion(), proteinSeq.getName(), map);
288 querySeq.getDatasetSequence().addDBRef(dbr);
289 DBRefEntry[] uprots = DBRefUtils.selectRefs(ds.getDBRefs(),
291 { DBRefSource.UNIPROT });
292 DBRefEntry[] upxrefs = DBRefUtils.selectRefs(querySeq.getDBRefs(),
294 { DBRefSource.UNIPROT });
297 for (DBRefEntry up : uprots)
299 // locate local uniprot ref and map
300 List<DBRefEntry> upx = DBRefUtils.searchRefs(upxrefs,
301 up.getAccessionId());
310 "Implementation issue - multiple uniprot acc on product sequence.");
315 upxref = new DBRefEntry(DBRefSource.UNIPROT,
316 getEnsemblDataVersion(), up.getAccessionId());
319 Mapping newMap = new Mapping(ds, mapList);
320 upxref.setVersion(getEnsemblDataVersion());
321 upxref.setMap(newMap);
324 // add the new uniprot ref
325 querySeq.getDatasetSequence().addDBRef(upxref);
332 * copy exon features to protein, compute peptide variants from dna
333 * variants and add as features on the protein sequence ta-da
335 AlignmentUtils.computeProteinFeatures(querySeq, proteinSeq,
338 } catch (Exception e)
341 .println(String.format("Error retrieving protein for %s: %s",
342 accId, e.getMessage()));
347 * Get database xrefs from Ensembl, and attach them to the sequence
351 protected void getCrossReferences(SequenceI seq)
353 while (seq.getDatasetSequence() != null)
355 seq = seq.getDatasetSequence();
358 EnsemblXref xrefFetcher = new EnsemblXref(getDomain(), getDbSource(),
359 getEnsemblDataVersion());
360 List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName());
361 for (DBRefEntry xref : xrefs)
367 * and add a reference to itself
369 DBRefEntry self = new DBRefEntry(getDbSource(), getEnsemblDataVersion(),
375 * Fetches sequences for the list of accession ids and adds them to the
376 * alignment. Returns the extended (or created) alignment.
381 * @throws JalviewException
382 * @throws IOException
384 protected AlignmentI fetchSequences(List<String> ids,
385 AlignmentI alignment) throws JalviewException, IOException
387 if (!isEnsemblAvailable())
390 throw new JalviewException("ENSEMBL Rest API not available.");
392 BufferedReader br = getSequenceReader(ids);
398 List<SequenceI> seqs = parseSequenceJson(br);
402 throw new IOException("No data returned for " + ids);
405 if (seqs.size() != ids.size())
407 System.out.println(String.format(
408 "Only retrieved %d sequences for %d query strings",
409 seqs.size(), ids.size()));
414 AlignmentI seqal = new Alignment(
415 seqs.toArray(new SequenceI[seqs.size()]));
416 for (SequenceI seq : seqs)
418 if (seq.getDescription() == null)
420 seq.setDescription(getDbName());
422 String name = seq.getName();
423 if (ids.contains(name)
424 || ids.contains(name.replace("ENSP", "ENST")))
426 // TODO JAL-3077 use true accession version in dbref
427 DBRefEntry dbref = DBRefUtils.parseToDbRef(seq, getDbSource(),
428 getEnsemblDataVersion(), name);
432 if (alignment == null)
438 alignment.append(seqal);
445 * Parses a JSON response into a list of sequences
449 * @see http://rest.ensembl.org/documentation/info/sequence_id
451 protected List<SequenceI> parseSequenceJson(BufferedReader br)
453 JSONParser jp = new JSONParser();
454 List<SequenceI> result = new ArrayList<>();
458 * for now, assumes only one sequence returned; refactor if needed
459 * in future to handle a JSONArray with more than one
461 final JSONObject val = (JSONObject) jp.parse(br);
462 Object s = val.get("desc");
463 String desc = s == null ? null : s.toString();
465 String id = s == null ? null : s.toString();
467 String seq = s == null ? null : s.toString();
468 Sequence sequence = new Sequence(id, seq);
471 sequence.setDescription(desc);
473 // todo JAL-3077 make a DBRefEntry with true accession version
474 // s = val.get("version");
475 // String version = s == null ? "0" : s.toString();
476 // DBRefEntry dbref = new DBRefEntry(getDbSource(), version, id);
477 // sequence.addDBRef(dbref);
478 result.add(sequence);
479 } catch (ParseException | IOException e)
481 System.err.println("Error processing JSON response: " + e.toString());
488 * Returns the URL for the REST call
491 * @throws MalformedURLException
494 protected URL getUrl(List<String> ids) throws MalformedURLException
497 * a single id is included in the URL path
498 * multiple ids go in the POST body instead
500 StringBuffer urlstring = new StringBuffer(128);
501 urlstring.append(getDomain() + "/sequence/id");
504 urlstring.append("/").append(ids.get(0));
506 // @see https://github.com/Ensembl/ensembl-rest/wiki/Output-formats
507 urlstring.append("?type=").append(getSourceEnsemblType().getType());
508 urlstring.append(("&Accept=application/json"));
509 urlstring.append(("&Content-Type=application/json"));
511 String objectType = getObjectType();
512 if (objectType != null)
514 urlstring.append("&").append(OBJECT_TYPE).append("=")
518 URL url = new URL(urlstring.toString());
523 * Override this method to specify object_type request parameter
527 protected String getObjectType()
533 * A sequence/id POST request currently allows up to 50 queries
535 * @see http://rest.ensembl.org/documentation/info/sequence_id_post
538 public int getMaximumQueryCount()
544 protected boolean useGetRequest()
551 * @return the configured sequence return type for this source
553 protected abstract EnsemblSeqType getSourceEnsemblType();
556 * Returns a list of [start, end] genomic ranges corresponding to the sequence
559 * The correspondence between the frames of reference is made by locating
560 * those features on the genomic sequence which identify the retrieved
561 * sequence. Specifically
563 * <li>genomic sequence is identified by "transcript" features with
564 * ID=transcript:transcriptId</li>
565 * <li>cdna sequence is identified by "exon" features with
566 * Parent=transcript:transcriptId</li>
567 * <li>cds sequence is identified by "CDS" features with
568 * Parent=transcript:transcriptId</li>
571 * The returned ranges are sorted to run forwards (for positive strand) or
572 * backwards (for negative strand). Aborts and returns null if both positive
573 * and negative strand are found (this should not normally happen).
575 * @param sourceSequence
578 * the start position of the sequence we are mapping to
581 protected MapList getGenomicRangesFromFeatures(SequenceI sourceSequence,
582 String accId, int start)
584 List<SequenceFeature> sfs = sourceSequence.getFeatures()
585 .getPositionalFeatures();
592 * generously initial size for number of cds regions
593 * (worst case titin Q8WZ42 has c. 313 exons)
595 List<int[]> regions = new ArrayList<>(100);
596 int mappedLength = 0;
597 int direction = 1; // forward
598 boolean directionSet = false;
600 for (SequenceFeature sf : sfs)
603 * accept the target feature type or a specialisation of it
604 * (e.g. coding_exon for exon)
606 if (identifiesSequence(sf, accId))
608 int strand = sf.getStrand();
609 strand = strand == 0 ? 1 : strand; // treat unknown as forward
611 if (directionSet && strand != direction)
613 // abort - mix of forward and backward
615 "Error: forward and backward strand for " + accId);
622 * add to CDS ranges, semi-sorted forwards/backwards
626 regions.add(0, new int[] { sf.getEnd(), sf.getBegin() });
630 regions.add(new int[] { sf.getBegin(), sf.getEnd() });
632 mappedLength += Math.abs(sf.getEnd() - sf.getBegin() + 1);
637 * 'gene' sequence is contiguous so we can stop as soon as its
638 * identifying feature has been found
645 if (regions.isEmpty())
647 System.out.println("Failed to identify target sequence for " + accId
648 + " from genomic features");
653 * a final sort is needed since Ensembl returns CDS sorted within source
654 * (havana / ensembl_havana)
656 Collections.sort(regions, direction == 1 ? IntRangeComparator.ASCENDING
657 : IntRangeComparator.DESCENDING);
659 List<int[]> to = Arrays
661 { start, start + mappedLength - 1 });
663 return new MapList(regions, to, 1, 1);
667 * Answers true if the sequence being retrieved may occupy discontiguous
668 * regions on the genomic sequence.
670 protected boolean isSpliceable()
676 * Returns true if the sequence feature marks positions of the genomic
677 * sequence feature which are within the sequence being retrieved. For
678 * example, an 'exon' feature whose parent is the target transcript marks the
679 * cdna positions of the transcript.
685 protected abstract boolean identifiesSequence(SequenceFeature sf,
689 * Answers a list of sequence features that mark positions of the genomic
690 * sequence feature which are within the sequence being retrieved. For
691 * example, an 'exon' feature whose parent is the target transcript marks the
692 * cdna positions of the transcript. For a gene sequence, this is trivially
693 * just the 'gene' feature with matching gene id.
699 protected abstract List<SequenceFeature> getIdentifyingFeatures(
700 SequenceI seq, String accId);
703 * Transfers the sequence feature to the target sequence, locating its start
704 * and end range based on the mapping. Features which do not overlap the
705 * target sequence are ignored.
708 * @param targetSequence
710 * mapping from the sequence feature's coordinates to the target
712 * @param forwardStrand
714 protected void transferFeature(SequenceFeature sf,
715 SequenceI targetSequence, MapList mapping, boolean forwardStrand)
717 int start = sf.getBegin();
718 int end = sf.getEnd();
719 int[] mappedRange = mapping.locateInTo(start, end);
721 if (mappedRange != null)
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 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
731 group, sf.getScore());
732 targetSequence.addSequenceFeature(copy);
735 * for sequence_variant on reverse strand, have to convert the allele
736 * values to their complements
738 if (!forwardStrand && SequenceOntologyFactory.getInstance()
739 .isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
741 reverseComplementAlleles(copy);
747 * Change the 'alleles' value of a feature by converting to complementary
748 * bases, and also update the feature description to match
752 static void reverseComplementAlleles(SequenceFeature sf)
754 final String alleles = (String) sf.getValue(Gff3Helper.ALLELES);
759 StringBuilder complement = new StringBuilder(alleles.length());
760 for (String allele : alleles.split(","))
762 reverseComplementAllele(complement, allele);
764 String comp = complement.toString();
765 sf.setValue(Gff3Helper.ALLELES, comp);
766 sf.setDescription(comp);
769 * replace value of "alleles=" in sf.ATTRIBUTES as well
770 * so 'output as GFF' shows reverse complement alleles
772 String atts = sf.getAttributes();
775 atts = atts.replace(Gff3Helper.ALLELES + "=" + alleles,
776 Gff3Helper.ALLELES + "=" + comp);
777 sf.setAttributes(atts);
782 * Makes the 'reverse complement' of the given allele and appends it to the
783 * buffer, after a comma separator if not the first
788 static void reverseComplementAllele(StringBuilder complement,
791 if (complement.length() > 0)
793 complement.append(",");
797 * some 'alleles' are actually descriptive terms
798 * e.g. HGMD_MUTATION, PhenCode_variation
799 * - we don't want to 'reverse complement' these
801 if (!Comparison.isNucleotideSequence(allele, true))
803 complement.append(allele);
807 for (int i = allele.length() - 1; i >= 0; i--)
809 complement.append(Dna.getComplement(allele.charAt(i)));
815 * Transfers features from sourceSequence to targetSequence
818 * @param sourceSequence
819 * @param targetSequence
820 * @return true if any features were transferred, else false
822 protected boolean transferFeatures(String accessionId,
823 SequenceI sourceSequence, SequenceI targetSequence)
825 if (sourceSequence == null || targetSequence == null)
830 // long start = System.currentTimeMillis();
831 List<SequenceFeature> sfs = sourceSequence.getFeatures()
832 .getPositionalFeatures();
833 MapList mapping = getGenomicRangesFromFeatures(sourceSequence,
834 accessionId, targetSequence.getStart());
840 boolean result = transferFeatures(sfs, targetSequence, mapping,
842 // System.out.println("transferFeatures (" + (sfs.size()) + " --> "
843 // + targetSequence.getFeatures().getFeatureCount(true) + ") to "
844 // + targetSequence.getName() + " took "
845 // + (System.currentTimeMillis() - start) + "ms");
850 * Transfer features to the target sequence. The start/end positions are
851 * converted using the mapping. Features which do not overlap are ignored.
852 * Features whose parent is not the specified identifier are also ignored.
855 * @param targetSequence
860 protected boolean transferFeatures(List<SequenceFeature> sfs,
861 SequenceI targetSequence, MapList mapping, String parentId)
863 final boolean forwardStrand = mapping.isFromForwardStrand();
866 * sort features by start position (which corresponds to end
867 * position descending if reverse strand) so as to add them in
868 * 'forwards' order to the target sequence
870 SequenceFeatures.sortFeatures(sfs, forwardStrand);
872 boolean transferred = false;
873 for (SequenceFeature sf : sfs)
875 if (retainFeature(sf, parentId))
877 transferFeature(sf, targetSequence, mapping, forwardStrand);
885 * Answers true if the feature type is one we want to keep for the sequence.
886 * Some features are only retrieved in order to identify the sequence range,
887 * and may then be discarded as redundant information (e.g. "CDS" feature for
890 @SuppressWarnings("unused")
891 protected boolean retainFeature(SequenceFeature sf, String accessionId)
893 return true; // override as required
897 * Answers true if the feature has a Parent which refers to the given
898 * accession id, or if the feature has no parent. Answers false if the
899 * feature's Parent is for a different accession id.
905 protected boolean featureMayBelong(SequenceFeature sf, String identifier)
907 String parent = (String) sf.getValue(PARENT);
908 // using contains to allow for prefix "gene:", "transcript:" etc
910 && !parent.toUpperCase().contains(identifier.toUpperCase()))
912 // this genomic feature belongs to a different transcript
919 public String getDescription()
921 return "Ensembl " + getSourceEnsemblType().getType()
922 + " sequence with variant features";
926 * Returns a (possibly empty) list of features on the sequence which have the
927 * specified sequence ontology term (or a sub-type of it), and the given
928 * identifier as parent
935 protected List<SequenceFeature> findFeatures(SequenceI sequence,
936 String term, String parentId)
938 List<SequenceFeature> result = new ArrayList<>();
940 List<SequenceFeature> sfs = sequence.getFeatures()
941 .getFeaturesByOntology(term);
942 for (SequenceFeature sf : sfs)
944 String parent = (String) sf.getValue(PARENT);
945 if (parent != null && parent.equalsIgnoreCase(parentId))
955 * Answers true if the feature type is either 'NMD_transcript_variant' or
956 * 'transcript' or one of its sub-types in the Sequence Ontology. This is
957 * needed because NMD_transcript_variant behaves like 'transcript' in Ensembl
958 * although strictly speaking it is not (it is a sub-type of
964 public static boolean isTranscript(String featureType)
966 return SequenceOntologyI.NMD_TRANSCRIPT_VARIANT.equals(featureType)
967 || SequenceOntologyFactory.getInstance().isA(featureType,
968 SequenceOntologyI.TRANSCRIPT);