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.bin.Console;
55 import jalview.datamodel.DBRefEntry;
56 import jalview.datamodel.GeneLociI;
57 import jalview.datamodel.Mapping;
58 import jalview.datamodel.SequenceFeature;
59 import jalview.datamodel.SequenceI;
60 import jalview.datamodel.features.FeatureAttributeType;
61 import jalview.datamodel.features.FeatureSource;
62 import jalview.datamodel.features.FeatureSources;
63 import jalview.ext.ensembl.EnsemblMap;
64 import jalview.ext.htsjdk.HtsContigDb;
65 import jalview.ext.htsjdk.VCFReader;
66 import jalview.io.gff.Gff3Helper;
67 import jalview.io.gff.SequenceOntologyI;
68 import jalview.util.MapList;
69 import jalview.util.MappingUtils;
70 import jalview.util.MessageManager;
71 import jalview.util.StringUtils;
74 * A class to read VCF data (using the htsjdk) and add variants as sequence
75 * features on dna and any related protein product sequences
79 public class VCFLoader
81 private static final String VCF_ENCODABLE = ":;=%,";
84 * Jalview feature attributes for VCF fixed column data
86 private static final String VCF_POS = "POS";
88 private static final String VCF_ID = "ID";
90 private static final String VCF_QUAL = "QUAL";
92 private static final String VCF_FILTER = "FILTER";
94 private static final String NO_VALUE = VCFConstants.MISSING_VALUE_v4; // '.'
96 private static final String DEFAULT_SPECIES = "homo_sapiens";
99 * A class to model the mapping from sequence to VCF coordinates. Cases include
101 * <li>a direct 1:1 mapping where the sequence is one of the VCF contigs</li>
102 * <li>a mapping of sequence to chromosomal coordinates, where sequence and VCF
103 * use the same reference assembly</li>
104 * <li>a modified mapping of sequence to chromosomal coordinates, where sequence
105 * and VCF use different reference assembles</li>
110 final String chromosome;
114 VCFMap(String chr, MapList m)
121 public String toString()
123 return chromosome + ":" + map.toString();
128 * Lookup keys, and default values, for Preference entries that describe
129 * patterns for VCF and VEP fields to capture
131 private static final String VEP_FIELDS_PREF = "VEP_FIELDS";
133 private static final String VCF_FIELDS_PREF = "VCF_FIELDS";
135 private static final String DEFAULT_VCF_FIELDS = ".*";
137 private static final String DEFAULT_VEP_FIELDS = ".*";// "Allele,Consequence,IMPACT,SWISSPROT,SIFT,PolyPhen,CLIN_SIG";
140 * Lookup keys, and default values, for Preference entries that give
141 * mappings from tokens in the 'reference' header to species or assembly
143 private static final String VCF_ASSEMBLY = "VCF_ASSEMBLY";
145 private static final String DEFAULT_VCF_ASSEMBLY = "assembly19=GRCh37,hs37=GRCh37,grch37=GRCh37,grch38=GRCh38";
147 private static final String VCF_SPECIES = "VCF_SPECIES"; // default is human
149 private static final String DEFAULT_REFERENCE = "grch37"; // fallback default is human GRCh37
152 * keys to fields of VEP CSQ consequence data
153 * see https://www.ensembl.org/info/docs/tools/vep/vep_formats.html
155 private static final String CSQ_CONSEQUENCE_KEY = "Consequence";
156 private static final String CSQ_ALLELE_KEY = "Allele";
157 private static final String CSQ_ALLELE_NUM_KEY = "ALLELE_NUM"; // 0 (ref), 1...
158 private static final String CSQ_FEATURE_KEY = "Feature"; // Ensembl stable id
161 * default VCF INFO key for VEP consequence data
162 * NB this can be overridden running VEP with --vcf_info_field
163 * - we don't handle this case (require identifier to be CSQ)
165 private static final String CSQ_FIELD = "CSQ";
168 * separator for fields in consequence data is '|'
170 private static final String PIPE_REGEX = "\\|";
173 * delimiter that separates multiple consequence data blocks
175 private static final String COMMA = ",";
178 * the feature group assigned to a VCF variant in Jalview
180 private static final String FEATURE_GROUP_VCF = "VCF";
183 * internal delimiter used to build keys for assemblyMappings
186 private static final String EXCL = "!";
189 * the VCF file we are processing
191 protected String vcfFilePath;
194 * mappings between VCF and sequence reference assembly regions, as
195 * key = "species!chromosome!fromAssembly!toAssembly
196 * value = Map{fromRange, toRange}
198 private Map<String, Map<int[], int[]>> assemblyMappings;
200 private VCFReader reader;
203 * holds details of the VCF header lines (metadata)
205 private VCFHeader header;
208 * species (as a valid Ensembl term) the VCF is for
210 private String vcfSpecies;
213 * genome assembly version (as a valid Ensembl identifier) the VCF is for
215 private String vcfAssembly;
218 * a Dictionary of contigs (if present) referenced in the VCF file
220 private SAMSequenceDictionary dictionary;
223 * the position (0...) of field in each block of
224 * CSQ (consequence) data (if declared in the VCF INFO header for CSQ)
225 * see http://www.ensembl.org/info/docs/tools/vep/vep_formats.html
227 private int csqConsequenceFieldIndex = -1;
228 private int csqAlleleFieldIndex = -1;
229 private int csqAlleleNumberFieldIndex = -1;
230 private int csqFeatureFieldIndex = -1;
232 // todo the same fields for SnpEff ANN data if wanted
233 // see http://snpeff.sourceforge.net/SnpEff_manual.html#input
236 * a unique identifier under which to save metadata about feature
237 * attributes (selected INFO field data)
239 private String sourceId;
242 * The INFO IDs of data that is both present in the VCF file, and
243 * also matched by any filters for data of interest
245 List<String> vcfFieldsOfInterest;
248 * The field offsets and identifiers for VEP (CSQ) data that is both present
249 * in the VCF file, and also matched by any filters for data of interest
250 * for example 0 -> Allele, 1 -> Consequence, ..., 36 -> SIFT, ...
252 Map<Integer, String> vepFieldsOfInterest;
255 * key:value for which rejected data has been seen
256 * (the error is logged only once for each combination)
258 private Set<String> badData;
261 * Constructor given a VCF file
265 public VCFLoader(String vcfFile)
270 } catch (IOException e)
272 System.err.println("Error opening VCF file: " + e.getMessage());
275 // map of species!chromosome!fromAssembly!toAssembly to {fromRange, toRange}
276 assemblyMappings = new HashMap<>();
280 * Starts a new thread to query and load VCF variant data on to the given
283 * This method is not thread safe - concurrent threads should use separate
284 * instances of this class.
289 public void loadVCF(SequenceI[] seqs, final AlignViewControllerGuiI gui)
293 gui.setStatus(MessageManager.getString("label.searching_vcf"));
301 VCFLoader.this.doLoad(seqs, gui);
307 * Reads the specified contig sequence and adds its VCF variants to it
310 * the id of a single sequence (contig) to load
313 public SequenceI loadVCFContig(String contig)
315 VCFHeaderLine headerLine = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
316 if (headerLine == null)
318 Console.error("VCF reference header not found");
321 String ref = headerLine.getValue();
322 if (ref.startsWith("file://"))
324 ref = ref.substring(7);
326 setSpeciesAndAssembly(ref);
328 SequenceI seq = null;
329 File dbFile = new File(ref);
333 HtsContigDb db = new HtsContigDb("", dbFile);
334 seq = db.getSequenceProxy(contig);
335 loadSequenceVCF(seq);
340 Console.error("VCF reference not found: " + ref);
347 * Loads VCF on to one or more sequences
351 * optional callback handler for messages
353 protected void doLoad(SequenceI[] seqs, AlignViewControllerGuiI gui)
357 VCFHeaderLine ref = header
358 .getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
359 String reference = ref == null ? null : ref.getValue();
361 setSpeciesAndAssembly(reference);
367 * query for VCF overlapping each sequence in turn
369 for (SequenceI seq : seqs)
371 int added = loadSequenceVCF(seq);
376 transferAddedFeatures(seq);
381 String msg = MessageManager.formatMessage("label.added_vcf",
384 if (gui.getFeatureSettingsUI() != null)
386 gui.getFeatureSettingsUI().discoverAllFeatureData();
389 } catch (Throwable e)
391 System.err.println("Error processing VCF: " + e.getMessage());
395 gui.setStatus("Error occurred - see console for details");
404 } catch (IOException e)
415 * Attempts to determine and save the species and genome assembly version to
416 * which the VCF data applies. This may be done by parsing the {@code reference}
417 * header line, configured in a property file, or (potentially) confirmed
418 * interactively by the user.
420 * The saved values should be identifiers valid for Ensembl's REST service
421 * {@code map} endpoint, so they can be used (if necessary) to retrieve the
422 * mapping between VCF coordinates and sequence coordinates.
425 * @see https://rest.ensembl.org/documentation/info/assembly_map
426 * @see https://rest.ensembl.org/info/assembly/human?content-type=text/xml
427 * @see https://rest.ensembl.org/info/species?content-type=text/xml
429 protected void setSpeciesAndAssembly(String reference)
431 if (reference == null)
433 Console.error("No VCF ##reference found, defaulting to "
434 + DEFAULT_REFERENCE + ":" + DEFAULT_SPECIES);
435 reference = DEFAULT_REFERENCE; // default to GRCh37 if not specified
437 reference = reference.toLowerCase(Locale.ROOT);
440 * for a non-human species, or other assembly identifier,
441 * specify as a Jalview property file entry e.g.
442 * VCF_ASSEMBLY = hs37=GRCh37,assembly19=GRCh37
443 * VCF_SPECIES = c_elegans=celegans
444 * to map a token in the reference header to a value
446 String prop = Cache.getDefault(VCF_ASSEMBLY, DEFAULT_VCF_ASSEMBLY);
447 for (String token : prop.split(","))
449 String[] tokens = token.split("=");
450 if (tokens.length == 2)
452 if (reference.contains(tokens[0].trim().toLowerCase(Locale.ROOT)))
454 vcfAssembly = tokens[1].trim();
460 vcfSpecies = DEFAULT_SPECIES;
461 prop = Cache.getProperty(VCF_SPECIES);
464 for (String token : prop.split(","))
466 String[] tokens = token.split("=");
467 if (tokens.length == 2)
469 if (reference.contains(tokens[0].trim().toLowerCase(Locale.ROOT)))
471 vcfSpecies = tokens[1].trim();
480 * Opens the VCF file and parses header data
483 * @throws IOException
485 private void initialise(String filePath) throws IOException
487 vcfFilePath = filePath;
489 reader = new VCFReader(filePath);
491 header = reader.getFileHeader();
495 dictionary = header.getSequenceDictionary();
496 } catch (SAMException e)
498 // ignore - thrown if any contig line lacks length info
503 saveMetadata(sourceId);
506 * get offset of CSQ ALLELE_NUM and Feature if declared
512 * Reads metadata (such as INFO field descriptions and datatypes) and saves
513 * them for future reference
517 void saveMetadata(String theSourceId)
519 List<Pattern> vcfFieldPatterns = getFieldMatchers(VCF_FIELDS_PREF,
521 vcfFieldsOfInterest = new ArrayList<>();
523 FeatureSource metadata = new FeatureSource(theSourceId);
525 for (VCFInfoHeaderLine info : header.getInfoHeaderLines())
527 String attributeId = info.getID();
528 String desc = info.getDescription();
529 VCFHeaderLineType type = info.getType();
530 FeatureAttributeType attType = null;
534 attType = FeatureAttributeType.Character;
537 attType = FeatureAttributeType.Flag;
540 attType = FeatureAttributeType.Float;
543 attType = FeatureAttributeType.Integer;
546 attType = FeatureAttributeType.String;
549 metadata.setAttributeName(attributeId, desc);
550 metadata.setAttributeType(attributeId, attType);
552 if (isFieldWanted(attributeId, vcfFieldPatterns))
554 vcfFieldsOfInterest.add(attributeId);
558 FeatureSources.getInstance().addSource(theSourceId, metadata);
562 * Answers true if the field id is matched by any of the filter patterns, else
563 * false. Matching is against regular expression patterns, and is not
570 private boolean isFieldWanted(String id, List<Pattern> filters)
572 for (Pattern p : filters)
574 if (p.matcher(id.toUpperCase(Locale.ROOT)).matches())
583 * Records 'wanted' fields defined in the CSQ INFO header (if there is one).
584 * Also records the position of selected fields (Allele, ALLELE_NUM, Feature)
585 * required for processing.
587 * CSQ fields are declared in the CSQ INFO Description e.g.
589 * Description="Consequence ...from ... VEP. Format: Allele|Consequence|...
591 protected void parseCsqHeader()
593 List<Pattern> vepFieldFilters = getFieldMatchers(VEP_FIELDS_PREF,
595 vepFieldsOfInterest = new HashMap<>();
597 VCFInfoHeaderLine csqInfo = header.getInfoHeaderLine(CSQ_FIELD);
604 * parse out the pipe-separated list of CSQ fields; we assume here that
605 * these form the last part of the description, and contain no spaces
607 String desc = csqInfo.getDescription();
608 int spacePos = desc.lastIndexOf(" ");
609 desc = desc.substring(spacePos + 1);
613 String[] format = desc.split(PIPE_REGEX);
615 for (String field : format)
617 if (CSQ_CONSEQUENCE_KEY.equals(field))
619 csqConsequenceFieldIndex = index;
621 if (CSQ_ALLELE_NUM_KEY.equals(field))
623 csqAlleleNumberFieldIndex = index;
625 if (CSQ_ALLELE_KEY.equals(field))
627 csqAlleleFieldIndex = index;
629 if (CSQ_FEATURE_KEY.equals(field))
631 csqFeatureFieldIndex = index;
634 if (isFieldWanted(field, vepFieldFilters))
636 vepFieldsOfInterest.put(index, field);
645 * Reads the Preference value for the given key, with default specified if no
646 * preference set. The value is interpreted as a comma-separated list of
647 * regular expressions, and converted into a list of compiled patterns ready
648 * for matching. Patterns are forced to upper-case for non-case-sensitive
651 * This supports user-defined filters for fields of interest to capture while
652 * processing data. For example, VCF_FIELDS = AF,AC* would mean that VCF INFO
653 * fields with an ID of AF, or starting with AC, would be matched.
659 private List<Pattern> getFieldMatchers(String key, String def)
661 String pref = Cache.getDefault(key, def);
662 List<Pattern> patterns = new ArrayList<>();
663 String[] tokens = pref.split(",");
664 for (String token : tokens)
668 patterns.add(Pattern.compile(token.toUpperCase(Locale.ROOT)));
669 } catch (PatternSyntaxException e)
671 System.err.println("Invalid pattern ignored: " + token);
678 * Transfers VCF features to sequences to which this sequence has a mapping.
682 protected void transferAddedFeatures(SequenceI seq)
684 List<DBRefEntry> dbrefs = seq.getDBRefs();
689 for (DBRefEntry dbref : dbrefs)
691 Mapping mapping = dbref.getMap();
692 if (mapping == null || mapping.getTo() == null)
697 SequenceI mapTo = mapping.getTo();
698 MapList map = mapping.getMap();
699 if (map.getFromRatio() == 3)
702 * dna-to-peptide product mapping
704 // JAL-3187 render on the fly instead
705 // AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
710 * nucleotide-to-nucleotide mapping e.g. transcript to CDS
712 List<SequenceFeature> features = seq.getFeatures()
713 .getPositionalFeatures(SequenceOntologyI.SEQUENCE_VARIANT);
714 for (SequenceFeature sf : features)
716 if (FEATURE_GROUP_VCF.equals(sf.getFeatureGroup()))
718 transferFeature(sf, mapTo, map);
726 * Tries to add overlapping variants read from a VCF file to the given sequence,
727 * and returns the number of variant features added
732 protected int loadSequenceVCF(SequenceI seq)
734 VCFMap vcfMap = getVcfMap(seq);
741 * work with the dataset sequence here
743 SequenceI dss = seq.getDatasetSequence();
748 return addVcfVariants(dss, vcfMap);
752 * Answers a map from sequence coordinates to VCF chromosome ranges
757 private VCFMap getVcfMap(SequenceI seq)
760 * simplest case: sequence has id and length matching a VCF contig
762 VCFMap vcfMap = null;
763 if (dictionary != null)
765 vcfMap = getContigMap(seq);
773 * otherwise, map to VCF from chromosomal coordinates
774 * of the sequence (if known)
776 GeneLociI seqCoords = seq.getGeneLoci();
777 if (seqCoords == null)
779 Console.warn(String.format(
780 "Can't query VCF for %s as chromosome coordinates not known",
785 String species = seqCoords.getSpeciesId();
786 String chromosome = seqCoords.getChromosomeId();
787 String seqRef = seqCoords.getAssemblyId();
788 MapList map = seqCoords.getMapping();
790 // note this requires the configured species to match that
791 // returned with the Ensembl sequence; todo: support aliases?
792 if (!vcfSpecies.equalsIgnoreCase(species))
794 Console.warn("No VCF loaded to " + seq.getName()
795 + " as species not matched");
799 if (seqRef.equalsIgnoreCase(vcfAssembly))
801 return new VCFMap(chromosome, map);
805 * VCF data has a different reference assembly to the sequence:
806 * query Ensembl to map chromosomal coordinates from sequence to VCF
808 List<int[]> toVcfRanges = new ArrayList<>();
809 List<int[]> fromSequenceRanges = new ArrayList<>();
811 for (int[] range : map.getToRanges())
813 int[] fromRange = map.locateInFrom(range[0], range[1]);
814 if (fromRange == null)
820 int[] newRange = mapReferenceRange(range, chromosome, "human", seqRef,
822 if (newRange == null)
825 String.format("Failed to map %s:%s:%s:%d:%d to %s", species,
826 chromosome, seqRef, range[0], range[1],
832 toVcfRanges.add(newRange);
833 fromSequenceRanges.add(fromRange);
837 return new VCFMap(chromosome,
838 new MapList(fromSequenceRanges, toVcfRanges, 1, 1));
842 * If the sequence id matches a contig declared in the VCF file, and the
843 * sequence length matches the contig length, then returns a 1:1 map of the
844 * sequence to the contig, else returns null
849 private VCFMap getContigMap(SequenceI seq)
851 String id = seq.getName();
852 SAMSequenceRecord contig = dictionary.getSequence(id);
855 int len = seq.getLength();
856 if (len == contig.getSequenceLength())
858 MapList map = new MapList(new int[] { 1, len },
861 return new VCFMap(id, map);
868 * Queries the VCF reader for any variants that overlap the mapped chromosome
869 * ranges of the sequence, and adds as variant features. Returns the number of
870 * overlapping variants found.
874 * mapping from sequence to VCF coordinates
877 protected int addVcfVariants(SequenceI seq, VCFMap map)
879 boolean forwardStrand = map.map.isToForwardStrand();
882 * query the VCF for overlaps of each contiguous chromosomal region
886 for (int[] range : map.map.getToRanges())
888 int vcfStart = Math.min(range[0], range[1]);
889 int vcfEnd = Math.max(range[0], range[1]);
892 CloseableIterator<VariantContext> variants = reader
893 .query(map.chromosome, vcfStart, vcfEnd);
894 while (variants.hasNext())
896 VariantContext variant = variants.next();
898 int[] featureRange = map.map.locateInFrom(variant.getStart(),
902 * only take features whose range is fully mappable to sequence positions
904 if (featureRange != null)
906 int featureStart = Math.min(featureRange[0], featureRange[1]);
907 int featureEnd = Math.max(featureRange[0], featureRange[1]);
908 if (featureEnd - featureStart == variant.getEnd()
909 - variant.getStart())
911 count += addAlleleFeatures(seq, variant, featureStart,
912 featureEnd, forwardStrand);
917 } catch (TribbleException e)
920 * RuntimeException throwable by htsjdk
922 String msg = String.format("Error reading VCF for %s:%d-%d: %s ",
923 map.chromosome, vcfStart, vcfEnd,e.getLocalizedMessage());
932 * A convenience method to get an attribute value for an alternate allele
935 * @param attributeName
939 protected String getAttributeValue(VariantContext variant,
940 String attributeName, int alleleIndex)
942 Object att = variant.getAttribute(attributeName);
944 if (att instanceof String)
948 else if (att instanceof ArrayList)
950 return ((List<String>) att).get(alleleIndex);
957 * Adds one variant feature for each allele in the VCF variant record, and
958 * returns the number of features added.
962 * @param featureStart
964 * @param forwardStrand
967 protected int addAlleleFeatures(SequenceI seq, VariantContext variant,
968 int featureStart, int featureEnd, boolean forwardStrand)
973 * Javadoc says getAlternateAlleles() imposes no order on the list returned
974 * so we proceed defensively to get them in strict order
976 int altAlleleCount = variant.getAlternateAlleles().size();
977 for (int i = 0; i < altAlleleCount; i++)
979 added += addAlleleFeature(seq, variant, i, featureStart, featureEnd,
986 * Inspects one allele and attempts to add a variant feature for it to the
987 * sequence. The additional data associated with this allele is extracted to
988 * store in the feature's key-value map. Answers the number of features added (0
993 * @param altAlleleIndex
995 * @param featureStart
997 * @param forwardStrand
1000 protected int addAlleleFeature(SequenceI seq, VariantContext variant,
1001 int altAlleleIndex, int featureStart, int featureEnd,
1002 boolean forwardStrand)
1004 String reference = variant.getReference().getBaseString();
1005 Allele alt = variant.getAlternateAllele(altAlleleIndex);
1006 String allele = alt.getBaseString();
1009 * insertion after a genomic base, if on reverse strand, has to be
1010 * converted to insertion of complement after the preceding position
1012 int referenceLength = reference.length();
1013 if (!forwardStrand && allele.length() > referenceLength
1014 && allele.startsWith(reference))
1016 featureStart -= referenceLength;
1017 featureEnd = featureStart;
1018 char insertAfter = seq.getCharAt(featureStart - seq.getStart());
1019 reference = Dna.reverseComplement(String.valueOf(insertAfter));
1020 allele = allele.substring(referenceLength) + reference;
1024 * build the ref,alt allele description e.g. "G,A", using the base
1025 * complement if the sequence is on the reverse strand
1027 StringBuilder sb = new StringBuilder();
1028 sb.append(forwardStrand ? reference : Dna.reverseComplement(reference));
1030 sb.append(forwardStrand ? allele : Dna.reverseComplement(allele));
1031 String alleles = sb.toString(); // e.g. G,A
1034 * pick out the consequence data (if any) that is for the current allele
1035 * and feature (transcript) that matches the current sequence
1037 String consequence = getConsequenceForAlleleAndFeature(variant, CSQ_FIELD,
1038 altAlleleIndex, csqAlleleFieldIndex,
1039 csqAlleleNumberFieldIndex, seq.getName().toLowerCase(Locale.ROOT),
1040 csqFeatureFieldIndex);
1043 * pick out the ontology term for the consequence type
1045 String type = SequenceOntologyI.SEQUENCE_VARIANT;
1046 if (consequence != null)
1048 type = getOntologyTerm(consequence);
1051 SequenceFeature sf = new SequenceFeature(type, alleles, featureStart,
1052 featureEnd, FEATURE_GROUP_VCF);
1053 sf.setSource(sourceId);
1056 * save the derived alleles as a named attribute; this will be
1057 * needed when Jalview computes derived peptide variants
1059 addFeatureAttribute(sf, Gff3Helper.ALLELES, alleles);
1062 * add selected VCF fixed column data as feature attributes
1064 addFeatureAttribute(sf, VCF_POS, String.valueOf(variant.getStart()));
1065 addFeatureAttribute(sf, VCF_ID, variant.getID());
1066 addFeatureAttribute(sf, VCF_QUAL,
1067 String.valueOf(variant.getPhredScaledQual()));
1068 addFeatureAttribute(sf, VCF_FILTER, getFilter(variant));
1070 addAlleleProperties(variant, sf, altAlleleIndex, consequence);
1072 seq.addSequenceFeature(sf);
1078 * Answers the VCF FILTER value for the variant - or an approximation to it.
1079 * This field is either PASS, or a semi-colon separated list of filters not
1080 * passed. htsjdk saves filters as a HashSet, so the order when reassembled into
1081 * a list may be different.
1086 String getFilter(VariantContext variant)
1088 Set<String> filters = variant.getFilters();
1089 if (filters.isEmpty())
1093 Iterator<String> iterator = filters.iterator();
1094 String first = iterator.next();
1095 if (filters.size() == 1)
1100 StringBuilder sb = new StringBuilder(first);
1101 while (iterator.hasNext())
1103 sb.append(";").append(iterator.next());
1106 return sb.toString();
1110 * Adds one feature attribute unless the value is null, empty or '.'
1116 void addFeatureAttribute(SequenceFeature sf, String key, String value)
1118 if (value != null && !value.isEmpty() && !NO_VALUE.equals(value))
1120 sf.setValue(key, value);
1125 * Determines the Sequence Ontology term to use for the variant feature type in
1126 * Jalview. The default is 'sequence_variant', but a more specific term is used
1129 * <li>VEP (or SnpEff) Consequence annotation is included in the VCF</li>
1130 * <li>sequence id can be matched to VEP Feature (or SnpEff Feature_ID)</li>
1133 * @param consequence
1135 * @see http://www.sequenceontology.org/browser/current_svn/term/SO:0001060
1137 String getOntologyTerm(String consequence)
1139 String type = SequenceOntologyI.SEQUENCE_VARIANT;
1142 * could we associate Consequence data with this allele and feature (transcript)?
1143 * if so, prefer the consequence term from that data
1145 if (csqAlleleFieldIndex == -1) // && snpEffAlleleFieldIndex == -1
1148 * no Consequence data so we can't refine the ontology term
1153 if (consequence != null)
1155 String[] csqFields = consequence.split(PIPE_REGEX);
1156 if (csqFields.length > csqConsequenceFieldIndex)
1158 type = csqFields[csqConsequenceFieldIndex];
1163 // todo the same for SnpEff consequence data matching if wanted
1167 * if of the form (e.g.) missense_variant&splice_region_variant,
1168 * just take the first ('most severe') consequence
1172 int pos = type.indexOf('&');
1175 type = type.substring(0, pos);
1182 * Returns matched consequence data if it can be found, else null.
1184 * <li>inspects the VCF data for key 'vcfInfoId'</li>
1185 * <li>splits this on comma (to distinct consequences)</li>
1186 * <li>returns the first consequence (if any) where</li>
1188 * <li>the allele matches the altAlleleIndex'th allele of variant</li>
1189 * <li>the feature matches the sequence name (e.g. transcript id)</li>
1192 * If matched, the consequence is returned (as pipe-delimited fields).
1196 * @param altAlleleIndex
1197 * @param alleleFieldIndex
1198 * @param alleleNumberFieldIndex
1200 * @param featureFieldIndex
1203 private String getConsequenceForAlleleAndFeature(VariantContext variant,
1204 String vcfInfoId, int altAlleleIndex, int alleleFieldIndex,
1205 int alleleNumberFieldIndex,
1206 String seqName, int featureFieldIndex)
1208 if (alleleFieldIndex == -1 || featureFieldIndex == -1)
1212 Object value = variant.getAttribute(vcfInfoId);
1214 if (value == null || !(value instanceof List<?>))
1220 * inspect each consequence in turn (comma-separated blocks
1221 * extracted by htsjdk)
1223 List<String> consequences = (List<String>) value;
1225 for (String consequence : consequences)
1227 String[] csqFields = consequence.split(PIPE_REGEX);
1228 if (csqFields.length > featureFieldIndex)
1230 String featureIdentifier = csqFields[featureFieldIndex];
1231 if (featureIdentifier.length() > 4
1232 && seqName.indexOf(featureIdentifier.toLowerCase(Locale.ROOT)) > -1)
1235 * feature (transcript) matched - now check for allele match
1237 if (matchAllele(variant, altAlleleIndex, csqFields,
1238 alleleFieldIndex, alleleNumberFieldIndex))
1248 private boolean matchAllele(VariantContext variant, int altAlleleIndex,
1249 String[] csqFields, int alleleFieldIndex,
1250 int alleleNumberFieldIndex)
1253 * if ALLELE_NUM is present, it must match altAlleleIndex
1254 * NB first alternate allele is 1 for ALLELE_NUM, 0 for altAlleleIndex
1256 if (alleleNumberFieldIndex > -1)
1258 if (csqFields.length <= alleleNumberFieldIndex)
1262 String alleleNum = csqFields[alleleNumberFieldIndex];
1263 return String.valueOf(altAlleleIndex + 1).equals(alleleNum);
1267 * else consequence allele must match variant allele
1269 if (alleleFieldIndex > -1 && csqFields.length > alleleFieldIndex)
1271 String csqAllele = csqFields[alleleFieldIndex];
1272 String vcfAllele = variant.getAlternateAllele(altAlleleIndex)
1274 return csqAllele.equals(vcfAllele);
1280 * Add any allele-specific VCF key-value data to the sequence feature
1284 * @param altAlelleIndex
1286 * @param consequence
1287 * if not null, the consequence specific to this sequence (transcript
1288 * feature) and allele
1290 protected void addAlleleProperties(VariantContext variant,
1291 SequenceFeature sf, final int altAlelleIndex, String consequence)
1293 Map<String, Object> atts = variant.getAttributes();
1295 for (Entry<String, Object> att : atts.entrySet())
1297 String key = att.getKey();
1300 * extract Consequence data (if present) that we are able to
1301 * associated with the allele for this variant feature
1303 if (CSQ_FIELD.equals(key))
1305 addConsequences(variant, sf, consequence);
1310 * filter out fields we don't want to capture
1312 if (!vcfFieldsOfInterest.contains(key))
1318 * we extract values for other data which are allele-specific;
1319 * these may be per alternate allele (INFO[key].Number = 'A')
1320 * or per allele including reference (INFO[key].Number = 'R')
1322 VCFInfoHeaderLine infoHeader = header.getInfoHeaderLine(key);
1323 if (infoHeader == null)
1326 * can't be sure what data belongs to this allele, so
1327 * play safe and don't take any
1332 VCFHeaderLineCount number = infoHeader.getCountType();
1333 int index = altAlelleIndex;
1334 if (number == VCFHeaderLineCount.R)
1337 * one value per allele including reference, so bump index
1338 * e.g. the 3rd value is for the 2nd alternate allele
1342 else if (number != VCFHeaderLineCount.A)
1345 * don't save other values as not allele-related
1351 * take the index'th value
1353 String value = getAttributeValue(variant, key, index);
1354 if (value != null && isValid(variant, key, value))
1357 * decode colon, semicolon, equals sign, percent sign, comma (only)
1358 * as required by the VCF specification (para 1.2)
1360 value = StringUtils.urlDecode(value, VCF_ENCODABLE);
1361 addFeatureAttribute(sf, key, value);
1367 * Answers true for '.', null, or an empty value, or if the INFO type is String.
1368 * If the INFO type is Integer or Float, answers false if the value is not in
1376 protected boolean isValid(VariantContext variant, String infoId,
1379 if (value == null || value.isEmpty() || NO_VALUE.equals(value))
1383 VCFInfoHeaderLine infoHeader = header.getInfoHeaderLine(infoId);
1384 if (infoHeader == null)
1386 Console.error("Field " + infoId + " has no INFO header");
1389 VCFHeaderLineType infoType = infoHeader.getType();
1392 if (infoType == VCFHeaderLineType.Integer)
1394 Integer.parseInt(value);
1396 else if (infoType == VCFHeaderLineType.Float)
1398 Float.parseFloat(value);
1400 } catch (NumberFormatException e)
1402 logInvalidValue(variant, infoId, value);
1409 * Logs an error message for malformed data; duplicate messages (same id and
1410 * value) are not logged
1416 private void logInvalidValue(VariantContext variant, String infoId,
1419 if (badData == null)
1421 badData = new HashSet<>();
1423 String token = infoId + ":" + value;
1424 if (!badData.contains(token))
1427 Console.error(String.format("Invalid VCF data at %s:%d %s=%s",
1428 variant.getContig(), variant.getStart(), infoId, value));
1433 * Inspects CSQ data blocks (consequences) and adds attributes on the sequence
1436 * If <code>myConsequence</code> is not null, then this is the specific
1437 * consequence data (pipe-delimited fields) that is for the current allele and
1438 * transcript (sequence) being processed)
1442 * @param myConsequence
1444 protected void addConsequences(VariantContext variant, SequenceFeature sf,
1445 String myConsequence)
1447 Object value = variant.getAttribute(CSQ_FIELD);
1449 if (value == null || !(value instanceof List<?>))
1454 List<String> consequences = (List<String>) value;
1457 * inspect CSQ consequences; restrict to the consequence
1458 * associated with the current transcript (Feature)
1460 Map<String, String> csqValues = new HashMap<>();
1462 for (String consequence : consequences)
1464 if (myConsequence == null || myConsequence.equals(consequence))
1466 String[] csqFields = consequence.split(PIPE_REGEX);
1469 * inspect individual fields of this consequence, copying non-null
1470 * values which are 'fields of interest'
1473 for (String field : csqFields)
1475 if (field != null && field.length() > 0)
1477 String id = vepFieldsOfInterest.get(i);
1481 * VCF spec requires encoding of special characters e.g. '='
1482 * so decode them here before storing
1484 field = StringUtils.urlDecode(field, VCF_ENCODABLE);
1485 csqValues.put(id, field);
1493 if (!csqValues.isEmpty())
1495 sf.setValue(CSQ_FIELD, csqValues);
1500 * A convenience method to complement a dna base and return the string value
1506 protected String complement(byte[] reference)
1508 return String.valueOf(Dna.getComplement((char) reference[0]));
1512 * Determines the location of the query range (chromosome positions) in a
1513 * different reference assembly.
1515 * If the range is just a subregion of one for which we already have a mapping
1516 * (for example, an exon sub-region of a gene), then the mapping is just
1517 * computed arithmetically.
1519 * Otherwise, calls the Ensembl REST service that maps from one assembly
1520 * reference's coordinates to another's
1523 * start-end chromosomal range in 'fromRef' coordinates
1527 * assembly reference for the query coordinates
1529 * assembly reference we wish to translate to
1530 * @return the start-end range in 'toRef' coordinates
1532 protected int[] mapReferenceRange(int[] queryRange, String chromosome,
1533 String species, String fromRef, String toRef)
1536 * first try shorcut of computing the mapping as a subregion of one
1537 * we already have (e.g. for an exon, if we have the gene mapping)
1539 int[] mappedRange = findSubsumedRangeMapping(queryRange, chromosome,
1540 species, fromRef, toRef);
1541 if (mappedRange != null)
1547 * call (e.g.) http://rest.ensembl.org/map/human/GRCh38/17:45051610..45109016:1/GRCh37
1549 EnsemblMap mapper = new EnsemblMap();
1550 int[] mapping = mapper.getAssemblyMapping(species, chromosome, fromRef,
1553 if (mapping == null)
1555 // mapping service failure
1560 * save mapping for possible future re-use
1562 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1563 if (!assemblyMappings.containsKey(key))
1565 assemblyMappings.put(key, new HashMap<int[], int[]>());
1568 assemblyMappings.get(key).put(queryRange, mapping);
1574 * If we already have a 1:1 contiguous mapping which subsumes the given query
1575 * range, this method just calculates and returns the subset of that mapping,
1576 * else it returns null. In practical terms, if a gene has a contiguous
1577 * mapping between (for example) GRCh37 and GRCh38, then we assume that its
1578 * subsidiary exons occupy unchanged relative positions, and just compute
1579 * these as offsets, rather than do another lookup of the mapping.
1581 * If in future these assumptions prove invalid (e.g. for bacterial dna?!),
1582 * simply remove this method or let it always return null.
1584 * Warning: many rapid calls to the /map service map result in a 429 overload
1594 protected int[] findSubsumedRangeMapping(int[] queryRange, String chromosome,
1595 String species, String fromRef, String toRef)
1597 String key = makeRangesKey(chromosome, species, fromRef, toRef);
1598 if (assemblyMappings.containsKey(key))
1600 Map<int[], int[]> mappedRanges = assemblyMappings.get(key);
1601 for (Entry<int[], int[]> mappedRange : mappedRanges.entrySet())
1603 int[] fromRange = mappedRange.getKey();
1604 int[] toRange = mappedRange.getValue();
1605 if (fromRange[1] - fromRange[0] == toRange[1] - toRange[0])
1608 * mapping is 1:1 in length, so we trust it to have no discontinuities
1610 if (MappingUtils.rangeContains(fromRange, queryRange))
1613 * fromRange subsumes our query range
1615 int offset = queryRange[0] - fromRange[0];
1616 int mappedRangeFrom = toRange[0] + offset;
1617 int mappedRangeTo = mappedRangeFrom + (queryRange[1] - queryRange[0]);
1618 return new int[] { mappedRangeFrom, mappedRangeTo };
1627 * Transfers the sequence feature to the target sequence, locating its start
1628 * and end range based on the mapping. Features which do not overlap the
1629 * target sequence are ignored.
1632 * @param targetSequence
1634 * mapping from the feature's coordinates to the target sequence
1636 protected void transferFeature(SequenceFeature sf,
1637 SequenceI targetSequence, MapList mapping)
1639 int[] mappedRange = mapping.locateInTo(sf.getBegin(), sf.getEnd());
1641 if (mappedRange != null)
1643 String group = sf.getFeatureGroup();
1644 int newBegin = Math.min(mappedRange[0], mappedRange[1]);
1645 int newEnd = Math.max(mappedRange[0], mappedRange[1]);
1646 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
1647 group, sf.getScore());
1648 targetSequence.addSequenceFeature(copy);
1653 * Formats a ranges map lookup key
1661 protected static String makeRangesKey(String chromosome, String species,
1662 String fromRef, String toRef)
1664 return species + EXCL + chromosome + EXCL + fromRef + EXCL