>();
+ assemblyMappings = new HashMap<>();
}
/**
- * Loads VCF on to an alignment - provided it can be related to one or more
- * sequence's chromosomal coordinates.
+ * Starts a new thread to query and load VCF variant data on to the alignment
*
* This method is not thread safe - concurrent threads should use separate
* instances of this class.
*
* @param filePath
- * @param status
+ * @param gui
*/
- public void loadVCF(String filePath, AlignViewControllerGuiI status)
+ public void loadVCF(final String filePath,
+ final AlignViewControllerGuiI gui)
{
- VCFReader reader = null;
+ if (gui != null)
+ {
+ gui.setStatus(MessageManager.getString("label.searching_vcf"));
+ }
+
+ new Thread()
+ {
+
+ @Override
+ public void run()
+ {
+ VCFLoader.this.doLoad(filePath, gui);
+ }
+
+ }.start();
+ }
+ /**
+ * Loads VCF on to an alignment - provided it can be related to one or more
+ * sequence's chromosomal coordinates
+ *
+ * @param filePath
+ * @param gui
+ * optional callback handler for messages
+ */
+ protected void doLoad(String filePath, AlignViewControllerGuiI gui)
+ {
+ VCFReader reader = null;
try
{
// 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;
@@ -97,25 +277,33 @@ public class VCFLoader
*/
for (SequenceI seq : al.getSequences())
{
- int added = loadVCF(seq, reader, isRefGrch37);
+ int added = loadSequenceVCF(seq, reader, vcfAssembly);
if (added > 0)
{
seqCount++;
varCount += added;
- computePeptideVariants(seq);
+ transferAddedFeatures(seq);
}
}
- // long elapsed = System.currentTimeMillis() - start;
- String msg = String.format("Added %d VCF variants to %d sequence(s)",
- varCount, seqCount);
- if (status != null)
+ if (gui != null)
{
- status.setStatus(msg);
+ // long elapsed = System.currentTimeMillis() - start;
+ String msg = MessageManager.formatMessage("label.added_vcf",
+ varCount, seqCount);
+ gui.setStatus(msg);
+ if (gui.getFeatureSettingsUI() != null)
+ {
+ gui.getFeatureSettingsUI().discoverAllFeatureData();
+ }
}
} catch (Throwable e)
{
System.err.println("Error processing VCF: " + e.getMessage());
e.printStackTrace();
+ if (gui != null)
+ {
+ gui.setStatus("Error occurred - see console for details");
+ }
} finally
{
if (reader != null)
@@ -128,20 +316,186 @@ public class VCFLoader
// ignore
}
}
+ header = null;
+ dictionary = null;
}
}
/**
- * (Re-)computes peptide variants from dna variants, for any protein sequence
- * to which the dna sequence has a mapping. Note that although duplicate
- * features may get computed, they will not be added, since duplicate sequence
- * features are ignored in Sequence.addSequenceFeature.
+ * Reads metadata (such as INFO field descriptions and datatypes) and saves
+ * them for future reference
*
- * @param dnaSeq
+ * @param theSourceId
*/
- protected void computePeptideVariants(SequenceI dnaSeq)
+ void saveMetadata(String theSourceId)
{
- DBRefEntry[] dbrefs = dnaSeq.getDBRefs();
+ List 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 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.
+ *
+ * CSQ fields are declared in the CSQ INFO Description e.g.
+ *
+ * Description="Consequence ...from ... VEP. Format: Allele|Consequence|...
+ */
+ protected void parseCsqHeader()
+ {
+ List 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.
+ *
+ * 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 getFieldMatchers(String key, String def)
+ {
+ String pref = Cache.getDefault(key, def);
+ List 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 3:1, computes peptide variants from nucleotide variants.
+ *
+ * @param seq
+ */
+ protected void transferAddedFeatures(SequenceI seq)
+ {
+ DBRefEntry[] dbrefs = seq.getDBRefs();
if (dbrefs == null)
{
return;
@@ -149,226 +503,777 @@ public class VCFLoader
for (DBRefEntry dbref : dbrefs)
{
Mapping mapping = dbref.getMap();
- if (mapping == null || mapping.getTo() == null
- || mapping.getMap().getFromRatio() != 3)
+ if (mapping == null || mapping.getTo() == null)
{
continue;
}
- AlignmentUtils.computeProteinFeatures(dnaSeq, mapping.getTo(),
- mapping.getMap());
+
+ SequenceI mapTo = mapping.getTo();
+ MapList map = mapping.getMap();
+ if (map.getFromRatio() == 3)
+ {
+ /*
+ * dna-to-peptide product mapping
+ */
+ AlignmentUtils.computeProteinFeatures(seq, mapTo, map);
+ }
+ else
+ {
+ /*
+ * nucleotide-to-nucleotide mapping e.g. transcript to CDS
+ */
+ List features = seq.getFeatures()
+ .getPositionalFeatures(SequenceOntologyI.SEQUENCE_VARIANT);
+ for (SequenceFeature sf : features)
+ {
+ if (FEATURE_GROUP_VCF.equals(sf.getFeatureGroup()))
+ {
+ transferFeature(sf, mapTo, map);
+ }
+ }
+ }
}
}
/**
* 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;
- GeneLoci seqCoords = ((Sequence) seq).getGeneLoci();
- if (seqCoords == null)
+ VCFMap vcfMap = getVcfMap(seq, vcfAssembly);
+ if (vcfMap == null)
{
return 0;
}
- List seqChromosomalContigs = seqCoords.mapping.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, 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)
{
- GeneLoci seqCoords = ((Sequence) seq).getGeneLoci();
+ /*
+ * 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();
+ MapList map = seqCoords.getMap();
+
+ if (!vcfSpeciesMatchesSequence(vcfAssembly, species))
+ {
+ return null;
+ }
- String chromosome = seqCoords.chromosome;
- String seqRef = seqCoords.assembly;
- String species = seqCoords.species;
+ if (vcfAssemblyMatchesSequence(vcfAssembly, seqRef))
+ {
+ return new VCFMap(chromosome, map);
+ }
- // TODO handle species properly
- if ("".equals(species))
+ if (!"GRCh38".equalsIgnoreCase(seqRef) // Ensembl
+ || !vcfAssembly.contains("Homo_sapiens_assembly19")) // gnomAD
{
- species = "human";
+ 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 toVcfRanges = new ArrayList<>();
+ List 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);
}
- offset = newRange[0] - range[0];
- range = newRange;
}
+ 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);
+ }
+ }
+ return null;
+ }
+
+ /**
+ * 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.mapping;
- int fromLocus = Math.min(range[0], range[1]);
- int toLocus = Math.max(range[0], range[1]);
- CloseableIterator 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 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
+ {
+ score = Float.parseFloat(attributeValue);
+ } catch (NumberFormatException e)
{
- addVariantFeatures(seq, variant, seqLocation[0], seqLocation[1]);
+ // 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) att).get(alleleIndex);
+ }
+
+ return null;
+ }
+
+ /**
+ * 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 int addAlleleFeatures(SequenceI seq, VariantContext variant,
+ int featureStart, int featureEnd, boolean forwardStrand)
+ {
+ 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 the VCF variant record, and adds variant features to the sequence.
- * Only SNP variants are added, not INDELs.
+ * 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 void addVariantFeatures(SequenceI seq, VariantContext variant,
- int featureStart, int featureEnd)
+ protected int addAlleleFeature(SequenceI seq, VariantContext variant,
+ int altAlleleIndex, int featureStart, int featureEnd,
+ boolean forwardStrand)
{
String reference = variant.getReference().getBaseString();
- if (reference.length() != 1)
+ Allele alt = variant.getAlternateAllele(altAlleleIndex);
+ String allele = alt.getBaseString();
+
+ /*
+ * 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))
+ {
+ 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
+ */
+ 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;
+ if (consequence != null)
+ {
+ type = getOntologyTerm(seq, variant, altAlleleIndex,
+ consequence);
+ }
+
+ 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, consequence);
+
+ 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:
+ *
+ * - VEP (or SnpEff) Consequence annotation is included in the VCF
+ * - sequence id can be matched to VEP Feature (or SnpEff Feature_ID)
+ *
+ *
+ * @param seq
+ * @param variant
+ * @param altAlleleIndex
+ * @param consequence
+ * @return
+ * @see http://www.sequenceontology.org/browser/current_svn/term/SO:0001060
+ */
+ String getOntologyTerm(SequenceI seq, VariantContext variant,
+ int altAlleleIndex, String consequence)
+ {
+ 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 if
- * multiple alleles at the locus; we extract for the simple case only,
- * since not sure how to match allele order with AF values
+ * 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)
+ if (consequence != null)
{
- try
- {
- score = Float.parseFloat((String) af);
- } catch (NumberFormatException e)
+ String[] csqFields = consequence.split(PIPE_REGEX);
+ if (csqFields.length > csqConsequenceFieldIndex)
{
- // leave as 0
+ type = csqFields[csqConsequenceFieldIndex];
}
}
-
- StringBuilder sb = new StringBuilder();
- sb.append(reference);
+ else
+ {
+ // todo the same for SnpEff consequence data matching if wanted
+ }
/*
- * inspect alleles and record SNP variants (as the variant
- * record could be MIXED and include INDEL and SNP alleles)
+ * if of the form (e.g.) missense_variant&splice_region_variant,
+ * just take the first ('most severe') consequence
*/
- int alleleCount = 0;
+ 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.
+ *
+ * - inspects the VCF data for key 'vcfInfoId'
+ * - splits this on comma (to distinct consequences)
+ * - returns the first consequence (if any) where
+ *
+ * - the allele matches the altAlleleIndex'th allele of variant
+ * - the feature matches the sequence name (e.g. transcript id)
+ *
+ *
+ * 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; 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 consequences = (List) value;
+
+ for (String consequence : consequences)
{
- if (!allele.isReference())
+ String[] csqFields = consequence.split(PIPE_REGEX);
+ if (csqFields.length > featureFieldIndex)
{
- String alleleBase = allele.getBaseString();
- if (alleleBase.length() == 1)
+ String featureIdentifier = csqFields[featureFieldIndex];
+ if (featureIdentifier.length() > 4
+ && seqName.indexOf(featureIdentifier.toLowerCase()) > -1)
{
- sb.append(",").append(alleleBase);
- alleleCount++;
+ /*
+ * 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, "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("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..)
+ * @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, String consequence)
+ {
Map atts = variant.getAttributes();
+
for (Entry 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, consequence);
+ 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.
+ *
+ * If myConsequence
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 myConsequence
+ */
+ protected void addConsequences(VariantContext variant, SequenceI seq,
+ SequenceFeature sf, String myConsequence)
+ {
+ Object value = variant.getAttribute(CSQ_FIELD);
+ // TODO if CSQ not present, try ANN (for SnpEff consequence data)?
+
+ if (value == null || !(value instanceof List>))
+ {
+ return;
+ }
+
+ List consequences = (List) value;
+
+ /*
+ * inspect CSQ consequences; restrict to the consequence
+ * associated with the current transcript (Feature)
+ */
+ Map 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);
+ }
+
+ /**
+ * A convenience method to complement a dna base and return the string value
+ * of its complement
+ *
+ * @param reference
+ * @return
+ */
+ protected String complement(byte[] reference)
+ {
+ return String.valueOf(Dna.getComplement((char) reference[0]));
}
/**
@@ -410,8 +1315,8 @@ public class VCFLoader
* 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)
{
@@ -487,6 +1392,32 @@ public class VCFLoader
}
/**
+ * Transfers the sequence feature to the target sequence, locating its start
+ * and end range based on the mapping. Features which do not overlap the
+ * target sequence are ignored.
+ *
+ * @param sf
+ * @param targetSequence
+ * @param mapping
+ * mapping from the feature's coordinates to the target sequence
+ */
+ protected void transferFeature(SequenceFeature sf,
+ SequenceI targetSequence, MapList mapping)
+ {
+ int[] mappedRange = mapping.locateInTo(sf.getBegin(), sf.getEnd());
+
+ if (mappedRange != null)
+ {
+ String group = sf.getFeatureGroup();
+ int newBegin = Math.min(mappedRange[0], mappedRange[1]);
+ int newEnd = Math.max(mappedRange[0], mappedRange[1]);
+ SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
+ group, sf.getScore());
+ targetSequence.addSequenceFeature(copy);
+ }
+ }
+
+ /**
* Formats a ranges map lookup key
*
* @param chromosome