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.SequenceFeature;
32 import jalview.datamodel.SequenceI;
33 import jalview.datamodel.features.SequenceFeatures;
34 import jalview.exceptions.JalviewException;
35 import jalview.io.FastaFile;
36 import jalview.io.FileParse;
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.IOException;
45 import java.net.MalformedURLException;
47 import java.util.ArrayList;
48 import java.util.Arrays;
49 import java.util.Collections;
50 import java.util.List;
53 * Base class for Ensembl sequence fetchers
55 * @see http://rest.ensembl.org/documentation/info/sequence_id
58 public abstract class EnsemblSeqProxy extends EnsemblRestClient
60 private static final String ALLELES = "alleles";
62 protected static final String PARENT = "Parent";
64 protected static final String ID = "ID";
66 protected static final String NAME = "Name";
68 protected static final String DESCRIPTION = "description";
71 * enum for 'type' parameter to the /sequence REST service
73 public enum EnsemblSeqType
76 * type=genomic to fetch full dna including introns
81 * type=cdna to fetch coding dna including UTRs
86 * type=cds to fetch coding dna excluding UTRs
91 * type=protein to fetch peptide product sequence
96 * the value of the 'type' parameter to fetch this version of
101 EnsemblSeqType(String t)
106 public String getType()
114 * Default constructor (to use rest.ensembl.org)
116 public EnsemblSeqProxy()
122 * Constructor given the target domain to fetch data from
124 public EnsemblSeqProxy(String d)
130 * Makes the sequence queries to Ensembl's REST service and returns an
131 * alignment consisting of the returned sequences.
134 public AlignmentI getSequenceRecords(String query) throws Exception
136 // TODO use a String... query vararg instead?
138 // danger: accession separator used as a regex here, a string elsewhere
139 // in this case it is ok (it is just a space), but (e.g.) '\' would not be
140 List<String> allIds = Arrays.asList(query
141 .split(getAccessionSeparator()));
142 AlignmentI alignment = null;
146 * execute queries, if necessary in batches of the
147 * maximum allowed number of ids
149 int maxQueryCount = getMaximumQueryCount();
150 for (int v = 0, vSize = allIds.size(); v < vSize; v += maxQueryCount)
152 int p = Math.min(vSize, v + maxQueryCount);
153 List<String> ids = allIds.subList(v, p);
156 alignment = fetchSequences(ids, alignment);
157 } catch (Throwable r)
160 String msg = "Aborting ID retrieval after " + v
161 + " chunks. Unexpected problem (" + r.getLocalizedMessage()
163 System.err.println(msg);
169 if (alignment == null)
175 * fetch and transfer genomic sequence features,
176 * fetch protein product and add as cross-reference
178 for (String accId : allIds)
180 addFeaturesAndProduct(accId, alignment);
183 for (SequenceI seq : alignment.getSequences())
185 getCrossReferences(seq);
192 * Fetches Ensembl features using the /overlap REST endpoint, and adds them to
193 * the sequence in the alignment. Also fetches the protein product, maps it
194 * from the CDS features of the sequence, and saves it as a cross-reference of
200 protected void addFeaturesAndProduct(String accId, AlignmentI alignment)
202 if (alignment == null)
210 * get 'dummy' genomic sequence with 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);
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)
239 System.err.println("Error transferring Ensembl features: "
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
279 .mapCdsToProtein(querySeq, proteinSeq);
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(),
290 new String[] { DBRefSource.UNIPROT });
291 DBRefEntry[] upxrefs = DBRefUtils.selectRefs(querySeq.getDBRefs(),
292 new String[] { 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 .warn("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
334 .computeProteinFeatures(querySeq, proteinSeq, mapList);
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();
356 EnsemblXref xrefFetcher = new EnsemblXref(getDomain(), getDbSource(),
357 getEnsemblDataVersion());
358 List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName());
359 for (DBRefEntry xref : xrefs)
365 * and add a reference to itself
367 DBRefEntry self = new DBRefEntry(getDbSource(),
368 getEnsemblDataVersion(), seq.getName());
373 * Fetches sequences for the list of accession ids and adds them to the
374 * alignment. Returns the extended (or created) alignment.
379 * @throws JalviewException
380 * @throws IOException
382 protected AlignmentI fetchSequences(List<String> ids, AlignmentI alignment)
383 throws JalviewException, IOException
385 if (!isEnsemblAvailable())
388 throw new JalviewException("ENSEMBL Rest API not available.");
390 FileParse fp = getSequenceReader(ids);
396 FastaFile fr = new FastaFile(fp);
397 if (fr.hasWarningMessage())
399 System.out.println(String.format(
400 "Warning when retrieving %d ids %s\n%s", ids.size(),
401 ids.toString(), fr.getWarningMessage()));
403 else if (fr.getSeqs().size() != ids.size())
405 System.out.println(String.format(
406 "Only retrieved %d sequences for %d query strings", fr
407 .getSeqs().size(), ids.size()));
410 if (fr.getSeqs().size() == 1 && fr.getSeqs().get(0).getLength() == 0)
413 * POST request has returned an empty FASTA file e.g. for invalid id
415 throw new IOException("No data returned for " + ids);
418 if (fr.getSeqs().size() > 0)
420 AlignmentI seqal = new Alignment(fr.getSeqsAsArray());
421 for (SequenceI sq : seqal.getSequences())
423 if (sq.getDescription() == null)
425 sq.setDescription(getDbName());
427 String name = sq.getName();
428 if (ids.contains(name)
429 || ids.contains(name.replace("ENSP", "ENST")))
431 DBRefEntry dbref = DBRefUtils.parseToDbRef(sq, getDbSource(),
432 getEnsemblDataVersion(), name);
436 if (alignment == null)
442 alignment.append(seqal);
449 * Returns the URL for the REST call
452 * @throws MalformedURLException
455 protected URL getUrl(List<String> ids) throws MalformedURLException
458 * a single id is included in the URL path
459 * multiple ids go in the POST body instead
461 StringBuffer urlstring = new StringBuffer(128);
462 urlstring.append(getDomain() + "/sequence/id");
465 urlstring.append("/").append(ids.get(0));
467 // @see https://github.com/Ensembl/ensembl-rest/wiki/Output-formats
468 urlstring.append("?type=").append(getSourceEnsemblType().getType());
469 urlstring.append(("&Accept=text/x-fasta"));
471 URL url = new URL(urlstring.toString());
476 * A sequence/id POST request currently allows up to 50 queries
478 * @see http://rest.ensembl.org/documentation/info/sequence_id_post
481 public int getMaximumQueryCount()
487 protected boolean useGetRequest()
493 protected String getRequestMimeType(boolean multipleIds)
495 return multipleIds ? "application/json" : "text/x-fasta";
499 protected String getResponseMimeType()
501 return "text/x-fasta";
506 * @return the configured sequence return type for this source
508 protected abstract EnsemblSeqType getSourceEnsemblType();
511 * Returns a list of [start, end] genomic ranges corresponding to the sequence
514 * The correspondence between the frames of reference is made by locating
515 * those features on the genomic sequence which identify the retrieved
516 * sequence. Specifically
518 * <li>genomic sequence is identified by "transcript" features with
519 * ID=transcript:transcriptId</li>
520 * <li>cdna sequence is identified by "exon" features with
521 * Parent=transcript:transcriptId</li>
522 * <li>cds sequence is identified by "CDS" features with
523 * Parent=transcript:transcriptId</li>
526 * The returned ranges are sorted to run forwards (for positive strand) or
527 * backwards (for negative strand). Aborts and returns null if both positive
528 * and negative strand are found (this should not normally happen).
530 * @param sourceSequence
533 * the start position of the sequence we are mapping to
536 protected MapList getGenomicRangesFromFeatures(SequenceI sourceSequence,
537 String accId, int start)
539 // SequenceFeature[] sfs = sourceSequence.getSequenceFeatures();
540 List<SequenceFeature> sfs = sourceSequence.getFeatures()
541 .getPositionalFeatures();
548 * generously initial size for number of cds regions
549 * (worst case titin Q8WZ42 has c. 313 exons)
551 List<int[]> regions = new ArrayList<int[]>(100);
552 int mappedLength = 0;
553 int direction = 1; // forward
554 boolean directionSet = false;
556 for (SequenceFeature sf : sfs)
559 * accept the target feature type or a specialisation of it
560 * (e.g. coding_exon for exon)
562 if (identifiesSequence(sf, accId))
564 int strand = sf.getStrand();
565 strand = strand == 0 ? 1 : strand; // treat unknown as forward
567 if (directionSet && strand != direction)
569 // abort - mix of forward and backward
570 System.err.println("Error: forward and backward strand for "
578 * add to CDS ranges, semi-sorted forwards/backwards
582 regions.add(0, new int[] { sf.getEnd(), sf.getBegin() });
586 regions.add(new int[] { sf.getBegin(), sf.getEnd() });
588 mappedLength += Math.abs(sf.getEnd() - sf.getBegin() + 1);
593 * 'gene' sequence is contiguous so we can stop as soon as its
594 * identifying feature has been found
601 if (regions.isEmpty())
603 System.out.println("Failed to identify target sequence for " + accId
604 + " from genomic features");
609 * a final sort is needed since Ensembl returns CDS sorted within source
610 * (havana / ensembl_havana)
612 Collections.sort(regions, direction == 1 ? IntRangeComparator.ASCENDING
613 : IntRangeComparator.DESCENDING);
615 List<int[]> to = Arrays.asList(new int[] { start,
616 start + mappedLength - 1 });
618 return new MapList(regions, to, 1, 1);
622 * Answers true if the sequence being retrieved may occupy discontiguous
623 * regions on the genomic sequence.
625 protected boolean isSpliceable()
631 * Returns true if the sequence feature marks positions of the genomic
632 * sequence feature which are within the sequence being retrieved. For
633 * example, an 'exon' feature whose parent is the target transcript marks the
634 * cdna positions of the transcript.
640 protected abstract boolean identifiesSequence(SequenceFeature sf,
644 * Transfers the sequence feature to the target sequence, locating its start
645 * and end range based on the mapping. Features which do not overlap the
646 * target sequence are ignored.
649 * @param targetSequence
651 * mapping from the sequence feature's coordinates to the target
653 * @param forwardStrand
655 protected void transferFeature(SequenceFeature sf,
656 SequenceI targetSequence, MapList mapping, boolean forwardStrand)
658 int start = sf.getBegin();
659 int end = sf.getEnd();
660 int[] mappedRange = mapping.locateInTo(start, end);
662 if (mappedRange != null)
664 String group = sf.getFeatureGroup();
665 if (".".equals(group))
667 group = getDbSource();
669 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
670 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
671 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
673 targetSequence.addSequenceFeature(copy);
676 * for sequence_variant on reverse strand, have to convert the allele
677 * values to their complements
680 && SequenceOntologyFactory.getInstance().isA(sf.getType(),
681 SequenceOntologyI.SEQUENCE_VARIANT))
683 reverseComplementAlleles(copy);
689 * Change the 'alleles' value of a feature by converting to complementary
690 * bases, and also update the feature description to match
694 static void reverseComplementAlleles(SequenceFeature sf)
696 final String alleles = (String) sf.getValue(ALLELES);
701 StringBuilder complement = new StringBuilder(alleles.length());
702 for (String allele : alleles.split(","))
704 reverseComplementAllele(complement, allele);
706 String comp = complement.toString();
707 sf.setValue(ALLELES, comp);
708 sf.setDescription(comp);
711 * replace value of "alleles=" in sf.ATTRIBUTES as well
712 * so 'output as GFF' shows reverse complement alleles
714 String atts = sf.getAttributes();
717 atts = atts.replace(ALLELES + "=" + alleles, ALLELES + "=" + comp);
718 sf.setAttributes(atts);
723 * Makes the 'reverse complement' of the given allele and appends it to the
724 * buffer, after a comma separator if not the first
729 static void reverseComplementAllele(StringBuilder complement,
732 if (complement.length() > 0)
734 complement.append(",");
738 * some 'alleles' are actually descriptive terms
739 * e.g. HGMD_MUTATION, PhenCode_variation
740 * - we don't want to 'reverse complement' these
742 if (!Comparison.isNucleotideSequence(allele, true))
744 complement.append(allele);
748 for (int i = allele.length() - 1; i >= 0; i--)
750 complement.append(Dna.getComplement(allele.charAt(i)));
756 * Transfers features from sourceSequence to targetSequence
759 * @param sourceSequence
760 * @param targetSequence
761 * @return true if any features were transferred, else false
763 protected boolean transferFeatures(String accessionId,
764 SequenceI sourceSequence, SequenceI targetSequence)
766 if (sourceSequence == null || targetSequence == null)
771 long start = System.currentTimeMillis();
772 // SequenceFeature[] sfs = sourceSequence.getSequenceFeatures();
773 List<SequenceFeature> sfs = sourceSequence.getFeatures()
774 .getPositionalFeatures();
775 MapList mapping = getGenomicRangesFromFeatures(sourceSequence,
776 accessionId, targetSequence.getStart());
782 boolean result = transferFeatures(sfs, targetSequence, mapping,
784 System.out.println("transferFeatures (" + (sfs.size()) + " --> "
785 + targetSequence.getFeatures().getFeatureCount(true) + ") to "
786 + targetSequence.getName() + " took "
787 + (System.currentTimeMillis() - start) + "ms");
792 * Transfer features to the target sequence. The start/end positions are
793 * converted using the mapping. Features which do not overlap are ignored.
794 * Features whose parent is not the specified identifier are also ignored.
797 * @param targetSequence
802 protected boolean transferFeatures(List<SequenceFeature> sfs,
803 SequenceI targetSequence, MapList mapping, String parentId)
805 final boolean forwardStrand = mapping.isFromForwardStrand();
808 * sort features by start position (which corresponds to end
809 * position descending if reverse strand) so as to add them in
810 * 'forwards' order to the target sequence
812 SequenceFeatures.sortFeatures(sfs, forwardStrand);
814 boolean transferred = false;
815 for (SequenceFeature sf : sfs)
817 if (retainFeature(sf, parentId))
819 transferFeature(sf, targetSequence, mapping, forwardStrand);
827 * Answers true if the feature type is one we want to keep for the sequence.
828 * Some features are only retrieved in order to identify the sequence range,
829 * and may then be discarded as redundant information (e.g. "CDS" feature for
832 @SuppressWarnings("unused")
833 protected boolean retainFeature(SequenceFeature sf, String accessionId)
835 return true; // override as required
839 * Answers true if the feature has a Parent which refers to the given
840 * accession id, or if the feature has no parent. Answers false if the
841 * feature's Parent is for a different accession id.
847 protected boolean featureMayBelong(SequenceFeature sf, String identifier)
849 String parent = (String) sf.getValue(PARENT);
850 // using contains to allow for prefix "gene:", "transcript:" etc
851 if (parent != null && !parent.contains(identifier))
853 // this genomic feature belongs to a different transcript
860 public String getDescription()
862 return "Ensembl " + getSourceEnsemblType().getType()
863 + " sequence with variant features";
867 * Returns a (possibly empty) list of features on the sequence which have the
868 * specified sequence ontology term (or a sub-type of it), and the given
869 * identifier as parent
876 protected List<SequenceFeature> findFeatures(SequenceI sequence,
877 String term, String parentId)
879 List<SequenceFeature> result = new ArrayList<SequenceFeature>();
881 List<SequenceFeature> sfs = sequence.getFeatures()
882 .getFeaturesByOntology(term);
883 for (SequenceFeature sf : sfs)
885 String parent = (String) sf.getValue(PARENT);
886 if (parent != null && parent.equals(parentId))
896 * Answers true if the feature type is either 'NMD_transcript_variant' or
897 * 'transcript' or one of its sub-types in the Sequence Ontology. This is
898 * needed because NMD_transcript_variant behaves like 'transcript' in Ensembl
899 * although strictly speaking it is not (it is a sub-type of
905 public static boolean isTranscript(String featureType)
907 return SequenceOntologyI.NMD_TRANSCRIPT_VARIANT.equals(featureType)
908 || SequenceOntologyFactory.getInstance().isA(featureType,
909 SequenceOntologyI.TRANSCRIPT);