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 htsjdk.variant.vcf.VCFHeaderLineCount;
-import htsjdk.variant.vcf.VCFInfoHeaderLine;
-
-import jalview.analysis.AlignmentUtils;
import jalview.analysis.Dna;
import jalview.api.AlignViewControllerGuiI;
-import jalview.datamodel.AlignmentI;
+import jalview.bin.Cache;
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.HtsContigDb;
import jalview.ext.htsjdk.VCFReader;
import jalview.io.gff.Gff3Helper;
import jalview.io.gff.SequenceOntologyI;
import jalview.util.MappingUtils;
import jalview.util.MessageManager;
+import java.io.File;
import java.io.IOException;
+import java.io.UnsupportedEncodingException;
+import java.net.URLDecoder;
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
{
- /*
- * keys to fields of VEP CSQ consequence data
- * see https://www.ensembl.org/info/docs/tools/vep/vep_formats.html
+ private static final String UTF_8 = "UTF-8";
+
+ private static final String DEFAULT_SPECIES = "homo_sapiens";
+
+ /**
+ * 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>
*/
- private static final String ALLELE_KEY = "Allele";
+ class VCFMap
+ {
+ final String chromosome;
+
+ final MapList map;
- private static final String ALLELE_NUM_KEY = "ALLELE_NUM"; // 0 (ref), 1...
- private static final String FEATURE_KEY = "Feature"; // Ensembl stable id
+ VCFMap(String chr, MapList m)
+ {
+ chromosome = chr;
+ map = m;
+ }
+
+ @Override
+ public String toString()
+ {
+ return chromosome + ":" + map.toString();
+ }
+ }
/*
- * what comes before column headings in CSQ Description field
+ * Lookup keys, and default values, for Preference entries that describe
+ * patterns for VCF and VEP fields to capture
*/
- private static final String FORMAT = "Format: ";
+ 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";
/*
- * 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 CSQ identifier)
+ * Lookup keys, and default values, for Preference entries that give
+ * mappings from tokens in the 'reference' header to species or assembly
*/
- private static final String CSQ = "CSQ";
+ private static final String VCF_ASSEMBLY = "VCF_ASSEMBLY";
+
+ private static final String DEFAULT_VCF_ASSEMBLY = "assembly19=GRCh37,hs37=GRCh37,grch37=GRCh37,grch38=GRCh38";
+
+ private static final String VCF_SPECIES = "VCF_SPECIES"; // default is human
+
+ private static final String DEFAULT_REFERENCE = "grch37"; // fallback default is human GRCh37
/*
- * separator for fields in consequence data
+ * keys to fields of VEP CSQ consequence data
+ * see https://www.ensembl.org/info/docs/tools/vep/vep_formats.html
*/
- private static final String PIPE = "|";
+ 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
- private static final String PIPE_REGEX = "\\" + PIPE;
+ /*
+ * 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";
/*
- * key for Allele Frequency output by VEP
- * see http://www.ensembl.org/info/docs/tools/vep/vep_formats.html
+ * separator for fields in consequence data is '|'
*/
- private static final String ALLELE_FREQUENCY_KEY = "AF";
+ private static final String PIPE_REGEX = "\\|";
/*
* delimiter that separates multiple consequence data blocks
private static final String EXCL = "!";
/*
- * the alignment we are associating VCF data with
+ * the VCF file we are processing
*/
- private AlignmentI al;
+ protected String vcfFilePath;
/*
* mappings between VCF and sequence reference assembly regions, as
*/
private Map<String, Map<int[], int[]>> assemblyMappings;
+ private VCFReader reader;
+
/*
* holds details of the VCF header lines (metadata)
*/
private VCFHeader header;
/*
+ * species (as a valid Ensembl term) the VCF is for
+ */
+ private String vcfSpecies;
+
+ /*
+ * genome assembly version (as a valid Ensembl identifier) the VCF is for
+ */
+ private String vcfAssembly;
+
+ /*
+ * 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
+ * Constructor given a VCF file
*
* @param alignment
*/
- public VCFLoader(AlignmentI alignment)
+ public VCFLoader(String vcfFile)
{
- al = alignment;
+ try
+ {
+ initialise(vcfFile);
+ } catch (IOException e)
+ {
+ System.err.println("Error opening VCF file: " + e.getMessage());
+ }
// map of species!chromosome!fromAssembly!toAssembly to {fromRange, toRange}
- assemblyMappings = new HashMap<String, Map<int[], int[]>>();
+ assemblyMappings = new HashMap<>();
}
/**
- * Starts a new thread to query and load VCF variant data on to the alignment
+ * Starts a new thread to query and load VCF variant data on to the given
+ * sequences
* <p>
* This method is not thread safe - concurrent threads should use separate
* instances of this class.
*
- * @param filePath
+ * @param seqs
* @param gui
*/
- public void loadVCF(final String filePath,
- final AlignViewControllerGuiI gui)
+ public void loadVCF(SequenceI[] seqs, final AlignViewControllerGuiI gui)
{
if (gui != null)
{
new Thread()
{
-
@Override
public void run()
{
- VCFLoader.this.doLoad(filePath, gui);
+ VCFLoader.this.doLoad(seqs, gui);
}
-
}.start();
}
/**
- * Loads VCF on to an alignment - provided it can be related to one or more
- * sequence's chromosomal coordinates
+ * Reads the specified contig sequence and adds its VCF variants to it
*
- * @param filePath
+ * @param contig
+ * the id of a single sequence (contig) to load
+ * @return
+ */
+ public SequenceI loadVCFContig(String contig)
+ {
+ VCFHeaderLine headerLine = header.getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
+ if (headerLine == null)
+ {
+ Cache.log.error("VCF reference header not found");
+ return null;
+ }
+ String ref = headerLine.getValue();
+ if (ref.startsWith("file://"))
+ {
+ ref = ref.substring(7);
+ }
+ setSpeciesAndAssembly(ref);
+
+ SequenceI seq = null;
+ File dbFile = new File(ref);
+
+ if (dbFile.exists())
+ {
+ HtsContigDb db = new HtsContigDb("", dbFile);
+ seq = db.getSequenceProxy(contig);
+ loadSequenceVCF(seq);
+ db.close();
+ }
+ else
+ {
+ Cache.log.error("VCF reference not found: " + ref);
+ }
+
+ return seq;
+ }
+
+ /**
+ * Loads VCF on to one or more sequences
+ *
+ * @param seqs
* @param gui
* optional callback handler for messages
*/
- protected void doLoad(String filePath, AlignViewControllerGuiI gui)
+ protected void doLoad(SequenceI[] seqs, AlignViewControllerGuiI gui)
{
- VCFReader reader = null;
try
{
- // long start = System.currentTimeMillis();
- reader = new VCFReader(filePath);
-
- header = reader.getFileHeader();
VCFHeaderLine ref = header
.getOtherHeaderLine(VCFHeader.REFERENCE_KEY);
+ String reference = ref == null ? null : ref.getValue();
- /*
- * get offset of CSQ ALLELE_NUM and Feature if declared
- */
- locateCsqFields();
-
- // 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"));
+ setSpeciesAndAssembly(reference);
int varCount = 0;
int seqCount = 0;
/*
* query for VCF overlapping each sequence in turn
*/
- for (SequenceI seq : al.getSequences())
+ for (SequenceI seq : seqs)
{
- int added = loadSequenceVCF(seq, reader, isRefGrch37);
+ int added = loadSequenceVCF(seq);
if (added > 0)
{
seqCount++;
}
if (gui != null)
{
- // long elapsed = System.currentTimeMillis() - start;
String msg = MessageManager.formatMessage("label.added_vcf",
varCount, seqCount);
gui.setStatus(msg);
// ignore
}
}
+ header = null;
+ dictionary = null;
}
}
/**
- * Records the position of selected fields defined in the CSQ INFO header (if
- * there is one). CSQ fields are declared in the CSQ INFO Description e.g.
+ * Attempts to determine and save the species and genome assembly version to
+ * which the VCF data applies. This may be done by parsing the {@code reference}
+ * header line, configured in a property file, or (potentially) confirmed
+ * interactively by the user.
+ * <p>
+ * The saved values should be identifiers valid for Ensembl's REST service
+ * {@code map} endpoint, so they can be used (if necessary) to retrieve the
+ * mapping between VCF coordinates and sequence coordinates.
+ *
+ * @param reference
+ * @see https://rest.ensembl.org/documentation/info/assembly_map
+ * @see https://rest.ensembl.org/info/assembly/human?content-type=text/xml
+ * @see https://rest.ensembl.org/info/species?content-type=text/xml
+ */
+ protected void setSpeciesAndAssembly(String reference)
+ {
+ if (reference == null)
+ {
+ Cache.log.error("No VCF ##reference found, defaulting to "
+ + DEFAULT_REFERENCE + ":" + DEFAULT_SPECIES);
+ reference = DEFAULT_REFERENCE; // default to GRCh37 if not specified
+ }
+ reference = reference.toLowerCase();
+
+ /*
+ * for a non-human species, or other assembly identifier,
+ * specify as a Jalview property file entry e.g.
+ * VCF_ASSEMBLY = hs37=GRCh37,assembly19=GRCh37
+ * VCF_SPECIES = c_elegans=celegans
+ * to map a token in the reference header to a value
+ */
+ String prop = Cache.getDefault(VCF_ASSEMBLY, DEFAULT_VCF_ASSEMBLY);
+ for (String token : prop.split(","))
+ {
+ String[] tokens = token.split("=");
+ if (tokens.length == 2)
+ {
+ if (reference.contains(tokens[0].trim().toLowerCase()))
+ {
+ vcfAssembly = tokens[1].trim();
+ break;
+ }
+ }
+ }
+
+ vcfSpecies = DEFAULT_SPECIES;
+ prop = Cache.getProperty(VCF_SPECIES);
+ if (prop != null)
+ {
+ for (String token : prop.split(","))
+ {
+ String[] tokens = token.split("=");
+ if (tokens.length == 2)
+ {
+ if (reference.contains(tokens[0].trim().toLowerCase()))
+ {
+ vcfSpecies = tokens[1].trim();
+ break;
+ }
+ }
+ }
+ }
+ }
+
+ /**
+ * Opens the VCF file and parses header data
+ *
+ * @param filePath
+ * @throws IOException
+ */
+ private void initialise(String filePath) throws IOException
+ {
+ vcfFilePath = filePath;
+
+ reader = new VCFReader(filePath);
+
+ 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();
+ }
+
+ /**
+ * 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 locateCsqFields()
+ protected void parseCsqHeader()
{
- VCFInfoHeaderLine csqInfo = header.getInfoHeaderLine(CSQ);
+ 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 formatPos = desc.indexOf(FORMAT);
- if (formatPos == -1)
- {
- System.err.println("Parse error, failed to find " + FORMAT
- + " in " + desc);
- return;
- }
- desc = desc.substring(formatPos + FORMAT.length());
+ int spacePos = desc.lastIndexOf(" ");
+ desc = desc.substring(spacePos + 1);
if (desc != null)
{
int index = 0;
for (String field : format)
{
- if (ALLELE_NUM_KEY.equals(field))
+ if (CSQ_CONSEQUENCE_KEY.equals(field))
+ {
+ csqConsequenceFieldIndex = index;
+ }
+ if (CSQ_ALLELE_NUM_KEY.equals(field))
{
csqAlleleNumberFieldIndex = index;
}
- if (ALLELE_KEY.equals(field))
+ if (CSQ_ALLELE_KEY.equals(field))
{
csqAlleleFieldIndex = index;
}
- if (FEATURE_KEY.equals(field))
+ 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
*/
/*
* dna-to-peptide product mapping
*/
- AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
+ // JAL-3187 render on the fly instead
+ // AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
}
else
{
}
/**
- * Tries to add overlapping variants read from a VCF file to the given
- * sequence, and returns the number of variant features added. 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.
+ * Tries to add overlapping variants read from a VCF file to the given sequence,
+ * and returns the number of variant features added
*
* @param seq
- * @param reader
- * @param isVcfRefGrch37
* @return
*/
- protected int loadSequenceVCF(SequenceI seq, VCFReader reader,
- boolean isVcfRefGrch37)
+ protected int loadSequenceVCF(SequenceI seq)
{
- int count = 0;
- GeneLociI seqCoords = seq.getGeneLoci();
- if (seqCoords == null)
+ VCFMap vcfMap = getVcfMap(seq);
+ if (vcfMap == null)
{
- System.out.println(String.format(
- "Can't query VCF for %s as chromosome coordinates not known",
- seq.getName()));
return 0;
}
- List<int[]> seqChromosomalContigs = seqCoords.getMap().getToRanges();
- for (int[] range : seqChromosomalContigs)
+ /*
+ * work with the dataset sequence here
+ */
+ SequenceI dss = seq.getDatasetSequence();
+ if (dss == null)
{
- count += addVcfVariants(seq, reader, range, isVcfRefGrch37);
+ dss = seq;
}
-
- return count;
+ return addVcfVariants(dss, vcfMap);
}
/**
- * 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
* @return
*/
- protected int addVcfVariants(SequenceI seq, VCFReader reader,
- int[] range, boolean isVcfRefGrch37)
+ private VCFMap getVcfMap(SequenceI seq)
{
+ /*
+ * 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.getMapping();
- /*
- * map chromosomal coordinates from GRCh38 (sequence) to
- * GRCh37 (VCF) if necessary
- */
- // 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, "human",
- fromRef, 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;
- }
- offset = newRange[0] - range[0];
- range = newRange;
+ // note this requires the configured species to match that
+ // returned with the Ensembl sequence; todo: support aliases?
+ if (!vcfSpecies.equalsIgnoreCase(species))
+ {
+ Cache.log.warn("No VCF loaded to " + seq.getName()
+ + " as species not matched");
+ return null;
}
- boolean forwardStrand = range[0] <= range[1];
+ if (seqRef.equalsIgnoreCase(vcfAssembly))
+ {
+ return new VCFMap(chromosome, map);
+ }
/*
- * query the VCF for overlaps
- * (convert a reverse strand range to forwards)
+ * VCF data has a different reference assembly to the sequence:
+ * query Ensembl to map chromosomal coordinates from sequence to VCF
*/
- int count = 0;
- MapList mapping = seqCoords.getMap();
+ List<int[]> toVcfRanges = new ArrayList<>();
+ List<int[]> fromSequenceRanges = new ArrayList<>();
- 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.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[] fromRange = map.locateInFrom(range[0], range[1]);
+ if (fromRange == null)
{
- // continue;
+ // corrupted map?!?
+ continue;
}
- int start = variant.getStart() - offset;
- int end = variant.getEnd() - offset;
-
- /*
- * convert chromosomal location to sequence coordinates
- * - null if a partially overlapping feature
- */
- int[] seqLocation = mapping.locateInFrom(start, end);
- if (seqLocation != null)
+ int[] newRange = mapReferenceRange(range, chromosome, "human", seqRef,
+ vcfAssembly);
+ if (newRange == null)
+ {
+ Cache.log.error(
+ String.format("Failed to map %s:%s:%s:%d:%d to %s", species,
+ chromosome, seqRef, range[0], range[1],
+ vcfAssembly));
+ continue;
+ }
+ else
{
- count += addAlleleFeatures(seq, variant, seqLocation[0],
- seqLocation[1], forwardStrand);
+ toVcfRanges.add(newRange);
+ fromSequenceRanges.add(fromRange);
}
}
- variants.close();
-
- return count;
+ return new VCFMap(chromosome,
+ new MapList(fromSequenceRanges, toVcfRanges, 1, 1));
}
/**
- * A convenience method to get the AF value for the given alternate allele
- * index
+ * 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 variant
- * @param alleleIndex
+ * @param seq
* @return
*/
- protected float getAlleleFrequency(VariantContext variant, int alleleIndex)
+ private VCFMap getContigMap(SequenceI seq)
{
- float score = 0f;
- String attributeValue = getAttributeValue(variant,
- ALLELE_FREQUENCY_KEY, alleleIndex);
- if (attributeValue != null)
+ String id = seq.getName();
+ SAMSequenceRecord contig = dictionary.getSequence(id);
+ if (contig != null)
{
- try
+ int len = seq.getLength();
+ if (len == contig.getSequenceLength())
{
- score = Float.parseFloat(attributeValue);
- } catch (NumberFormatException e)
+ MapList map = new MapList(new int[] { 1, len },
+ new int[]
+ { 1, len }, 1, 1);
+ return new VCFMap(id, map);
+ }
+ }
+ return null;
+ }
+
+ /**
+ * 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 map
+ * mapping from sequence to VCF coordinates
+ * @return
+ */
+ protected int addVcfVariants(SequenceI seq, VCFMap map)
+ {
+ boolean forwardStrand = map.map.isToForwardStrand();
+
+ /*
+ * query the VCF for overlaps of each contiguous chromosomal region
+ */
+ int count = 0;
+
+ for (int[] range : map.map.getToRanges())
+ {
+ 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())
{
- // leave as 0
+ 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();
}
- return score;
+ return count;
}
/**
/**
* Inspects one allele and attempts to add a variant feature for it to the
- * sequence. We extract as much as possible of the additional data associated
- * with this allele to store in the feature's key-value map. Answers the
- * number of features added (0 or 1).
+ * 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
String reference = variant.getReference().getBaseString();
Allele alt = variant.getAlternateAllele(altAlleleIndex);
String allele = alt.getBaseString();
- if (allele.length() != 1)
+
+ /*
+ * insertion after a genomic base, if on reverse strand, has to be
+ * converted to insertion of complement after the preceding position
+ */
+ int referenceLength = reference.length();
+ if (!forwardStrand && allele.length() > referenceLength
+ && allele.startsWith(reference))
{
- /*
- * not a SNP variant
- */
- // return 0;
+ featureStart -= referenceLength;
+ featureEnd = featureStart;
+ char insertAfter = seq.getCharAt(featureStart - seq.getStart());
+ reference = Dna.reverseComplement(String.valueOf(insertAfter));
+ allele = allele.substring(referenceLength) + reference;
}
/*
* build the ref,alt allele description e.g. "G,A", using the base
* complement if the sequence is on the reverse strand
*/
- // TODO check how structural variants are shown on reverse strand
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
+ /*
+ * pick out the consequence data (if any) that is for the current allele
+ * and feature (transcript) that matches the current sequence
+ */
+ String consequence = getConsequenceForAlleleAndFeature(variant, CSQ_FIELD,
+ altAlleleIndex, csqAlleleFieldIndex,
+ csqAlleleNumberFieldIndex, seq.getName().toLowerCase(),
+ csqFeatureFieldIndex);
+
+ /*
+ * pick out the ontology term for the consequence type
+ */
String type = SequenceOntologyI.SEQUENCE_VARIANT;
- float score = getAlleleFrequency(variant, altAlleleIndex);
+ if (consequence != null)
+ {
+ type = getOntologyTerm(consequence);
+ }
SequenceFeature sf = new SequenceFeature(type, alleles, featureStart,
- featureEnd, score, FEATURE_GROUP_VCF);
+ featureEnd, FEATURE_GROUP_VCF);
+ sf.setSource(sourceId);
sf.setValue(Gff3Helper.ALLELES, alleles);
- addAlleleProperties(variant, seq, sf, altAlleleIndex);
+ addAlleleProperties(variant, sf, altAlleleIndex, consequence);
seq.addSequenceFeature(sf);
}
/**
+ * 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 consequence
+ * @return
+ * @see http://www.sequenceontology.org/browser/current_svn/term/SO:0001060
+ */
+ String getOntologyTerm(String consequence)
+ {
+ String type = SequenceOntologyI.SEQUENCE_VARIANT;
+
+ /*
+ * could we associate Consequence data with this allele and feature (transcript)?
+ * if so, prefer the consequence term from that data
+ */
+ if (csqAlleleFieldIndex == -1) // && snpEffAlleleFieldIndex == -1
+ {
+ /*
+ * no Consequence data so we can't refine the ontology term
+ */
+ return type;
+ }
+
+ if (consequence != null)
+ {
+ String[] csqFields = consequence.split(PIPE_REGEX);
+ if (csqFields.length > csqConsequenceFieldIndex)
+ {
+ 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)
+ {
+ type = type.substring(0, pos);
+ }
+ }
+ return type;
+ }
+
+ /**
+ * 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 each consequence in turn (comma-separated blocks
+ * extracted by htsjdk)
+ */
+ List<String> consequences = (List<String>) value;
+
+ for (String consequence : consequences)
+ {
+ String[] csqFields = consequence.split(PIPE_REGEX);
+ if (csqFields.length > featureFieldIndex)
+ {
+ String featureIdentifier = csqFields[featureFieldIndex];
+ if (featureIdentifier.length() > 4
+ && seqName.indexOf(featureIdentifier.toLowerCase()) > -1)
+ {
+ /*
+ * feature (transcript) matched - now check for allele match
+ */
+ if (matchAllele(variant, altAlleleIndex, csqFields,
+ alleleFieldIndex, alleleNumberFieldIndex))
+ {
+ return consequence;
+ }
+ }
+ }
+ }
+ return null;
+ }
+
+ 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);
+ }
+
+ /*
+ * 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..)
+ * @param consequence
+ * if not null, the consequence specific to this sequence (transcript
+ * feature) and allele
*/
- protected void addAlleleProperties(VariantContext variant, SequenceI seq,
- SequenceFeature sf, final int altAlelleIndex)
+ protected void addAlleleProperties(VariantContext variant,
+ SequenceFeature sf, final int altAlelleIndex, String consequence)
{
Map<String, Object> atts = variant.getAttributes();
* extract Consequence data (if present) that we are able to
* associated with the allele for this variant feature
*/
- if (CSQ.equals(key))
+ if (CSQ_FIELD.equals(key))
+ {
+ addConsequences(variant, sf, consequence);
+ continue;
+ }
+
+ /*
+ * filter out fields we don't want to capture
+ */
+ if (!vcfFieldsOfInterest.contains(key))
{
- addConsequences(variant, seq, sf, altAlelleIndex);
continue;
}
String value = getAttributeValue(variant, key, index);
if (value != null)
{
+ /*
+ * VCF spec requires encoding of special characters e.g. '='
+ * so decode them here before storing
+ */
+ try
+ {
+ value = URLDecoder.decode(value, UTF_8);
+ } catch (UnsupportedEncodingException e)
+ {
+ }
sf.setValue(key, value);
}
}
/**
* Inspects CSQ data blocks (consequences) and adds attributes on the sequence
- * feature for the current allele (and transcript if applicable)
+ * feature.
* <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).
+ * If <code>myConsequence</code> is not null, then this is the specific
+ * consequence data (pipe-delimited fields) that is for the current allele and
+ * transcript (sequence) being processed)
*
* @param variant
- * @param seq
* @param sf
- * @param altAlelleIndex
- * (0, 1..)
+ * @param myConsequence
*/
- protected void addConsequences(VariantContext variant, SequenceI seq,
- SequenceFeature sf, int altAlelleIndex)
+ protected void addConsequences(VariantContext variant, SequenceFeature sf,
+ String myConsequence)
{
- Object value = variant.getAttribute(CSQ);
+ Object value = variant.getAttribute(CSQ_FIELD);
- if (value == null || !(value instanceof ArrayList<?>))
+ if (value == null || !(value instanceof List<?>))
{
return;
}
List<String> consequences = (List<String>) value;
/*
- * if CSQ data includes 'Feature', and any value matches the sequence name,
- * then restrict consequence data to only the matching value (transcript)
- * i.e. just pick out consequences for the transcript the variant feature is on
+ * inspect CSQ consequences; restrict to the consequence
+ * associated with the current transcript (Feature)
*/
- String seqName = seq.getName()== null ? "" : seq.getName().toLowerCase();
- String matchFeature = null;
- if (csqFeatureFieldIndex > -1)
+ Map<String, String> csqValues = new HashMap<>();
+
+ for (String consequence : consequences)
{
- for (String consequence : consequences)
+ if (myConsequence == null || myConsequence.equals(consequence))
{
String[] csqFields = consequence.split(PIPE_REGEX);
- if (csqFields.length > csqFeatureFieldIndex)
+
+ /*
+ * inspect individual fields of this consequence, copying non-null
+ * values which are 'fields of interest'
+ */
+ int i = 0;
+ for (String field : csqFields)
{
- String featureIdentifier = csqFields[csqFeatureFieldIndex];
- if (featureIdentifier.length() > 4
- && seqName.indexOf(featureIdentifier.toLowerCase()) > -1)
+ if (field != null && field.length() > 0)
{
- matchFeature = featureIdentifier;
+ String id = vepFieldsOfInterest.get(i);
+ if (id != null)
+ {
+ /*
+ * VCF spec requires encoding of special characters e.g. '='
+ * so decode them here before storing
+ */
+ try
+ {
+ field = URLDecoder.decode(field, UTF_8);
+ } catch (UnsupportedEncodingException e)
+ {
+ }
+ csqValues.put(id, field);
+ }
}
+ i++;
}
}
}
- StringBuilder sb = new StringBuilder(128);
- boolean found = false;
-
- for (String consequence : consequences)
- {
- String[] csqFields = consequence.split(PIPE_REGEX);
-
- if (includeConsequence(csqFields, matchFeature, variant,
- altAlelleIndex))
- {
- if (found)
- {
- sb.append(COMMA);
- }
- found = true;
- sb.append(consequence);
- }
- }
-
- if (found)
- {
- sf.setValue(CSQ, sb.toString());
- }
- }
-
- /**
- * Answers true if we want to associate this block of consequence data with
- * the specified alternate allele of the VCF variant.
- * <p>
- * If consequence data includes the ALLELE_NUM field, then this has to match
- * altAlleleIndex. Otherwise the Allele field of the consequence data has to
- * match the allele value.
- * <p>
- * Optionally (if matchFeature is not null), restrict to only include
- * consequences whose Feature value matches. This allows us to attach
- * consequences to their respective transcripts.
- *
- * @param csqFields
- * @param matchFeature
- * @param variant
- * @param altAlelleIndex
- * (0, 1..)
- * @return
- */
- protected boolean includeConsequence(String[] csqFields,
- String matchFeature, VariantContext variant, int altAlelleIndex)
- {
- /*
- * check consequence is for the current transcript
- */
- if (matchFeature != null)
- {
- if (csqFields.length <= csqFeatureFieldIndex)
- {
- return false;
- }
- String featureIdentifier = csqFields[csqFeatureFieldIndex];
- if (!featureIdentifier.equals(matchFeature))
- {
- return false; // consequence is for a different transcript
- }
- }
-
- /*
- * if ALLELE_NUM is present, it must match altAlleleIndex
- * NB first alternate allele is 1 for ALLELE_NUM, 0 for altAlleleIndex
- */
- if (csqAlleleNumberFieldIndex > -1)
- {
- if (csqFields.length <= csqAlleleNumberFieldIndex)
- {
- return false;
- }
- String alleleNum = csqFields[csqAlleleNumberFieldIndex];
- return String.valueOf(altAlelleIndex + 1).equals(alleleNum);
- }
-
- /*
- * else consequence allele must match variant allele
- */
- if (csqAlleleFieldIndex > -1 && csqFields.length > csqAlleleFieldIndex)
+ if (!csqValues.isEmpty())
{
- String csqAllele = csqFields[csqAlleleFieldIndex];
- String vcfAllele = variant.getAlternateAllele(altAlelleIndex)
- .getBaseString();
- return csqAllele.equals(vcfAllele);
+ sf.setValue(CSQ_FIELD, csqValues);
}
-
- return false;
}
/**
* 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)
{
git://source.jalview.org / jalview.git / blobdiff