1 package jalview.io.vcf;
3 import htsjdk.samtools.util.CloseableIterator;
4 import htsjdk.variant.variantcontext.Allele;
5 import htsjdk.variant.variantcontext.VariantContext;
6 import htsjdk.variant.vcf.VCFHeader;
7 import htsjdk.variant.vcf.VCFHeaderLine;
9 import jalview.analysis.AlignmentUtils;
10 import jalview.analysis.Dna;
11 import jalview.api.AlignViewControllerGuiI;
12 import jalview.datamodel.AlignmentI;
13 import jalview.datamodel.DBRefEntry;
14 import jalview.datamodel.GeneLociI;
15 import jalview.datamodel.Mapping;
16 import jalview.datamodel.SequenceFeature;
17 import jalview.datamodel.SequenceI;
18 import jalview.ext.ensembl.EnsemblMap;
19 import jalview.ext.htsjdk.VCFReader;
20 import jalview.io.gff.Gff3Helper;
21 import jalview.io.gff.SequenceOntologyI;
22 import jalview.util.MapList;
23 import jalview.util.MappingUtils;
25 import java.io.IOException;
26 import java.util.HashMap;
27 import java.util.List;
29 import java.util.Map.Entry;
32 * A class to read VCF data (using the htsjdk) and add variants as sequence
33 * features on dna and any related protein product sequences
37 public class VCFLoader
39 private static final String FEATURE_GROUP_VCF = "VCF";
41 private static final String EXCL = "!";
44 * the alignment we are associated VCF data with
46 private AlignmentI al;
49 * mappings between VCF and sequence reference assembly regions, as
50 * key = "species!chromosome!fromAssembly!toAssembly
51 * value = Map{fromRange, toRange}
53 private Map<String, Map<int[], int[]>> assemblyMappings;
56 * Constructor given an alignment context
60 public VCFLoader(AlignmentI alignment)
64 // map of species!chromosome!fromAssembly!toAssembly to {fromRange, toRange}
65 assemblyMappings = new HashMap<String, Map<int[], int[]>>();
69 * Loads VCF on to an alignment - provided it can be related to one or more
70 * sequence's chromosomal coordinates.
72 * This method is not thread safe - concurrent threads should use separate
73 * instances of this class.
78 public void loadVCF(String filePath, AlignViewControllerGuiI status)
80 VCFReader reader = null;
84 // long start = System.currentTimeMillis();
85 reader = new VCFReader(filePath);
87 VCFHeader header = reader.getFileHeader();
88 VCFHeaderLine ref = header
89 .getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
90 // check if reference is wrt assembly19 (GRCh37)
91 // todo may need to allow user to specify reference assembly?
92 boolean isRefGrch37 = (ref != null && ref.getValue().contains(
99 * query for VCF overlapping each sequence in turn
101 for (SequenceI seq : al.getSequences())
103 int added = loadVCF(seq, reader, isRefGrch37);
108 transferAddedFeatures(seq);
111 // long elapsed = System.currentTimeMillis() - start;
112 String msg = String.format("Added %d VCF variants to %d sequence(s)",
116 status.setStatus(msg);
118 } catch (Throwable e)
120 System.err.println("Error processing VCF: " + e.getMessage());
129 } catch (IOException e)
138 * Transfers VCF features to sequences to which this sequence has a mapping.
139 * If the mapping is 1:3, computes peptide variants from nucleotide variants.
143 protected void transferAddedFeatures(SequenceI seq)
145 DBRefEntry[] dbrefs = seq.getDBRefs();
150 for (DBRefEntry dbref : dbrefs)
152 Mapping mapping = dbref.getMap();
153 if (mapping == null || mapping.getTo() == null)
158 SequenceI mapTo = mapping.getTo();
159 MapList map = mapping.getMap();
160 if (map.getFromRatio() == 3)
163 * dna-to-peptide product mapping
165 AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
170 * nucleotide-to-nucleotide mapping e.g. transcript to CDS
172 // TODO no DBRef to CDS is added to transcripts
173 List<SequenceFeature> features = seq.getFeatures()
174 .getPositionalFeatures(SequenceOntologyI.SEQUENCE_VARIANT);
175 for (SequenceFeature sf : features)
177 if (FEATURE_GROUP_VCF.equals(sf.getFeatureGroup()))
179 transferFeature(sf, mapTo, map);
187 * Tries to add overlapping variants read from a VCF file to the given
188 * sequence, and returns the number of overlapping variants found. Note that
189 * this requires the sequence to hold information as to its chromosomal
190 * positions and reference, in order to be able to map the VCF variants to the
195 * @param isVcfRefGrch37
198 protected int loadVCF(SequenceI seq, VCFReader reader,
199 boolean isVcfRefGrch37)
202 GeneLociI seqCoords = seq.getGeneLoci();
203 if (seqCoords == null)
208 List<int[]> seqChromosomalContigs = seqCoords.getMap().getToRanges();
209 for (int[] range : seqChromosomalContigs)
211 count += addVcfVariants(seq, reader, range, isVcfRefGrch37);
218 * Queries the VCF reader for any variants that overlap the given chromosome
219 * region of the sequence, and adds as variant features. Returns the number of
220 * overlapping variants found.
225 * start-end range of a sequence region in its chromosomal
227 * @param isVcfRefGrch37
228 * true if the VCF is with reference to GRCh37
231 protected int addVcfVariants(SequenceI seq, VCFReader reader,
232 int[] range, boolean isVcfRefGrch37)
234 GeneLociI seqCoords = seq.getGeneLoci();
236 String chromosome = seqCoords.getChromosomeId();
237 String seqRef = seqCoords.getAssemblyId();
238 String species = seqCoords.getSpeciesId();
240 // TODO handle species properly
241 if ("".equals(species))
247 * map chromosomal coordinates from GRCh38 (sequence) to
248 * GRCh37 (VCF) if necessary
250 // TODO generalise for other assemblies and species
252 String fromRef = "GRCh38";
253 if (fromRef.equalsIgnoreCase(seqRef) && isVcfRefGrch37)
255 String toRef = "GRCh37";
256 int[] newRange = mapReferenceRange(range, chromosome, species,
258 if (newRange == null)
260 System.err.println(String.format(
261 "Failed to map %s:%s:%s:%d:%d to %s", species, chromosome,
262 fromRef, range[0], range[1], toRef));
265 offset = newRange[0] - range[0];
269 boolean forwardStrand = range[0] <= range[1];
272 * query the VCF for overlaps
273 * (convert a reverse strand range to forwards)
276 MapList mapping = seqCoords.getMap();
278 int fromLocus = Math.min(range[0], range[1]);
279 int toLocus = Math.max(range[0], range[1]);
280 CloseableIterator<VariantContext> variants = reader.query(chromosome,
282 while (variants.hasNext())
285 * get variant location in sequence chromosomal coordinates
287 VariantContext variant = variants.next();
290 * we can only process SNP variants (which can be reported
291 * as part of a MIXED variant record
293 if (!variant.isSNP() && !variant.isMixed())
299 int start = variant.getStart() - offset;
300 int end = variant.getEnd() - offset;
303 * convert chromosomal location to sequence coordinates
304 * - null if a partially overlapping feature
306 int[] seqLocation = mapping.locateInFrom(start, end);
307 if (seqLocation != null)
309 addVariantFeatures(seq, variant, seqLocation[0], seqLocation[1],
320 * Inspects the VCF variant record, and adds variant features to the sequence.
321 * Only SNP variants are added, not INDELs.
323 * If the sequence maps to the reverse strand of the chromosome, reference and
324 * variant bases are recorded as their complements (C/G, A/T).
328 * @param featureStart
330 * @param forwardStrand
332 protected void addVariantFeatures(SequenceI seq, VariantContext variant,
333 int featureStart, int featureEnd, boolean forwardStrand)
335 byte[] reference = variant.getReference().getBases();
336 if (reference.length != 1)
339 * sorry, we don't handle INDEL variants
345 * for now we extract allele frequency as feature score; note
346 * this attribute is String for a simple SNP, but List<String> if
347 * multiple alleles at the locus; we extract for the simple case only
349 Object af = variant.getAttribute("AF");
351 if (af instanceof String)
355 score = Float.parseFloat((String) af);
356 } catch (NumberFormatException e)
362 StringBuilder sb = new StringBuilder();
363 sb.append(forwardStrand ? (char) reference[0] : complement(reference));
366 * inspect alleles and record SNP variants (as the variant
367 * record could be MIXED and include INDEL and SNP alleles)
368 * warning: getAlleles gives no guarantee as to the order
369 * in which they are returned
371 for (Allele allele : variant.getAlleles())
373 if (!allele.isReference())
375 byte[] alleleBase = allele.getBases();
376 if (alleleBase.length == 1)
378 sb.append(",").append(
379 forwardStrand ? (char) alleleBase[0]
380 : complement(alleleBase));
384 String alleles = sb.toString(); // e.g. G,A,C
386 String type = SequenceOntologyI.SEQUENCE_VARIANT;
388 SequenceFeature sf = new SequenceFeature(type, alleles, featureStart,
389 featureEnd, score, FEATURE_GROUP_VCF);
391 sf.setValue(Gff3Helper.ALLELES, alleles);
393 Map<String, Object> atts = variant.getAttributes();
394 for (Entry<String, Object> att : atts.entrySet())
396 sf.setValue(att.getKey(), att.getValue());
398 seq.addSequenceFeature(sf);
402 * A convenience method to complement a dna base and return the string value
408 protected String complement(byte[] reference)
410 return String.valueOf(Dna.getComplement((char) reference[0]));
414 * Determines the location of the query range (chromosome positions) in a
415 * different reference assembly.
417 * If the range is just a subregion of one for which we already have a mapping
418 * (for example, an exon sub-region of a gene), then the mapping is just
419 * computed arithmetically.
421 * Otherwise, calls the Ensembl REST service that maps from one assembly
422 * reference's coordinates to another's
425 * start-end chromosomal range in 'fromRef' coordinates
429 * assembly reference for the query coordinates
431 * assembly reference we wish to translate to
432 * @return the start-end range in 'toRef' coordinates
434 protected int[] mapReferenceRange(int[] queryRange, String chromosome,
435 String species, String fromRef, String toRef)
438 * first try shorcut of computing the mapping as a subregion of one
439 * we already have (e.g. for an exon, if we have the gene mapping)
441 int[] mappedRange = findSubsumedRangeMapping(queryRange, chromosome,
442 species, fromRef, toRef);
443 if (mappedRange != null)
449 * call (e.g.) http://rest.ensembl.org/map/human/GRCh38/17:45051610..45109016:1/GRCh37
451 EnsemblMap mapper = new EnsemblMap();
452 int[] mapping = mapper.getMapping(species, chromosome, fromRef, toRef,
457 // mapping service failure
462 * save mapping for possible future re-use
464 String key = makeRangesKey(chromosome, species, fromRef, toRef);
465 if (!assemblyMappings.containsKey(key))
467 assemblyMappings.put(key, new HashMap<int[], int[]>());
470 assemblyMappings.get(key).put(queryRange, mapping);
476 * If we already have a 1:1 contiguous mapping which subsumes the given query
477 * range, this method just calculates and returns the subset of that mapping,
478 * else it returns null. In practical terms, if a gene has a contiguous
479 * mapping between (for example) GRCh37 and GRCh38, then we assume that its
480 * subsidiary exons occupy unchanged relative positions, and just compute
481 * these as offsets, rather than do another lookup of the mapping.
483 * If in future these assumptions prove invalid (e.g. for bacterial dna?!),
484 * simply remove this method or let it always return null.
486 * Warning: many rapid calls to the /map service map result in a 429 overload
496 protected int[] findSubsumedRangeMapping(int[] queryRange, String chromosome,
497 String species, String fromRef, String toRef)
499 String key = makeRangesKey(chromosome, species, fromRef, toRef);
500 if (assemblyMappings.containsKey(key))
502 Map<int[], int[]> mappedRanges = assemblyMappings.get(key);
503 for (Entry<int[], int[]> mappedRange : mappedRanges.entrySet())
505 int[] fromRange = mappedRange.getKey();
506 int[] toRange = mappedRange.getValue();
507 if (fromRange[1] - fromRange[0] == toRange[1] - toRange[0])
510 * mapping is 1:1 in length, so we trust it to have no discontinuities
512 if (MappingUtils.rangeContains(fromRange, queryRange))
515 * fromRange subsumes our query range
517 int offset = queryRange[0] - fromRange[0];
518 int mappedRangeFrom = toRange[0] + offset;
519 int mappedRangeTo = mappedRangeFrom + (queryRange[1] - queryRange[0]);
520 return new int[] { mappedRangeFrom, mappedRangeTo };
529 * Transfers the sequence feature to the target sequence, locating its start
530 * and end range based on the mapping. Features which do not overlap the
531 * target sequence are ignored.
534 * @param targetSequence
536 * mapping from the feature's coordinates to the target sequence
538 protected void transferFeature(SequenceFeature sf,
539 SequenceI targetSequence, MapList mapping)
541 int[] mappedRange = mapping.locateInTo(sf.getBegin(), sf.getEnd());
543 if (mappedRange != null)
545 String group = sf.getFeatureGroup();
546 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
547 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
548 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
549 group, sf.getScore());
550 targetSequence.addSequenceFeature(copy);
555 * Formats a ranges map lookup key
563 protected static String makeRangesKey(String chromosome, String species,
564 String fromRef, String toRef)
566 return species + EXCL + chromosome + EXCL + fromRef + EXCL