1 package jalview.io.vcf;
3 import jalview.analysis.AlignmentUtils;
4 import jalview.analysis.Dna;
5 import jalview.api.AlignViewControllerGuiI;
6 import jalview.bin.Cache;
7 import jalview.datamodel.DBRefEntry;
8 import jalview.datamodel.GeneLociI;
9 import jalview.datamodel.Mapping;
10 import jalview.datamodel.SequenceFeature;
11 import jalview.datamodel.SequenceI;
12 import jalview.datamodel.features.FeatureAttributeType;
13 import jalview.datamodel.features.FeatureSource;
14 import jalview.datamodel.features.FeatureSources;
15 import jalview.ext.ensembl.EnsemblMap;
16 import jalview.ext.htsjdk.HtsContigDb;
17 import jalview.ext.htsjdk.VCFReader;
18 import jalview.io.gff.Gff3Helper;
19 import jalview.io.gff.SequenceOntologyI;
20 import jalview.util.MapList;
21 import jalview.util.MappingUtils;
22 import jalview.util.MessageManager;
25 import java.io.IOException;
26 import java.util.ArrayList;
27 import java.util.HashMap;
28 import java.util.List;
30 import java.util.Map.Entry;
31 import java.util.regex.Pattern;
32 import java.util.regex.PatternSyntaxException;
34 import htsjdk.samtools.SAMException;
35 import htsjdk.samtools.SAMSequenceDictionary;
36 import htsjdk.samtools.SAMSequenceRecord;
37 import htsjdk.samtools.util.CloseableIterator;
38 import htsjdk.variant.variantcontext.Allele;
39 import htsjdk.variant.variantcontext.VariantContext;
40 import htsjdk.variant.vcf.VCFHeader;
41 import htsjdk.variant.vcf.VCFHeaderLine;
42 import htsjdk.variant.vcf.VCFHeaderLineCount;
43 import htsjdk.variant.vcf.VCFHeaderLineType;
44 import htsjdk.variant.vcf.VCFInfoHeaderLine;
47 * A class to read VCF data (using the htsjdk) and add variants as sequence
48 * features on dna and any related protein product sequences
52 public class VCFLoader
54 private static final String NO_VALUE = ".";
56 private static final String DEFAULT_SPECIES = "homo_sapiens";
59 * A class to model the mapping from sequence to VCF coordinates. Cases include
61 * <li>a direct 1:1 mapping where the sequence is one of the VCF contigs</li>
62 * <li>a mapping of sequence to chromosomal coordinates, where sequence and VCF
63 * use the same reference assembly</li>
64 * <li>a modified mapping of sequence to chromosomal coordinates, where sequence
65 * and VCF use different reference assembles</li>
70 final String chromosome;
74 VCFMap(String chr, MapList m)
81 public String toString()
83 return chromosome + ":" + map.toString();
88 * Lookup keys, and default values, for Preference entries that describe
89 * patterns for VCF and VEP fields to capture
91 private static final String VEP_FIELDS_PREF = "VEP_FIELDS";
93 private static final String VCF_FIELDS_PREF = "VCF_FIELDS";
95 private static final String DEFAULT_VCF_FIELDS = ".*";
97 private static final String DEFAULT_VEP_FIELDS = ".*";// "Allele,Consequence,IMPACT,SWISSPROT,SIFT,PolyPhen,CLIN_SIG";
100 * Lookup keys, and default values, for Preference entries that give
101 * mappings from tokens in the 'reference' header to species or assembly
103 private static final String VCF_ASSEMBLY = "VCF_ASSEMBLY";
105 private static final String DEFAULT_VCF_ASSEMBLY = "assembly19=GRCh37,hs37=GRCh37,grch37=GRCh37,grch38=GRCh38";
107 private static final String VCF_SPECIES = "VCF_SPECIES"; // default is human
109 private static final String DEFAULT_REFERENCE = "grch37"; // fallback default is human GRCh37
112 * keys to fields of VEP CSQ consequence data
113 * see https://www.ensembl.org/info/docs/tools/vep/vep_formats.html
115 private static final String CSQ_CONSEQUENCE_KEY = "Consequence";
116 private static final String CSQ_ALLELE_KEY = "Allele";
117 private static final String CSQ_ALLELE_NUM_KEY = "ALLELE_NUM"; // 0 (ref), 1...
118 private static final String CSQ_FEATURE_KEY = "Feature"; // Ensembl stable id
121 * default VCF INFO key for VEP consequence data
122 * NB this can be overridden running VEP with --vcf_info_field
123 * - we don't handle this case (require identifier to be CSQ)
125 private static final String CSQ_FIELD = "CSQ";
128 * separator for fields in consequence data is '|'
130 private static final String PIPE_REGEX = "\\|";
133 * delimiter that separates multiple consequence data blocks
135 private static final String COMMA = ",";
138 * the feature group assigned to a VCF variant in Jalview
140 private static final String FEATURE_GROUP_VCF = "VCF";
143 * internal delimiter used to build keys for assemblyMappings
146 private static final String EXCL = "!";
149 * the VCF file we are processing
151 protected String vcfFilePath;
154 * mappings between VCF and sequence reference assembly regions, as
155 * key = "species!chromosome!fromAssembly!toAssembly
156 * value = Map{fromRange, toRange}
158 private Map<String, Map<int[], int[]>> assemblyMappings;
160 private VCFReader reader;
163 * holds details of the VCF header lines (metadata)
165 private VCFHeader header;
168 * species (as a valid Ensembl term) the VCF is for
170 private String vcfSpecies;
173 * genome assembly version (as a valid Ensembl identifier) the VCF is for
175 private String vcfAssembly;
178 * a Dictionary of contigs (if present) referenced in the VCF file
180 private SAMSequenceDictionary dictionary;
183 * the position (0...) of field in each block of
184 * CSQ (consequence) data (if declared in the VCF INFO header for CSQ)
185 * see http://www.ensembl.org/info/docs/tools/vep/vep_formats.html
187 private int csqConsequenceFieldIndex = -1;
188 private int csqAlleleFieldIndex = -1;
189 private int csqAlleleNumberFieldIndex = -1;
190 private int csqFeatureFieldIndex = -1;
192 // todo the same fields for SnpEff ANN data if wanted
193 // see http://snpeff.sourceforge.net/SnpEff_manual.html#input
196 * a unique identifier under which to save metadata about feature
197 * attributes (selected INFO field data)
199 private String sourceId;
202 * The INFO IDs of data that is both present in the VCF file, and
203 * also matched by any filters for data of interest
205 List<String> vcfFieldsOfInterest;
208 * The field offsets and identifiers for VEP (CSQ) data that is both present
209 * in the VCF file, and also matched by any filters for data of interest
210 * for example 0 -> Allele, 1 -> Consequence, ..., 36 -> SIFT, ...
212 Map<Integer, String> vepFieldsOfInterest;
215 * Constructor given a VCF file
219 public VCFLoader(String vcfFile)
224 } catch (IOException e)
226 System.err.println("Error opening VCF file: " + e.getMessage());
229 // map of species!chromosome!fromAssembly!toAssembly to {fromRange, toRange}
230 assemblyMappings = new HashMap<>();
234 * Starts a new thread to query and load VCF variant data on to the given
237 * This method is not thread safe - concurrent threads should use separate
238 * instances of this class.
243 public void loadVCF(SequenceI[] seqs, final AlignViewControllerGuiI gui)
247 gui.setStatus(MessageManager.getString("label.searching_vcf"));
255 VCFLoader.this.doLoad(seqs, gui);
261 * Reads the specified contig sequence and adds its VCF variants to it
264 * the id of a single sequence (contig) to load
267 public SequenceI loadVCFContig(String contig)
269 VCFHeaderLine headerLine = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
270 if (headerLine == null)
272 Cache.log.error("VCF reference header not found");
275 String ref = headerLine.getValue();
276 if (ref.startsWith("file://"))
278 ref = ref.substring(7);
280 setSpeciesAndAssembly(ref);
282 SequenceI seq = null;
283 File dbFile = new File(ref);
287 HtsContigDb db = new HtsContigDb("", dbFile);
288 seq = db.getSequenceProxy(contig);
289 loadSequenceVCF(seq);
294 Cache.log.error("VCF reference not found: " + ref);
301 * Loads VCF on to one or more sequences
305 * optional callback handler for messages
307 protected void doLoad(SequenceI[] seqs, AlignViewControllerGuiI gui)
311 VCFHeaderLine ref = header
312 .getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
313 String reference = ref == null ? null : ref.getValue();
315 setSpeciesAndAssembly(reference);
321 * query for VCF overlapping each sequence in turn
323 for (SequenceI seq : seqs)
325 int added = loadSequenceVCF(seq);
330 transferAddedFeatures(seq);
335 String msg = MessageManager.formatMessage("label.added_vcf",
338 if (gui.getFeatureSettingsUI() != null)
340 gui.getFeatureSettingsUI().discoverAllFeatureData();
343 } catch (Throwable e)
345 System.err.println("Error processing VCF: " + e.getMessage());
349 gui.setStatus("Error occurred - see console for details");
358 } catch (IOException e)
369 * Attempts to determine and save the species and genome assembly version to
370 * which the VCF data applies. This may be done by parsing the {@code reference}
371 * header line, configured in a property file, or (potentially) confirmed
372 * interactively by the user.
374 * The saved values should be identifiers valid for Ensembl's REST service
375 * {@code map} endpoint, so they can be used (if necessary) to retrieve the
376 * mapping between VCF coordinates and sequence coordinates.
379 * @see https://rest.ensembl.org/documentation/info/assembly_map
380 * @see https://rest.ensembl.org/info/assembly/human?content-type=text/xml
381 * @see https://rest.ensembl.org/info/species?content-type=text/xml
383 protected void setSpeciesAndAssembly(String reference)
385 if (reference == null)
387 Cache.log.error("No VCF ##reference found, defaulting to "
388 + DEFAULT_REFERENCE + ":" + DEFAULT_SPECIES);
389 reference = DEFAULT_REFERENCE; // default to GRCh37 if not specified
391 reference = reference.toLowerCase();
394 * for a non-human species, or other assembly identifier,
395 * specify as a Jalview property file entry e.g.
396 * VCF_ASSEMBLY = hs37=GRCh37,assembly19=GRCh37
397 * VCF_SPECIES = c_elegans=celegans
398 * to map a token in the reference header to a value
400 String prop = Cache.getDefault(VCF_ASSEMBLY, DEFAULT_VCF_ASSEMBLY);
401 for (String token : prop.split(","))
403 String[] tokens = token.split("=");
404 if (tokens.length == 2)
406 if (reference.contains(tokens[0].trim().toLowerCase()))
408 vcfAssembly = tokens[1].trim();
414 vcfSpecies = DEFAULT_SPECIES;
415 prop = Cache.getProperty(VCF_SPECIES);
418 for (String token : prop.split(","))
420 String[] tokens = token.split("=");
421 if (tokens.length == 2)
423 if (reference.contains(tokens[0].trim().toLowerCase()))
425 vcfSpecies = tokens[1].trim();
434 * Opens the VCF file and parses header data
437 * @throws IOException
439 private void initialise(String filePath) throws IOException
441 vcfFilePath = filePath;
443 reader = new VCFReader(filePath);
445 header = reader.getFileHeader();
449 dictionary = header.getSequenceDictionary();
450 } catch (SAMException e)
452 // ignore - thrown if any contig line lacks length info
457 saveMetadata(sourceId);
460 * get offset of CSQ ALLELE_NUM and Feature if declared
466 * Reads metadata (such as INFO field descriptions and datatypes) and saves
467 * them for future reference
471 void saveMetadata(String theSourceId)
473 List<Pattern> vcfFieldPatterns = getFieldMatchers(VCF_FIELDS_PREF,
475 vcfFieldsOfInterest = new ArrayList<>();
477 FeatureSource metadata = new FeatureSource(theSourceId);
479 for (VCFInfoHeaderLine info : header.getInfoHeaderLines())
481 String attributeId = info.getID();
482 String desc = info.getDescription();
483 VCFHeaderLineType type = info.getType();
484 FeatureAttributeType attType = null;
488 attType = FeatureAttributeType.Character;
491 attType = FeatureAttributeType.Flag;
494 attType = FeatureAttributeType.Float;
497 attType = FeatureAttributeType.Integer;
500 attType = FeatureAttributeType.String;
503 metadata.setAttributeName(attributeId, desc);
504 metadata.setAttributeType(attributeId, attType);
506 if (isFieldWanted(attributeId, vcfFieldPatterns))
508 vcfFieldsOfInterest.add(attributeId);
512 FeatureSources.getInstance().addSource(theSourceId, metadata);
516 * Answers true if the field id is matched by any of the filter patterns, else
517 * false. Matching is against regular expression patterns, and is not
524 private boolean isFieldWanted(String id, List<Pattern> filters)
526 for (Pattern p : filters)
528 if (p.matcher(id.toUpperCase()).matches())
537 * Records 'wanted' fields defined in the CSQ INFO header (if there is one).
538 * Also records the position of selected fields (Allele, ALLELE_NUM, Feature)
539 * required for processing.
541 * CSQ fields are declared in the CSQ INFO Description e.g.
543 * Description="Consequence ...from ... VEP. Format: Allele|Consequence|...
545 protected void parseCsqHeader()
547 List<Pattern> vepFieldFilters = getFieldMatchers(VEP_FIELDS_PREF,
549 vepFieldsOfInterest = new HashMap<>();
551 VCFInfoHeaderLine csqInfo = header.getInfoHeaderLine(CSQ_FIELD);
558 * parse out the pipe-separated list of CSQ fields; we assume here that
559 * these form the last part of the description, and contain no spaces
561 String desc = csqInfo.getDescription();
562 int spacePos = desc.lastIndexOf(" ");
563 desc = desc.substring(spacePos + 1);
567 String[] format = desc.split(PIPE_REGEX);
569 for (String field : format)
571 if (CSQ_CONSEQUENCE_KEY.equals(field))
573 csqConsequenceFieldIndex = index;
575 if (CSQ_ALLELE_NUM_KEY.equals(field))
577 csqAlleleNumberFieldIndex = index;
579 if (CSQ_ALLELE_KEY.equals(field))
581 csqAlleleFieldIndex = index;
583 if (CSQ_FEATURE_KEY.equals(field))
585 csqFeatureFieldIndex = index;
588 if (isFieldWanted(field, vepFieldFilters))
590 vepFieldsOfInterest.put(index, field);
599 * Reads the Preference value for the given key, with default specified if no
600 * preference set. The value is interpreted as a comma-separated list of
601 * regular expressions, and converted into a list of compiled patterns ready
602 * for matching. Patterns are forced to upper-case for non-case-sensitive
605 * This supports user-defined filters for fields of interest to capture while
606 * processing data. For example, VCF_FIELDS = AF,AC* would mean that VCF INFO
607 * fields with an ID of AF, or starting with AC, would be matched.
613 private List<Pattern> getFieldMatchers(String key, String def)
615 String pref = Cache.getDefault(key, def);
616 List<Pattern> patterns = new ArrayList<>();
617 String[] tokens = pref.split(",");
618 for (String token : tokens)
622 patterns.add(Pattern.compile(token.toUpperCase()));
623 } catch (PatternSyntaxException e)
625 System.err.println("Invalid pattern ignored: " + token);
632 * Transfers VCF features to sequences to which this sequence has a mapping.
633 * If the mapping is 3:1, computes peptide variants from nucleotide variants.
637 protected void transferAddedFeatures(SequenceI seq)
639 DBRefEntry[] dbrefs = seq.getDBRefs();
644 for (DBRefEntry dbref : dbrefs)
646 Mapping mapping = dbref.getMap();
647 if (mapping == null || mapping.getTo() == null)
652 SequenceI mapTo = mapping.getTo();
653 MapList map = mapping.getMap();
654 if (map.getFromRatio() == 3)
657 * dna-to-peptide product mapping
659 AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
664 * nucleotide-to-nucleotide mapping e.g. transcript to CDS
666 List<SequenceFeature> features = seq.getFeatures()
667 .getPositionalFeatures(SequenceOntologyI.SEQUENCE_VARIANT);
668 for (SequenceFeature sf : features)
670 if (FEATURE_GROUP_VCF.equals(sf.getFeatureGroup()))
672 transferFeature(sf, mapTo, map);
680 * Tries to add overlapping variants read from a VCF file to the given sequence,
681 * and returns the number of variant features added
686 protected int loadSequenceVCF(SequenceI seq)
688 VCFMap vcfMap = getVcfMap(seq);
695 * work with the dataset sequence here
697 SequenceI dss = seq.getDatasetSequence();
702 return addVcfVariants(dss, vcfMap);
706 * Answers a map from sequence coordinates to VCF chromosome ranges
711 private VCFMap getVcfMap(SequenceI seq)
714 * simplest case: sequence has id and length matching a VCF contig
716 VCFMap vcfMap = null;
717 if (dictionary != null)
719 vcfMap = getContigMap(seq);
727 * otherwise, map to VCF from chromosomal coordinates
728 * of the sequence (if known)
730 GeneLociI seqCoords = seq.getGeneLoci();
731 if (seqCoords == null)
733 Cache.log.warn(String.format(
734 "Can't query VCF for %s as chromosome coordinates not known",
739 String species = seqCoords.getSpeciesId();
740 String chromosome = seqCoords.getChromosomeId();
741 String seqRef = seqCoords.getAssemblyId();
742 MapList map = seqCoords.getMapping();
744 // note this requires the configured species to match that
745 // returned with the Ensembl sequence; todo: support aliases?
746 if (!vcfSpecies.equalsIgnoreCase(species))
748 Cache.log.warn("No VCF loaded to " + seq.getName()
749 + " as species not matched");
753 if (seqRef.equalsIgnoreCase(vcfAssembly))
755 return new VCFMap(chromosome, map);
759 * VCF data has a different reference assembly to the sequence:
760 * query Ensembl to map chromosomal coordinates from sequence to VCF
762 List<int[]> toVcfRanges = new ArrayList<>();
763 List<int[]> fromSequenceRanges = new ArrayList<>();
765 for (int[] range : map.getToRanges())
767 int[] fromRange = map.locateInFrom(range[0], range[1]);
768 if (fromRange == null)
774 int[] newRange = mapReferenceRange(range, chromosome, "human", seqRef,
776 if (newRange == null)
779 String.format("Failed to map %s:%s:%s:%d:%d to %s", species,
780 chromosome, seqRef, range[0], range[1],
786 toVcfRanges.add(newRange);
787 fromSequenceRanges.add(fromRange);
791 return new VCFMap(chromosome,
792 new MapList(fromSequenceRanges, toVcfRanges, 1, 1));
796 * If the sequence id matches a contig declared in the VCF file, and the
797 * sequence length matches the contig length, then returns a 1:1 map of the
798 * sequence to the contig, else returns null
803 private VCFMap getContigMap(SequenceI seq)
805 String id = seq.getName();
806 SAMSequenceRecord contig = dictionary.getSequence(id);
809 int len = seq.getLength();
810 if (len == contig.getSequenceLength())
812 MapList map = new MapList(new int[] { 1, len },
815 return new VCFMap(id, map);
822 * Queries the VCF reader for any variants that overlap the mapped chromosome
823 * ranges of the sequence, and adds as variant features. Returns the number of
824 * overlapping variants found.
828 * mapping from sequence to VCF coordinates
831 protected int addVcfVariants(SequenceI seq, VCFMap map)
833 boolean forwardStrand = map.map.isToForwardStrand();
836 * query the VCF for overlaps of each contiguous chromosomal region
840 for (int[] range : map.map.getToRanges())
842 int vcfStart = Math.min(range[0], range[1]);
843 int vcfEnd = Math.max(range[0], range[1]);
844 CloseableIterator<VariantContext> variants = reader
845 .query(map.chromosome, vcfStart, vcfEnd);
846 while (variants.hasNext())
848 VariantContext variant = variants.next();
850 int[] featureRange = map.map.locateInFrom(variant.getStart(),
853 if (featureRange != null)
855 int featureStart = Math.min(featureRange[0], featureRange[1]);
856 int featureEnd = Math.max(featureRange[0], featureRange[1]);
857 count += addAlleleFeatures(seq, variant, featureStart, featureEnd,
868 * A convenience method to get an attribute value for an alternate allele
871 * @param attributeName
875 protected String getAttributeValue(VariantContext variant,
876 String attributeName, int alleleIndex)
878 Object att = variant.getAttribute(attributeName);
880 if (att instanceof String)
882 return NO_VALUE.equals(att) ? null : (String) att;
884 else if (att instanceof ArrayList)
886 return ((List<String>) att).get(alleleIndex);
893 * Adds one variant feature for each allele in the VCF variant record, and
894 * returns the number of features added.
898 * @param featureStart
900 * @param forwardStrand
903 protected int addAlleleFeatures(SequenceI seq, VariantContext variant,
904 int featureStart, int featureEnd, boolean forwardStrand)
909 * Javadoc says getAlternateAlleles() imposes no order on the list returned
910 * so we proceed defensively to get them in strict order
912 int altAlleleCount = variant.getAlternateAlleles().size();
913 for (int i = 0; i < altAlleleCount; i++)
915 added += addAlleleFeature(seq, variant, i, featureStart, featureEnd,
922 * Inspects one allele and attempts to add a variant feature for it to the
923 * sequence. The additional data associated with this allele is extracted to
924 * store in the feature's key-value map. Answers the number of features added (0
929 * @param altAlleleIndex
931 * @param featureStart
933 * @param forwardStrand
936 protected int addAlleleFeature(SequenceI seq, VariantContext variant,
937 int altAlleleIndex, int featureStart, int featureEnd,
938 boolean forwardStrand)
940 String reference = variant.getReference().getBaseString();
941 Allele alt = variant.getAlternateAllele(altAlleleIndex);
942 String allele = alt.getBaseString();
945 * insertion after a genomic base, if on reverse strand, has to be
946 * converted to insertion of complement after the preceding position
948 int referenceLength = reference.length();
949 if (!forwardStrand && allele.length() > referenceLength
950 && allele.startsWith(reference))
952 featureStart -= referenceLength;
953 featureEnd = featureStart;
954 char insertAfter = seq.getCharAt(featureStart - seq.getStart());
955 reference = Dna.reverseComplement(String.valueOf(insertAfter));
956 allele = allele.substring(referenceLength) + reference;
960 * build the ref,alt allele description e.g. "G,A", using the base
961 * complement if the sequence is on the reverse strand
963 StringBuilder sb = new StringBuilder();
964 sb.append(forwardStrand ? reference : Dna.reverseComplement(reference));
966 sb.append(forwardStrand ? allele : Dna.reverseComplement(allele));
967 String alleles = sb.toString(); // e.g. G,A
970 * pick out the consequence data (if any) that is for the current allele
971 * and feature (transcript) that matches the current sequence
973 String consequence = getConsequenceForAlleleAndFeature(variant, CSQ_FIELD,
974 altAlleleIndex, csqAlleleFieldIndex,
975 csqAlleleNumberFieldIndex, seq.getName().toLowerCase(),
976 csqFeatureFieldIndex);
979 * pick out the ontology term for the consequence type
981 String type = SequenceOntologyI.SEQUENCE_VARIANT;
982 if (consequence != null)
984 type = getOntologyTerm(consequence);
987 SequenceFeature sf = new SequenceFeature(type, alleles, featureStart,
988 featureEnd, FEATURE_GROUP_VCF);
989 sf.setSource(sourceId);
991 sf.setValue(Gff3Helper.ALLELES, alleles);
993 addAlleleProperties(variant, sf, altAlleleIndex, consequence);
995 seq.addSequenceFeature(sf);
1001 * Determines the Sequence Ontology term to use for the variant feature type in
1002 * Jalview. The default is 'sequence_variant', but a more specific term is used
1005 * <li>VEP (or SnpEff) Consequence annotation is included in the VCF</li>
1006 * <li>sequence id can be matched to VEP Feature (or SnpEff Feature_ID)</li>
1009 * @param consequence
1011 * @see http://www.sequenceontology.org/browser/current_svn/term/SO:0001060
1013 String getOntologyTerm(String consequence)
1015 String type = SequenceOntologyI.SEQUENCE_VARIANT;
1018 * could we associate Consequence data with this allele and feature (transcript)?
1019 * if so, prefer the consequence term from that data
1021 if (csqAlleleFieldIndex == -1) // && snpEffAlleleFieldIndex == -1
1024 * no Consequence data so we can't refine the ontology term
1029 if (consequence != null)
1031 String[] csqFields = consequence.split(PIPE_REGEX);
1032 if (csqFields.length > csqConsequenceFieldIndex)
1034 type = csqFields[csqConsequenceFieldIndex];
1039 // todo the same for SnpEff consequence data matching if wanted
1043 * if of the form (e.g.) missense_variant&splice_region_variant,
1044 * just take the first ('most severe') consequence
1048 int pos = type.indexOf('&');
1051 type = type.substring(0, pos);
1058 * Returns matched consequence data if it can be found, else null.
1060 * <li>inspects the VCF data for key 'vcfInfoId'</li>
1061 * <li>splits this on comma (to distinct consequences)</li>
1062 * <li>returns the first consequence (if any) where</li>
1064 * <li>the allele matches the altAlleleIndex'th allele of variant</li>
1065 * <li>the feature matches the sequence name (e.g. transcript id)</li>
1068 * If matched, the consequence is returned (as pipe-delimited fields).
1072 * @param altAlleleIndex
1073 * @param alleleFieldIndex
1074 * @param alleleNumberFieldIndex
1076 * @param featureFieldIndex
1079 private String getConsequenceForAlleleAndFeature(VariantContext variant,
1080 String vcfInfoId, int altAlleleIndex, int alleleFieldIndex,
1081 int alleleNumberFieldIndex,
1082 String seqName, int featureFieldIndex)
1084 if (alleleFieldIndex == -1 || featureFieldIndex == -1)
1088 Object value = variant.getAttribute(vcfInfoId);
1090 if (value == null || !(value instanceof List<?>))
1096 * inspect each consequence in turn (comma-separated blocks
1097 * extracted by htsjdk)
1099 List<String> consequences = (List<String>) value;
1101 for (String consequence : consequences)
1103 String[] csqFields = consequence.split(PIPE_REGEX);
1104 if (csqFields.length > featureFieldIndex)
1106 String featureIdentifier = csqFields[featureFieldIndex];
1107 if (featureIdentifier.length() > 4
1108 && seqName.indexOf(featureIdentifier.toLowerCase()) > -1)
1111 * feature (transcript) matched - now check for allele match
1113 if (matchAllele(variant, altAlleleIndex, csqFields,
1114 alleleFieldIndex, alleleNumberFieldIndex))
1124 private boolean matchAllele(VariantContext variant, int altAlleleIndex,
1125 String[] csqFields, int alleleFieldIndex,
1126 int alleleNumberFieldIndex)
1129 * if ALLELE_NUM is present, it must match altAlleleIndex
1130 * NB first alternate allele is 1 for ALLELE_NUM, 0 for altAlleleIndex
1132 if (alleleNumberFieldIndex > -1)
1134 if (csqFields.length <= alleleNumberFieldIndex)
1138 String alleleNum = csqFields[alleleNumberFieldIndex];
1139 return String.valueOf(altAlleleIndex + 1).equals(alleleNum);
1143 * else consequence allele must match variant allele
1145 if (alleleFieldIndex > -1 && csqFields.length > alleleFieldIndex)
1147 String csqAllele = csqFields[alleleFieldIndex];
1148 String vcfAllele = variant.getAlternateAllele(altAlleleIndex)
1150 return csqAllele.equals(vcfAllele);
1156 * Add any allele-specific VCF key-value data to the sequence feature
1160 * @param altAlelleIndex
1162 * @param consequence
1163 * if not null, the consequence specific to this sequence (transcript
1164 * feature) and allele
1166 protected void addAlleleProperties(VariantContext variant,
1167 SequenceFeature sf, final int altAlelleIndex, String consequence)
1169 Map<String, Object> atts = variant.getAttributes();
1171 for (Entry<String, Object> att : atts.entrySet())
1173 String key = att.getKey();
1176 * extract Consequence data (if present) that we are able to
1177 * associated with the allele for this variant feature
1179 if (CSQ_FIELD.equals(key))
1181 addConsequences(variant, sf, consequence);
1186 * filter out fields we don't want to capture
1188 if (!vcfFieldsOfInterest.contains(key))
1194 * filter out fields we don't want to capture
1196 if (!vcfFieldsOfInterest.contains(key))
1202 * we extract values for other data which are allele-specific;
1203 * these may be per alternate allele (INFO[key].Number = 'A')
1204 * or per allele including reference (INFO[key].Number = 'R')
1206 VCFInfoHeaderLine infoHeader = header.getInfoHeaderLine(key);
1207 if (infoHeader == null)
1210 * can't be sure what data belongs to this allele, so
1211 * play safe and don't take any
1216 VCFHeaderLineCount number = infoHeader.getCountType();
1217 int index = altAlelleIndex;
1218 if (number == VCFHeaderLineCount.R)
1221 * one value per allele including reference, so bump index
1222 * e.g. the 3rd value is for the 2nd alternate allele
1226 else if (number != VCFHeaderLineCount.A)
1229 * don't save other values as not allele-related
1235 * take the index'th value
1237 String value = getAttributeValue(variant, key, index);
1240 sf.setValue(key, value);
1246 * Inspects CSQ data blocks (consequences) and adds attributes on the sequence
1249 * If <code>myConsequence</code> is not null, then this is the specific
1250 * consequence data (pipe-delimited fields) that is for the current allele and
1251 * transcript (sequence) being processed)
1255 * @param myConsequence
1257 protected void addConsequences(VariantContext variant, SequenceFeature sf,
1258 String myConsequence)
1260 Object value = variant.getAttribute(CSQ_FIELD);
1262 if (value == null || !(value instanceof List<?>))
1267 List<String> consequences = (List<String>) value;
1270 * inspect CSQ consequences; restrict to the consequence
1271 * associated with the current transcript (Feature)
1273 Map<String, String> csqValues = new HashMap<>();
1275 for (String consequence : consequences)
1277 if (myConsequence == null || myConsequence.equals(consequence))
1279 String[] csqFields = consequence.split(PIPE_REGEX);
1282 * inspect individual fields of this consequence, copying non-null
1283 * values which are 'fields of interest'
1286 for (String field : csqFields)
1288 if (field != null && field.length() > 0)
1290 String id = vepFieldsOfInterest.get(i);
1293 csqValues.put(id, field);
1301 if (!csqValues.isEmpty())
1303 sf.setValue(CSQ_FIELD, csqValues);
1308 * A convenience method to complement a dna base and return the string value
1314 protected String complement(byte[] reference)
1316 return String.valueOf(Dna.getComplement((char) reference[0]));
1320 * Determines the location of the query range (chromosome positions) in a
1321 * different reference assembly.
1323 * If the range is just a subregion of one for which we already have a mapping
1324 * (for example, an exon sub-region of a gene), then the mapping is just
1325 * computed arithmetically.
1327 * Otherwise, calls the Ensembl REST service that maps from one assembly
1328 * reference's coordinates to another's
1331 * start-end chromosomal range in 'fromRef' coordinates
1335 * assembly reference for the query coordinates
1337 * assembly reference we wish to translate to
1338 * @return the start-end range in 'toRef' coordinates
1340 protected int[] mapReferenceRange(int[] queryRange, String chromosome,
1341 String species, String fromRef, String toRef)
1344 * first try shorcut of computing the mapping as a subregion of one
1345 * we already have (e.g. for an exon, if we have the gene mapping)
1347 int[] mappedRange = findSubsumedRangeMapping(queryRange, chromosome,
1348 species, fromRef, toRef);
1349 if (mappedRange != null)
1355 * call (e.g.) http://rest.ensembl.org/map/human/GRCh38/17:45051610..45109016:1/GRCh37
1357 EnsemblMap mapper = new EnsemblMap();
1358 int[] mapping = mapper.getAssemblyMapping(species, chromosome, fromRef,
1361 if (mapping == null)
1363 // mapping service failure
1368 * save mapping for possible future re-use
1370 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1371 if (!assemblyMappings.containsKey(key))
1373 assemblyMappings.put(key, new HashMap<int[], int[]>());
1376 assemblyMappings.get(key).put(queryRange, mapping);
1382 * If we already have a 1:1 contiguous mapping which subsumes the given query
1383 * range, this method just calculates and returns the subset of that mapping,
1384 * else it returns null. In practical terms, if a gene has a contiguous
1385 * mapping between (for example) GRCh37 and GRCh38, then we assume that its
1386 * subsidiary exons occupy unchanged relative positions, and just compute
1387 * these as offsets, rather than do another lookup of the mapping.
1389 * If in future these assumptions prove invalid (e.g. for bacterial dna?!),
1390 * simply remove this method or let it always return null.
1392 * Warning: many rapid calls to the /map service map result in a 429 overload
1402 protected int[] findSubsumedRangeMapping(int[] queryRange, String chromosome,
1403 String species, String fromRef, String toRef)
1405 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1406 if (assemblyMappings.containsKey(key))
1408 Map<int[], int[]> mappedRanges = assemblyMappings.get(key);
1409 for (Entry<int[], int[]> mappedRange : mappedRanges.entrySet())
1411 int[] fromRange = mappedRange.getKey();
1412 int[] toRange = mappedRange.getValue();
1413 if (fromRange[1] - fromRange[0] == toRange[1] - toRange[0])
1416 * mapping is 1:1 in length, so we trust it to have no discontinuities
1418 if (MappingUtils.rangeContains(fromRange, queryRange))
1421 * fromRange subsumes our query range
1423 int offset = queryRange[0] - fromRange[0];
1424 int mappedRangeFrom = toRange[0] + offset;
1425 int mappedRangeTo = mappedRangeFrom + (queryRange[1] - queryRange[0]);
1426 return new int[] { mappedRangeFrom, mappedRangeTo };
1435 * Transfers the sequence feature to the target sequence, locating its start
1436 * and end range based on the mapping. Features which do not overlap the
1437 * target sequence are ignored.
1440 * @param targetSequence
1442 * mapping from the feature's coordinates to the target sequence
1444 protected void transferFeature(SequenceFeature sf,
1445 SequenceI targetSequence, MapList mapping)
1447 int[] mappedRange = mapping.locateInTo(sf.getBegin(), sf.getEnd());
1449 if (mappedRange != null)
1451 String group = sf.getFeatureGroup();
1452 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
1453 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
1454 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
1455 group, sf.getScore());
1456 targetSequence.addSequenceFeature(copy);
1461 * Formats a ranges map lookup key
1469 protected static String makeRangesKey(String chromosome, String species,
1470 String fromRef, String toRef)
1472 return species + EXCL + chromosome + EXCL + fromRef + EXCL