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
55 * A class to model the mapping from sequence to VCF coordinates. Cases include
57 * <li>a direct 1:1 mapping where the sequence is one of the VCF contigs</li>
58 * <li>a mapping of sequence to chromosomal coordinates, where sequence and VCF
59 * use the same reference assembly</li>
60 * <li>a modified mapping of sequence to chromosomal coordinates, where sequence
61 * and VCF use different reference assembles</li>
66 final String chromosome;
70 VCFMap(String chr, MapList m)
77 public String toString()
79 return chromosome + ":" + map.toString();
84 * Lookup keys, and default values, for Preference entries that describe
85 * patterns for VCF and VEP fields to capture
87 private static final String VEP_FIELDS_PREF = "VEP_FIELDS";
89 private static final String VCF_FIELDS_PREF = "VCF_FIELDS";
91 private static final String DEFAULT_VCF_FIELDS = ".*";
93 private static final String DEFAULT_VEP_FIELDS = ".*";// "Allele,Consequence,IMPACT,SWISSPROT,SIFT,PolyPhen,CLIN_SIG";
96 * keys to fields of VEP CSQ consequence data
97 * see https://www.ensembl.org/info/docs/tools/vep/vep_formats.html
99 private static final String CSQ_CONSEQUENCE_KEY = "Consequence";
100 private static final String CSQ_ALLELE_KEY = "Allele";
101 private static final String CSQ_ALLELE_NUM_KEY = "ALLELE_NUM"; // 0 (ref), 1...
102 private static final String CSQ_FEATURE_KEY = "Feature"; // Ensembl stable id
105 * default VCF INFO key for VEP consequence data
106 * NB this can be overridden running VEP with --vcf_info_field
107 * - we don't handle this case (require identifier to be CSQ)
109 private static final String CSQ_FIELD = "CSQ";
112 * separator for fields in consequence data is '|'
114 private static final String PIPE_REGEX = "\\|";
117 * key for Allele Frequency output by VEP
118 * see http://www.ensembl.org/info/docs/tools/vep/vep_formats.html
120 private static final String ALLELE_FREQUENCY_KEY = "AF";
123 * delimiter that separates multiple consequence data blocks
125 private static final String COMMA = ",";
128 * the feature group assigned to a VCF variant in Jalview
130 private static final String FEATURE_GROUP_VCF = "VCF";
133 * internal delimiter used to build keys for assemblyMappings
136 private static final String EXCL = "!";
139 * the VCF file we are processing
141 protected String vcfFilePath;
144 * mappings between VCF and sequence reference assembly regions, as
145 * key = "species!chromosome!fromAssembly!toAssembly
146 * value = Map{fromRange, toRange}
148 private Map<String, Map<int[], int[]>> assemblyMappings;
150 private VCFReader reader;
153 * holds details of the VCF header lines (metadata)
155 private VCFHeader header;
158 * a Dictionary of contigs (if present) referenced in the VCF file
160 private SAMSequenceDictionary dictionary;
163 * the position (0...) of field in each block of
164 * CSQ (consequence) data (if declared in the VCF INFO header for CSQ)
165 * see http://www.ensembl.org/info/docs/tools/vep/vep_formats.html
167 private int csqConsequenceFieldIndex = -1;
168 private int csqAlleleFieldIndex = -1;
169 private int csqAlleleNumberFieldIndex = -1;
170 private int csqFeatureFieldIndex = -1;
172 // todo the same fields for SnpEff ANN data if wanted
173 // see http://snpeff.sourceforge.net/SnpEff_manual.html#input
176 * a unique identifier under which to save metadata about feature
177 * attributes (selected INFO field data)
179 private String sourceId;
182 * The INFO IDs of data that is both present in the VCF file, and
183 * also matched by any filters for data of interest
185 List<String> vcfFieldsOfInterest;
188 * The field offsets and identifiers for VEP (CSQ) data that is both present
189 * in the VCF file, and also matched by any filters for data of interest
190 * for example 0 -> Allele, 1 -> Consequence, ..., 36 -> SIFT, ...
192 Map<Integer, String> vepFieldsOfInterest;
195 * Constructor given a VCF file
199 public VCFLoader(String vcfFile)
204 } catch (IOException e)
206 System.err.println("Error opening VCF file: " + e.getMessage());
209 // map of species!chromosome!fromAssembly!toAssembly to {fromRange, toRange}
210 assemblyMappings = new HashMap<>();
214 * Starts a new thread to query and load VCF variant data on to the given
217 * This method is not thread safe - concurrent threads should use separate
218 * instances of this class.
223 public void loadVCF(SequenceI[] seqs, final AlignViewControllerGuiI gui)
227 gui.setStatus(MessageManager.getString("label.searching_vcf"));
235 VCFLoader.this.doLoad(seqs, gui);
241 * Reads the specified contig sequence and adds its VCF variants to it
244 * the id of a single sequence (contig) to load
247 public SequenceI loadVCFContig(String contig)
249 String ref = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY)
251 if (ref.startsWith("file://"))
253 ref = ref.substring(7);
256 SequenceI seq = null;
257 File dbFile = new File(ref);
261 HtsContigDb db = new HtsContigDb("", dbFile);
262 seq = db.getSequenceProxy(contig);
263 loadSequenceVCF(seq, ref);
268 System.err.println("VCF reference not found: " + ref);
275 * Loads VCF on to one or more sequences
279 * optional callback handler for messages
281 protected void doLoad(SequenceI[] seqs, AlignViewControllerGuiI gui)
285 VCFHeaderLine ref = header
286 .getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
287 String vcfAssembly = ref.getValue();
293 * query for VCF overlapping each sequence in turn
295 for (SequenceI seq : seqs)
297 int added = loadSequenceVCF(seq, vcfAssembly);
302 transferAddedFeatures(seq);
307 String msg = MessageManager.formatMessage("label.added_vcf",
310 if (gui.getFeatureSettingsUI() != null)
312 gui.getFeatureSettingsUI().discoverAllFeatureData();
315 } catch (Throwable e)
317 System.err.println("Error processing VCF: " + e.getMessage());
321 gui.setStatus("Error occurred - see console for details");
330 } catch (IOException e)
341 * Opens the VCF file and parses header data
344 * @throws IOException
346 private void initialise(String filePath) throws IOException
348 vcfFilePath = filePath;
350 reader = new VCFReader(filePath);
352 header = reader.getFileHeader();
356 dictionary = header.getSequenceDictionary();
357 } catch (SAMException e)
359 // ignore - thrown if any contig line lacks length info
364 saveMetadata(sourceId);
367 * get offset of CSQ ALLELE_NUM and Feature if declared
373 * Reads metadata (such as INFO field descriptions and datatypes) and saves
374 * them for future reference
378 void saveMetadata(String theSourceId)
380 List<Pattern> vcfFieldPatterns = getFieldMatchers(VCF_FIELDS_PREF,
382 vcfFieldsOfInterest = new ArrayList<>();
384 FeatureSource metadata = new FeatureSource(theSourceId);
386 for (VCFInfoHeaderLine info : header.getInfoHeaderLines())
388 String attributeId = info.getID();
389 String desc = info.getDescription();
390 VCFHeaderLineType type = info.getType();
391 FeatureAttributeType attType = null;
395 attType = FeatureAttributeType.Character;
398 attType = FeatureAttributeType.Flag;
401 attType = FeatureAttributeType.Float;
404 attType = FeatureAttributeType.Integer;
407 attType = FeatureAttributeType.String;
410 metadata.setAttributeName(attributeId, desc);
411 metadata.setAttributeType(attributeId, attType);
413 if (isFieldWanted(attributeId, vcfFieldPatterns))
415 vcfFieldsOfInterest.add(attributeId);
419 FeatureSources.getInstance().addSource(theSourceId, metadata);
423 * Answers true if the field id is matched by any of the filter patterns, else
424 * false. Matching is against regular expression patterns, and is not
431 private boolean isFieldWanted(String id, List<Pattern> filters)
433 for (Pattern p : filters)
435 if (p.matcher(id.toUpperCase()).matches())
444 * Records 'wanted' fields defined in the CSQ INFO header (if there is one).
445 * Also records the position of selected fields (Allele, ALLELE_NUM, Feature)
446 * required for processing.
448 * CSQ fields are declared in the CSQ INFO Description e.g.
450 * Description="Consequence ...from ... VEP. Format: Allele|Consequence|...
452 protected void parseCsqHeader()
454 List<Pattern> vepFieldFilters = getFieldMatchers(VEP_FIELDS_PREF,
456 vepFieldsOfInterest = new HashMap<>();
458 VCFInfoHeaderLine csqInfo = header.getInfoHeaderLine(CSQ_FIELD);
465 * parse out the pipe-separated list of CSQ fields; we assume here that
466 * these form the last part of the description, and contain no spaces
468 String desc = csqInfo.getDescription();
469 int spacePos = desc.lastIndexOf(" ");
470 desc = desc.substring(spacePos + 1);
474 String[] format = desc.split(PIPE_REGEX);
476 for (String field : format)
478 if (CSQ_CONSEQUENCE_KEY.equals(field))
480 csqConsequenceFieldIndex = index;
482 if (CSQ_ALLELE_NUM_KEY.equals(field))
484 csqAlleleNumberFieldIndex = index;
486 if (CSQ_ALLELE_KEY.equals(field))
488 csqAlleleFieldIndex = index;
490 if (CSQ_FEATURE_KEY.equals(field))
492 csqFeatureFieldIndex = index;
495 if (isFieldWanted(field, vepFieldFilters))
497 vepFieldsOfInterest.put(index, field);
506 * Reads the Preference value for the given key, with default specified if no
507 * preference set. The value is interpreted as a comma-separated list of
508 * regular expressions, and converted into a list of compiled patterns ready
509 * for matching. Patterns are forced to upper-case for non-case-sensitive
512 * This supports user-defined filters for fields of interest to capture while
513 * processing data. For example, VCF_FIELDS = AF,AC* would mean that VCF INFO
514 * fields with an ID of AF, or starting with AC, would be matched.
520 private List<Pattern> getFieldMatchers(String key, String def)
522 String pref = Cache.getDefault(key, def);
523 List<Pattern> patterns = new ArrayList<>();
524 String[] tokens = pref.split(",");
525 for (String token : tokens)
529 patterns.add(Pattern.compile(token.toUpperCase()));
530 } catch (PatternSyntaxException e)
532 System.err.println("Invalid pattern ignored: " + token);
539 * Transfers VCF features to sequences to which this sequence has a mapping.
540 * If the mapping is 3:1, computes peptide variants from nucleotide variants.
544 protected void transferAddedFeatures(SequenceI seq)
546 DBRefEntry[] dbrefs = seq.getDBRefs();
551 for (DBRefEntry dbref : dbrefs)
553 Mapping mapping = dbref.getMap();
554 if (mapping == null || mapping.getTo() == null)
559 SequenceI mapTo = mapping.getTo();
560 MapList map = mapping.getMap();
561 if (map.getFromRatio() == 3)
564 * dna-to-peptide product mapping
566 AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
571 * nucleotide-to-nucleotide mapping e.g. transcript to CDS
573 List<SequenceFeature> features = seq.getFeatures()
574 .getPositionalFeatures(SequenceOntologyI.SEQUENCE_VARIANT);
575 for (SequenceFeature sf : features)
577 if (FEATURE_GROUP_VCF.equals(sf.getFeatureGroup()))
579 transferFeature(sf, mapTo, map);
587 * Tries to add overlapping variants read from a VCF file to the given sequence,
588 * and returns the number of variant features added
594 protected int loadSequenceVCF(SequenceI seq, String vcfAssembly)
596 VCFMap vcfMap = getVcfMap(seq, vcfAssembly);
603 * work with the dataset sequence here
605 SequenceI dss = seq.getDatasetSequence();
610 return addVcfVariants(dss, vcfMap);
614 * Answers a map from sequence coordinates to VCF chromosome ranges
620 private VCFMap getVcfMap(SequenceI seq, String vcfAssembly)
623 * simplest case: sequence has id and length matching a VCF contig
625 VCFMap vcfMap = null;
626 if (dictionary != null)
628 vcfMap = getContigMap(seq);
636 * otherwise, map to VCF from chromosomal coordinates
637 * of the sequence (if known)
639 GeneLociI seqCoords = seq.getGeneLoci();
640 if (seqCoords == null)
642 Cache.log.warn(String.format(
643 "Can't query VCF for %s as chromosome coordinates not known",
648 String species = seqCoords.getSpeciesId();
649 String chromosome = seqCoords.getChromosomeId();
650 String seqRef = seqCoords.getAssemblyId();
651 MapList map = seqCoords.getMap();
653 if (!vcfSpeciesMatchesSequence(vcfAssembly, species))
658 if (vcfAssemblyMatchesSequence(vcfAssembly, seqRef))
660 return new VCFMap(chromosome, map);
663 if (!"GRCh38".equalsIgnoreCase(seqRef) // Ensembl
664 || !vcfAssembly.contains("Homo_sapiens_assembly19")) // gnomAD
670 * map chromosomal coordinates from sequence to VCF if the VCF
671 * data has a different reference assembly to the sequence
673 // TODO generalise for cases other than GRCh38 -> GRCh37 !
674 // - or get the user to choose in a dialog
676 List<int[]> toVcfRanges = new ArrayList<>();
677 List<int[]> fromSequenceRanges = new ArrayList<>();
678 String toRef = "GRCh37";
680 for (int[] range : map.getToRanges())
682 int[] fromRange = map.locateInFrom(range[0], range[1]);
683 if (fromRange == null)
689 int[] newRange = mapReferenceRange(range, chromosome, "human", seqRef,
691 if (newRange == null)
694 String.format("Failed to map %s:%s:%s:%d:%d to %s", species,
695 chromosome, seqRef, range[0], range[1], toRef));
700 toVcfRanges.add(newRange);
701 fromSequenceRanges.add(fromRange);
705 return new VCFMap(chromosome,
706 new MapList(fromSequenceRanges, toVcfRanges, 1, 1));
710 * If the sequence id matches a contig declared in the VCF file, and the
711 * sequence length matches the contig length, then returns a 1:1 map of the
712 * sequence to the contig, else returns null
717 private VCFMap getContigMap(SequenceI seq)
719 String id = seq.getName();
720 SAMSequenceRecord contig = dictionary.getSequence(id);
723 int len = seq.getLength();
724 if (len == contig.getSequenceLength())
726 MapList map = new MapList(new int[] { 1, len },
729 return new VCFMap(id, map);
736 * Answers true if we determine that the VCF data uses the same reference
737 * assembly as the sequence, else false
743 private boolean vcfAssemblyMatchesSequence(String vcfAssembly,
746 // TODO improve on this stub, which handles gnomAD and
747 // hopes for the best for other cases
749 if ("GRCh38".equalsIgnoreCase(seqRef) // Ensembl
750 && vcfAssembly.contains("Homo_sapiens_assembly19")) // gnomAD
758 * Answers true if the species inferred from the VCF reference identifier
759 * matches that for the sequence
765 boolean vcfSpeciesMatchesSequence(String vcfAssembly, String speciesId)
768 // there are many aliases for species - how to equate one with another?
770 // VCF ##reference header is an unstructured URI - how to extract species?
771 // perhaps check if ref includes any (Ensembl) alias of speciesId??
772 // TODO ask the user to confirm this??
774 if (vcfAssembly.contains("Homo_sapiens") // gnomAD exome data example
775 && "HOMO_SAPIENS".equals(speciesId)) // Ensembl species id
780 if (vcfAssembly.contains("c_elegans") // VEP VCF response example
781 && "CAENORHABDITIS_ELEGANS".equals(speciesId)) // Ensembl
786 // this is not a sustainable solution...
792 * Queries the VCF reader for any variants that overlap the mapped chromosome
793 * ranges of the sequence, and adds as variant features. Returns the number of
794 * overlapping variants found.
798 * mapping from sequence to VCF coordinates
801 protected int addVcfVariants(SequenceI seq, VCFMap map)
803 boolean forwardStrand = map.map.isToForwardStrand();
806 * query the VCF for overlaps of each contiguous chromosomal region
810 for (int[] range : map.map.getToRanges())
812 int vcfStart = Math.min(range[0], range[1]);
813 int vcfEnd = Math.max(range[0], range[1]);
814 CloseableIterator<VariantContext> variants = reader
815 .query(map.chromosome, vcfStart, vcfEnd);
816 while (variants.hasNext())
818 VariantContext variant = variants.next();
820 int[] featureRange = map.map.locateInFrom(variant.getStart(),
823 if (featureRange != null)
825 int featureStart = Math.min(featureRange[0], featureRange[1]);
826 int featureEnd = Math.max(featureRange[0], featureRange[1]);
827 count += addAlleleFeatures(seq, variant, featureStart, featureEnd,
838 * A convenience method to get the AF value for the given alternate allele
845 protected float getAlleleFrequency(VariantContext variant, int alleleIndex)
848 String attributeValue = getAttributeValue(variant,
849 ALLELE_FREQUENCY_KEY, alleleIndex);
850 if (attributeValue != null)
854 score = Float.parseFloat(attributeValue);
855 } catch (NumberFormatException e)
865 * A convenience method to get an attribute value for an alternate allele
868 * @param attributeName
872 protected String getAttributeValue(VariantContext variant,
873 String attributeName, int alleleIndex)
875 Object att = variant.getAttribute(attributeName);
877 if (att instanceof String)
881 else if (att instanceof ArrayList)
883 return ((List<String>) att).get(alleleIndex);
890 * Adds one variant feature for each allele in the VCF variant record, and
891 * returns the number of features added.
895 * @param featureStart
897 * @param forwardStrand
900 protected int addAlleleFeatures(SequenceI seq, VariantContext variant,
901 int featureStart, int featureEnd, boolean forwardStrand)
906 * Javadoc says getAlternateAlleles() imposes no order on the list returned
907 * so we proceed defensively to get them in strict order
909 int altAlleleCount = variant.getAlternateAlleles().size();
910 for (int i = 0; i < altAlleleCount; i++)
912 added += addAlleleFeature(seq, variant, i, featureStart, featureEnd,
919 * Inspects one allele and attempts to add a variant feature for it to the
920 * sequence. The additional data associated with this allele is extracted to
921 * store in the feature's key-value map. Answers the number of features added (0
926 * @param altAlleleIndex
928 * @param featureStart
930 * @param forwardStrand
933 protected int addAlleleFeature(SequenceI seq, VariantContext variant,
934 int altAlleleIndex, int featureStart, int featureEnd,
935 boolean forwardStrand)
937 String reference = variant.getReference().getBaseString();
938 Allele alt = variant.getAlternateAllele(altAlleleIndex);
939 String allele = alt.getBaseString();
942 * insertion after a genomic base, if on reverse strand, has to be
943 * converted to insertion of complement after the preceding position
945 int referenceLength = reference.length();
946 if (!forwardStrand && allele.length() > referenceLength
947 && allele.startsWith(reference))
949 featureStart -= referenceLength;
950 featureEnd = featureStart;
951 char insertAfter = seq.getCharAt(featureStart - seq.getStart());
952 reference = Dna.reverseComplement(String.valueOf(insertAfter));
953 allele = allele.substring(referenceLength) + reference;
957 * build the ref,alt allele description e.g. "G,A", using the base
958 * complement if the sequence is on the reverse strand
960 StringBuilder sb = new StringBuilder();
961 sb.append(forwardStrand ? reference : Dna.reverseComplement(reference));
963 sb.append(forwardStrand ? allele : Dna.reverseComplement(allele));
964 String alleles = sb.toString(); // e.g. G,A
967 * pick out the consequence data (if any) that is for the current allele
968 * and feature (transcript) that matches the current sequence
970 String consequence = getConsequenceForAlleleAndFeature(variant, CSQ_FIELD,
971 altAlleleIndex, csqAlleleFieldIndex,
972 csqAlleleNumberFieldIndex, seq.getName().toLowerCase(),
973 csqFeatureFieldIndex);
976 * pick out the ontology term for the consequence type
978 String type = SequenceOntologyI.SEQUENCE_VARIANT;
979 if (consequence != null)
981 type = getOntologyTerm(consequence);
984 float score = getAlleleFrequency(variant, altAlleleIndex);
986 SequenceFeature sf = new SequenceFeature(type, alleles, featureStart,
987 featureEnd, score, FEATURE_GROUP_VCF);
988 sf.setSource(sourceId);
990 sf.setValue(Gff3Helper.ALLELES, alleles);
992 addAlleleProperties(variant, sf, altAlleleIndex, consequence);
994 seq.addSequenceFeature(sf);
1000 * Determines the Sequence Ontology term to use for the variant feature type in
1001 * Jalview. The default is 'sequence_variant', but a more specific term is used
1004 * <li>VEP (or SnpEff) Consequence annotation is included in the VCF</li>
1005 * <li>sequence id can be matched to VEP Feature (or SnpEff Feature_ID)</li>
1008 * @param consequence
1010 * @see http://www.sequenceontology.org/browser/current_svn/term/SO:0001060
1012 String getOntologyTerm(String consequence)
1014 String type = SequenceOntologyI.SEQUENCE_VARIANT;
1017 * could we associate Consequence data with this allele and feature (transcript)?
1018 * if so, prefer the consequence term from that data
1020 if (csqAlleleFieldIndex == -1) // && snpEffAlleleFieldIndex == -1
1023 * no Consequence data so we can't refine the ontology term
1028 if (consequence != null)
1030 String[] csqFields = consequence.split(PIPE_REGEX);
1031 if (csqFields.length > csqConsequenceFieldIndex)
1033 type = csqFields[csqConsequenceFieldIndex];
1038 // todo the same for SnpEff consequence data matching if wanted
1042 * if of the form (e.g.) missense_variant&splice_region_variant,
1043 * just take the first ('most severe') consequence
1047 int pos = type.indexOf('&');
1050 type = type.substring(0, pos);
1057 * Returns matched consequence data if it can be found, else null.
1059 * <li>inspects the VCF data for key 'vcfInfoId'</li>
1060 * <li>splits this on comma (to distinct consequences)</li>
1061 * <li>returns the first consequence (if any) where</li>
1063 * <li>the allele matches the altAlleleIndex'th allele of variant</li>
1064 * <li>the feature matches the sequence name (e.g. transcript id)</li>
1067 * If matched, the consequence is returned (as pipe-delimited fields).
1071 * @param altAlleleIndex
1072 * @param alleleFieldIndex
1073 * @param alleleNumberFieldIndex
1075 * @param featureFieldIndex
1078 private String getConsequenceForAlleleAndFeature(VariantContext variant,
1079 String vcfInfoId, int altAlleleIndex, int alleleFieldIndex,
1080 int alleleNumberFieldIndex,
1081 String seqName, int featureFieldIndex)
1083 if (alleleFieldIndex == -1 || featureFieldIndex == -1)
1087 Object value = variant.getAttribute(vcfInfoId);
1089 if (value == null || !(value instanceof List<?>))
1095 * inspect each consequence in turn (comma-separated blocks
1096 * extracted by htsjdk)
1098 List<String> consequences = (List<String>) value;
1100 for (String consequence : consequences)
1102 String[] csqFields = consequence.split(PIPE_REGEX);
1103 if (csqFields.length > featureFieldIndex)
1105 String featureIdentifier = csqFields[featureFieldIndex];
1106 if (featureIdentifier.length() > 4
1107 && seqName.indexOf(featureIdentifier.toLowerCase()) > -1)
1110 * feature (transcript) matched - now check for allele match
1112 if (matchAllele(variant, altAlleleIndex, csqFields,
1113 alleleFieldIndex, alleleNumberFieldIndex))
1123 private boolean matchAllele(VariantContext variant, int altAlleleIndex,
1124 String[] csqFields, int alleleFieldIndex,
1125 int alleleNumberFieldIndex)
1128 * if ALLELE_NUM is present, it must match altAlleleIndex
1129 * NB first alternate allele is 1 for ALLELE_NUM, 0 for altAlleleIndex
1131 if (alleleNumberFieldIndex > -1)
1133 if (csqFields.length <= alleleNumberFieldIndex)
1137 String alleleNum = csqFields[alleleNumberFieldIndex];
1138 return String.valueOf(altAlleleIndex + 1).equals(alleleNum);
1142 * else consequence allele must match variant allele
1144 if (alleleFieldIndex > -1 && csqFields.length > alleleFieldIndex)
1146 String csqAllele = csqFields[alleleFieldIndex];
1147 String vcfAllele = variant.getAlternateAllele(altAlleleIndex)
1149 return csqAllele.equals(vcfAllele);
1155 * Add any allele-specific VCF key-value data to the sequence feature
1159 * @param altAlelleIndex
1161 * @param consequence
1162 * if not null, the consequence specific to this sequence (transcript
1163 * feature) and allele
1165 protected void addAlleleProperties(VariantContext variant,
1166 SequenceFeature sf, final int altAlelleIndex, String consequence)
1168 Map<String, Object> atts = variant.getAttributes();
1170 for (Entry<String, Object> att : atts.entrySet())
1172 String key = att.getKey();
1175 * extract Consequence data (if present) that we are able to
1176 * associated with the allele for this variant feature
1178 if (CSQ_FIELD.equals(key))
1180 addConsequences(variant, sf, consequence);
1185 * filter out fields we don't want to capture
1187 if (!vcfFieldsOfInterest.contains(key))
1193 * filter out fields we don't want to capture
1195 if (!vcfFieldsOfInterest.contains(key))
1201 * we extract values for other data which are allele-specific;
1202 * these may be per alternate allele (INFO[key].Number = 'A')
1203 * or per allele including reference (INFO[key].Number = 'R')
1205 VCFInfoHeaderLine infoHeader = header.getInfoHeaderLine(key);
1206 if (infoHeader == null)
1209 * can't be sure what data belongs to this allele, so
1210 * play safe and don't take any
1215 VCFHeaderLineCount number = infoHeader.getCountType();
1216 int index = altAlelleIndex;
1217 if (number == VCFHeaderLineCount.R)
1220 * one value per allele including reference, so bump index
1221 * e.g. the 3rd value is for the 2nd alternate allele
1225 else if (number != VCFHeaderLineCount.A)
1228 * don't save other values as not allele-related
1234 * take the index'th value
1236 String value = getAttributeValue(variant, key, index);
1239 sf.setValue(key, value);
1245 * Inspects CSQ data blocks (consequences) and adds attributes on the sequence
1248 * If <code>myConsequence</code> is not null, then this is the specific
1249 * consequence data (pipe-delimited fields) that is for the current allele and
1250 * transcript (sequence) being processed)
1254 * @param myConsequence
1256 protected void addConsequences(VariantContext variant, SequenceFeature sf,
1257 String myConsequence)
1259 Object value = variant.getAttribute(CSQ_FIELD);
1261 if (value == null || !(value instanceof List<?>))
1266 List<String> consequences = (List<String>) value;
1269 * inspect CSQ consequences; restrict to the consequence
1270 * associated with the current transcript (Feature)
1272 Map<String, String> csqValues = new HashMap<>();
1274 for (String consequence : consequences)
1276 if (myConsequence == null || myConsequence.equals(consequence))
1278 String[] csqFields = consequence.split(PIPE_REGEX);
1281 * inspect individual fields of this consequence, copying non-null
1282 * values which are 'fields of interest'
1285 for (String field : csqFields)
1287 if (field != null && field.length() > 0)
1289 String id = vepFieldsOfInterest.get(i);
1292 csqValues.put(id, field);
1300 if (!csqValues.isEmpty())
1302 sf.setValue(CSQ_FIELD, csqValues);
1307 * A convenience method to complement a dna base and return the string value
1313 protected String complement(byte[] reference)
1315 return String.valueOf(Dna.getComplement((char) reference[0]));
1319 * Determines the location of the query range (chromosome positions) in a
1320 * different reference assembly.
1322 * If the range is just a subregion of one for which we already have a mapping
1323 * (for example, an exon sub-region of a gene), then the mapping is just
1324 * computed arithmetically.
1326 * Otherwise, calls the Ensembl REST service that maps from one assembly
1327 * reference's coordinates to another's
1330 * start-end chromosomal range in 'fromRef' coordinates
1334 * assembly reference for the query coordinates
1336 * assembly reference we wish to translate to
1337 * @return the start-end range in 'toRef' coordinates
1339 protected int[] mapReferenceRange(int[] queryRange, String chromosome,
1340 String species, String fromRef, String toRef)
1343 * first try shorcut of computing the mapping as a subregion of one
1344 * we already have (e.g. for an exon, if we have the gene mapping)
1346 int[] mappedRange = findSubsumedRangeMapping(queryRange, chromosome,
1347 species, fromRef, toRef);
1348 if (mappedRange != null)
1354 * call (e.g.) http://rest.ensembl.org/map/human/GRCh38/17:45051610..45109016:1/GRCh37
1356 EnsemblMap mapper = new EnsemblMap();
1357 int[] mapping = mapper.getAssemblyMapping(species, chromosome, fromRef,
1360 if (mapping == null)
1362 // mapping service failure
1367 * save mapping for possible future re-use
1369 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1370 if (!assemblyMappings.containsKey(key))
1372 assemblyMappings.put(key, new HashMap<int[], int[]>());
1375 assemblyMappings.get(key).put(queryRange, mapping);
1381 * If we already have a 1:1 contiguous mapping which subsumes the given query
1382 * range, this method just calculates and returns the subset of that mapping,
1383 * else it returns null. In practical terms, if a gene has a contiguous
1384 * mapping between (for example) GRCh37 and GRCh38, then we assume that its
1385 * subsidiary exons occupy unchanged relative positions, and just compute
1386 * these as offsets, rather than do another lookup of the mapping.
1388 * If in future these assumptions prove invalid (e.g. for bacterial dna?!),
1389 * simply remove this method or let it always return null.
1391 * Warning: many rapid calls to the /map service map result in a 429 overload
1401 protected int[] findSubsumedRangeMapping(int[] queryRange, String chromosome,
1402 String species, String fromRef, String toRef)
1404 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1405 if (assemblyMappings.containsKey(key))
1407 Map<int[], int[]> mappedRanges = assemblyMappings.get(key);
1408 for (Entry<int[], int[]> mappedRange : mappedRanges.entrySet())
1410 int[] fromRange = mappedRange.getKey();
1411 int[] toRange = mappedRange.getValue();
1412 if (fromRange[1] - fromRange[0] == toRange[1] - toRange[0])
1415 * mapping is 1:1 in length, so we trust it to have no discontinuities
1417 if (MappingUtils.rangeContains(fromRange, queryRange))
1420 * fromRange subsumes our query range
1422 int offset = queryRange[0] - fromRange[0];
1423 int mappedRangeFrom = toRange[0] + offset;
1424 int mappedRangeTo = mappedRangeFrom + (queryRange[1] - queryRange[0]);
1425 return new int[] { mappedRangeFrom, mappedRangeTo };
1434 * Transfers the sequence feature to the target sequence, locating its start
1435 * and end range based on the mapping. Features which do not overlap the
1436 * target sequence are ignored.
1439 * @param targetSequence
1441 * mapping from the feature's coordinates to the target sequence
1443 protected void transferFeature(SequenceFeature sf,
1444 SequenceI targetSequence, MapList mapping)
1446 int[] mappedRange = mapping.locateInTo(sf.getBegin(), sf.getEnd());
1448 if (mappedRange != null)
1450 String group = sf.getFeatureGroup();
1451 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
1452 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
1453 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
1454 group, sf.getScore());
1455 targetSequence.addSequenceFeature(copy);
1460 * Formats a ranges map lookup key
1468 protected static String makeRangesKey(String chromosome, String species,
1469 String fromRef, String toRef)
1471 return species + EXCL + chromosome + EXCL + fromRef + EXCL