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 DEFAULT_SPECIES = "homo_sapiens";
57 * A class to model the mapping from sequence to VCF coordinates. Cases include
59 * <li>a direct 1:1 mapping where the sequence is one of the VCF contigs</li>
60 * <li>a mapping of sequence to chromosomal coordinates, where sequence and VCF
61 * use the same reference assembly</li>
62 * <li>a modified mapping of sequence to chromosomal coordinates, where sequence
63 * and VCF use different reference assembles</li>
68 final String chromosome;
72 VCFMap(String chr, MapList m)
79 public String toString()
81 return chromosome + ":" + map.toString();
86 * Lookup keys, and default values, for Preference entries that describe
87 * patterns for VCF and VEP fields to capture
89 private static final String VEP_FIELDS_PREF = "VEP_FIELDS";
91 private static final String VCF_FIELDS_PREF = "VCF_FIELDS";
93 private static final String DEFAULT_VCF_FIELDS = ".*";
95 private static final String DEFAULT_VEP_FIELDS = ".*";// "Allele,Consequence,IMPACT,SWISSPROT,SIFT,PolyPhen,CLIN_SIG";
98 * Lookup keys, and default values, for Preference entries that give
99 * mappings from tokens in the 'reference' header to species or assembly
101 private static final String VCF_ASSEMBLY = "VCF_ASSEMBLY";
103 private static final String DEFAULT_VCF_ASSEMBLY = "assembly19=GRCh37,hs37=GRCh37,grch37=GRCh37,grch38=GRCh38";
105 private static final String VCF_SPECIES = "VCF_SPECIES"; // default is human
107 private static final String DEFAULT_REFERENCE = "grch37"; // fallback default is human GRCh37
110 * keys to fields of VEP CSQ consequence data
111 * see https://www.ensembl.org/info/docs/tools/vep/vep_formats.html
113 private static final String CSQ_CONSEQUENCE_KEY = "Consequence";
114 private static final String CSQ_ALLELE_KEY = "Allele";
115 private static final String CSQ_ALLELE_NUM_KEY = "ALLELE_NUM"; // 0 (ref), 1...
116 private static final String CSQ_FEATURE_KEY = "Feature"; // Ensembl stable id
119 * default VCF INFO key for VEP consequence data
120 * NB this can be overridden running VEP with --vcf_info_field
121 * - we don't handle this case (require identifier to be CSQ)
123 private static final String CSQ_FIELD = "CSQ";
126 * separator for fields in consequence data is '|'
128 private static final String PIPE_REGEX = "\\|";
131 * delimiter that separates multiple consequence data blocks
133 private static final String COMMA = ",";
136 * the feature group assigned to a VCF variant in Jalview
138 private static final String FEATURE_GROUP_VCF = "VCF";
141 * internal delimiter used to build keys for assemblyMappings
144 private static final String EXCL = "!";
147 * the VCF file we are processing
149 protected String vcfFilePath;
152 * mappings between VCF and sequence reference assembly regions, as
153 * key = "species!chromosome!fromAssembly!toAssembly
154 * value = Map{fromRange, toRange}
156 private Map<String, Map<int[], int[]>> assemblyMappings;
158 private VCFReader reader;
161 * holds details of the VCF header lines (metadata)
163 private VCFHeader header;
166 * species (as a valid Ensembl term) the VCF is for
168 private String vcfSpecies;
171 * genome assembly version (as a valid Ensembl identifier) the VCF is for
173 private String vcfAssembly;
176 * a Dictionary of contigs (if present) referenced in the VCF file
178 private SAMSequenceDictionary dictionary;
181 * the position (0...) of field in each block of
182 * CSQ (consequence) data (if declared in the VCF INFO header for CSQ)
183 * see http://www.ensembl.org/info/docs/tools/vep/vep_formats.html
185 private int csqConsequenceFieldIndex = -1;
186 private int csqAlleleFieldIndex = -1;
187 private int csqAlleleNumberFieldIndex = -1;
188 private int csqFeatureFieldIndex = -1;
190 // todo the same fields for SnpEff ANN data if wanted
191 // see http://snpeff.sourceforge.net/SnpEff_manual.html#input
194 * a unique identifier under which to save metadata about feature
195 * attributes (selected INFO field data)
197 private String sourceId;
200 * The INFO IDs of data that is both present in the VCF file, and
201 * also matched by any filters for data of interest
203 List<String> vcfFieldsOfInterest;
206 * The field offsets and identifiers for VEP (CSQ) data that is both present
207 * in the VCF file, and also matched by any filters for data of interest
208 * for example 0 -> Allele, 1 -> Consequence, ..., 36 -> SIFT, ...
210 Map<Integer, String> vepFieldsOfInterest;
213 * Constructor given a VCF file
217 public VCFLoader(String vcfFile)
222 } catch (IOException e)
224 System.err.println("Error opening VCF file: " + e.getMessage());
227 // map of species!chromosome!fromAssembly!toAssembly to {fromRange, toRange}
228 assemblyMappings = new HashMap<>();
232 * Starts a new thread to query and load VCF variant data on to the given
235 * This method is not thread safe - concurrent threads should use separate
236 * instances of this class.
241 public void loadVCF(SequenceI[] seqs, final AlignViewControllerGuiI gui)
245 gui.setStatus(MessageManager.getString("label.searching_vcf"));
253 VCFLoader.this.doLoad(seqs, gui);
259 * Reads the specified contig sequence and adds its VCF variants to it
262 * the id of a single sequence (contig) to load
265 public SequenceI loadVCFContig(String contig)
267 VCFHeaderLine headerLine = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
268 if (headerLine == null)
270 Cache.log.error("VCF reference header not found");
273 String ref = headerLine.getValue();
274 if (ref.startsWith("file://"))
276 ref = ref.substring(7);
278 setSpeciesAndAssembly(ref);
280 SequenceI seq = null;
281 File dbFile = new File(ref);
285 HtsContigDb db = new HtsContigDb("", dbFile);
286 seq = db.getSequenceProxy(contig);
287 loadSequenceVCF(seq);
292 Cache.log.error("VCF reference not found: " + ref);
299 * Loads VCF on to one or more sequences
303 * optional callback handler for messages
305 protected void doLoad(SequenceI[] seqs, AlignViewControllerGuiI gui)
309 VCFHeaderLine ref = header
310 .getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
311 String reference = ref == null ? null : ref.getValue();
313 setSpeciesAndAssembly(reference);
319 * query for VCF overlapping each sequence in turn
321 for (SequenceI seq : seqs)
323 int added = loadSequenceVCF(seq);
328 transferAddedFeatures(seq);
333 String msg = MessageManager.formatMessage("label.added_vcf",
336 if (gui.getFeatureSettingsUI() != null)
338 gui.getFeatureSettingsUI().discoverAllFeatureData();
341 } catch (Throwable e)
343 System.err.println("Error processing VCF: " + e.getMessage());
347 gui.setStatus("Error occurred - see console for details");
356 } catch (IOException e)
367 * Attempts to determine and save the species and genome assembly version to
368 * which the VCF data applies. This may be done by parsing the {@code reference}
369 * header line, configured in a property file, or (potentially) confirmed
370 * interactively by the user.
372 * The saved values should be identifiers valid for Ensembl's REST service
373 * {@code map} endpoint, so they can be used (if necessary) to retrieve the
374 * mapping between VCF coordinates and sequence coordinates.
377 * @see https://rest.ensembl.org/documentation/info/assembly_map
378 * @see https://rest.ensembl.org/info/assembly/human?content-type=text/xml
379 * @see https://rest.ensembl.org/info/species?content-type=text/xml
381 protected void setSpeciesAndAssembly(String reference)
383 if (reference == null)
385 reference = DEFAULT_REFERENCE; // default to GRCh37 if not specified
387 reference = reference.toLowerCase();
390 * for a non-human species, or other assembly identifier,
391 * specify as a Jalview property file entry e.g.
392 * VCF_ASSEMBLY = hs37=GRCh37,assembly19=GRCh37
393 * VCF_SPECIES = c_elegans=celegans
394 * to map a token in the reference header to a value
396 String prop = Cache.getDefault(VCF_ASSEMBLY, DEFAULT_VCF_ASSEMBLY);
397 for (String token : prop.split(","))
399 String[] tokens = token.split("=");
400 if (tokens.length == 2)
402 if (reference.contains(tokens[0].trim().toLowerCase()))
404 vcfAssembly = tokens[1].trim();
410 vcfSpecies = DEFAULT_SPECIES;
411 prop = Cache.getProperty(VCF_SPECIES);
414 for (String token : prop.split(","))
416 String[] tokens = token.split("=");
417 if (tokens.length == 2)
419 if (reference.contains(tokens[0].trim().toLowerCase()))
421 vcfSpecies = tokens[1].trim();
430 * Opens the VCF file and parses header data
433 * @throws IOException
435 private void initialise(String filePath) throws IOException
437 vcfFilePath = filePath;
439 reader = new VCFReader(filePath);
441 header = reader.getFileHeader();
445 dictionary = header.getSequenceDictionary();
446 } catch (SAMException e)
448 // ignore - thrown if any contig line lacks length info
453 saveMetadata(sourceId);
456 * get offset of CSQ ALLELE_NUM and Feature if declared
462 * Reads metadata (such as INFO field descriptions and datatypes) and saves
463 * them for future reference
467 void saveMetadata(String theSourceId)
469 List<Pattern> vcfFieldPatterns = getFieldMatchers(VCF_FIELDS_PREF,
471 vcfFieldsOfInterest = new ArrayList<>();
473 FeatureSource metadata = new FeatureSource(theSourceId);
475 for (VCFInfoHeaderLine info : header.getInfoHeaderLines())
477 String attributeId = info.getID();
478 String desc = info.getDescription();
479 VCFHeaderLineType type = info.getType();
480 FeatureAttributeType attType = null;
484 attType = FeatureAttributeType.Character;
487 attType = FeatureAttributeType.Flag;
490 attType = FeatureAttributeType.Float;
493 attType = FeatureAttributeType.Integer;
496 attType = FeatureAttributeType.String;
499 metadata.setAttributeName(attributeId, desc);
500 metadata.setAttributeType(attributeId, attType);
502 if (isFieldWanted(attributeId, vcfFieldPatterns))
504 vcfFieldsOfInterest.add(attributeId);
508 FeatureSources.getInstance().addSource(theSourceId, metadata);
512 * Answers true if the field id is matched by any of the filter patterns, else
513 * false. Matching is against regular expression patterns, and is not
520 private boolean isFieldWanted(String id, List<Pattern> filters)
522 for (Pattern p : filters)
524 if (p.matcher(id.toUpperCase()).matches())
533 * Records 'wanted' fields defined in the CSQ INFO header (if there is one).
534 * Also records the position of selected fields (Allele, ALLELE_NUM, Feature)
535 * required for processing.
537 * CSQ fields are declared in the CSQ INFO Description e.g.
539 * Description="Consequence ...from ... VEP. Format: Allele|Consequence|...
541 protected void parseCsqHeader()
543 List<Pattern> vepFieldFilters = getFieldMatchers(VEP_FIELDS_PREF,
545 vepFieldsOfInterest = new HashMap<>();
547 VCFInfoHeaderLine csqInfo = header.getInfoHeaderLine(CSQ_FIELD);
554 * parse out the pipe-separated list of CSQ fields; we assume here that
555 * these form the last part of the description, and contain no spaces
557 String desc = csqInfo.getDescription();
558 int spacePos = desc.lastIndexOf(" ");
559 desc = desc.substring(spacePos + 1);
563 String[] format = desc.split(PIPE_REGEX);
565 for (String field : format)
567 if (CSQ_CONSEQUENCE_KEY.equals(field))
569 csqConsequenceFieldIndex = index;
571 if (CSQ_ALLELE_NUM_KEY.equals(field))
573 csqAlleleNumberFieldIndex = index;
575 if (CSQ_ALLELE_KEY.equals(field))
577 csqAlleleFieldIndex = index;
579 if (CSQ_FEATURE_KEY.equals(field))
581 csqFeatureFieldIndex = index;
584 if (isFieldWanted(field, vepFieldFilters))
586 vepFieldsOfInterest.put(index, field);
595 * Reads the Preference value for the given key, with default specified if no
596 * preference set. The value is interpreted as a comma-separated list of
597 * regular expressions, and converted into a list of compiled patterns ready
598 * for matching. Patterns are forced to upper-case for non-case-sensitive
601 * This supports user-defined filters for fields of interest to capture while
602 * processing data. For example, VCF_FIELDS = AF,AC* would mean that VCF INFO
603 * fields with an ID of AF, or starting with AC, would be matched.
609 private List<Pattern> getFieldMatchers(String key, String def)
611 String pref = Cache.getDefault(key, def);
612 List<Pattern> patterns = new ArrayList<>();
613 String[] tokens = pref.split(",");
614 for (String token : tokens)
618 patterns.add(Pattern.compile(token.toUpperCase()));
619 } catch (PatternSyntaxException e)
621 System.err.println("Invalid pattern ignored: " + token);
628 * Transfers VCF features to sequences to which this sequence has a mapping.
629 * If the mapping is 3:1, computes peptide variants from nucleotide variants.
633 protected void transferAddedFeatures(SequenceI seq)
635 DBRefEntry[] dbrefs = seq.getDBRefs();
640 for (DBRefEntry dbref : dbrefs)
642 Mapping mapping = dbref.getMap();
643 if (mapping == null || mapping.getTo() == null)
648 SequenceI mapTo = mapping.getTo();
649 MapList map = mapping.getMap();
650 if (map.getFromRatio() == 3)
653 * dna-to-peptide product mapping
655 AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
660 * nucleotide-to-nucleotide mapping e.g. transcript to CDS
662 List<SequenceFeature> features = seq.getFeatures()
663 .getPositionalFeatures(SequenceOntologyI.SEQUENCE_VARIANT);
664 for (SequenceFeature sf : features)
666 if (FEATURE_GROUP_VCF.equals(sf.getFeatureGroup()))
668 transferFeature(sf, mapTo, map);
676 * Tries to add overlapping variants read from a VCF file to the given sequence,
677 * and returns the number of variant features added
682 protected int loadSequenceVCF(SequenceI seq)
684 VCFMap vcfMap = getVcfMap(seq);
691 * work with the dataset sequence here
693 SequenceI dss = seq.getDatasetSequence();
698 return addVcfVariants(dss, vcfMap);
702 * Answers a map from sequence coordinates to VCF chromosome ranges
707 private VCFMap getVcfMap(SequenceI seq)
710 * simplest case: sequence has id and length matching a VCF contig
712 VCFMap vcfMap = null;
713 if (dictionary != null)
715 vcfMap = getContigMap(seq);
723 * otherwise, map to VCF from chromosomal coordinates
724 * of the sequence (if known)
726 GeneLociI seqCoords = seq.getGeneLoci();
727 if (seqCoords == null)
729 Cache.log.warn(String.format(
730 "Can't query VCF for %s as chromosome coordinates not known",
735 String species = seqCoords.getSpeciesId();
736 String chromosome = seqCoords.getChromosomeId();
737 String seqRef = seqCoords.getAssemblyId();
738 MapList map = seqCoords.getMap();
740 // note this requires the configured species to match that
741 // returned with the Ensembl sequence; todo: support aliases?
742 if (!vcfSpecies.equalsIgnoreCase(species))
744 Cache.log.warn("No VCF loaded to " + seq.getName()
745 + " as species not matched");
749 if (seqRef.equalsIgnoreCase(vcfAssembly))
751 return new VCFMap(chromosome, map);
755 * VCF data has a different reference assembly to the sequence:
756 * query Ensembl to map chromosomal coordinates from sequence to VCF
758 List<int[]> toVcfRanges = new ArrayList<>();
759 List<int[]> fromSequenceRanges = new ArrayList<>();
761 for (int[] range : map.getToRanges())
763 int[] fromRange = map.locateInFrom(range[0], range[1]);
764 if (fromRange == null)
770 int[] newRange = mapReferenceRange(range, chromosome, "human", seqRef,
772 if (newRange == null)
775 String.format("Failed to map %s:%s:%s:%d:%d to %s", species,
776 chromosome, seqRef, range[0], range[1],
782 toVcfRanges.add(newRange);
783 fromSequenceRanges.add(fromRange);
787 return new VCFMap(chromosome,
788 new MapList(fromSequenceRanges, toVcfRanges, 1, 1));
792 * If the sequence id matches a contig declared in the VCF file, and the
793 * sequence length matches the contig length, then returns a 1:1 map of the
794 * sequence to the contig, else returns null
799 private VCFMap getContigMap(SequenceI seq)
801 String id = seq.getName();
802 SAMSequenceRecord contig = dictionary.getSequence(id);
805 int len = seq.getLength();
806 if (len == contig.getSequenceLength())
808 MapList map = new MapList(new int[] { 1, len },
811 return new VCFMap(id, map);
818 * Queries the VCF reader for any variants that overlap the mapped chromosome
819 * ranges of the sequence, and adds as variant features. Returns the number of
820 * overlapping variants found.
824 * mapping from sequence to VCF coordinates
827 protected int addVcfVariants(SequenceI seq, VCFMap map)
829 boolean forwardStrand = map.map.isToForwardStrand();
832 * query the VCF for overlaps of each contiguous chromosomal region
836 for (int[] range : map.map.getToRanges())
838 int vcfStart = Math.min(range[0], range[1]);
839 int vcfEnd = Math.max(range[0], range[1]);
840 CloseableIterator<VariantContext> variants = reader
841 .query(map.chromosome, vcfStart, vcfEnd);
842 while (variants.hasNext())
844 VariantContext variant = variants.next();
846 int[] featureRange = map.map.locateInFrom(variant.getStart(),
849 if (featureRange != null)
851 int featureStart = Math.min(featureRange[0], featureRange[1]);
852 int featureEnd = Math.max(featureRange[0], featureRange[1]);
853 count += addAlleleFeatures(seq, variant, featureStart, featureEnd,
864 * A convenience method to get an attribute value for an alternate allele
867 * @param attributeName
871 protected String getAttributeValue(VariantContext variant,
872 String attributeName, int alleleIndex)
874 Object att = variant.getAttribute(attributeName);
876 if (att instanceof String)
880 else if (att instanceof ArrayList)
882 return ((List<String>) att).get(alleleIndex);
889 * Adds one variant feature for each allele in the VCF variant record, and
890 * returns the number of features added.
894 * @param featureStart
896 * @param forwardStrand
899 protected int addAlleleFeatures(SequenceI seq, VariantContext variant,
900 int featureStart, int featureEnd, boolean forwardStrand)
905 * Javadoc says getAlternateAlleles() imposes no order on the list returned
906 * so we proceed defensively to get them in strict order
908 int altAlleleCount = variant.getAlternateAlleles().size();
909 for (int i = 0; i < altAlleleCount; i++)
911 added += addAlleleFeature(seq, variant, i, featureStart, featureEnd,
918 * Inspects one allele and attempts to add a variant feature for it to the
919 * sequence. The additional data associated with this allele is extracted to
920 * store in the feature's key-value map. Answers the number of features added (0
925 * @param altAlleleIndex
927 * @param featureStart
929 * @param forwardStrand
932 protected int addAlleleFeature(SequenceI seq, VariantContext variant,
933 int altAlleleIndex, int featureStart, int featureEnd,
934 boolean forwardStrand)
936 String reference = variant.getReference().getBaseString();
937 Allele alt = variant.getAlternateAllele(altAlleleIndex);
938 String allele = alt.getBaseString();
941 * insertion after a genomic base, if on reverse strand, has to be
942 * converted to insertion of complement after the preceding position
944 int referenceLength = reference.length();
945 if (!forwardStrand && allele.length() > referenceLength
946 && allele.startsWith(reference))
948 featureStart -= referenceLength;
949 featureEnd = featureStart;
950 char insertAfter = seq.getCharAt(featureStart - seq.getStart());
951 reference = Dna.reverseComplement(String.valueOf(insertAfter));
952 allele = allele.substring(referenceLength) + reference;
956 * build the ref,alt allele description e.g. "G,A", using the base
957 * complement if the sequence is on the reverse strand
959 StringBuilder sb = new StringBuilder();
960 sb.append(forwardStrand ? reference : Dna.reverseComplement(reference));
962 sb.append(forwardStrand ? allele : Dna.reverseComplement(allele));
963 String alleles = sb.toString(); // e.g. G,A
966 * pick out the consequence data (if any) that is for the current allele
967 * and feature (transcript) that matches the current sequence
969 String consequence = getConsequenceForAlleleAndFeature(variant, CSQ_FIELD,
970 altAlleleIndex, csqAlleleFieldIndex,
971 csqAlleleNumberFieldIndex, seq.getName().toLowerCase(),
972 csqFeatureFieldIndex);
975 * pick out the ontology term for the consequence type
977 String type = SequenceOntologyI.SEQUENCE_VARIANT;
978 if (consequence != null)
980 type = getOntologyTerm(consequence);
983 SequenceFeature sf = new SequenceFeature(type, alleles, featureStart,
984 featureEnd, FEATURE_GROUP_VCF);
985 sf.setSource(sourceId);
987 sf.setValue(Gff3Helper.ALLELES, alleles);
989 addAlleleProperties(variant, sf, altAlleleIndex, consequence);
991 seq.addSequenceFeature(sf);
997 * Determines the Sequence Ontology term to use for the variant feature type in
998 * Jalview. The default is 'sequence_variant', but a more specific term is used
1001 * <li>VEP (or SnpEff) Consequence annotation is included in the VCF</li>
1002 * <li>sequence id can be matched to VEP Feature (or SnpEff Feature_ID)</li>
1005 * @param consequence
1007 * @see http://www.sequenceontology.org/browser/current_svn/term/SO:0001060
1009 String getOntologyTerm(String consequence)
1011 String type = SequenceOntologyI.SEQUENCE_VARIANT;
1014 * could we associate Consequence data with this allele and feature (transcript)?
1015 * if so, prefer the consequence term from that data
1017 if (csqAlleleFieldIndex == -1) // && snpEffAlleleFieldIndex == -1
1020 * no Consequence data so we can't refine the ontology term
1025 if (consequence != null)
1027 String[] csqFields = consequence.split(PIPE_REGEX);
1028 if (csqFields.length > csqConsequenceFieldIndex)
1030 type = csqFields[csqConsequenceFieldIndex];
1035 // todo the same for SnpEff consequence data matching if wanted
1039 * if of the form (e.g.) missense_variant&splice_region_variant,
1040 * just take the first ('most severe') consequence
1044 int pos = type.indexOf('&');
1047 type = type.substring(0, pos);
1054 * Returns matched consequence data if it can be found, else null.
1056 * <li>inspects the VCF data for key 'vcfInfoId'</li>
1057 * <li>splits this on comma (to distinct consequences)</li>
1058 * <li>returns the first consequence (if any) where</li>
1060 * <li>the allele matches the altAlleleIndex'th allele of variant</li>
1061 * <li>the feature matches the sequence name (e.g. transcript id)</li>
1064 * If matched, the consequence is returned (as pipe-delimited fields).
1068 * @param altAlleleIndex
1069 * @param alleleFieldIndex
1070 * @param alleleNumberFieldIndex
1072 * @param featureFieldIndex
1075 private String getConsequenceForAlleleAndFeature(VariantContext variant,
1076 String vcfInfoId, int altAlleleIndex, int alleleFieldIndex,
1077 int alleleNumberFieldIndex,
1078 String seqName, int featureFieldIndex)
1080 if (alleleFieldIndex == -1 || featureFieldIndex == -1)
1084 Object value = variant.getAttribute(vcfInfoId);
1086 if (value == null || !(value instanceof List<?>))
1092 * inspect each consequence in turn (comma-separated blocks
1093 * extracted by htsjdk)
1095 List<String> consequences = (List<String>) value;
1097 for (String consequence : consequences)
1099 String[] csqFields = consequence.split(PIPE_REGEX);
1100 if (csqFields.length > featureFieldIndex)
1102 String featureIdentifier = csqFields[featureFieldIndex];
1103 if (featureIdentifier.length() > 4
1104 && seqName.indexOf(featureIdentifier.toLowerCase()) > -1)
1107 * feature (transcript) matched - now check for allele match
1109 if (matchAllele(variant, altAlleleIndex, csqFields,
1110 alleleFieldIndex, alleleNumberFieldIndex))
1120 private boolean matchAllele(VariantContext variant, int altAlleleIndex,
1121 String[] csqFields, int alleleFieldIndex,
1122 int alleleNumberFieldIndex)
1125 * if ALLELE_NUM is present, it must match altAlleleIndex
1126 * NB first alternate allele is 1 for ALLELE_NUM, 0 for altAlleleIndex
1128 if (alleleNumberFieldIndex > -1)
1130 if (csqFields.length <= alleleNumberFieldIndex)
1134 String alleleNum = csqFields[alleleNumberFieldIndex];
1135 return String.valueOf(altAlleleIndex + 1).equals(alleleNum);
1139 * else consequence allele must match variant allele
1141 if (alleleFieldIndex > -1 && csqFields.length > alleleFieldIndex)
1143 String csqAllele = csqFields[alleleFieldIndex];
1144 String vcfAllele = variant.getAlternateAllele(altAlleleIndex)
1146 return csqAllele.equals(vcfAllele);
1152 * Add any allele-specific VCF key-value data to the sequence feature
1156 * @param altAlelleIndex
1158 * @param consequence
1159 * if not null, the consequence specific to this sequence (transcript
1160 * feature) and allele
1162 protected void addAlleleProperties(VariantContext variant,
1163 SequenceFeature sf, final int altAlelleIndex, String consequence)
1165 Map<String, Object> atts = variant.getAttributes();
1167 for (Entry<String, Object> att : atts.entrySet())
1169 String key = att.getKey();
1172 * extract Consequence data (if present) that we are able to
1173 * associated with the allele for this variant feature
1175 if (CSQ_FIELD.equals(key))
1177 addConsequences(variant, sf, consequence);
1182 * filter out fields we don't want to capture
1184 if (!vcfFieldsOfInterest.contains(key))
1190 * filter out fields we don't want to capture
1192 if (!vcfFieldsOfInterest.contains(key))
1198 * we extract values for other data which are allele-specific;
1199 * these may be per alternate allele (INFO[key].Number = 'A')
1200 * or per allele including reference (INFO[key].Number = 'R')
1202 VCFInfoHeaderLine infoHeader = header.getInfoHeaderLine(key);
1203 if (infoHeader == null)
1206 * can't be sure what data belongs to this allele, so
1207 * play safe and don't take any
1212 VCFHeaderLineCount number = infoHeader.getCountType();
1213 int index = altAlelleIndex;
1214 if (number == VCFHeaderLineCount.R)
1217 * one value per allele including reference, so bump index
1218 * e.g. the 3rd value is for the 2nd alternate allele
1222 else if (number != VCFHeaderLineCount.A)
1225 * don't save other values as not allele-related
1231 * take the index'th value
1233 String value = getAttributeValue(variant, key, index);
1236 sf.setValue(key, value);
1242 * Inspects CSQ data blocks (consequences) and adds attributes on the sequence
1245 * If <code>myConsequence</code> is not null, then this is the specific
1246 * consequence data (pipe-delimited fields) that is for the current allele and
1247 * transcript (sequence) being processed)
1251 * @param myConsequence
1253 protected void addConsequences(VariantContext variant, SequenceFeature sf,
1254 String myConsequence)
1256 Object value = variant.getAttribute(CSQ_FIELD);
1258 if (value == null || !(value instanceof List<?>))
1263 List<String> consequences = (List<String>) value;
1266 * inspect CSQ consequences; restrict to the consequence
1267 * associated with the current transcript (Feature)
1269 Map<String, String> csqValues = new HashMap<>();
1271 for (String consequence : consequences)
1273 if (myConsequence == null || myConsequence.equals(consequence))
1275 String[] csqFields = consequence.split(PIPE_REGEX);
1278 * inspect individual fields of this consequence, copying non-null
1279 * values which are 'fields of interest'
1282 for (String field : csqFields)
1284 if (field != null && field.length() > 0)
1286 String id = vepFieldsOfInterest.get(i);
1289 csqValues.put(id, field);
1297 if (!csqValues.isEmpty())
1299 sf.setValue(CSQ_FIELD, csqValues);
1304 * A convenience method to complement a dna base and return the string value
1310 protected String complement(byte[] reference)
1312 return String.valueOf(Dna.getComplement((char) reference[0]));
1316 * Determines the location of the query range (chromosome positions) in a
1317 * different reference assembly.
1319 * If the range is just a subregion of one for which we already have a mapping
1320 * (for example, an exon sub-region of a gene), then the mapping is just
1321 * computed arithmetically.
1323 * Otherwise, calls the Ensembl REST service that maps from one assembly
1324 * reference's coordinates to another's
1327 * start-end chromosomal range in 'fromRef' coordinates
1331 * assembly reference for the query coordinates
1333 * assembly reference we wish to translate to
1334 * @return the start-end range in 'toRef' coordinates
1336 protected int[] mapReferenceRange(int[] queryRange, String chromosome,
1337 String species, String fromRef, String toRef)
1340 * first try shorcut of computing the mapping as a subregion of one
1341 * we already have (e.g. for an exon, if we have the gene mapping)
1343 int[] mappedRange = findSubsumedRangeMapping(queryRange, chromosome,
1344 species, fromRef, toRef);
1345 if (mappedRange != null)
1351 * call (e.g.) http://rest.ensembl.org/map/human/GRCh38/17:45051610..45109016:1/GRCh37
1353 EnsemblMap mapper = new EnsemblMap();
1354 int[] mapping = mapper.getAssemblyMapping(species, chromosome, fromRef,
1357 if (mapping == null)
1359 // mapping service failure
1364 * save mapping for possible future re-use
1366 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1367 if (!assemblyMappings.containsKey(key))
1369 assemblyMappings.put(key, new HashMap<int[], int[]>());
1372 assemblyMappings.get(key).put(queryRange, mapping);
1378 * If we already have a 1:1 contiguous mapping which subsumes the given query
1379 * range, this method just calculates and returns the subset of that mapping,
1380 * else it returns null. In practical terms, if a gene has a contiguous
1381 * mapping between (for example) GRCh37 and GRCh38, then we assume that its
1382 * subsidiary exons occupy unchanged relative positions, and just compute
1383 * these as offsets, rather than do another lookup of the mapping.
1385 * If in future these assumptions prove invalid (e.g. for bacterial dna?!),
1386 * simply remove this method or let it always return null.
1388 * Warning: many rapid calls to the /map service map result in a 429 overload
1398 protected int[] findSubsumedRangeMapping(int[] queryRange, String chromosome,
1399 String species, String fromRef, String toRef)
1401 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1402 if (assemblyMappings.containsKey(key))
1404 Map<int[], int[]> mappedRanges = assemblyMappings.get(key);
1405 for (Entry<int[], int[]> mappedRange : mappedRanges.entrySet())
1407 int[] fromRange = mappedRange.getKey();
1408 int[] toRange = mappedRange.getValue();
1409 if (fromRange[1] - fromRange[0] == toRange[1] - toRange[0])
1412 * mapping is 1:1 in length, so we trust it to have no discontinuities
1414 if (MappingUtils.rangeContains(fromRange, queryRange))
1417 * fromRange subsumes our query range
1419 int offset = queryRange[0] - fromRange[0];
1420 int mappedRangeFrom = toRange[0] + offset;
1421 int mappedRangeTo = mappedRangeFrom + (queryRange[1] - queryRange[0]);
1422 return new int[] { mappedRangeFrom, mappedRangeTo };
1431 * Transfers the sequence feature to the target sequence, locating its start
1432 * and end range based on the mapping. Features which do not overlap the
1433 * target sequence are ignored.
1436 * @param targetSequence
1438 * mapping from the feature's coordinates to the target sequence
1440 protected void transferFeature(SequenceFeature sf,
1441 SequenceI targetSequence, MapList mapping)
1443 int[] mappedRange = mapping.locateInTo(sf.getBegin(), sf.getEnd());
1445 if (mappedRange != null)
1447 String group = sf.getFeatureGroup();
1448 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
1449 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
1450 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
1451 group, sf.getScore());
1452 targetSequence.addSequenceFeature(copy);
1457 * Formats a ranges map lookup key
1465 protected static String makeRangesKey(String chromosome, String species,
1466 String fromRef, String toRef)
1468 return species + EXCL + chromosome + EXCL + fromRef + EXCL