package jalview.io.vcf;
-import htsjdk.samtools.util.CloseableIterator;
-import htsjdk.variant.variantcontext.Allele;
-import htsjdk.variant.variantcontext.VariantContext;
-import htsjdk.variant.vcf.VCFHeader;
-import htsjdk.variant.vcf.VCFHeaderLine;
-
import jalview.analysis.AlignmentUtils;
import jalview.analysis.Dna;
import jalview.api.AlignViewControllerGuiI;
+import jalview.bin.Cache;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.DBRefEntry;
import jalview.datamodel.GeneLociI;
import jalview.datamodel.Mapping;
import jalview.datamodel.SequenceFeature;
import jalview.datamodel.SequenceI;
+import jalview.datamodel.features.FeatureAttributeType;
+import jalview.datamodel.features.FeatureSource;
+import jalview.datamodel.features.FeatureSources;
import jalview.ext.ensembl.EnsemblMap;
import jalview.ext.htsjdk.VCFReader;
import jalview.io.gff.Gff3Helper;
import jalview.util.MessageManager;
import java.io.IOException;
+import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
+import java.util.regex.Pattern;
+import java.util.regex.PatternSyntaxException;
+
+import htsjdk.samtools.SAMException;
+import htsjdk.samtools.SAMSequenceDictionary;
+import htsjdk.samtools.SAMSequenceRecord;
+import htsjdk.samtools.util.CloseableIterator;
+import htsjdk.variant.variantcontext.Allele;
+import htsjdk.variant.variantcontext.VariantContext;
+import htsjdk.variant.vcf.VCFHeader;
+import htsjdk.variant.vcf.VCFHeaderLine;
+import htsjdk.variant.vcf.VCFHeaderLineCount;
+import htsjdk.variant.vcf.VCFHeaderLineType;
+import htsjdk.variant.vcf.VCFInfoHeaderLine;
/**
* A class to read VCF data (using the htsjdk) and add variants as sequence
*/
public class VCFLoader
{
+ /**
+ * A class to model the mapping from sequence to VCF coordinates. Cases include
+ * <ul>
+ * <li>a direct 1:1 mapping where the sequence is one of the VCF contigs</li>
+ * <li>a mapping of sequence to chromosomal coordinates, where sequence and VCF
+ * use the same reference assembly</li>
+ * <li>a modified mapping of sequence to chromosomal coordinates, where sequence
+ * and VCF use different reference assembles</li>
+ * </ul>
+ */
+ class VCFMap
+ {
+ final String chromosome;
+
+ final MapList map;
+
+ VCFMap(String chr, MapList m)
+ {
+ chromosome = chr;
+ map = m;
+ }
+
+ @Override
+ public String toString()
+ {
+ return chromosome + ":" + map.toString();
+ }
+ }
+
+ /*
+ * Lookup keys, and default values, for Preference entries that describe
+ * patterns for VCF and VEP fields to capture
+ */
+ private static final String VEP_FIELDS_PREF = "VEP_FIELDS";
+
+ private static final String VCF_FIELDS_PREF = "VCF_FIELDS";
+
+ private static final String DEFAULT_VCF_FIELDS = ".*";
+
+ private static final String DEFAULT_VEP_FIELDS = ".*";// "Allele,Consequence,IMPACT,SWISSPROT,SIFT,PolyPhen,CLIN_SIG";
+
+ /*
+ * keys to fields of VEP CSQ consequence data
+ * see https://www.ensembl.org/info/docs/tools/vep/vep_formats.html
+ */
+ private static final String CSQ_CONSEQUENCE_KEY = "Consequence";
+ private static final String CSQ_ALLELE_KEY = "Allele";
+ private static final String CSQ_ALLELE_NUM_KEY = "ALLELE_NUM"; // 0 (ref), 1...
+ private static final String CSQ_FEATURE_KEY = "Feature"; // Ensembl stable id
+
+ /*
+ * default VCF INFO key for VEP consequence data
+ * NB this can be overridden running VEP with --vcf_info_field
+ * - we don't handle this case (require identifier to be CSQ)
+ */
+ private static final String CSQ_FIELD = "CSQ";
+
+ /*
+ * separator for fields in consequence data is '|'
+ */
+ private static final String PIPE_REGEX = "\\|";
+
+ /*
+ * key for Allele Frequency output by VEP
+ * see http://www.ensembl.org/info/docs/tools/vep/vep_formats.html
+ */
+ private static final String ALLELE_FREQUENCY_KEY = "AF";
+
+ /*
+ * delimiter that separates multiple consequence data blocks
+ */
+ private static final String COMMA = ",";
+
+ /*
+ * the feature group assigned to a VCF variant in Jalview
+ */
private static final String FEATURE_GROUP_VCF = "VCF";
+ /*
+ * internal delimiter used to build keys for assemblyMappings
+ *
+ */
private static final String EXCL = "!";
/*
- * the alignment we are associated VCF data with
+ * the alignment we are associating VCF data with
*/
private AlignmentI al;
*/
private Map<String, Map<int[], int[]>> assemblyMappings;
+ /*
+ * holds details of the VCF header lines (metadata)
+ */
+ private VCFHeader header;
+
+ /*
+ * a Dictionary of contigs (if present) referenced in the VCF file
+ */
+ private SAMSequenceDictionary dictionary;
+
+ /*
+ * the position (0...) of field in each block of
+ * CSQ (consequence) data (if declared in the VCF INFO header for CSQ)
+ * see http://www.ensembl.org/info/docs/tools/vep/vep_formats.html
+ */
+ private int csqConsequenceFieldIndex = -1;
+ private int csqAlleleFieldIndex = -1;
+ private int csqAlleleNumberFieldIndex = -1;
+ private int csqFeatureFieldIndex = -1;
+
+ // todo the same fields for SnpEff ANN data if wanted
+ // see http://snpeff.sourceforge.net/SnpEff_manual.html#input
+
+ /*
+ * a unique identifier under which to save metadata about feature
+ * attributes (selected INFO field data)
+ */
+ private String sourceId;
+
+ /*
+ * The INFO IDs of data that is both present in the VCF file, and
+ * also matched by any filters for data of interest
+ */
+ List<String> vcfFieldsOfInterest;
+
+ /*
+ * The field offsets and identifiers for VEP (CSQ) data that is both present
+ * in the VCF file, and also matched by any filters for data of interest
+ * for example 0 -> Allele, 1 -> Consequence, ..., 36 -> SIFT, ...
+ */
+ Map<Integer, String> vepFieldsOfInterest;
+
/**
* Constructor given an alignment context
*
al = alignment;
// map of species!chromosome!fromAssembly!toAssembly to {fromRange, toRange}
- assemblyMappings = new HashMap<String, Map<int[], int[]>>();
+ assemblyMappings = new HashMap<>();
}
/**
/**
* Loads VCF on to an alignment - provided it can be related to one or more
- * sequence's chromosomal coordinates.
+ * sequence's chromosomal coordinates
*
* @param filePath
* @param gui
+ * optional callback handler for messages
*/
protected void doLoad(String filePath, AlignViewControllerGuiI gui)
{
// long start = System.currentTimeMillis();
reader = new VCFReader(filePath);
- VCFHeader header = reader.getFileHeader();
+ header = reader.getFileHeader();
+
+ try
+ {
+ dictionary = header.getSequenceDictionary();
+ } catch (SAMException e)
+ {
+ // ignore - thrown if any contig line lacks length info
+ }
+
+ sourceId = filePath;
+
+ saveMetadata(sourceId);
+
+ /*
+ * get offset of CSQ ALLELE_NUM and Feature if declared
+ */
+ parseCsqHeader();
+
VCFHeaderLine ref = header
.getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
- // check if reference is wrt assembly19 (GRCh37)
- // todo may need to allow user to specify reference assembly?
- boolean isRefGrch37 = (ref != null && ref.getValue().contains(
- "assembly19"));
+ String vcfAssembly = ref.getValue();
int varCount = 0;
int seqCount = 0;
*/
for (SequenceI seq : al.getSequences())
{
- int added = loadVCF(seq, reader, isRefGrch37);
+ int added = loadSequenceVCF(seq, reader, vcfAssembly);
if (added > 0)
{
seqCount++;
String msg = MessageManager.formatMessage("label.added_vcf",
varCount, seqCount);
gui.setStatus(msg);
- gui.getFeatureSettingsUI().discoverAllFeatureData();
+ if (gui.getFeatureSettingsUI() != null)
+ {
+ gui.getFeatureSettingsUI().discoverAllFeatureData();
+ }
}
} catch (Throwable e)
{
// ignore
}
}
+ header = null;
+ dictionary = null;
+ }
+ }
+
+ /**
+ * Reads metadata (such as INFO field descriptions and datatypes) and saves
+ * them for future reference
+ *
+ * @param theSourceId
+ */
+ void saveMetadata(String theSourceId)
+ {
+ List<Pattern> vcfFieldPatterns = getFieldMatchers(VCF_FIELDS_PREF,
+ DEFAULT_VCF_FIELDS);
+ vcfFieldsOfInterest = new ArrayList<>();
+
+ FeatureSource metadata = new FeatureSource(theSourceId);
+
+ for (VCFInfoHeaderLine info : header.getInfoHeaderLines())
+ {
+ String attributeId = info.getID();
+ String desc = info.getDescription();
+ VCFHeaderLineType type = info.getType();
+ FeatureAttributeType attType = null;
+ switch (type)
+ {
+ case Character:
+ attType = FeatureAttributeType.Character;
+ break;
+ case Flag:
+ attType = FeatureAttributeType.Flag;
+ break;
+ case Float:
+ attType = FeatureAttributeType.Float;
+ break;
+ case Integer:
+ attType = FeatureAttributeType.Integer;
+ break;
+ case String:
+ attType = FeatureAttributeType.String;
+ break;
+ }
+ metadata.setAttributeName(attributeId, desc);
+ metadata.setAttributeType(attributeId, attType);
+
+ if (isFieldWanted(attributeId, vcfFieldPatterns))
+ {
+ vcfFieldsOfInterest.add(attributeId);
+ }
+ }
+
+ FeatureSources.getInstance().addSource(theSourceId, metadata);
+ }
+
+ /**
+ * Answers true if the field id is matched by any of the filter patterns, else
+ * false. Matching is against regular expression patterns, and is not
+ * case-sensitive.
+ *
+ * @param id
+ * @param filters
+ * @return
+ */
+ private boolean isFieldWanted(String id, List<Pattern> filters)
+ {
+ for (Pattern p : filters)
+ {
+ if (p.matcher(id.toUpperCase()).matches())
+ {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ /**
+ * Records 'wanted' fields defined in the CSQ INFO header (if there is one).
+ * Also records the position of selected fields (Allele, ALLELE_NUM, Feature)
+ * required for processing.
+ * <p>
+ * CSQ fields are declared in the CSQ INFO Description e.g.
+ * <p>
+ * Description="Consequence ...from ... VEP. Format: Allele|Consequence|...
+ */
+ protected void parseCsqHeader()
+ {
+ List<Pattern> vepFieldFilters = getFieldMatchers(VEP_FIELDS_PREF,
+ DEFAULT_VEP_FIELDS);
+ vepFieldsOfInterest = new HashMap<>();
+
+ VCFInfoHeaderLine csqInfo = header.getInfoHeaderLine(CSQ_FIELD);
+ if (csqInfo == null)
+ {
+ return;
+ }
+
+ /*
+ * parse out the pipe-separated list of CSQ fields; we assume here that
+ * these form the last part of the description, and contain no spaces
+ */
+ String desc = csqInfo.getDescription();
+ int spacePos = desc.lastIndexOf(" ");
+ desc = desc.substring(spacePos + 1);
+
+ if (desc != null)
+ {
+ String[] format = desc.split(PIPE_REGEX);
+ int index = 0;
+ for (String field : format)
+ {
+ if (CSQ_CONSEQUENCE_KEY.equals(field))
+ {
+ csqConsequenceFieldIndex = index;
+ }
+ if (CSQ_ALLELE_NUM_KEY.equals(field))
+ {
+ csqAlleleNumberFieldIndex = index;
+ }
+ if (CSQ_ALLELE_KEY.equals(field))
+ {
+ csqAlleleFieldIndex = index;
+ }
+ if (CSQ_FEATURE_KEY.equals(field))
+ {
+ csqFeatureFieldIndex = index;
+ }
+
+ if (isFieldWanted(field, vepFieldFilters))
+ {
+ vepFieldsOfInterest.put(index, field);
+ }
+
+ index++;
+ }
}
}
/**
+ * Reads the Preference value for the given key, with default specified if no
+ * preference set. The value is interpreted as a comma-separated list of
+ * regular expressions, and converted into a list of compiled patterns ready
+ * for matching. Patterns are forced to upper-case for non-case-sensitive
+ * matching.
+ * <p>
+ * This supports user-defined filters for fields of interest to capture while
+ * processing data. For example, VCF_FIELDS = AF,AC* would mean that VCF INFO
+ * fields with an ID of AF, or starting with AC, would be matched.
+ *
+ * @param key
+ * @param def
+ * @return
+ */
+ private List<Pattern> getFieldMatchers(String key, String def)
+ {
+ String pref = Cache.getDefault(key, def);
+ List<Pattern> patterns = new ArrayList<>();
+ String[] tokens = pref.split(",");
+ for (String token : tokens)
+ {
+ try
+ {
+ patterns.add(Pattern.compile(token.toUpperCase()));
+ } catch (PatternSyntaxException e)
+ {
+ System.err.println("Invalid pattern ignored: " + token);
+ }
+ }
+ return patterns;
+ }
+
+ /**
* Transfers VCF features to sequences to which this sequence has a mapping.
- * If the mapping is 1:3, computes peptide variants from nucleotide variants.
+ * If the mapping is 3:1, computes peptide variants from nucleotide variants.
*
* @param seq
*/
/*
* nucleotide-to-nucleotide mapping e.g. transcript to CDS
*/
- // TODO no DBRef to CDS is added to transcripts
List<SequenceFeature> features = seq.getFeatures()
.getPositionalFeatures(SequenceOntologyI.SEQUENCE_VARIANT);
for (SequenceFeature sf : features)
/**
* Tries to add overlapping variants read from a VCF file to the given
- * sequence, and returns the number of overlapping variants found. Note that
- * this requires the sequence to hold information as to its chromosomal
- * positions and reference, in order to be able to map the VCF variants to the
- * sequence.
+ * sequence, and returns the number of variant features added. Note that this
+ * requires the sequence to hold information as to its species, chromosomal
+ * positions and reference assembly, in order to be able to map the VCF
+ * variants to the sequence (or not)
*
* @param seq
* @param reader
- * @param isVcfRefGrch37
+ * @param vcfAssembly
* @return
*/
- protected int loadVCF(SequenceI seq, VCFReader reader,
- boolean isVcfRefGrch37)
+ protected int loadSequenceVCF(SequenceI seq, VCFReader reader,
+ String vcfAssembly)
{
- int count = 0;
- GeneLociI seqCoords = seq.getGeneLoci();
- if (seqCoords == null)
+ VCFMap vcfMap = getVcfMap(seq, vcfAssembly);
+ if (vcfMap == null)
{
return 0;
}
- List<int[]> seqChromosomalContigs = seqCoords.getMap().getToRanges();
- for (int[] range : seqChromosomalContigs)
- {
- count += addVcfVariants(seq, reader, range, isVcfRefGrch37);
- }
-
- return count;
+ return addVcfVariants(seq, reader, vcfMap, vcfAssembly);
}
/**
- * Queries the VCF reader for any variants that overlap the given chromosome
- * region of the sequence, and adds as variant features. Returns the number of
- * overlapping variants found.
+ * Answers a map from sequence coordinates to VCF chromosome ranges
*
* @param seq
- * @param reader
- * @param range
- * start-end range of a sequence region in its chromosomal
- * coordinates
- * @param isVcfRefGrch37
- * true if the VCF is with reference to GRCh37
+ * @param vcfAssembly
* @return
*/
- protected int addVcfVariants(SequenceI seq, VCFReader reader,
- int[] range, boolean isVcfRefGrch37)
+ private VCFMap getVcfMap(SequenceI seq, String vcfAssembly)
{
+ /*
+ * simplest case: sequence has id and length matching a VCF contig
+ */
+ VCFMap vcfMap = null;
+ if (dictionary != null)
+ {
+ vcfMap = getContigMap(seq);
+ }
+ if (vcfMap != null)
+ {
+ return vcfMap;
+ }
+
+ /*
+ * otherwise, map to VCF from chromosomal coordinates
+ * of the sequence (if known)
+ */
GeneLociI seqCoords = seq.getGeneLoci();
+ if (seqCoords == null)
+ {
+ Cache.log.warn(String.format(
+ "Can't query VCF for %s as chromosome coordinates not known",
+ seq.getName()));
+ return null;
+ }
+ String species = seqCoords.getSpeciesId();
String chromosome = seqCoords.getChromosomeId();
String seqRef = seqCoords.getAssemblyId();
- String species = seqCoords.getSpeciesId();
+ MapList map = seqCoords.getMap();
+
+ if (!vcfSpeciesMatchesSequence(vcfAssembly, species))
+ {
+ return null;
+ }
- // TODO handle species properly
- if ("".equals(species))
+ if (vcfAssemblyMatchesSequence(vcfAssembly, seqRef))
{
- species = "human";
+ return new VCFMap(chromosome, map);
+ }
+
+ if (!"GRCh38".equalsIgnoreCase(seqRef) // Ensembl
+ || !vcfAssembly.contains("Homo_sapiens_assembly19")) // gnomAD
+ {
+ return null;
}
/*
- * map chromosomal coordinates from GRCh38 (sequence) to
- * GRCh37 (VCF) if necessary
+ * map chromosomal coordinates from sequence to VCF if the VCF
+ * data has a different reference assembly to the sequence
*/
- // TODO generalise for other assemblies and species
- int offset = 0;
- String fromRef = "GRCh38";
- if (fromRef.equalsIgnoreCase(seqRef) && isVcfRefGrch37)
- {
- String toRef = "GRCh37";
- int[] newRange = mapReferenceRange(range, chromosome, species,
- fromRef, toRef);
+ // TODO generalise for cases other than GRCh38 -> GRCh37 !
+ // - or get the user to choose in a dialog
+
+ List<int[]> toVcfRanges = new ArrayList<>();
+ List<int[]> fromSequenceRanges = new ArrayList<>();
+ String toRef = "GRCh37";
+
+ for (int[] range : map.getToRanges())
+ {
+ int[] fromRange = map.locateInFrom(range[0], range[1]);
+ if (fromRange == null)
+ {
+ // corrupted map?!?
+ continue;
+ }
+
+ int[] newRange = mapReferenceRange(range, chromosome, "human", seqRef,
+ toRef);
if (newRange == null)
{
- System.err.println(String.format(
- "Failed to map %s:%s:%s:%d:%d to %s", species, chromosome,
- fromRef, range[0], range[1], toRef));
- return 0;
+ Cache.log.error(
+ String.format("Failed to map %s:%s:%s:%d:%d to %s", species,
+ chromosome, seqRef, range[0], range[1], toRef));
+ continue;
+ }
+ else
+ {
+ toVcfRanges.add(newRange);
+ fromSequenceRanges.add(fromRange);
+ }
+ }
+
+ return new VCFMap(chromosome,
+ new MapList(fromSequenceRanges, toVcfRanges, 1, 1));
+ }
+
+ /**
+ * If the sequence id matches a contig declared in the VCF file, and the
+ * sequence length matches the contig length, then returns a 1:1 map of the
+ * sequence to the contig, else returns null
+ *
+ * @param seq
+ * @return
+ */
+ private VCFMap getContigMap(SequenceI seq)
+ {
+ String id = seq.getName();
+ SAMSequenceRecord contig = dictionary.getSequence(id);
+ if (contig != null)
+ {
+ int len = seq.getLength();
+ if (len == contig.getSequenceLength())
+ {
+ MapList map = new MapList(new int[] { 1, len },
+ new int[]
+ { 1, len }, 1, 1);
+ return new VCFMap(id, map);
}
- offset = newRange[0] - range[0];
- range = newRange;
}
+ return null;
+ }
- boolean forwardStrand = range[0] <= range[1];
+ /**
+ * Answers true if we determine that the VCF data uses the same reference
+ * assembly as the sequence, else false
+ *
+ * @param vcfAssembly
+ * @param seqRef
+ * @return
+ */
+ private boolean vcfAssemblyMatchesSequence(String vcfAssembly,
+ String seqRef)
+ {
+ // TODO improve on this stub, which handles gnomAD and
+ // hopes for the best for other cases
+
+ if ("GRCh38".equalsIgnoreCase(seqRef) // Ensembl
+ && vcfAssembly.contains("Homo_sapiens_assembly19")) // gnomAD
+ {
+ return false;
+ }
+ return true;
+ }
+
+ /**
+ * Answers true if the species inferred from the VCF reference identifier
+ * matches that for the sequence
+ *
+ * @param vcfAssembly
+ * @param speciesId
+ * @return
+ */
+ boolean vcfSpeciesMatchesSequence(String vcfAssembly, String speciesId)
+ {
+ // PROBLEM 1
+ // there are many aliases for species - how to equate one with another?
+ // PROBLEM 2
+ // VCF ##reference header is an unstructured URI - how to extract species?
+ // perhaps check if ref includes any (Ensembl) alias of speciesId??
+ // TODO ask the user to confirm this??
+
+ if (vcfAssembly.contains("Homo_sapiens") // gnomAD exome data example
+ && "HOMO_SAPIENS".equals(speciesId)) // Ensembl species id
+ {
+ return true;
+ }
+
+ if (vcfAssembly.contains("c_elegans") // VEP VCF response example
+ && "CAENORHABDITIS_ELEGANS".equals(speciesId)) // Ensembl
+ {
+ return true;
+ }
+
+ // this is not a sustainable solution...
+
+ return false;
+ }
+
+ /**
+ * Queries the VCF reader for any variants that overlap the mapped chromosome
+ * ranges of the sequence, and adds as variant features. Returns the number of
+ * overlapping variants found.
+ *
+ * @param seq
+ * @param reader
+ * @param map
+ * mapping from sequence to VCF coordinates
+ * @param vcfAssembly
+ * the '##reference' identifier for the VCF reference assembly
+ * @return
+ */
+ protected int addVcfVariants(SequenceI seq, VCFReader reader,
+ VCFMap map, String vcfAssembly)
+ {
+ boolean forwardStrand = map.map.isToForwardStrand();
/*
- * query the VCF for overlaps
- * (convert a reverse strand range to forwards)
+ * query the VCF for overlaps of each contiguous chromosomal region
*/
int count = 0;
- MapList mapping = seqCoords.getMap();
- int fromLocus = Math.min(range[0], range[1]);
- int toLocus = Math.max(range[0], range[1]);
- CloseableIterator<VariantContext> variants = reader.query(chromosome,
- fromLocus, toLocus);
- while (variants.hasNext())
+ for (int[] range : map.map.getToRanges())
{
- /*
- * get variant location in sequence chromosomal coordinates
- */
- VariantContext variant = variants.next();
-
- /*
- * we can only process SNP variants (which can be reported
- * as part of a MIXED variant record
- */
- if (!variant.isSNP() && !variant.isMixed())
+ int vcfStart = Math.min(range[0], range[1]);
+ int vcfEnd = Math.max(range[0], range[1]);
+ CloseableIterator<VariantContext> variants = reader
+ .query(map.chromosome, vcfStart, vcfEnd);
+ while (variants.hasNext())
{
- continue;
+ VariantContext variant = variants.next();
+
+ int[] featureRange = map.map.locateInFrom(variant.getStart(),
+ variant.getEnd());
+
+ if (featureRange != null)
+ {
+ int featureStart = Math.min(featureRange[0], featureRange[1]);
+ int featureEnd = Math.max(featureRange[0], featureRange[1]);
+ count += addAlleleFeatures(seq, variant, featureStart, featureEnd,
+ forwardStrand);
+ }
}
+ variants.close();
+ }
- count++;
- int start = variant.getStart() - offset;
- int end = variant.getEnd() - offset;
+ return count;
+ }
- /*
- * convert chromosomal location to sequence coordinates
- * - null if a partially overlapping feature
- */
- int[] seqLocation = mapping.locateInFrom(start, end);
- if (seqLocation != null)
+ /**
+ * A convenience method to get the AF value for the given alternate allele
+ * index
+ *
+ * @param variant
+ * @param alleleIndex
+ * @return
+ */
+ protected float getAlleleFrequency(VariantContext variant, int alleleIndex)
+ {
+ float score = 0f;
+ String attributeValue = getAttributeValue(variant,
+ ALLELE_FREQUENCY_KEY, alleleIndex);
+ if (attributeValue != null)
+ {
+ try
{
- addVariantFeatures(seq, variant, seqLocation[0], seqLocation[1],
- forwardStrand);
+ score = Float.parseFloat(attributeValue);
+ } catch (NumberFormatException e)
+ {
+ // leave as 0
}
}
- variants.close();
+ return score;
+ }
- return count;
+ /**
+ * A convenience method to get an attribute value for an alternate allele
+ *
+ * @param variant
+ * @param attributeName
+ * @param alleleIndex
+ * @return
+ */
+ protected String getAttributeValue(VariantContext variant,
+ String attributeName, int alleleIndex)
+ {
+ Object att = variant.getAttribute(attributeName);
+
+ if (att instanceof String)
+ {
+ return (String) att;
+ }
+ else if (att instanceof ArrayList)
+ {
+ return ((List<String>) att).get(alleleIndex);
+ }
+
+ return null;
}
/**
- * Inspects the VCF variant record, and adds variant features to the sequence.
- * Only SNP variants are added, not INDELs.
- * <p>
- * If the sequence maps to the reverse strand of the chromosome, reference and
- * variant bases are recorded as their complements (C/G, A/T).
+ * Adds one variant feature for each allele in the VCF variant record, and
+ * returns the number of features added.
*
* @param seq
* @param variant
* @param featureStart
* @param featureEnd
* @param forwardStrand
+ * @return
*/
- protected void addVariantFeatures(SequenceI seq, VariantContext variant,
+ protected int addAlleleFeatures(SequenceI seq, VariantContext variant,
int featureStart, int featureEnd, boolean forwardStrand)
{
- byte[] reference = variant.getReference().getBases();
- if (reference.length != 1)
+ int added = 0;
+
+ /*
+ * Javadoc says getAlternateAlleles() imposes no order on the list returned
+ * so we proceed defensively to get them in strict order
+ */
+ int altAlleleCount = variant.getAlternateAlleles().size();
+ for (int i = 0; i < altAlleleCount; i++)
+ {
+ added += addAlleleFeature(seq, variant, i, featureStart, featureEnd,
+ forwardStrand);
+ }
+ return added;
+ }
+
+ /**
+ * Inspects one allele and attempts to add a variant feature for it to the
+ * sequence. The additional data associated with this allele is extracted to
+ * store in the feature's key-value map. Answers the number of features added (0
+ * or 1).
+ *
+ * @param seq
+ * @param variant
+ * @param altAlleleIndex
+ * (0, 1..)
+ * @param featureStart
+ * @param featureEnd
+ * @param forwardStrand
+ * @return
+ */
+ protected int addAlleleFeature(SequenceI seq, VariantContext variant,
+ int altAlleleIndex, int featureStart, int featureEnd,
+ boolean forwardStrand)
+ {
+ String reference = variant.getReference().getBaseString();
+ Allele alt = variant.getAlternateAllele(altAlleleIndex);
+ String allele = alt.getBaseString();
+
+ /*
+ * build the ref,alt allele description e.g. "G,A", using the base
+ * complement if the sequence is on the reverse strand
+ */
+ // FIXME correctly handle insertions on reverse strand JAL-2845
+ StringBuilder sb = new StringBuilder();
+ sb.append(forwardStrand ? reference : Dna.reverseComplement(reference));
+ sb.append(COMMA);
+ sb.append(forwardStrand ? allele : Dna.reverseComplement(allele));
+ String alleles = sb.toString(); // e.g. G,A
+
+ String type = getOntologyTerm(seq, variant, altAlleleIndex);
+
+ float score = getAlleleFrequency(variant, altAlleleIndex);
+
+ SequenceFeature sf = new SequenceFeature(type, alleles, featureStart,
+ featureEnd, score, FEATURE_GROUP_VCF);
+ sf.setSource(sourceId);
+
+ sf.setValue(Gff3Helper.ALLELES, alleles);
+
+ addAlleleProperties(variant, seq, sf, altAlleleIndex);
+
+ seq.addSequenceFeature(sf);
+
+ return 1;
+ }
+
+ /**
+ * Determines the Sequence Ontology term to use for the variant feature type in
+ * Jalview. The default is 'sequence_variant', but a more specific term is used
+ * if:
+ * <ul>
+ * <li>VEP (or SnpEff) Consequence annotation is included in the VCF</li>
+ * <li>sequence id can be matched to VEP Feature (or SnpEff Feature_ID)</li>
+ * </ul>
+ *
+ * @param seq
+ * @param variant
+ * @param altAlleleIndex
+ * @return
+ * @see http://www.sequenceontology.org/browser/current_svn/term/SO:0001060
+ */
+ String getOntologyTerm(SequenceI seq, VariantContext variant,
+ int altAlleleIndex)
+ {
+ String type = SequenceOntologyI.SEQUENCE_VARIANT;
+
+ if (csqAlleleFieldIndex == -1) // && snpEffAlleleFieldIndex == -1
{
/*
- * sorry, we don't handle INDEL variants
+ * no Consequence data so we can't refine the ontology term
*/
- return;
+ return type;
}
/*
- * for now we extract allele frequency as feature score; note
- * this attribute is String for a simple SNP, but List<String> if
- * multiple alleles at the locus; we extract for the simple case only
+ * can we associate Consequence data with this allele and feature (transcript)?
+ * if so, prefer the consequence term from that data
*/
- Object af = variant.getAttribute("AF");
- float score = 0f;
- if (af instanceof String)
+ String consequence = getConsequenceForAlleleAndFeature(variant,
+ CSQ_FIELD,
+ altAlleleIndex, csqAlleleFieldIndex, csqAlleleNumberFieldIndex,
+ seq.getName().toLowerCase(), csqFeatureFieldIndex);
+ if (consequence != null)
{
- try
+ String[] csqFields = consequence.split(PIPE_REGEX);
+ if (csqFields.length > csqConsequenceFieldIndex)
{
- score = Float.parseFloat((String) af);
- } catch (NumberFormatException e)
+ type = csqFields[csqConsequenceFieldIndex];
+ }
+ }
+ else
+ {
+ // todo the same for SnpEff consequence data matching if wanted
+ }
+
+ /*
+ * if of the form (e.g.) missense_variant&splice_region_variant,
+ * just take the first ('most severe') consequence
+ */
+ if (type != null)
+ {
+ int pos = type.indexOf('&');
+ if (pos > 0)
{
- // leave as 0
+ type = type.substring(0, pos);
}
}
+ return type;
+ }
- StringBuilder sb = new StringBuilder();
- sb.append(forwardStrand ? (char) reference[0] : complement(reference));
+ /**
+ * Returns matched consequence data if it can be found, else null.
+ * <ul>
+ * <li>inspects the VCF data for key 'vcfInfoId'</li>
+ * <li>splits this on comma (to distinct consequences)</li>
+ * <li>returns the first consequence (if any) where</li>
+ * <ul>
+ * <li>the allele matches the altAlleleIndex'th allele of variant</li>
+ * <li>the feature matches the sequence name (e.g. transcript id)</li>
+ * </ul>
+ * </ul>
+ * If matched, the consequence is returned (as pipe-delimited fields).
+ *
+ * @param variant
+ * @param vcfInfoId
+ * @param altAlleleIndex
+ * @param alleleFieldIndex
+ * @param alleleNumberFieldIndex
+ * @param seqName
+ * @param featureFieldIndex
+ * @return
+ */
+ private String getConsequenceForAlleleAndFeature(VariantContext variant,
+ String vcfInfoId, int altAlleleIndex, int alleleFieldIndex,
+ int alleleNumberFieldIndex,
+ String seqName, int featureFieldIndex)
+ {
+ if (alleleFieldIndex == -1 || featureFieldIndex == -1)
+ {
+ return null;
+ }
+ Object value = variant.getAttribute(vcfInfoId);
+
+ if (value == null || !(value instanceof List<?>))
+ {
+ return null;
+ }
/*
- * inspect alleles and record SNP variants (as the variant
- * record could be MIXED and include INDEL and SNP alleles)
- * warning: getAlleles gives no guarantee as to the order
- * in which they are returned
+ * inspect each consequence in turn (comma-separated blocks
+ * extracted by htsjdk)
*/
- for (Allele allele : variant.getAlleles())
+ List<String> consequences = (List<String>) value;
+
+ for (String consequence : consequences)
{
- if (!allele.isReference())
+ String[] csqFields = consequence.split(PIPE_REGEX);
+ if (csqFields.length > featureFieldIndex)
{
- byte[] alleleBase = allele.getBases();
- if (alleleBase.length == 1)
+ String featureIdentifier = csqFields[featureFieldIndex];
+ if (featureIdentifier.length() > 4
+ && seqName.indexOf(featureIdentifier.toLowerCase()) > -1)
{
- sb.append(",").append(
- forwardStrand ? (char) alleleBase[0]
- : complement(alleleBase));
+ /*
+ * feature (transcript) matched - now check for allele match
+ */
+ if (matchAllele(variant, altAlleleIndex, csqFields,
+ alleleFieldIndex, alleleNumberFieldIndex))
+ {
+ return consequence;
+ }
}
}
}
- String alleles = sb.toString(); // e.g. G,A,C
-
- String type = SequenceOntologyI.SEQUENCE_VARIANT;
+ return null;
+ }
- SequenceFeature sf = new SequenceFeature(type, alleles, featureStart,
- featureEnd, score, FEATURE_GROUP_VCF);
+ private boolean matchAllele(VariantContext variant, int altAlleleIndex,
+ String[] csqFields, int alleleFieldIndex,
+ int alleleNumberFieldIndex)
+ {
+ /*
+ * if ALLELE_NUM is present, it must match altAlleleIndex
+ * NB first alternate allele is 1 for ALLELE_NUM, 0 for altAlleleIndex
+ */
+ if (alleleNumberFieldIndex > -1)
+ {
+ if (csqFields.length <= alleleNumberFieldIndex)
+ {
+ return false;
+ }
+ String alleleNum = csqFields[alleleNumberFieldIndex];
+ return String.valueOf(altAlleleIndex + 1).equals(alleleNum);
+ }
- sf.setValue(Gff3Helper.ALLELES, alleles);
+ /*
+ * else consequence allele must match variant allele
+ */
+ if (alleleFieldIndex > -1 && csqFields.length > alleleFieldIndex)
+ {
+ String csqAllele = csqFields[alleleFieldIndex];
+ String vcfAllele = variant.getAlternateAllele(altAlleleIndex)
+ .getBaseString();
+ return csqAllele.equals(vcfAllele);
+ }
+ return false;
+ }
+ /**
+ * Add any allele-specific VCF key-value data to the sequence feature
+ *
+ * @param variant
+ * @param seq
+ * @param sf
+ * @param altAlelleIndex
+ * (0, 1..)
+ */
+ protected void addAlleleProperties(VariantContext variant, SequenceI seq,
+ SequenceFeature sf, final int altAlelleIndex)
+ {
Map<String, Object> atts = variant.getAttributes();
+
for (Entry<String, Object> att : atts.entrySet())
{
- sf.setValue(att.getKey(), att.getValue());
+ String key = att.getKey();
+
+ /*
+ * extract Consequence data (if present) that we are able to
+ * associated with the allele for this variant feature
+ */
+ if (CSQ_FIELD.equals(key))
+ {
+ addConsequences(variant, seq, sf, altAlelleIndex);
+ continue;
+ }
+
+ /*
+ * filter out fields we don't want to capture
+ */
+ if (!vcfFieldsOfInterest.contains(key))
+ {
+ continue;
+ }
+
+ /*
+ * we extract values for other data which are allele-specific;
+ * these may be per alternate allele (INFO[key].Number = 'A')
+ * or per allele including reference (INFO[key].Number = 'R')
+ */
+ VCFInfoHeaderLine infoHeader = header.getInfoHeaderLine(key);
+ if (infoHeader == null)
+ {
+ /*
+ * can't be sure what data belongs to this allele, so
+ * play safe and don't take any
+ */
+ continue;
+ }
+
+ VCFHeaderLineCount number = infoHeader.getCountType();
+ int index = altAlelleIndex;
+ if (number == VCFHeaderLineCount.R)
+ {
+ /*
+ * one value per allele including reference, so bump index
+ * e.g. the 3rd value is for the 2nd alternate allele
+ */
+ index++;
+ }
+ else if (number != VCFHeaderLineCount.A)
+ {
+ /*
+ * don't save other values as not allele-related
+ */
+ continue;
+ }
+
+ /*
+ * take the index'th value
+ */
+ String value = getAttributeValue(variant, key, index);
+ if (value != null)
+ {
+ sf.setValue(key, value);
+ }
+ }
+ }
+
+ /**
+ * Inspects CSQ data blocks (consequences) and adds attributes on the sequence
+ * feature for the current allele (and transcript if applicable)
+ * <p>
+ * Allele matching: if field ALLELE_NUM is present, it must match
+ * altAlleleIndex. If not present, then field Allele value must match the VCF
+ * Allele.
+ * <p>
+ * Transcript matching: if sequence name can be identified to at least one of
+ * the consequences' Feature values, then select only consequences that match
+ * the value (i.e. consequences for the current transcript sequence). If not,
+ * take all consequences (this is the case when adding features to the gene
+ * sequence).
+ *
+ * @param variant
+ * @param seq
+ * @param sf
+ * @param altAlleleIndex
+ * (0, 1..)
+ */
+ protected void addConsequences(VariantContext variant, SequenceI seq,
+ SequenceFeature sf, int altAlleleIndex)
+ {
+ /*
+ * first try to identify the matching consequence
+ */
+ String myConsequence = getConsequenceForAlleleAndFeature(variant,
+ CSQ_FIELD, altAlleleIndex, csqAlleleFieldIndex,
+ csqAlleleNumberFieldIndex, seq.getName().toLowerCase(),
+ csqFeatureFieldIndex);
+
+ Object value = variant.getAttribute(CSQ_FIELD);
+
+ if (value == null || !(value instanceof List<?>))
+ {
+ return;
+ }
+
+ List<String> consequences = (List<String>) value;
+
+ /*
+ * inspect CSQ consequences; restrict to the consequence
+ * associated with the current transcript (Feature)
+ */
+ Map<String, String> csqValues = new HashMap<>();
+
+ for (String consequence : consequences)
+ {
+ if (myConsequence == null || myConsequence.equals(consequence))
+ {
+ String[] csqFields = consequence.split(PIPE_REGEX);
+
+ /*
+ * inspect individual fields of this consequence, copying non-null
+ * values which are 'fields of interest'
+ */
+ int i = 0;
+ for (String field : csqFields)
+ {
+ if (field != null && field.length() > 0)
+ {
+ String id = vepFieldsOfInterest.get(i);
+ if (id != null)
+ {
+ csqValues.put(id, field);
+ }
+ }
+ i++;
+ }
+ }
+ }
+
+ if (!csqValues.isEmpty())
+ {
+ sf.setValue(CSQ_FIELD, csqValues);
}
- seq.addSequenceFeature(sf);
}
/**
* call (e.g.) http://rest.ensembl.org/map/human/GRCh38/17:45051610..45109016:1/GRCh37
*/
EnsemblMap mapper = new EnsemblMap();
- int[] mapping = mapper.getMapping(species, chromosome, fromRef, toRef,
- queryRange);
+ int[] mapping = mapper.getAssemblyMapping(species, chromosome, fromRef,
+ toRef, queryRange);
if (mapping == null)
{