2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.io.vcf;
23 import java.util.Locale;
26 import java.io.IOException;
27 import java.util.ArrayList;
28 import java.util.HashMap;
29 import java.util.HashSet;
30 import java.util.Iterator;
31 import java.util.List;
33 import java.util.Map.Entry;
35 import java.util.regex.Pattern;
36 import java.util.regex.PatternSyntaxException;
38 import htsjdk.samtools.SAMException;
39 import htsjdk.samtools.SAMSequenceDictionary;
40 import htsjdk.samtools.SAMSequenceRecord;
41 import htsjdk.samtools.util.CloseableIterator;
42 import htsjdk.tribble.TribbleException;
43 import htsjdk.variant.variantcontext.Allele;
44 import htsjdk.variant.variantcontext.VariantContext;
45 import htsjdk.variant.vcf.VCFConstants;
46 import htsjdk.variant.vcf.VCFHeader;
47 import htsjdk.variant.vcf.VCFHeaderLine;
48 import htsjdk.variant.vcf.VCFHeaderLineCount;
49 import htsjdk.variant.vcf.VCFHeaderLineType;
50 import htsjdk.variant.vcf.VCFInfoHeaderLine;
51 import jalview.analysis.Dna;
52 import jalview.api.AlignViewControllerGuiI;
53 import jalview.bin.Cache;
54 import jalview.datamodel.DBRefEntry;
55 import jalview.datamodel.GeneLociI;
56 import jalview.datamodel.Mapping;
57 import jalview.datamodel.SequenceFeature;
58 import jalview.datamodel.SequenceI;
59 import jalview.datamodel.features.FeatureAttributeType;
60 import jalview.datamodel.features.FeatureSource;
61 import jalview.datamodel.features.FeatureSources;
62 import jalview.ext.ensembl.EnsemblMap;
63 import jalview.ext.htsjdk.HtsContigDb;
64 import jalview.ext.htsjdk.VCFReader;
65 import jalview.io.gff.Gff3Helper;
66 import jalview.io.gff.SequenceOntologyI;
67 import jalview.util.MapList;
68 import jalview.util.MappingUtils;
69 import jalview.util.MessageManager;
70 import jalview.util.StringUtils;
73 * A class to read VCF data (using the htsjdk) and add variants as sequence
74 * features on dna and any related protein product sequences
78 public class VCFLoader
80 private static final String VCF_ENCODABLE = ":;=%,";
83 * Jalview feature attributes for VCF fixed column data
85 private static final String VCF_POS = "POS";
87 private static final String VCF_ID = "ID";
89 private static final String VCF_QUAL = "QUAL";
91 private static final String VCF_FILTER = "FILTER";
93 private static final String NO_VALUE = VCFConstants.MISSING_VALUE_v4; // '.'
95 private static final String DEFAULT_SPECIES = "homo_sapiens";
98 * A class to model the mapping from sequence to VCF coordinates. Cases include
100 * <li>a direct 1:1 mapping where the sequence is one of the VCF contigs</li>
101 * <li>a mapping of sequence to chromosomal coordinates, where sequence and VCF
102 * use the same reference assembly</li>
103 * <li>a modified mapping of sequence to chromosomal coordinates, where sequence
104 * and VCF use different reference assembles</li>
109 final String chromosome;
113 VCFMap(String chr, MapList m)
120 public String toString()
122 return chromosome + ":" + map.toString();
127 * Lookup keys, and default values, for Preference entries that describe
128 * patterns for VCF and VEP fields to capture
130 private static final String VEP_FIELDS_PREF = "VEP_FIELDS";
132 private static final String VCF_FIELDS_PREF = "VCF_FIELDS";
134 private static final String DEFAULT_VCF_FIELDS = ".*";
136 private static final String DEFAULT_VEP_FIELDS = ".*";// "Allele,Consequence,IMPACT,SWISSPROT,SIFT,PolyPhen,CLIN_SIG";
139 * Lookup keys, and default values, for Preference entries that give
140 * mappings from tokens in the 'reference' header to species or assembly
142 private static final String VCF_ASSEMBLY = "VCF_ASSEMBLY";
144 private static final String DEFAULT_VCF_ASSEMBLY = "assembly19=GRCh37,hs37=GRCh37,grch37=GRCh37,grch38=GRCh38";
146 private static final String VCF_SPECIES = "VCF_SPECIES"; // default is human
148 private static final String DEFAULT_REFERENCE = "grch37"; // fallback default is human GRCh37
151 * keys to fields of VEP CSQ consequence data
152 * see https://www.ensembl.org/info/docs/tools/vep/vep_formats.html
154 private static final String CSQ_CONSEQUENCE_KEY = "Consequence";
155 private static final String CSQ_ALLELE_KEY = "Allele";
156 private static final String CSQ_ALLELE_NUM_KEY = "ALLELE_NUM"; // 0 (ref), 1...
157 private static final String CSQ_FEATURE_KEY = "Feature"; // Ensembl stable id
160 * default VCF INFO key for VEP consequence data
161 * NB this can be overridden running VEP with --vcf_info_field
162 * - we don't handle this case (require identifier to be CSQ)
164 private static final String CSQ_FIELD = "CSQ";
167 * separator for fields in consequence data is '|'
169 private static final String PIPE_REGEX = "\\|";
172 * delimiter that separates multiple consequence data blocks
174 private static final String COMMA = ",";
177 * the feature group assigned to a VCF variant in Jalview
179 private static final String FEATURE_GROUP_VCF = "VCF";
182 * internal delimiter used to build keys for assemblyMappings
185 private static final String EXCL = "!";
188 * the VCF file we are processing
190 protected String vcfFilePath;
193 * mappings between VCF and sequence reference assembly regions, as
194 * key = "species!chromosome!fromAssembly!toAssembly
195 * value = Map{fromRange, toRange}
197 private Map<String, Map<int[], int[]>> assemblyMappings;
199 private VCFReader reader;
202 * holds details of the VCF header lines (metadata)
204 private VCFHeader header;
207 * species (as a valid Ensembl term) the VCF is for
209 private String vcfSpecies;
212 * genome assembly version (as a valid Ensembl identifier) the VCF is for
214 private String vcfAssembly;
217 * a Dictionary of contigs (if present) referenced in the VCF file
219 private SAMSequenceDictionary dictionary;
222 * the position (0...) of field in each block of
223 * CSQ (consequence) data (if declared in the VCF INFO header for CSQ)
224 * see http://www.ensembl.org/info/docs/tools/vep/vep_formats.html
226 private int csqConsequenceFieldIndex = -1;
227 private int csqAlleleFieldIndex = -1;
228 private int csqAlleleNumberFieldIndex = -1;
229 private int csqFeatureFieldIndex = -1;
231 // todo the same fields for SnpEff ANN data if wanted
232 // see http://snpeff.sourceforge.net/SnpEff_manual.html#input
235 * a unique identifier under which to save metadata about feature
236 * attributes (selected INFO field data)
238 private String sourceId;
241 * The INFO IDs of data that is both present in the VCF file, and
242 * also matched by any filters for data of interest
244 List<String> vcfFieldsOfInterest;
247 * The field offsets and identifiers for VEP (CSQ) data that is both present
248 * in the VCF file, and also matched by any filters for data of interest
249 * for example 0 -> Allele, 1 -> Consequence, ..., 36 -> SIFT, ...
251 Map<Integer, String> vepFieldsOfInterest;
254 * key:value for which rejected data has been seen
255 * (the error is logged only once for each combination)
257 private Set<String> badData;
260 * Constructor given a VCF file
264 public VCFLoader(String vcfFile)
269 } catch (IOException e)
271 System.err.println("Error opening VCF file: " + e.getMessage());
274 // map of species!chromosome!fromAssembly!toAssembly to {fromRange, toRange}
275 assemblyMappings = new HashMap<>();
279 * Starts a new thread to query and load VCF variant data on to the given
282 * This method is not thread safe - concurrent threads should use separate
283 * instances of this class.
288 public void loadVCF(SequenceI[] seqs, final AlignViewControllerGuiI gui)
292 gui.setStatus(MessageManager.getString("label.searching_vcf"));
300 VCFLoader.this.doLoad(seqs, gui);
306 * Reads the specified contig sequence and adds its VCF variants to it
309 * the id of a single sequence (contig) to load
312 public SequenceI loadVCFContig(String contig)
314 VCFHeaderLine headerLine = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
315 if (headerLine == null)
317 Cache.log.error("VCF reference header not found");
320 String ref = headerLine.getValue();
321 if (ref.startsWith("file://"))
323 ref = ref.substring(7);
325 setSpeciesAndAssembly(ref);
327 SequenceI seq = null;
328 File dbFile = new File(ref);
332 HtsContigDb db = new HtsContigDb("", dbFile);
333 seq = db.getSequenceProxy(contig);
334 loadSequenceVCF(seq);
339 Cache.log.error("VCF reference not found: " + ref);
346 * Loads VCF on to one or more sequences
350 * optional callback handler for messages
352 protected void doLoad(SequenceI[] seqs, AlignViewControllerGuiI gui)
356 VCFHeaderLine ref = header
357 .getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
358 String reference = ref == null ? null : ref.getValue();
360 setSpeciesAndAssembly(reference);
366 * query for VCF overlapping each sequence in turn
368 for (SequenceI seq : seqs)
370 int added = loadSequenceVCF(seq);
375 transferAddedFeatures(seq);
380 String msg = MessageManager.formatMessage("label.added_vcf",
383 if (gui.getFeatureSettingsUI() != null)
385 gui.getFeatureSettingsUI().discoverAllFeatureData();
388 } catch (Throwable e)
390 System.err.println("Error processing VCF: " + e.getMessage());
394 gui.setStatus("Error occurred - see console for details");
403 } catch (IOException e)
414 * Attempts to determine and save the species and genome assembly version to
415 * which the VCF data applies. This may be done by parsing the {@code reference}
416 * header line, configured in a property file, or (potentially) confirmed
417 * interactively by the user.
419 * The saved values should be identifiers valid for Ensembl's REST service
420 * {@code map} endpoint, so they can be used (if necessary) to retrieve the
421 * mapping between VCF coordinates and sequence coordinates.
424 * @see https://rest.ensembl.org/documentation/info/assembly_map
425 * @see https://rest.ensembl.org/info/assembly/human?content-type=text/xml
426 * @see https://rest.ensembl.org/info/species?content-type=text/xml
428 protected void setSpeciesAndAssembly(String reference)
430 if (reference == null)
432 Cache.log.error("No VCF ##reference found, defaulting to "
433 + DEFAULT_REFERENCE + ":" + DEFAULT_SPECIES);
434 reference = DEFAULT_REFERENCE; // default to GRCh37 if not specified
436 reference = reference.toLowerCase(Locale.ROOT);
439 * for a non-human species, or other assembly identifier,
440 * specify as a Jalview property file entry e.g.
441 * VCF_ASSEMBLY = hs37=GRCh37,assembly19=GRCh37
442 * VCF_SPECIES = c_elegans=celegans
443 * to map a token in the reference header to a value
445 String prop = Cache.getDefault(VCF_ASSEMBLY, DEFAULT_VCF_ASSEMBLY);
446 for (String token : prop.split(","))
448 String[] tokens = token.split("=");
449 if (tokens.length == 2)
451 if (reference.contains(tokens[0].trim().toLowerCase(Locale.ROOT)))
453 vcfAssembly = tokens[1].trim();
459 vcfSpecies = DEFAULT_SPECIES;
460 prop = Cache.getProperty(VCF_SPECIES);
463 for (String token : prop.split(","))
465 String[] tokens = token.split("=");
466 if (tokens.length == 2)
468 if (reference.contains(tokens[0].trim().toLowerCase(Locale.ROOT)))
470 vcfSpecies = tokens[1].trim();
479 * Opens the VCF file and parses header data
482 * @throws IOException
484 private void initialise(String filePath) throws IOException
486 vcfFilePath = filePath;
488 reader = new VCFReader(filePath);
490 header = reader.getFileHeader();
494 dictionary = header.getSequenceDictionary();
495 } catch (SAMException e)
497 // ignore - thrown if any contig line lacks length info
502 saveMetadata(sourceId);
505 * get offset of CSQ ALLELE_NUM and Feature if declared
511 * Reads metadata (such as INFO field descriptions and datatypes) and saves
512 * them for future reference
516 void saveMetadata(String theSourceId)
518 List<Pattern> vcfFieldPatterns = getFieldMatchers(VCF_FIELDS_PREF,
520 vcfFieldsOfInterest = new ArrayList<>();
522 FeatureSource metadata = new FeatureSource(theSourceId);
524 for (VCFInfoHeaderLine info : header.getInfoHeaderLines())
526 String attributeId = info.getID();
527 String desc = info.getDescription();
528 VCFHeaderLineType type = info.getType();
529 FeatureAttributeType attType = null;
533 attType = FeatureAttributeType.Character;
536 attType = FeatureAttributeType.Flag;
539 attType = FeatureAttributeType.Float;
542 attType = FeatureAttributeType.Integer;
545 attType = FeatureAttributeType.String;
548 metadata.setAttributeName(attributeId, desc);
549 metadata.setAttributeType(attributeId, attType);
551 if (isFieldWanted(attributeId, vcfFieldPatterns))
553 vcfFieldsOfInterest.add(attributeId);
557 FeatureSources.getInstance().addSource(theSourceId, metadata);
561 * Answers true if the field id is matched by any of the filter patterns, else
562 * false. Matching is against regular expression patterns, and is not
569 private boolean isFieldWanted(String id, List<Pattern> filters)
571 for (Pattern p : filters)
573 if (p.matcher(id.toUpperCase(Locale.ROOT)).matches())
582 * Records 'wanted' fields defined in the CSQ INFO header (if there is one).
583 * Also records the position of selected fields (Allele, ALLELE_NUM, Feature)
584 * required for processing.
586 * CSQ fields are declared in the CSQ INFO Description e.g.
588 * Description="Consequence ...from ... VEP. Format: Allele|Consequence|...
590 protected void parseCsqHeader()
592 List<Pattern> vepFieldFilters = getFieldMatchers(VEP_FIELDS_PREF,
594 vepFieldsOfInterest = new HashMap<>();
596 VCFInfoHeaderLine csqInfo = header.getInfoHeaderLine(CSQ_FIELD);
603 * parse out the pipe-separated list of CSQ fields; we assume here that
604 * these form the last part of the description, and contain no spaces
606 String desc = csqInfo.getDescription();
607 int spacePos = desc.lastIndexOf(" ");
608 desc = desc.substring(spacePos + 1);
612 String[] format = desc.split(PIPE_REGEX);
614 for (String field : format)
616 if (CSQ_CONSEQUENCE_KEY.equals(field))
618 csqConsequenceFieldIndex = index;
620 if (CSQ_ALLELE_NUM_KEY.equals(field))
622 csqAlleleNumberFieldIndex = index;
624 if (CSQ_ALLELE_KEY.equals(field))
626 csqAlleleFieldIndex = index;
628 if (CSQ_FEATURE_KEY.equals(field))
630 csqFeatureFieldIndex = index;
633 if (isFieldWanted(field, vepFieldFilters))
635 vepFieldsOfInterest.put(index, field);
644 * Reads the Preference value for the given key, with default specified if no
645 * preference set. The value is interpreted as a comma-separated list of
646 * regular expressions, and converted into a list of compiled patterns ready
647 * for matching. Patterns are forced to upper-case for non-case-sensitive
650 * This supports user-defined filters for fields of interest to capture while
651 * processing data. For example, VCF_FIELDS = AF,AC* would mean that VCF INFO
652 * fields with an ID of AF, or starting with AC, would be matched.
658 private List<Pattern> getFieldMatchers(String key, String def)
660 String pref = Cache.getDefault(key, def);
661 List<Pattern> patterns = new ArrayList<>();
662 String[] tokens = pref.split(",");
663 for (String token : tokens)
667 patterns.add(Pattern.compile(token.toUpperCase(Locale.ROOT)));
668 } catch (PatternSyntaxException e)
670 System.err.println("Invalid pattern ignored: " + token);
677 * Transfers VCF features to sequences to which this sequence has a mapping.
681 protected void transferAddedFeatures(SequenceI seq)
683 List<DBRefEntry> dbrefs = seq.getDBRefs();
688 for (DBRefEntry dbref : dbrefs)
690 Mapping mapping = dbref.getMap();
691 if (mapping == null || mapping.getTo() == null)
696 SequenceI mapTo = mapping.getTo();
697 MapList map = mapping.getMap();
698 if (map.getFromRatio() == 3)
701 * dna-to-peptide product mapping
703 // JAL-3187 render on the fly instead
704 // AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
709 * nucleotide-to-nucleotide mapping e.g. transcript to CDS
711 List<SequenceFeature> features = seq.getFeatures()
712 .getPositionalFeatures(SequenceOntologyI.SEQUENCE_VARIANT);
713 for (SequenceFeature sf : features)
715 if (FEATURE_GROUP_VCF.equals(sf.getFeatureGroup()))
717 transferFeature(sf, mapTo, map);
725 * Tries to add overlapping variants read from a VCF file to the given sequence,
726 * and returns the number of variant features added
731 protected int loadSequenceVCF(SequenceI seq)
733 VCFMap vcfMap = getVcfMap(seq);
740 * work with the dataset sequence here
742 SequenceI dss = seq.getDatasetSequence();
747 return addVcfVariants(dss, vcfMap);
751 * Answers a map from sequence coordinates to VCF chromosome ranges
756 private VCFMap getVcfMap(SequenceI seq)
759 * simplest case: sequence has id and length matching a VCF contig
761 VCFMap vcfMap = null;
762 if (dictionary != null)
764 vcfMap = getContigMap(seq);
772 * otherwise, map to VCF from chromosomal coordinates
773 * of the sequence (if known)
775 GeneLociI seqCoords = seq.getGeneLoci();
776 if (seqCoords == null)
778 Cache.log.warn(String.format(
779 "Can't query VCF for %s as chromosome coordinates not known",
784 String species = seqCoords.getSpeciesId();
785 String chromosome = seqCoords.getChromosomeId();
786 String seqRef = seqCoords.getAssemblyId();
787 MapList map = seqCoords.getMapping();
789 // note this requires the configured species to match that
790 // returned with the Ensembl sequence; todo: support aliases?
791 if (!vcfSpecies.equalsIgnoreCase(species))
793 Cache.log.warn("No VCF loaded to " + seq.getName()
794 + " as species not matched");
798 if (seqRef.equalsIgnoreCase(vcfAssembly))
800 return new VCFMap(chromosome, map);
804 * VCF data has a different reference assembly to the sequence:
805 * query Ensembl to map chromosomal coordinates from sequence to VCF
807 List<int[]> toVcfRanges = new ArrayList<>();
808 List<int[]> fromSequenceRanges = new ArrayList<>();
810 for (int[] range : map.getToRanges())
812 int[] fromRange = map.locateInFrom(range[0], range[1]);
813 if (fromRange == null)
819 int[] newRange = mapReferenceRange(range, chromosome, "human", seqRef,
821 if (newRange == null)
824 String.format("Failed to map %s:%s:%s:%d:%d to %s", species,
825 chromosome, seqRef, range[0], range[1],
831 toVcfRanges.add(newRange);
832 fromSequenceRanges.add(fromRange);
836 return new VCFMap(chromosome,
837 new MapList(fromSequenceRanges, toVcfRanges, 1, 1));
841 * If the sequence id matches a contig declared in the VCF file, and the
842 * sequence length matches the contig length, then returns a 1:1 map of the
843 * sequence to the contig, else returns null
848 private VCFMap getContigMap(SequenceI seq)
850 String id = seq.getName();
851 SAMSequenceRecord contig = dictionary.getSequence(id);
854 int len = seq.getLength();
855 if (len == contig.getSequenceLength())
857 MapList map = new MapList(new int[] { 1, len },
860 return new VCFMap(id, map);
867 * Queries the VCF reader for any variants that overlap the mapped chromosome
868 * ranges of the sequence, and adds as variant features. Returns the number of
869 * overlapping variants found.
873 * mapping from sequence to VCF coordinates
876 protected int addVcfVariants(SequenceI seq, VCFMap map)
878 boolean forwardStrand = map.map.isToForwardStrand();
881 * query the VCF for overlaps of each contiguous chromosomal region
885 for (int[] range : map.map.getToRanges())
887 int vcfStart = Math.min(range[0], range[1]);
888 int vcfEnd = Math.max(range[0], range[1]);
891 CloseableIterator<VariantContext> variants = reader
892 .query(map.chromosome, vcfStart, vcfEnd);
893 while (variants.hasNext())
895 VariantContext variant = variants.next();
897 int[] featureRange = map.map.locateInFrom(variant.getStart(),
901 * only take features whose range is fully mappable to sequence positions
903 if (featureRange != null)
905 int featureStart = Math.min(featureRange[0], featureRange[1]);
906 int featureEnd = Math.max(featureRange[0], featureRange[1]);
907 if (featureEnd - featureStart == variant.getEnd()
908 - variant.getStart())
910 count += addAlleleFeatures(seq, variant, featureStart,
911 featureEnd, forwardStrand);
916 } catch (TribbleException e)
919 * RuntimeException throwable by htsjdk
921 String msg = String.format("Error reading VCF for %s:%d-%d: %s ",
922 map.chromosome, vcfStart, vcfEnd,e.getLocalizedMessage());
923 Cache.log.error(msg);
931 * A convenience method to get an attribute value for an alternate allele
934 * @param attributeName
938 protected String getAttributeValue(VariantContext variant,
939 String attributeName, int alleleIndex)
941 Object att = variant.getAttribute(attributeName);
943 if (att instanceof String)
947 else if (att instanceof ArrayList)
949 return ((List<String>) att).get(alleleIndex);
956 * Adds one variant feature for each allele in the VCF variant record, and
957 * returns the number of features added.
961 * @param featureStart
963 * @param forwardStrand
966 protected int addAlleleFeatures(SequenceI seq, VariantContext variant,
967 int featureStart, int featureEnd, boolean forwardStrand)
972 * Javadoc says getAlternateAlleles() imposes no order on the list returned
973 * so we proceed defensively to get them in strict order
975 int altAlleleCount = variant.getAlternateAlleles().size();
976 for (int i = 0; i < altAlleleCount; i++)
978 added += addAlleleFeature(seq, variant, i, featureStart, featureEnd,
985 * Inspects one allele and attempts to add a variant feature for it to the
986 * sequence. The additional data associated with this allele is extracted to
987 * store in the feature's key-value map. Answers the number of features added (0
992 * @param altAlleleIndex
994 * @param featureStart
996 * @param forwardStrand
999 protected int addAlleleFeature(SequenceI seq, VariantContext variant,
1000 int altAlleleIndex, int featureStart, int featureEnd,
1001 boolean forwardStrand)
1003 String reference = variant.getReference().getBaseString();
1004 Allele alt = variant.getAlternateAllele(altAlleleIndex);
1005 String allele = alt.getBaseString();
1008 * insertion after a genomic base, if on reverse strand, has to be
1009 * converted to insertion of complement after the preceding position
1011 int referenceLength = reference.length();
1012 if (!forwardStrand && allele.length() > referenceLength
1013 && allele.startsWith(reference))
1015 featureStart -= referenceLength;
1016 featureEnd = featureStart;
1017 char insertAfter = seq.getCharAt(featureStart - seq.getStart());
1018 reference = Dna.reverseComplement(String.valueOf(insertAfter));
1019 allele = allele.substring(referenceLength) + reference;
1023 * build the ref,alt allele description e.g. "G,A", using the base
1024 * complement if the sequence is on the reverse strand
1026 StringBuilder sb = new StringBuilder();
1027 sb.append(forwardStrand ? reference : Dna.reverseComplement(reference));
1029 sb.append(forwardStrand ? allele : Dna.reverseComplement(allele));
1030 String alleles = sb.toString(); // e.g. G,A
1033 * pick out the consequence data (if any) that is for the current allele
1034 * and feature (transcript) that matches the current sequence
1036 String consequence = getConsequenceForAlleleAndFeature(variant, CSQ_FIELD,
1037 altAlleleIndex, csqAlleleFieldIndex,
1038 csqAlleleNumberFieldIndex, seq.getName().toLowerCase(Locale.ROOT),
1039 csqFeatureFieldIndex);
1042 * pick out the ontology term for the consequence type
1044 String type = SequenceOntologyI.SEQUENCE_VARIANT;
1045 if (consequence != null)
1047 type = getOntologyTerm(consequence);
1050 SequenceFeature sf = new SequenceFeature(type, alleles, featureStart,
1051 featureEnd, FEATURE_GROUP_VCF);
1052 sf.setSource(sourceId);
1055 * save the derived alleles as a named attribute; this will be
1056 * needed when Jalview computes derived peptide variants
1058 addFeatureAttribute(sf, Gff3Helper.ALLELES, alleles);
1061 * add selected VCF fixed column data as feature attributes
1063 addFeatureAttribute(sf, VCF_POS, String.valueOf(variant.getStart()));
1064 addFeatureAttribute(sf, VCF_ID, variant.getID());
1065 addFeatureAttribute(sf, VCF_QUAL,
1066 String.valueOf(variant.getPhredScaledQual()));
1067 addFeatureAttribute(sf, VCF_FILTER, getFilter(variant));
1069 addAlleleProperties(variant, sf, altAlleleIndex, consequence);
1071 seq.addSequenceFeature(sf);
1077 * Answers the VCF FILTER value for the variant - or an approximation to it.
1078 * This field is either PASS, or a semi-colon separated list of filters not
1079 * passed. htsjdk saves filters as a HashSet, so the order when reassembled into
1080 * a list may be different.
1085 String getFilter(VariantContext variant)
1087 Set<String> filters = variant.getFilters();
1088 if (filters.isEmpty())
1092 Iterator<String> iterator = filters.iterator();
1093 String first = iterator.next();
1094 if (filters.size() == 1)
1099 StringBuilder sb = new StringBuilder(first);
1100 while (iterator.hasNext())
1102 sb.append(";").append(iterator.next());
1105 return sb.toString();
1109 * Adds one feature attribute unless the value is null, empty or '.'
1115 void addFeatureAttribute(SequenceFeature sf, String key, String value)
1117 if (value != null && !value.isEmpty() && !NO_VALUE.equals(value))
1119 sf.setValue(key, value);
1124 * Determines the Sequence Ontology term to use for the variant feature type in
1125 * Jalview. The default is 'sequence_variant', but a more specific term is used
1128 * <li>VEP (or SnpEff) Consequence annotation is included in the VCF</li>
1129 * <li>sequence id can be matched to VEP Feature (or SnpEff Feature_ID)</li>
1132 * @param consequence
1134 * @see http://www.sequenceontology.org/browser/current_svn/term/SO:0001060
1136 String getOntologyTerm(String consequence)
1138 String type = SequenceOntologyI.SEQUENCE_VARIANT;
1141 * could we associate Consequence data with this allele and feature (transcript)?
1142 * if so, prefer the consequence term from that data
1144 if (csqAlleleFieldIndex == -1) // && snpEffAlleleFieldIndex == -1
1147 * no Consequence data so we can't refine the ontology term
1152 if (consequence != null)
1154 String[] csqFields = consequence.split(PIPE_REGEX);
1155 if (csqFields.length > csqConsequenceFieldIndex)
1157 type = csqFields[csqConsequenceFieldIndex];
1162 // todo the same for SnpEff consequence data matching if wanted
1166 * if of the form (e.g.) missense_variant&splice_region_variant,
1167 * just take the first ('most severe') consequence
1171 int pos = type.indexOf('&');
1174 type = type.substring(0, pos);
1181 * Returns matched consequence data if it can be found, else null.
1183 * <li>inspects the VCF data for key 'vcfInfoId'</li>
1184 * <li>splits this on comma (to distinct consequences)</li>
1185 * <li>returns the first consequence (if any) where</li>
1187 * <li>the allele matches the altAlleleIndex'th allele of variant</li>
1188 * <li>the feature matches the sequence name (e.g. transcript id)</li>
1191 * If matched, the consequence is returned (as pipe-delimited fields).
1195 * @param altAlleleIndex
1196 * @param alleleFieldIndex
1197 * @param alleleNumberFieldIndex
1199 * @param featureFieldIndex
1202 private String getConsequenceForAlleleAndFeature(VariantContext variant,
1203 String vcfInfoId, int altAlleleIndex, int alleleFieldIndex,
1204 int alleleNumberFieldIndex,
1205 String seqName, int featureFieldIndex)
1207 if (alleleFieldIndex == -1 || featureFieldIndex == -1)
1211 Object value = variant.getAttribute(vcfInfoId);
1213 if (value == null || !(value instanceof List<?>))
1219 * inspect each consequence in turn (comma-separated blocks
1220 * extracted by htsjdk)
1222 List<String> consequences = (List<String>) value;
1224 for (String consequence : consequences)
1226 String[] csqFields = consequence.split(PIPE_REGEX);
1227 if (csqFields.length > featureFieldIndex)
1229 String featureIdentifier = csqFields[featureFieldIndex];
1230 if (featureIdentifier.length() > 4
1231 && seqName.indexOf(featureIdentifier.toLowerCase(Locale.ROOT)) > -1)
1234 * feature (transcript) matched - now check for allele match
1236 if (matchAllele(variant, altAlleleIndex, csqFields,
1237 alleleFieldIndex, alleleNumberFieldIndex))
1247 private boolean matchAllele(VariantContext variant, int altAlleleIndex,
1248 String[] csqFields, int alleleFieldIndex,
1249 int alleleNumberFieldIndex)
1252 * if ALLELE_NUM is present, it must match altAlleleIndex
1253 * NB first alternate allele is 1 for ALLELE_NUM, 0 for altAlleleIndex
1255 if (alleleNumberFieldIndex > -1)
1257 if (csqFields.length <= alleleNumberFieldIndex)
1261 String alleleNum = csqFields[alleleNumberFieldIndex];
1262 return String.valueOf(altAlleleIndex + 1).equals(alleleNum);
1266 * else consequence allele must match variant allele
1268 if (alleleFieldIndex > -1 && csqFields.length > alleleFieldIndex)
1270 String csqAllele = csqFields[alleleFieldIndex];
1271 String vcfAllele = variant.getAlternateAllele(altAlleleIndex)
1273 return csqAllele.equals(vcfAllele);
1279 * Add any allele-specific VCF key-value data to the sequence feature
1283 * @param altAlelleIndex
1285 * @param consequence
1286 * if not null, the consequence specific to this sequence (transcript
1287 * feature) and allele
1289 protected void addAlleleProperties(VariantContext variant,
1290 SequenceFeature sf, final int altAlelleIndex, String consequence)
1292 Map<String, Object> atts = variant.getAttributes();
1294 for (Entry<String, Object> att : atts.entrySet())
1296 String key = att.getKey();
1299 * extract Consequence data (if present) that we are able to
1300 * associated with the allele for this variant feature
1302 if (CSQ_FIELD.equals(key))
1304 addConsequences(variant, sf, consequence);
1309 * filter out fields we don't want to capture
1311 if (!vcfFieldsOfInterest.contains(key))
1317 * we extract values for other data which are allele-specific;
1318 * these may be per alternate allele (INFO[key].Number = 'A')
1319 * or per allele including reference (INFO[key].Number = 'R')
1321 VCFInfoHeaderLine infoHeader = header.getInfoHeaderLine(key);
1322 if (infoHeader == null)
1325 * can't be sure what data belongs to this allele, so
1326 * play safe and don't take any
1331 VCFHeaderLineCount number = infoHeader.getCountType();
1332 int index = altAlelleIndex;
1333 if (number == VCFHeaderLineCount.R)
1336 * one value per allele including reference, so bump index
1337 * e.g. the 3rd value is for the 2nd alternate allele
1341 else if (number != VCFHeaderLineCount.A)
1344 * don't save other values as not allele-related
1350 * take the index'th value
1352 String value = getAttributeValue(variant, key, index);
1353 if (value != null && isValid(variant, key, value))
1356 * decode colon, semicolon, equals sign, percent sign, comma (only)
1357 * as required by the VCF specification (para 1.2)
1359 value = StringUtils.urlDecode(value, VCF_ENCODABLE);
1360 addFeatureAttribute(sf, key, value);
1366 * Answers true for '.', null, or an empty value, or if the INFO type is String.
1367 * If the INFO type is Integer or Float, answers false if the value is not in
1375 protected boolean isValid(VariantContext variant, String infoId,
1378 if (value == null || value.isEmpty() || NO_VALUE.equals(value))
1382 VCFInfoHeaderLine infoHeader = header.getInfoHeaderLine(infoId);
1383 if (infoHeader == null)
1385 Cache.log.error("Field " + infoId + " has no INFO header");
1388 VCFHeaderLineType infoType = infoHeader.getType();
1391 if (infoType == VCFHeaderLineType.Integer)
1393 Integer.parseInt(value);
1395 else if (infoType == VCFHeaderLineType.Float)
1397 Float.parseFloat(value);
1399 } catch (NumberFormatException e)
1401 logInvalidValue(variant, infoId, value);
1408 * Logs an error message for malformed data; duplicate messages (same id and
1409 * value) are not logged
1415 private void logInvalidValue(VariantContext variant, String infoId,
1418 if (badData == null)
1420 badData = new HashSet<>();
1422 String token = infoId + ":" + value;
1423 if (!badData.contains(token))
1426 Cache.log.error(String.format("Invalid VCF data at %s:%d %s=%s",
1427 variant.getContig(), variant.getStart(), infoId, value));
1432 * Inspects CSQ data blocks (consequences) and adds attributes on the sequence
1435 * If <code>myConsequence</code> is not null, then this is the specific
1436 * consequence data (pipe-delimited fields) that is for the current allele and
1437 * transcript (sequence) being processed)
1441 * @param myConsequence
1443 protected void addConsequences(VariantContext variant, SequenceFeature sf,
1444 String myConsequence)
1446 Object value = variant.getAttribute(CSQ_FIELD);
1448 if (value == null || !(value instanceof List<?>))
1453 List<String> consequences = (List<String>) value;
1456 * inspect CSQ consequences; restrict to the consequence
1457 * associated with the current transcript (Feature)
1459 Map<String, String> csqValues = new HashMap<>();
1461 for (String consequence : consequences)
1463 if (myConsequence == null || myConsequence.equals(consequence))
1465 String[] csqFields = consequence.split(PIPE_REGEX);
1468 * inspect individual fields of this consequence, copying non-null
1469 * values which are 'fields of interest'
1472 for (String field : csqFields)
1474 if (field != null && field.length() > 0)
1476 String id = vepFieldsOfInterest.get(i);
1480 * VCF spec requires encoding of special characters e.g. '='
1481 * so decode them here before storing
1483 field = StringUtils.urlDecode(field, VCF_ENCODABLE);
1484 csqValues.put(id, field);
1492 if (!csqValues.isEmpty())
1494 sf.setValue(CSQ_FIELD, csqValues);
1499 * A convenience method to complement a dna base and return the string value
1505 protected String complement(byte[] reference)
1507 return String.valueOf(Dna.getComplement((char) reference[0]));
1511 * Determines the location of the query range (chromosome positions) in a
1512 * different reference assembly.
1514 * If the range is just a subregion of one for which we already have a mapping
1515 * (for example, an exon sub-region of a gene), then the mapping is just
1516 * computed arithmetically.
1518 * Otherwise, calls the Ensembl REST service that maps from one assembly
1519 * reference's coordinates to another's
1522 * start-end chromosomal range in 'fromRef' coordinates
1526 * assembly reference for the query coordinates
1528 * assembly reference we wish to translate to
1529 * @return the start-end range in 'toRef' coordinates
1531 protected int[] mapReferenceRange(int[] queryRange, String chromosome,
1532 String species, String fromRef, String toRef)
1535 * first try shorcut of computing the mapping as a subregion of one
1536 * we already have (e.g. for an exon, if we have the gene mapping)
1538 int[] mappedRange = findSubsumedRangeMapping(queryRange, chromosome,
1539 species, fromRef, toRef);
1540 if (mappedRange != null)
1546 * call (e.g.) http://rest.ensembl.org/map/human/GRCh38/17:45051610..45109016:1/GRCh37
1548 EnsemblMap mapper = new EnsemblMap();
1549 int[] mapping = mapper.getAssemblyMapping(species, chromosome, fromRef,
1552 if (mapping == null)
1554 // mapping service failure
1559 * save mapping for possible future re-use
1561 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1562 if (!assemblyMappings.containsKey(key))
1564 assemblyMappings.put(key, new HashMap<int[], int[]>());
1567 assemblyMappings.get(key).put(queryRange, mapping);
1573 * If we already have a 1:1 contiguous mapping which subsumes the given query
1574 * range, this method just calculates and returns the subset of that mapping,
1575 * else it returns null. In practical terms, if a gene has a contiguous
1576 * mapping between (for example) GRCh37 and GRCh38, then we assume that its
1577 * subsidiary exons occupy unchanged relative positions, and just compute
1578 * these as offsets, rather than do another lookup of the mapping.
1580 * If in future these assumptions prove invalid (e.g. for bacterial dna?!),
1581 * simply remove this method or let it always return null.
1583 * Warning: many rapid calls to the /map service map result in a 429 overload
1593 protected int[] findSubsumedRangeMapping(int[] queryRange, String chromosome,
1594 String species, String fromRef, String toRef)
1596 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1597 if (assemblyMappings.containsKey(key))
1599 Map<int[], int[]> mappedRanges = assemblyMappings.get(key);
1600 for (Entry<int[], int[]> mappedRange : mappedRanges.entrySet())
1602 int[] fromRange = mappedRange.getKey();
1603 int[] toRange = mappedRange.getValue();
1604 if (fromRange[1] - fromRange[0] == toRange[1] - toRange[0])
1607 * mapping is 1:1 in length, so we trust it to have no discontinuities
1609 if (MappingUtils.rangeContains(fromRange, queryRange))
1612 * fromRange subsumes our query range
1614 int offset = queryRange[0] - fromRange[0];
1615 int mappedRangeFrom = toRange[0] + offset;
1616 int mappedRangeTo = mappedRangeFrom + (queryRange[1] - queryRange[0]);
1617 return new int[] { mappedRangeFrom, mappedRangeTo };
1626 * Transfers the sequence feature to the target sequence, locating its start
1627 * and end range based on the mapping. Features which do not overlap the
1628 * target sequence are ignored.
1631 * @param targetSequence
1633 * mapping from the feature's coordinates to the target sequence
1635 protected void transferFeature(SequenceFeature sf,
1636 SequenceI targetSequence, MapList mapping)
1638 int[] mappedRange = mapping.locateInTo(sf.getBegin(), sf.getEnd());
1640 if (mappedRange != null)
1642 String group = sf.getFeatureGroup();
1643 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
1644 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
1645 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
1646 group, sf.getScore());
1647 targetSequence.addSequenceFeature(copy);
1652 * Formats a ranges map lookup key
1660 protected static String makeRangesKey(String chromosome, String species,
1661 String fromRef, String toRef)
1663 return species + EXCL + chromosome + EXCL + fromRef + EXCL