package jalview.ext.ensembl;
+import jalview.analysis.AlignmentUtils;
import jalview.datamodel.Alignment;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.DBRefEntry;
import jalview.exceptions.JalviewException;
import jalview.io.FastaFile;
import jalview.io.FileParse;
-import jalview.io.gff.SequenceOntology;
+import jalview.io.gff.SequenceOntologyFactory;
+import jalview.io.gff.SequenceOntologyI;
+import jalview.schemes.ResidueProperties;
import jalview.util.DBRefUtils;
import jalview.util.MapList;
+import jalview.util.MappingUtils;
+import jalview.util.StringUtils;
import java.io.IOException;
import java.net.MalformedURLException;
import java.util.Arrays;
import java.util.Collections;
import java.util.Comparator;
+import java.util.LinkedHashMap;
import java.util.List;
+import java.util.Map.Entry;
/**
* Base class for Ensembl sequence fetchers
*/
public abstract class EnsemblSeqProxy extends EnsemblRestClient
{
+ private static final List<String> CROSS_REFERENCES = Arrays
+ .asList(new String[] { "CCDS" });
+
protected static final String CONSEQUENCE_TYPE = "consequence_type";
protected static final String PARENT = "Parent";
protected static final String ID = "ID";
- /*
- * this needs special handling, as it isA sequence_variant in the
- * Sequence Ontology, but behaves in Ensembl as if it isA transcript
- */
- protected static final String NMD_VARIANT = "NMD_transcript_variant";
-
protected static final String NAME = "Name";
+ /*
+ * enum for 'type' parameter to the /sequence REST service
+ */
public enum EnsemblSeqType
{
/**
- * type=genomic for the full dna including introns
+ * type=genomic to fetch full dna including introns
*/
GENOMIC("genomic"),
/**
- * type=cdna for transcribed dna including UTRs
+ * type=cdna to fetch dna including UTRs
*/
CDNA("cdna"),
/**
- * type=cds for coding dna excluding UTRs
+ * type=cds to fetch coding dna excluding UTRs
*/
CDS("cds"),
/**
- * type=protein for the peptide product sequence
+ * type=protein to fetch peptide product sequence
*/
PROTEIN("protein");
@Override
public AlignmentI getSequenceRecords(String query) throws Exception
{
- long now = System.currentTimeMillis();
// TODO use a String... query vararg instead?
// danger: accession separator used as a regex here, a string elsewhere
+ " chunks. Unexpected problem (" + r.getLocalizedMessage()
+ ")";
System.err.println(msg);
- if (alignment != null)
- {
- break; // return what we got
- }
- else
- {
- throw new JalviewException(msg, r);
- }
+ break;
+ // if (alignment != null)
+ // {
+ // break; // return what we got
+ // }
+ // else
+ // {
+ // throw new JalviewException(msg, r);
+ // }
}
}
addFeaturesAndProduct(accId, alignment);
}
- inProgress = false;
- System.out.println(getClass().getName() + " took "
- + (System.currentTimeMillis() - now) + "ms to fetch");
+ for (SequenceI seq : alignment.getSequences())
+ {
+ getCrossReferences(seq);
+ }
+
return alignment;
}
* get 'dummy' genomic sequence with exon, cds and variation features
*/
SequenceI genomicSequence = null;
- EnsemblOverlap gffFetcher = new EnsemblOverlap();
+ EnsemblFeatures gffFetcher = new EnsemblFeatures();
EnsemblFeatureType[] features = getFeaturesToFetch();
AlignmentI geneFeatures = gffFetcher.getSequenceRecords(accId,
features);
MapList mapList = mapCdsToProtein(querySeq, proteinSeq);
if (mapList != null)
{
- Mapping map = new Mapping(proteinSeq.getDatasetSequence(), mapList);
+ // clunky: ensure Uniprot xref if we have one is on mapped sequence
+ SequenceI ds = proteinSeq.getDatasetSequence();
+ ds.setSourceDBRef(proteinSeq.getSourceDBRef());
+ Mapping map = new Mapping(ds, mapList);
DBRefEntry dbr = new DBRefEntry(getDbSource(), getDbVersion(),
accId, map);
querySeq.getDatasetSequence().addDBRef(dbr);
+
+ /*
+ * compute peptide variants from dna variants and add as
+ * sequence features on the protein sequence ta-da
+ */
+ computeProteinFeatures(querySeq, proteinSeq, mapList);
}
} catch (Exception e)
{
}
/**
+ * Get Uniprot and PDB xrefs from Ensembl, and attach them to the protein
+ * sequence
+ *
+ * @param seq
+ */
+ protected void getCrossReferences(SequenceI seq)
+ {
+ while (seq.getDatasetSequence() != null)
+ {
+ seq = seq.getDatasetSequence();
+ }
+
+ EnsemblXref xrefFetcher = new EnsemblXref();
+ List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName(),
+ getCrossReferenceDatabases());
+ for (DBRefEntry xref : xrefs)
+ {
+ seq.addDBRef(xref);
+ /*
+ * Save any Uniprot xref to be the reference for SIFTS mapping
+ */
+ if (DBRefSource.UNIPROT.equals(xref.getSource()))
+ {
+ seq.setSourceDBRef(xref);
+ }
+ }
+ }
+
+ /**
+ * Returns a list of database names to be used when fetching cross-references.
+ *
+ * @return
+ */
+ protected List<String> getCrossReferenceDatabases()
+ {
+ return CROSS_REFERENCES;
+ }
+
+ /**
* Returns a mapping from dna to protein by inspecting sequence features of
* type "CDS" on the dna.
*
*/
protected MapList mapCdsToProtein(SequenceI dnaSeq, SequenceI proteinSeq)
{
- SequenceFeature[] sfs = dnaSeq.getSequenceFeatures();
- if (sfs == null)
+ List<int[]> ranges = new ArrayList<int[]>(50);
+
+ int mappedDnaLength = getCdsRanges(dnaSeq, ranges);
+
+ int proteinLength = proteinSeq.getLength();
+ List<int[]> proteinRange = new ArrayList<int[]>();
+ int proteinStart = 1;
+
+ /*
+ * incomplete start codon may mean X at start of peptide
+ * we ignore both for mapping purposes
+ */
+ if (proteinSeq.getCharAt(0) == 'X')
{
- return null;
+ proteinStart = 2;
+ proteinLength--;
}
+ proteinRange.add(new int[] { proteinStart, proteinLength });
- List<int[]> ranges = new ArrayList<int[]>(50);
- SequenceOntology so = SequenceOntology.getInstance();
-
- int mappedDnaLength = 0;
-
/*
- * Map CDS columns of dna to peptide. No need to worry about reverse strand
- * dna here since the retrieved sequence is as transcribed (reverse
- * complement for reverse strand), i.e in the same sense as the peptide.
+ * dna length should map to protein (or protein plus stop codon)
*/
- boolean fivePrimeIncomplete = false;
+ int codesForResidues = mappedDnaLength / 3;
+ if (codesForResidues == proteinLength
+ || codesForResidues == (proteinLength + 1))
+ {
+ return new MapList(ranges, proteinRange, 3, 1);
+ }
+ return null;
+ }
+
+ /**
+ * Adds CDS ranges to the ranges list, and returns the total length mapped
+ * from.
+ *
+ * No need to worry about reverse strand dna, here since the retrieved
+ * sequence is as transcribed (reverse complement for reverse strand), i.e in
+ * the same sense as the peptide.
+ *
+ * @param dnaSeq
+ * @param ranges
+ * @return
+ */
+ protected int getCdsRanges(SequenceI dnaSeq, List<int[]> ranges)
+ {
+ SequenceFeature[] sfs = dnaSeq.getSequenceFeatures();
+ if (sfs == null)
+ {
+ return 0;
+ }
+ int mappedDnaLength = 0;
for (SequenceFeature sf : sfs)
{
/*
* process a CDS feature (or a sub-type of CDS)
*/
- if (so.isA(sf.getType(), SequenceOntology.CDS))
+ if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
+ SequenceOntologyI.CDS))
{
int phase = 0;
try {
int end = sf.getEnd();
if (ranges.isEmpty() && phase > 0)
{
- fivePrimeIncomplete = true;
begin += phase;
if (begin > end)
{
mappedDnaLength += Math.abs(end - begin) + 1;
}
}
- int proteinLength = proteinSeq.getLength();
- List<int[]> proteinRange = new ArrayList<int[]>();
- int proteinStart = 1;
- if (fivePrimeIncomplete && proteinSeq.getCharAt(0) == 'X')
- {
- proteinStart = 2;
- proteinLength--;
- }
- proteinRange.add(new int[] { proteinStart, proteinLength });
-
- /*
- * dna length should map to protein (or protein plus stop codon)
- */
- int codesForResidues = mappedDnaLength / 3;
- if (codesForResidues == proteinLength
- || codesForResidues == (proteinLength + 1))
- {
- return new MapList(ranges, proteinRange, 3, 1);
- }
- return null;
+ return mappedDnaLength;
}
/**
* the start position of the sequence we are mapping to
* @return
*/
- protected MapList getGenomicRanges(SequenceI sourceSequence,
+ protected MapList getGenomicRangesFromFeatures(SequenceI sourceSequence,
String accId, int start)
{
SequenceFeature[] sfs = sourceSequence.getSequenceFeatures();
*/
if (identifiesSequence(sf, accId))
{
- int strand = sf.getStrand();
-
- if (directionSet && strand != direction)
- {
- // abort - mix of forward and backward
+ int strand = sf.getStrand();
+ strand = strand == 0 ? 1 : strand; // treat unknown as forward
+
+ if (directionSet && strand != direction)
+ {
+ // abort - mix of forward and backward
System.err.println("Error: forward and backward strand for "
+ accId);
return null;
*/
Collections.sort(regions, new RangeSorter(direction == 1));
- List<int[]> to = new ArrayList<int[]>();
- to.add(new int[] { start, start + mappedLength - 1 });
+ List<int[]> to = Arrays.asList(new int[] { start,
+ start + mappedLength - 1 });
return new MapList(regions, to, 1, 1);
}
/*
* for sequence_variant, make an additional feature with consequence
*/
- if (SequenceOntology.getInstance().isSequenceVariant(sf.getType()))
+ if (SequenceOntologyFactory.getInstance().isA(sf.getType(),
+ SequenceOntologyI.SEQUENCE_VARIANT))
{
String consequence = (String) sf.getValue(CONSEQUENCE_TYPE);
if (consequence != null)
}
SequenceFeature[] sfs = sourceSequence.getSequenceFeatures();
- MapList mapping = getGenomicRanges(sourceSequence, accessionId,
+ MapList mapping = getGenomicRangesFromFeatures(sourceSequence, accessionId,
targetSequence.getStart());
if (mapping == null)
{
SequenceFeature[] sfs = sequence.getSequenceFeatures();
if (sfs != null) {
- SequenceOntology so = SequenceOntology.getInstance();
+ SequenceOntologyI so = SequenceOntologyFactory.getInstance();
for (SequenceFeature sf :sfs) {
if (so.isA(sf.getType(), type))
{
}
/**
+ * Maps exon features from dna to protein, and computes variants in peptide
+ * product generated by variants in dna, and adds them as sequence_variant
+ * features on the protein sequence. Returns the number of variant features
+ * added.
+ *
+ * @param dnaSeq
+ * @param peptide
+ * @param dnaToProtein
+ */
+ static int computeProteinFeatures(SequenceI dnaSeq,
+ SequenceI peptide, MapList dnaToProtein)
+ {
+ while (dnaSeq.getDatasetSequence() != null)
+ {
+ dnaSeq = dnaSeq.getDatasetSequence();
+ }
+ while (peptide.getDatasetSequence() != null)
+ {
+ peptide = peptide.getDatasetSequence();
+ }
+
+ AlignmentUtils.transferFeatures(dnaSeq, peptide, dnaToProtein,
+ SequenceOntologyI.EXON);
+
+ LinkedHashMap<Integer, String[][]> variants = buildDnaVariantsMap(
+ dnaSeq, dnaToProtein);
+
+ /*
+ * scan codon variations, compute peptide variants and add to peptide sequence
+ */
+ int count = 0;
+ for (Entry<Integer, String[][]> variant : variants.entrySet())
+ {
+ int peptidePos = variant.getKey();
+ String[][] codonVariants = variant.getValue();
+ String residue = String.valueOf(peptide.getCharAt(peptidePos - 1)); // 0-based
+ List<String> peptideVariants = computePeptideVariants(codonVariants,
+ residue);
+ if (!peptideVariants.isEmpty())
+ {
+ String desc = StringUtils.listToDelimitedString(peptideVariants,
+ ", ");
+ SequenceFeature sf = new SequenceFeature(
+ SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
+ peptidePos, 0f, null);
+ peptide.addSequenceFeature(sf);
+ count++;
+ }
+ }
+
+ /*
+ * ugly sort to get sequence features in start position order
+ * - would be better to store in Sequence as a TreeSet instead?
+ */
+ Arrays.sort(peptide.getSequenceFeatures(),
+ new Comparator<SequenceFeature>()
+ {
+ @Override
+ public int compare(SequenceFeature o1, SequenceFeature o2)
+ {
+ int c = Integer.compare(o1.getBegin(), o2.getBegin());
+ return c == 0 ? Integer.compare(o1.getEnd(), o2.getEnd())
+ : c;
+ }
+ });
+ return count;
+ }
+
+ /**
+ * Builds a map whose key is position in the protein sequence, and value is an
+ * array of all variants for the coding codon positions
+ *
+ * @param dnaSeq
+ * @param dnaToProtein
+ * @return
+ */
+ static LinkedHashMap<Integer, String[][]> buildDnaVariantsMap(
+ SequenceI dnaSeq, MapList dnaToProtein)
+ {
+ /*
+ * map from peptide position to all variant features of the codon for it
+ * LinkedHashMap ensures we add the peptide features in sequence order
+ */
+ LinkedHashMap<Integer, String[][]> variants = new LinkedHashMap<Integer, String[][]>();
+ SequenceOntologyI so = SequenceOntologyFactory.getInstance();
+
+ SequenceFeature[] dnaFeatures = dnaSeq.getSequenceFeatures();
+ if (dnaFeatures == null)
+ {
+ return variants;
+ }
+
+ int dnaStart = dnaSeq.getStart();
+ int[] lastCodon = null;
+ int lastPeptidePostion = 0;
+
+ /*
+ * build a map of codon variations for peptides
+ */
+ for (SequenceFeature sf : dnaFeatures)
+ {
+ int dnaCol = sf.getBegin();
+ if (dnaCol != sf.getEnd())
+ {
+ // not handling multi-locus variant features
+ continue;
+ }
+ if (so.isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
+ {
+ int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
+ if (mapsTo == null)
+ {
+ // feature doesn't lie within coding region
+ continue;
+ }
+ int peptidePosition = mapsTo[0];
+ String[][] codonVariants = variants.get(peptidePosition);
+ if (codonVariants == null)
+ {
+ codonVariants = new String[3][];
+ variants.put(peptidePosition, codonVariants);
+ }
+
+ /*
+ * extract dna variants to a string array
+ */
+ String alls = (String) sf.getValue("alleles");
+ if (alls == null)
+ {
+ continue;
+ }
+ String[] alleles = alls.split(",");
+
+ /*
+ * get this peptides codon positions e.g. [3, 4, 5] or [4, 7, 10]
+ */
+ int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
+ : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
+ peptidePosition, peptidePosition));
+ lastPeptidePostion = peptidePosition;
+ lastCodon = codon;
+
+ /*
+ * save nucleotide (and this variant) for each codon position
+ */
+ for (int codonPos = 0; codonPos < 3; codonPos++)
+ {
+ String nucleotide = String.valueOf(dnaSeq
+ .getCharAt(codon[codonPos] - dnaStart));
+ if (codon[codonPos] == dnaCol)
+ {
+ /*
+ * record current dna base and its alleles
+ */
+ String[] dnaVariants = new String[alleles.length + 1];
+ dnaVariants[0] = nucleotide;
+ System.arraycopy(alleles, 0, dnaVariants, 1, alleles.length);
+ codonVariants[codonPos] = dnaVariants;
+ }
+ else if (codonVariants[codonPos] == null)
+ {
+ /*
+ * record current dna base only
+ * (at least until we find any variation and overwrite it)
+ */
+ codonVariants[codonPos] = new String[] { nucleotide };
+ }
+ }
+ }
+ }
+ return variants;
+ }
+
+ /**
+ * Returns a sorted, non-redundant list of all peptide translations generated
+ * by the given dna variants, excluding the current residue value
+ *
+ * @param codonVariants
+ * an array of base values (acgtACGT) for codon positions 1, 2, 3
+ * @param residue
+ * the current residue translation
+ * @return
+ */
+ static List<String> computePeptideVariants(
+ String[][] codonVariants, String residue)
+ {
+ List<String> result = new ArrayList<String>();
+ for (String base1 : codonVariants[0])
+ {
+ for (String base2 : codonVariants[1])
+ {
+ for (String base3 : codonVariants[2])
+ {
+ String codon = base1 + base2 + base3;
+ // TODO: report frameshift/insertion/deletion
+ // and multiple-base variants?!
+ String peptide = codon.contains("-") ? "-" : ResidueProperties
+ .codonTranslate(codon);
+ if (peptide != null && !result.contains(peptide)
+ && !peptide.equalsIgnoreCase(residue))
+ {
+ result.add(peptide);
+ }
+ }
+ }
+ }
+
+ /*
+ * sort alphabetically with STOP at the end
+ */
+ Collections.sort(result, new Comparator<String>()
+ {
+
+ @Override
+ public int compare(String o1, String o2)
+ {
+ if ("STOP".equals(o1))
+ {
+ return 1;
+ }
+ else if ("STOP".equals(o2))
+ {
+ return -1;
+ }
+ else
+ {
+ return o1.compareTo(o2);
+ }
+ }
+ });
+ return result;
+ }
+
+ /**
* Answers true if the feature type is either 'NMD_transcript_variant' or
* 'transcript' or one of its sub-types in the Sequence Ontology. This is
* needed because NMD_transcript_variant behaves like 'transcript' in Ensembl
*/
public static boolean isTranscript(String featureType)
{
- return NMD_VARIANT.equals(featureType)
- || SequenceOntology.getInstance().isA(featureType, SequenceOntology.TRANSCRIPT);
+ return SequenceOntologyI.NMD_TRANSCRIPT_VARIANT.equals(featureType)
+ || SequenceOntologyFactory.getInstance().isA(featureType,
+ SequenceOntologyI.TRANSCRIPT);
}
}