*/
package jalview.analysis;
+import static jalview.io.gff.GffConstants.CLINICAL_SIGNIFICANCE;
+
import jalview.datamodel.AlignedCodon;
import jalview.datamodel.AlignedCodonFrame;
+import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
import jalview.datamodel.Alignment;
import jalview.datamodel.AlignmentAnnotation;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.DBRefEntry;
-import jalview.datamodel.DBRefSource;
-import jalview.datamodel.FeatureProperties;
+import jalview.datamodel.IncompleteCodonException;
import jalview.datamodel.Mapping;
-import jalview.datamodel.SearchResults;
import jalview.datamodel.Sequence;
+import jalview.datamodel.SequenceFeature;
import jalview.datamodel.SequenceGroup;
import jalview.datamodel.SequenceI;
+import jalview.io.gff.SequenceOntologyFactory;
+import jalview.io.gff.SequenceOntologyI;
import jalview.schemes.ResidueProperties;
-import jalview.util.DBRefUtils;
+import jalview.util.Comparison;
import jalview.util.MapList;
import jalview.util.MappingUtils;
+import jalview.util.StringUtils;
+import java.io.UnsupportedEncodingException;
+import java.net.URLEncoder;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collection;
+import java.util.Collections;
+import java.util.Comparator;
import java.util.HashMap;
import java.util.HashSet;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
+import java.util.NoSuchElementException;
import java.util.Set;
import java.util.TreeMap;
public class AlignmentUtils
{
+ private static final String SEQUENCE_VARIANT = "sequence_variant:";
+ private static final String ID = "ID";
+
+ /**
+ * A data model to hold the 'normal' base value at a position, and an optional
+ * sequence variant feature
+ */
+ static class DnaVariant
+ {
+ String base;
+
+ SequenceFeature variant;
+
+ DnaVariant(String nuc)
+ {
+ base = nuc;
+ }
+
+ DnaVariant(String nuc, SequenceFeature var)
+ {
+ base = nuc;
+ variant = var;
+ }
+ }
+
/**
* given an existing alignment, create a new alignment including all, or up to
* flankSize additional symbols from each sequence's dataset sequence
}
else
{
- MapList map = mapProteinSequenceToCdna(aaSeq, cdnaSeq);
+ MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
if (map != null)
{
acf.addMap(cdnaSeq, aaSeq, map);
}
/**
- * Build a mapping (if possible) of a protein to a cDNA sequence. The cDNA
- * must be three times the length of the protein, possibly after ignoring
- * start and/or stop codons, and must translate to the protein. Returns null
- * if no mapping is determined.
+ * Builds a mapping (if possible) of a cDNA to a protein sequence.
+ * <ul>
+ * <li>first checks if the cdna translates exactly to the protein sequence</li>
+ * <li>else checks for translation after removing a STOP codon</li>
+ * <li>else checks for translation after removing a START codon</li>
+ * <li>if that fails, inspect CDS features on the cDNA sequence</li>
+ * </ul>
+ * Returns null if no mapping is determined.
*
- * @param proteinSeqs
+ * @param proteinSeq
+ * the aligned protein sequence
* @param cdnaSeq
+ * the aligned cdna sequence
* @return
*/
- public static MapList mapProteinSequenceToCdna(SequenceI proteinSeq,
+ public static MapList mapCdnaToProtein(SequenceI proteinSeq,
SequenceI cdnaSeq)
{
/*
final int proteinEnd = proteinSeq.getEnd();
/*
- * If lengths don't match, try ignoring stop codon.
+ * If lengths don't match, try ignoring stop codon (if present)
*/
if (cdnaLength != mappedLength && cdnaLength > 2)
{
cdnaLength -= 3;
}
- if (cdnaLength != mappedLength)
- {
- return null;
- }
- if (!translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
+ if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
{
- return null;
+ /*
+ * protein is translation of dna (+/- start/stop codons)
+ */
+ MapList map = new MapList(new int[] { cdnaStart, cdnaEnd }, new int[]
+ { proteinStart, proteinEnd }, 3, 1);
+ return map;
}
- MapList map = new MapList(new int[] { cdnaStart, cdnaEnd }, new int[] {
- proteinStart, proteinEnd }, 3, 1);
- return map;
+
+ /*
+ * translation failed - try mapping CDS annotated regions of dna
+ */
+ return mapCdsToProtein(cdnaSeq, proteinSeq);
}
/**
return false;
}
- int aaResidue = 0;
- for (int i = cdnaStart; i < cdnaSeqChars.length - 2
- && aaResidue < aaSeqChars.length; i += 3, aaResidue++)
+ int aaPos = 0;
+ int dnaPos = cdnaStart;
+ for (; dnaPos < cdnaSeqChars.length - 2
+ && aaPos < aaSeqChars.length; dnaPos += 3, aaPos++)
{
- String codon = String.valueOf(cdnaSeqChars, i, 3);
+ String codon = String.valueOf(cdnaSeqChars, dnaPos, 3);
final String translated = ResidueProperties.codonTranslate(codon);
+
/*
* allow * in protein to match untranslatable in dna
*/
- final char aaRes = aaSeqChars[aaResidue];
+ final char aaRes = aaSeqChars[aaPos];
if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
{
continue;
return false;
}
}
- // fail if we didn't match all of the aa sequence
- return (aaResidue == aaSeqChars.length);
+
+ /*
+ * check we matched all of the protein sequence
+ */
+ if (aaPos != aaSeqChars.length)
+ {
+ return false;
+ }
+
+ /*
+ * check we matched all of the dna except
+ * for optional trailing STOP codon
+ */
+ if (dnaPos == cdnaSeqChars.length)
+ {
+ return true;
+ }
+ if (dnaPos == cdnaSeqChars.length - 3)
+ {
+ String codon = String.valueOf(cdnaSeqChars, dnaPos, 3);
+ if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
+ {
+ return true;
+ }
+ }
+ return false;
}
/**
AlignedCodonFrame mapping = null;
for (AlignedCodonFrame mp : mappings)
{
- alignFrom = mp.findAlignedSequence(seq.getDatasetSequence(), al);
+ alignFrom = mp.findAlignedSequence(seq, al);
if (alignFrom != null)
{
mapping = mp;
}
/**
- * Returns a list of sequences mapped from the given sequences and aligned
- * (gapped) in the same way. For example, the cDNA for aligned protein, where
- * a single gap in protein generates three gaps in cDNA.
+ * Realigns the given protein to match the alignment of the dna, using codon
+ * mappings to translate aligned codon positions to protein residues.
*
- * @param sequences
- * @param gapCharacter
- * @param mappings
- * @return
+ * @param protein
+ * the alignment whose sequences are realigned by this method
+ * @param dna
+ * the dna alignment whose alignment we are 'copying'
+ * @return the number of sequences that were realigned
*/
- public static List<SequenceI> getAlignedTranslation(
- List<SequenceI> sequences, char gapCharacter,
- Set<AlignedCodonFrame> mappings)
+ public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
{
- List<SequenceI> alignedSeqs = new ArrayList<SequenceI>();
-
- for (SequenceI seq : sequences)
- {
- List<SequenceI> mapped = getAlignedTranslation(seq, gapCharacter,
- mappings);
- alignedSeqs.addAll(mapped);
- }
- return alignedSeqs;
+ List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
+ Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
+ protein, dna, unmappedProtein);
+ return alignProteinAs(protein, alignedCodons, unmappedProtein);
}
/**
- * Returns sequences aligned 'like' the source sequence, as mapped by the
- * given mappings. Normally we expect zero or one 'mapped' sequences, but this
- * will support 1-to-many as well.
+ * Builds a map whose key is an aligned codon position (3 alignment column
+ * numbers base 0), and whose value is a map from protein sequence to each
+ * protein's peptide residue for that codon. The map generates an ordering of
+ * the codons, and allows us to read off the peptides at each position in
+ * order to assemble 'aligned' protein sequences.
*
- * @param seq
- * @param gapCharacter
- * @param mappings
+ * @param protein
+ * the protein alignment
+ * @param dna
+ * the coding dna alignment
+ * @param unmappedProtein
+ * any unmapped proteins are added to this list
* @return
*/
- protected static List<SequenceI> getAlignedTranslation(SequenceI seq,
- char gapCharacter, Set<AlignedCodonFrame> mappings)
+ protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
+ AlignmentI protein, AlignmentI dna,
+ List<SequenceI> unmappedProtein)
{
- List<SequenceI> result = new ArrayList<SequenceI>();
- for (AlignedCodonFrame mapping : mappings)
+ /*
+ * maintain a list of any proteins with no mappings - these will be
+ * rendered 'as is' in the protein alignment as we can't align them
+ */
+ unmappedProtein.addAll(protein.getSequences());
+
+ List<AlignedCodonFrame> mappings = protein.getCodonFrames();
+
+ /*
+ * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
+ * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
+ * comparator keeps the codon positions ordered.
+ */
+ Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
+ new CodonComparator());
+
+ for (SequenceI dnaSeq : dna.getSequences())
{
- if (mapping.involvesSequence(seq))
+ for (AlignedCodonFrame mapping : mappings)
{
- SequenceI mapped = getAlignedTranslation(seq, gapCharacter, mapping);
- if (mapped != null)
+ SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
+ if (prot != null)
{
- result.add(mapped);
+ Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
+ addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
+ seqMap, alignedCodons);
+ unmappedProtein.remove(prot);
}
}
}
- return result;
+
+ /*
+ * Finally add any unmapped peptide start residues (e.g. for incomplete
+ * codons) as if at the codon position before the second residue
+ */
+ // TODO resolve JAL-2022 so this fudge can be removed
+ int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
+ addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
+
+ return alignedCodons;
}
/**
- * Returns the translation of 'seq' (as held in the mapping) with
- * corresponding alignment (gaps).
+ * Scans for any protein mapped from position 2 (meaning unmapped start
+ * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
+ * preceding position in the alignment
*
- * @param seq
- * @param gapCharacter
- * @param mapping
- * @return
+ * @param alignedCodons
+ * the codon-to-peptide map
+ * @param mappedSequenceCount
+ * the number of distinct sequences in the map
*/
- protected static SequenceI getAlignedTranslation(SequenceI seq,
- char gapCharacter, AlignedCodonFrame mapping)
+ protected static void addUnmappedPeptideStarts(
+ Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
+ int mappedSequenceCount)
{
- String gap = String.valueOf(gapCharacter);
- boolean toDna = false;
- int fromRatio = 1;
- SequenceI mapTo = mapping.getDnaForAaSeq(seq);
- if (mapTo != null)
- {
- // mapping is from protein to nucleotide
- toDna = true;
- // should ideally get gap count ratio from mapping
- gap = String.valueOf(new char[] { gapCharacter, gapCharacter,
- gapCharacter });
- }
- else
- {
- // mapping is from nucleotide to protein
- mapTo = mapping.getAaForDnaSeq(seq);
- fromRatio = 3;
- }
- StringBuilder newseq = new StringBuilder(seq.getLength()
- * (toDna ? 3 : 1));
+ // TODO delete this ugly hack once JAL-2022 is resolved
+ // i.e. we can model startPhase > 0 (incomplete start codon)
- int residueNo = 0; // in seq, base 1
- int[] phrase = new int[fromRatio];
- int phraseOffset = 0;
- int gapWidth = 0;
- boolean first = true;
- final Sequence alignedSeq = new Sequence("", "");
+ List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
+ AlignedCodon lastCodon = null;
+ Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
- for (char c : seq.getSequence())
+ for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
+ .entrySet())
{
- if (c == gapCharacter)
+ for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
+ .entrySet())
{
- gapWidth++;
- if (gapWidth >= fromRatio)
+ SequenceI seq = sequenceCodon.getKey();
+ if (sequencesChecked.contains(seq))
{
- newseq.append(gap);
- gapWidth = 0;
+ continue;
}
- }
- else
- {
- phrase[phraseOffset++] = residueNo + 1;
- if (phraseOffset == fromRatio)
+ sequencesChecked.add(seq);
+ AlignedCodon codon = sequenceCodon.getValue();
+ if (codon.peptideCol > 1)
+ {
+ System.err
+ .println("Problem mapping protein with >1 unmapped start positions: "
+ + seq.getName());
+ }
+ else if (codon.peptideCol == 1)
{
/*
- * Have read a whole codon (or protein residue), now translate: map
- * source phrase to positions in target sequence add characters at
- * these positions to newseq Note mapping positions are base 1, our
- * sequence positions base 0.
+ * first position (peptideCol == 0) was unmapped - add it
*/
- SearchResults sr = new SearchResults();
- for (int pos : phrase)
+ if (lastCodon != null)
{
- mapping.markMappedRegion(seq, pos, sr);
+ AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
+ lastCodon.pos2, lastCodon.pos3, String.valueOf(seq
+ .getCharAt(0)), 0);
+ toAdd.put(seq, firstPeptide);
}
- newseq.append(sr.getCharacters());
- if (first)
+ else
{
- first = false;
- // Hack: Copy sequence dataset, name and description from
- // SearchResults.match[0].sequence
- // TODO? carry over sequence names from original 'complement'
- // alignment
- SequenceI mappedTo = sr.getResultSequence(0);
- alignedSeq.setName(mappedTo.getName());
- alignedSeq.setDescription(mappedTo.getDescription());
- alignedSeq.setDatasetSequence(mappedTo);
+ /*
+ * unmapped residue at start of alignment (no prior column) -
+ * 'insert' at nominal codon [0, 0, 0]
+ */
+ AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
+ String.valueOf(seq.getCharAt(0)), 0);
+ toAdd.put(seq, firstPeptide);
}
- phraseOffset = 0;
}
- residueNo++;
+ if (sequencesChecked.size() == mappedSequenceCount)
+ {
+ // no need to check past first mapped position in all sequences
+ break;
+ }
}
+ lastCodon = entry.getKey();
}
- alignedSeq.setSequence(newseq.toString());
- return alignedSeq;
- }
-
- /**
- * Realigns the given protein to match the alignment of the dna, using codon
- * mappings to translate aligned codon positions to protein residues.
- *
- * @param protein
- * the alignment whose sequences are realigned by this method
- * @param dna
- * the dna alignment whose alignment we are 'copying'
- * @return the number of sequences that were realigned
- */
- public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
- {
- List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
- unmappedProtein.addAll(protein.getSequences());
-
- List<AlignedCodonFrame> mappings = protein.getCodonFrames();
/*
- * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
- * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
- * comparator keeps the codon positions ordered.
+ * add any new codons safely after iterating over the map
*/
- Map<AlignedCodon, Map<SequenceI, String>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, String>>(
- new CodonComparator());
- for (SequenceI dnaSeq : dna.getSequences())
+ for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
{
- for (AlignedCodonFrame mapping : mappings)
- {
- Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
- SequenceI prot = mapping.findAlignedSequence(
- dnaSeq.getDatasetSequence(), protein);
- if (prot != null)
- {
- addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
- seqMap, alignedCodons);
- unmappedProtein.remove(prot);
- }
- }
+ addCodonToMap(alignedCodons, startCodon.getValue(),
+ startCodon.getKey());
}
- return alignProteinAs(protein, alignedCodons, unmappedProtein);
}
/**
* @return
*/
protected static int alignProteinAs(AlignmentI protein,
- Map<AlignedCodon, Map<SequenceI, String>> alignedCodons,
+ Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
List<SequenceI> unmappedProtein)
{
/*
int column = 0;
for (AlignedCodon codon : alignedCodons.keySet())
{
- final Map<SequenceI, String> columnResidues = alignedCodons
+ final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
.get(codon);
- for (Entry<SequenceI, String> entry : columnResidues.entrySet())
+ for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
{
// place translated codon at its column position in sequence
- entry.getKey().getSequence()[column] = entry.getValue().charAt(0);
+ entry.getKey().getSequence()[column] = entry.getValue().product
+ .charAt(0);
}
column++;
}
*/
static void addCodonPositions(SequenceI dna, SequenceI protein,
char gapChar, Mapping seqMap,
- Map<AlignedCodon, Map<SequenceI, String>> alignedCodons)
+ Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
{
Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
+
+ /*
+ * add codon positions, and their peptide translations, to the alignment
+ * map, while remembering the first codon mapped
+ */
while (codons.hasNext())
{
- AlignedCodon codon = codons.next();
- Map<SequenceI, String> seqProduct = alignedCodons.get(codon);
- if (seqProduct == null)
+ try
+ {
+ AlignedCodon codon = codons.next();
+ addCodonToMap(alignedCodons, codon, protein);
+ } catch (IncompleteCodonException e)
+ {
+ // possible incomplete trailing codon - ignore
+ } catch (NoSuchElementException e)
{
- seqProduct = new HashMap<SequenceI, String>();
- alignedCodons.put(codon, seqProduct);
+ // possibly peptide lacking STOP
}
- seqProduct.put(protein, codon.product);
}
}
/**
+ * Helper method to add a codon-to-peptide entry to the aligned codons map
+ *
+ * @param alignedCodons
+ * @param codon
+ * @param protein
+ */
+ protected static void addCodonToMap(
+ Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
+ AlignedCodon codon, SequenceI protein)
+ {
+ Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
+ if (seqProduct == null)
+ {
+ seqProduct = new HashMap<SequenceI, AlignedCodon>();
+ alignedCodons.put(codon, seqProduct);
+ }
+ seqProduct.put(protein, codon);
+ }
+
+ /**
* Returns true if a cDNA/Protein mapping either exists, or could be made,
* between at least one pair of sequences in the two alignments. Currently,
* the logic is:
* Just try to make a mapping (it is not yet stored), test whether
* successful.
*/
- return mapProteinSequenceToCdna(proteinDs, dnaDs) != null;
+ return mapCdnaToProtein(proteinDs, dnaDs) != null;
}
/**
Collection<String> types, List<SequenceI> forSequences,
boolean anyType, boolean doShow)
{
- for (AlignmentAnnotation aa : al.getAlignmentAnnotation())
+ AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
+ if (anns != null)
{
- if (anyType || types.contains(aa.label))
+ for (AlignmentAnnotation aa : anns)
{
- if ((aa.sequenceRef != null)
- && (forSequences == null || forSequences
- .contains(aa.sequenceRef)))
+ if (anyType || types.contains(aa.label))
{
- aa.visible = doShow;
+ if ((aa.sequenceRef != null)
+ && (forSequences == null || forSequences
+ .contains(aa.sequenceRef)))
+ {
+ aa.visible = doShow;
+ }
}
}
}
return false;
}
String name = seq2.getName();
- final DBRefEntry[] xrefs = seq1.getDBRef();
+ final DBRefEntry[] xrefs = seq1.getDBRefs();
if (xrefs != null)
{
for (DBRefEntry xref : xrefs)
}
/**
- * Constructs an alignment consisting of the mapped exon regions in the given
- * nucleotide sequences, and updates mappings to match.
+ * Constructs an alignment consisting of the mapped (CDS) regions in the given
+ * nucleotide sequences, and updates mappings to match. The CDS sequences are
+ * added to the original alignment's dataset, which is shared by the new
+ * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
+ * added to the alignment dataset.
*
* @param dna
* aligned dna sequences
- * @param mappings
- * from dna to protein; these are replaced with new mappings
- * @return an alignment whose sequences are the exon-only parts of the dna
- * sequences (or null if no exons are found)
+ * @param dataset
+ * - throws error if not given a dataset
+ * @param products
+ * (optional) to restrict results to CDS that map to specified
+ * protein products
+ * @return an alignment whose sequences are the cds-only parts of the dna
+ * sequences (or null if no mappings are found)
*/
- public static AlignmentI makeExonAlignment(SequenceI[] dna,
- List<AlignedCodonFrame> mappings)
+ public static AlignmentI makeCdsAlignment(SequenceI[] dna,
+ AlignmentI dataset, AlignmentI products)
{
- List<AlignedCodonFrame> newMappings = new ArrayList<AlignedCodonFrame>();
- List<SequenceI> exonSequences = new ArrayList<SequenceI>();
+ if (dataset.getDataset() != null)
+ {
+ throw new Error(
+ "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
+ }
+ List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
+ List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
+ HashSet<SequenceI> productSeqs = null;
+ if (products != null)
+ {
+ productSeqs = new HashSet<SequenceI>();
+ for (SequenceI seq : products.getSequences())
+ {
+ productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
+ .getDatasetSequence());
+ }
+ }
- for (SequenceI dnaSeq : dna)
+ /*
+ * construct CDS sequences from the (cds-to-protein) mappings made earlier;
+ * this makes it possible to model multiple products from dna (e.g. EMBL);
+ * however it does mean we don't have the EMBL protein_id (a property on
+ * the CDS features) in order to make the CDS sequence name :-(
+ */
+ for (SequenceI seq : dna)
{
- final SequenceI ds = dnaSeq.getDatasetSequence();
+ SequenceI seqDss = seq.getDatasetSequence() == null ? seq : seq
+ .getDatasetSequence();
List<AlignedCodonFrame> seqMappings = MappingUtils
- .findMappingsForSequence(ds, mappings);
- for (AlignedCodonFrame acf : seqMappings)
+ .findMappingsForSequence(seq, mappings);
+ for (AlignedCodonFrame mapping : seqMappings)
{
- AlignedCodonFrame newMapping = new AlignedCodonFrame();
- final List<SequenceI> mappedExons = makeExonSequences(ds, acf,
- newMapping);
- if (!mappedExons.isEmpty())
+ List<Mapping> mappingsFromSequence = mapping.getMappingsFromSequence(seq);
+
+ for (Mapping aMapping : mappingsFromSequence)
{
- exonSequences.addAll(mappedExons);
- newMappings.add(newMapping);
+ if (aMapping.getMap().getFromRatio() == 1)
+ {
+ /*
+ * not a dna-to-protein mapping (likely dna-to-cds)
+ */
+ continue;
+ }
+
+ /*
+ * check for an existing CDS sequence i.e. a 3:1 mapping to
+ * the dna mapping's product
+ */
+ SequenceI cdsSeq = null;
+
+ // TODO better mappings collection data model so we can do
+ // a direct lookup instead of double loops to find mappings
+
+ SequenceI proteinProduct = aMapping.getTo();
+
+ /*
+ * skip if not mapped to one of a specified set of proteins
+ */
+ if (productSeqs != null && !productSeqs.contains(proteinProduct))
+ {
+ continue;
+ }
+
+ for (AlignedCodonFrame acf : MappingUtils
+ .findMappingsForSequence(proteinProduct, mappings))
+ {
+ for (SequenceToSequenceMapping map : acf.getMappings())
+ {
+ if (map.getMapping().getMap().getFromRatio() == 3
+ && proteinProduct == map.getMapping().getTo()
+ && seqDss != map.getFromSeq())
+ {
+ /*
+ * found a 3:1 mapping to the protein product which is not
+ * from the dna sequence...assume it is from the CDS sequence
+ * TODO mappings data model that brings together related
+ * dna-cds-protein mappings in one object
+ */
+ cdsSeq = map.getFromSeq();
+ }
+ }
+ }
+ if (cdsSeq != null)
+ {
+ /*
+ * mappings are always to dataset sequences so create an aligned
+ * sequence to own it; add the dataset sequence to the dataset
+ */
+ SequenceI derivedSequence = cdsSeq.deriveSequence();
+ cdsSeqs.add(derivedSequence);
+ if (!dataset.getSequences().contains(cdsSeq))
+ {
+ dataset.addSequence(cdsSeq);
+ }
+ continue;
+ }
+
+ /*
+ * didn't find mapped CDS sequence - construct it and add
+ * its dataset sequence to the dataset
+ */
+ cdsSeq = makeCdsSequence(seq.getDatasetSequence(), aMapping);
+ SequenceI cdsSeqDss = cdsSeq.createDatasetSequence();
+ cdsSeqs.add(cdsSeq);
+ if (!dataset.getSequences().contains(cdsSeqDss))
+ {
+ dataset.addSequence(cdsSeqDss);
+ }
+
+ /*
+ * add a mapping from CDS to the (unchanged) mapped to range
+ */
+ List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
+ cdsSeq.getLength() });
+ MapList map = new MapList(cdsRange, aMapping.getMap()
+ .getToRanges(), aMapping.getMap().getFromRatio(),
+ aMapping.getMap().getToRatio());
+ AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
+ cdsToProteinMapping.addMap(cdsSeq, proteinProduct, map);
+
+ /*
+ * guard against duplicating the mapping if repeating this action
+ */
+ if (!mappings.contains(cdsToProteinMapping))
+ {
+ mappings.add(cdsToProteinMapping);
+ }
+
+ /*
+ * add another mapping from original 'from' range to CDS
+ */
+ AlignedCodonFrame dnaToProteinMapping = new AlignedCodonFrame();
+ map = new MapList(aMapping.getMap().getFromRanges(), cdsRange, 1,
+ 1);
+ dnaToProteinMapping.addMap(seq.getDatasetSequence(), cdsSeq, map);
+ if (!mappings.contains(dnaToProteinMapping))
+ {
+ mappings.add(dnaToProteinMapping);
+ }
+
+
+ /*
+ * transfer any features on dna that overlap the CDS
+ */
+ transferFeatures(seq, cdsSeq, map, null, SequenceOntologyI.CDS);
}
}
}
- AlignmentI al = new Alignment(
- exonSequences.toArray(new SequenceI[exonSequences.size()]));
- al.setDataset(null);
- /*
- * Replace the old mappings with the new ones
- */
- mappings.clear();
- mappings.addAll(newMappings);
+ AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
+ .size()]));
+ cds.setDataset(dataset);
- return al;
+ return cds;
}
/**
- * Helper method to make exon-only sequences and populate their mappings to
- * protein products
- * <p>
- * For example, if ggCCaTTcGAg has mappings [3, 4, 6, 7, 9, 10] to protein
- * then generate a sequence CCTTGA with mapping [1, 6] to the same protein
- * residues
- * <p>
- * Typically eukaryotic dna will include exons encoding for a single peptide
- * sequence i.e. return a single result. Bacterial dna may have overlapping
- * exon mappings coding for multiple peptides so return multiple results
- * (example EMBL KF591215).
+ * Helper method that makes a CDS sequence as defined by the mappings from the
+ * given sequence i.e. extracts the 'mapped from' ranges (which may be on
+ * forward or reverse strand).
*
- * @param dnaSeq
- * a dna dataset sequence
+ * @param seq
* @param mapping
- * containing one or more mappings of the sequence to protein
- * @param newMapping
- * the new mapping to populate, from the exon-only sequences to their
- * mapped protein sequences
- * @return
+ * @return CDS sequence (as a dataset sequence)
*/
- protected static List<SequenceI> makeExonSequences(SequenceI dnaSeq,
- AlignedCodonFrame mapping, AlignedCodonFrame newMapping)
+ static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping)
{
- List<SequenceI> exonSequences = new ArrayList<SequenceI>();
- List<Mapping> seqMappings = mapping.getMappingsForSequence(dnaSeq);
- final char[] dna = dnaSeq.getSequence();
- for (Mapping seqMapping : seqMappings)
- {
- StringBuilder newSequence = new StringBuilder(dnaSeq.getLength());
+ char[] seqChars = seq.getSequence();
+ List<int[]> fromRanges = mapping.getMap().getFromRanges();
+ int cdsWidth = MappingUtils.getLength(fromRanges);
+ char[] newSeqChars = new char[cdsWidth];
- /*
- * Get the codon regions as { [2, 5], [7, 12], [14, 14] etc }
- */
- final List<int[]> dnaExonRanges = seqMapping.getMap().getFromRanges();
- for (int[] range : dnaExonRanges)
+ int newPos = 0;
+ for (int[] range : fromRanges)
+ {
+ if (range[0] <= range[1])
+ {
+ // forward strand mapping - just copy the range
+ int length = range[1] - range[0] + 1;
+ System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
+ length);
+ newPos += length;
+ }
+ else
{
- for (int pos = range[0]; pos <= range[1]; pos++)
+ // reverse strand mapping - copy and complement one by one
+ for (int i = range[0]; i >= range[1]; i--)
{
- newSequence.append(dna[pos - 1]);
+ newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
}
}
+ }
- SequenceI exon = new Sequence(dnaSeq.getName(),
- newSequence.toString());
+ SequenceI newSeq = new Sequence(seq.getName() + "|"
+ + mapping.getTo().getName(), newSeqChars, 1, newPos);
+ return newSeq;
+ }
- /*
- * Locate any xrefs to CDS database on the protein product and attach to
- * the CDS sequence. Also add as a sub-token of the sequence name.
- */
- // default to "CDS" if we can't locate an actual gene id
- String cdsAccId = FeatureProperties
- .getCodingFeature(DBRefSource.EMBL);
- DBRefEntry[] cdsRefs = DBRefUtils.selectRefs(seqMapping.getTo()
- .getDBRef(), DBRefSource.CODINGDBS);
- if (cdsRefs != null)
+ /**
+ * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
+ * feature start/end ranges, optionally omitting specified feature types.
+ * Returns the number of features copied.
+ *
+ * @param fromSeq
+ * @param toSeq
+ * @param select
+ * if not null, only features of this type are copied (including
+ * subtypes in the Sequence Ontology)
+ * @param mapping
+ * the mapping from 'fromSeq' to 'toSeq'
+ * @param omitting
+ */
+ public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
+ MapList mapping, String select, String... omitting)
+ {
+ SequenceI copyTo = toSeq;
+ while (copyTo.getDatasetSequence() != null)
+ {
+ copyTo = copyTo.getDatasetSequence();
+ }
+
+ SequenceOntologyI so = SequenceOntologyFactory.getInstance();
+ int count = 0;
+ SequenceFeature[] sfs = fromSeq.getSequenceFeatures();
+ if (sfs != null)
+ {
+ for (SequenceFeature sf : sfs)
{
- for (DBRefEntry cdsRef : cdsRefs)
+ String type = sf.getType();
+ if (select != null && !so.isA(type, select))
{
- exon.addDBRef(new DBRefEntry(cdsRef));
- cdsAccId = cdsRef.getAccessionId();
+ continue;
+ }
+ boolean omit = false;
+ for (String toOmit : omitting)
+ {
+ if (type.equals(toOmit))
+ {
+ omit = true;
+ }
+ }
+ if (omit)
+ {
+ continue;
}
- }
- exon.setName(exon.getName() + "|" + cdsAccId);
- exon.createDatasetSequence();
-
- /*
- * Build new mappings - from the same protein regions, but now to
- * contiguous exons
- */
- List<int[]> exonRange = new ArrayList<int[]>();
- exonRange.add(new int[] { 1, newSequence.length() });
- MapList map = new MapList(exonRange, seqMapping.getMap()
- .getToRanges(), 3, 1);
- newMapping.addMap(exon.getDatasetSequence(), seqMapping.getTo(), map);
- MapList cdsToDnaMap = new MapList(dnaExonRanges, exonRange, 1, 1);
- newMapping.addMap(dnaSeq, exon.getDatasetSequence(), cdsToDnaMap);
- exonSequences.add(exon);
- }
- return exonSequences;
+ /*
+ * locate the mapped range - null if either start or end is
+ * not mapped (no partial overlaps are calculated)
+ */
+ int start = sf.getBegin();
+ int end = sf.getEnd();
+ int[] mappedTo = mapping.locateInTo(start, end);
+ /*
+ * if whole exon range doesn't map, try interpreting it
+ * as 5' or 3' exon overlapping the CDS range
+ */
+ if (mappedTo == null)
+ {
+ mappedTo = mapping.locateInTo(end, end);
+ if (mappedTo != null)
+ {
+ /*
+ * end of exon is in CDS range - 5' overlap
+ * to a range from the start of the peptide
+ */
+ mappedTo[0] = 1;
+ }
+ }
+ if (mappedTo == null)
+ {
+ mappedTo = mapping.locateInTo(start, start);
+ if (mappedTo != null)
+ {
+ /*
+ * start of exon is in CDS range - 3' overlap
+ * to a range up to the end of the peptide
+ */
+ mappedTo[1] = toSeq.getLength();
+ }
+ }
+ if (mappedTo != null)
+ {
+ SequenceFeature copy = new SequenceFeature(sf);
+ copy.setBegin(Math.min(mappedTo[0], mappedTo[1]));
+ copy.setEnd(Math.max(mappedTo[0], mappedTo[1]));
+ copyTo.addSequenceFeature(copy);
+ count++;
+ }
+ }
+ }
+ return count;
+ }
+
+ /**
+ * Returns a mapping from dna to protein by inspecting sequence features of
+ * type "CDS" on the dna.
+ *
+ * @param dnaSeq
+ * @param proteinSeq
+ * @return
+ */
+ public static MapList mapCdsToProtein(SequenceI dnaSeq,
+ SequenceI proteinSeq)
+ {
+ List<int[]> ranges = findCdsPositions(dnaSeq);
+ int mappedDnaLength = MappingUtils.getLength(ranges);
+
+ int proteinLength = proteinSeq.getLength();
+ int proteinStart = proteinSeq.getStart();
+ int proteinEnd = proteinSeq.getEnd();
+
+ /*
+ * incomplete start codon may mean X at start of peptide
+ * we ignore both for mapping purposes
+ */
+ if (proteinSeq.getCharAt(0) == 'X')
+ {
+ // todo JAL-2022 support startPhase > 0
+ proteinStart++;
+ proteinLength--;
+ }
+ List<int[]> proteinRange = new ArrayList<int[]>();
+
+ /*
+ * dna length should map to protein (or protein plus stop codon)
+ */
+ int codesForResidues = mappedDnaLength / 3;
+ if (codesForResidues == (proteinLength + 1))
+ {
+ // assuming extra codon is for STOP and not in peptide
+ codesForResidues--;
+ }
+ if (codesForResidues == proteinLength)
+ {
+ proteinRange.add(new int[] { proteinStart, proteinEnd });
+ return new MapList(ranges, proteinRange, 3, 1);
+ }
+ return null;
+ }
+
+ /**
+ * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
+ * start/end positions of sequence features of type "CDS" (or a sub-type of
+ * CDS in the Sequence Ontology). The ranges are sorted into ascending start
+ * position order, so this method is only valid for linear CDS in the same
+ * sense as the protein product.
+ *
+ * @param dnaSeq
+ * @return
+ */
+ public static List<int[]> findCdsPositions(SequenceI dnaSeq)
+ {
+ List<int[]> result = new ArrayList<int[]>();
+ SequenceFeature[] sfs = dnaSeq.getSequenceFeatures();
+ if (sfs == null)
+ {
+ return result;
+ }
+
+ SequenceOntologyI so = SequenceOntologyFactory.getInstance();
+ int startPhase = 0;
+
+ for (SequenceFeature sf : sfs)
+ {
+ /*
+ * process a CDS feature (or a sub-type of CDS)
+ */
+ if (so.isA(sf.getType(), SequenceOntologyI.CDS))
+ {
+ int phase = 0;
+ try
+ {
+ phase = Integer.parseInt(sf.getPhase());
+ } catch (NumberFormatException e)
+ {
+ // ignore
+ }
+ /*
+ * phase > 0 on first codon means 5' incomplete - skip to the start
+ * of the next codon; example ENST00000496384
+ */
+ int begin = sf.getBegin();
+ int end = sf.getEnd();
+ if (result.isEmpty())
+ {
+ begin += phase;
+ if (begin > end)
+ {
+ // shouldn't happen!
+ System.err
+ .println("Error: start phase extends beyond start CDS in "
+ + dnaSeq.getName());
+ }
+ }
+ result.add(new int[] { begin, end });
+ }
+ }
+
+ /*
+ * remove 'startPhase' positions (usually 0) from the first range
+ * so we begin at the start of a complete codon
+ */
+ if (!result.isEmpty())
+ {
+ // TODO JAL-2022 correctly model start phase > 0
+ result.get(0)[0] += startPhase;
+ }
+
+ /*
+ * Finally sort ranges by start position. This avoids a dependency on
+ * keeping features in order on the sequence (if they are in order anyway,
+ * the sort will have almost no work to do). The implicit assumption is CDS
+ * ranges are assembled in order. Other cases should not use this method,
+ * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
+ */
+ Collections.sort(result, new Comparator<int[]>()
+ {
+ @Override
+ public int compare(int[] o1, int[] o2)
+ {
+ return Integer.compare(o1[0], o2[0]);
+ }
+ });
+ return result;
+ }
+
+ /**
+ * 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
+ */
+ public static int computeProteinFeatures(SequenceI dnaSeq,
+ SequenceI peptide, MapList dnaToProtein)
+ {
+ while (dnaSeq.getDatasetSequence() != null)
+ {
+ dnaSeq = dnaSeq.getDatasetSequence();
+ }
+ while (peptide.getDatasetSequence() != null)
+ {
+ peptide = peptide.getDatasetSequence();
+ }
+
+ transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
+
+ /*
+ * compute protein variants from dna variants and codon mappings;
+ * NB - alternatively we could retrieve this using the REST service e.g.
+ * http://rest.ensembl.org/overlap/translation
+ * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
+ * which would be a bit slower but possibly more reliable
+ */
+
+ /*
+ * build a map with codon variations for each potentially varying peptide
+ */
+ LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
+ dnaSeq, dnaToProtein);
+
+ /*
+ * scan codon variations, compute peptide variants and add to peptide sequence
+ */
+ int count = 0;
+ for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
+ {
+ int peptidePos = variant.getKey();
+ List<DnaVariant>[] codonVariants = variant.getValue();
+ count += computePeptideVariants(peptide, peptidePos, codonVariants);
+ }
+
+ /*
+ * sort to get sequence features in start position order
+ * - would be better to store in Sequence as a TreeSet or NCList?
+ */
+ if (peptide.getSequenceFeatures() != null)
+ {
+ 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;
+ }
+
+ /**
+ * Computes non-synonymous peptide variants from codon variants and adds them
+ * as sequence_variant features on the protein sequence (one feature per
+ * allele variant). Selected attributes (variant id, clinical significance)
+ * are copied over to the new features.
+ *
+ * @param peptide
+ * the protein sequence
+ * @param peptidePos
+ * the position to compute peptide variants for
+ * @param codonVariants
+ * a list of dna variants per codon position
+ * @return the number of features added
+ */
+ static int computePeptideVariants(SequenceI peptide, int peptidePos,
+ List<DnaVariant>[] codonVariants)
+ {
+ String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
+ int count = 0;
+ String base1 = codonVariants[0].get(0).base;
+ String base2 = codonVariants[1].get(0).base;
+ String base3 = codonVariants[2].get(0).base;
+
+ /*
+ * variants in first codon base
+ */
+ for (DnaVariant var : codonVariants[0])
+ {
+ if (var.variant != null)
+ {
+ String alleles = (String) var.variant.getValue("alleles");
+ if (alleles != null)
+ {
+ for (String base : alleles.split(","))
+ {
+ String codon = base + base2 + base3;
+ if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
+ {
+ count++;
+ }
+ }
+ }
+ }
+ }
+
+ /*
+ * variants in second codon base
+ */
+ for (DnaVariant var : codonVariants[1])
+ {
+ if (var.variant != null)
+ {
+ String alleles = (String) var.variant.getValue("alleles");
+ if (alleles != null)
+ {
+ for (String base : alleles.split(","))
+ {
+ String codon = base1 + base + base3;
+ if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
+ {
+ count++;
+ }
+ }
+ }
+ }
+ }
+
+ /*
+ * variants in third codon base
+ */
+ for (DnaVariant var : codonVariants[2])
+ {
+ if (var.variant != null)
+ {
+ String alleles = (String) var.variant.getValue("alleles");
+ if (alleles != null)
+ {
+ for (String base : alleles.split(","))
+ {
+ String codon = base1 + base2 + base;
+ if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
+ {
+ count++;
+ }
+ }
+ }
+ }
+ }
+
+ return count;
+ }
+
+ /**
+ * Helper method that adds a peptide variant feature, provided the given codon
+ * translates to a value different to the current residue (is a non-synonymous
+ * variant). ID and clinical_significance attributes of the dna variant (if
+ * present) are copied to the new feature.
+ *
+ * @param peptide
+ * @param peptidePos
+ * @param residue
+ * @param var
+ * @param codon
+ * @return true if a feature was added, else false
+ */
+ static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
+ String residue, DnaVariant var, String codon)
+ {
+ /*
+ * get peptide translation of codon e.g. GAT -> D
+ * note that variants which are not single alleles,
+ * e.g. multibase variants or HGMD_MUTATION etc
+ * are currently ignored here
+ */
+ String trans = codon.contains("-") ? "-"
+ : (codon.length() > 3 ? null : ResidueProperties
+ .codonTranslate(codon));
+ if (trans != null && !trans.equals(residue))
+ {
+ String residue3Char = StringUtils
+ .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
+ String trans3Char = StringUtils
+ .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
+ String desc = "p." + residue3Char + peptidePos + trans3Char;
+ // set score to 0f so 'graduated colour' option is offered! JAL-2060
+ SequenceFeature sf = new SequenceFeature(
+ SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
+ peptidePos, 0f, "Jalview");
+ StringBuilder attributes = new StringBuilder(32);
+ String id = (String) var.variant.getValue(ID);
+ if (id != null)
+ {
+ if (id.startsWith(SEQUENCE_VARIANT))
+ {
+ id = id.substring(SEQUENCE_VARIANT.length());
+ }
+ sf.setValue(ID, id);
+ attributes.append(ID).append("=").append(id);
+ // TODO handle other species variants
+ StringBuilder link = new StringBuilder(32);
+ try
+ {
+ link.append(desc).append(" ").append(id)
+ .append("|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
+ .append(URLEncoder.encode(id, "UTF-8"));
+ sf.addLink(link.toString());
+ } catch (UnsupportedEncodingException e)
+ {
+ // as if
+ }
+ }
+ String clinSig = (String) var.variant
+ .getValue(CLINICAL_SIGNIFICANCE);
+ if (clinSig != null)
+ {
+ sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
+ attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
+ .append(clinSig);
+ }
+ peptide.addSequenceFeature(sf);
+ if (attributes.length() > 0)
+ {
+ sf.setAttributes(attributes.toString());
+ }
+ return true;
+ }
+ return false;
+ }
+
+ /**
+ * Builds a map whose key is position in the protein sequence, and value is a
+ * list of the base and all variants for each corresponding codon position
+ *
+ * @param dnaSeq
+ * @param dnaToProtein
+ * @return
+ */
+ static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
+ SequenceI dnaSeq, MapList dnaToProtein)
+ {
+ /*
+ * map from peptide position to all variants of the codon which codes for it
+ * LinkedHashMap ensures we keep the peptide features in sequence order
+ */
+ LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
+ 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];
+ List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
+ if (codonVariants == null)
+ {
+ codonVariants = new ArrayList[3];
+ codonVariants[0] = new ArrayList<DnaVariant>();
+ codonVariants[1] = new ArrayList<DnaVariant>();
+ codonVariants[2] = new ArrayList<DnaVariant>();
+ variants.put(peptidePosition, codonVariants);
+ }
+
+ /*
+ * extract dna variants to a string array
+ */
+ String alls = (String) sf.getValue("alleles");
+ if (alls == null)
+ {
+ continue;
+ }
+ String[] alleles = alls.toUpperCase().split(",");
+ int i = 0;
+ for (String allele : alleles)
+ {
+ alleles[i++] = allele.trim(); // lose any space characters "A, G"
+ }
+
+ /*
+ * get this peptide's 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 any variant) for each codon position
+ */
+ for (int codonPos = 0; codonPos < 3; codonPos++)
+ {
+ String nucleotide = String.valueOf(
+ dnaSeq.getCharAt(codon[codonPos] - dnaStart))
+ .toUpperCase();
+ List<DnaVariant> codonVariant = codonVariants[codonPos];
+ if (codon[codonPos] == dnaCol)
+ {
+ if (!codonVariant.isEmpty()
+ && codonVariant.get(0).variant == null)
+ {
+ /*
+ * already recorded base value, add this variant
+ */
+ codonVariant.get(0).variant = sf;
+ }
+ else
+ {
+ /*
+ * add variant with base value
+ */
+ codonVariant.add(new DnaVariant(nucleotide, sf));
+ }
+ }
+ else if (codonVariant.isEmpty())
+ {
+ /*
+ * record (possibly non-varying) base value
+ */
+ codonVariant.add(new DnaVariant(nucleotide));
+ }
+ }
+ }
+ }
+ return variants;
+ }
+
+ /**
+ * Makes an alignment with a copy of the given sequences, adding in any
+ * non-redundant sequences which are mapped to by the cross-referenced
+ * sequences.
+ *
+ * @param seqs
+ * @param xrefs
+ * @return
+ */
+ public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
+ SequenceI[] xrefs)
+ {
+ AlignmentI copy = new Alignment(new Alignment(seqs));
+
+ /*
+ * add mappings between sequences to the new alignment
+ */
+ AlignedCodonFrame mappings = new AlignedCodonFrame();
+ copy.addCodonFrame(mappings);
+ for (int i = 0; i < copy.getHeight(); i++)
+ {
+ SequenceI from = seqs[i];
+ SequenceI to = copy.getSequenceAt(i);
+ if (to.getDatasetSequence() != null)
+ {
+ to = to.getDatasetSequence();
+ }
+ int start = from.getStart();
+ int end = from.getEnd();
+ MapList map = new MapList(new int[] { start, end }, new int[] {
+ start, end }, 1, 1);
+ mappings.addMap(to, from, map);
+ }
+
+ SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
+ if (xrefs != null)
+ {
+ for (SequenceI xref : xrefs)
+ {
+ DBRefEntry[] dbrefs = xref.getDBRefs();
+ if (dbrefs != null)
+ {
+ for (DBRefEntry dbref : dbrefs)
+ {
+ if (dbref.getMap() == null || dbref.getMap().getTo() == null)
+ {
+ continue;
+ }
+ SequenceI mappedTo = dbref.getMap().getTo();
+ SequenceI match = matcher.findIdMatch(mappedTo);
+ if (match == null)
+ {
+ matcher.add(mappedTo);
+ copy.addSequence(mappedTo);
+ }
+ }
+ }
+ }
+ }
+ return copy;
+ }
+
+ /**
+ * Try to align sequences in 'unaligned' to match the alignment of their
+ * mapped regions in 'aligned'. For example, could use this to align CDS
+ * sequences which are mapped to their parent cDNA sequences.
+ *
+ * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
+ * dna-to-protein or protein-to-dna use alternative methods.
+ *
+ * @param unaligned
+ * sequences to be aligned
+ * @param aligned
+ * holds aligned sequences and their mappings
+ * @return
+ */
+ public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
+ {
+ List<SequenceI> unmapped = new ArrayList<SequenceI>();
+ Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
+ unaligned, aligned, unmapped);
+ int width = columnMap.size();
+ char gap = unaligned.getGapCharacter();
+ int realignedCount = 0;
+
+ for (SequenceI seq : unaligned.getSequences())
+ {
+ if (!unmapped.contains(seq))
+ {
+ char[] newSeq = new char[width];
+ Arrays.fill(newSeq, gap);
+ int newCol = 0;
+ int lastCol = 0;
+
+ /*
+ * traverse the map to find columns populated
+ * by our sequence
+ */
+ for (Integer column : columnMap.keySet())
+ {
+ Character c = columnMap.get(column).get(seq);
+ if (c != null)
+ {
+ /*
+ * sequence has a character at this position
+ *
+ */
+ newSeq[newCol] = c;
+ lastCol = newCol;
+ }
+ newCol++;
+ }
+
+ /*
+ * trim trailing gaps
+ */
+ if (lastCol < width)
+ {
+ char[] tmp = new char[lastCol + 1];
+ System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
+ newSeq = tmp;
+ }
+ seq.setSequence(String.valueOf(newSeq));
+ realignedCount++;
+ }
+ }
+ return realignedCount;
+ }
+
+ /**
+ * Returns a map whose key is alignment column number (base 1), and whose
+ * values are a map of sequence characters in that column.
+ *
+ * @param unaligned
+ * @param aligned
+ * @param unmapped
+ * @return
+ */
+ static Map<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
+ AlignmentI unaligned, AlignmentI aligned, List<SequenceI> unmapped)
+ {
+ /*
+ * Map will hold, for each aligned column position, a map of
+ * {unalignedSequence, sequenceCharacter} at that position.
+ * TreeMap keeps the entries in ascending column order.
+ */
+ Map<Integer, Map<SequenceI, Character>> map = new TreeMap<Integer, Map<SequenceI, Character>>();
+
+ /*
+ * r any sequences that have no mapping so can't be realigned
+ */
+ unmapped.addAll(unaligned.getSequences());
+
+ List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
+
+ for (SequenceI seq : unaligned.getSequences())
+ {
+ for (AlignedCodonFrame mapping : mappings)
+ {
+ SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
+ if (fromSeq != null)
+ {
+ Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
+ if (addMappedPositions(seq, fromSeq, seqMap, map))
+ {
+ unmapped.remove(seq);
+ }
+ }
+ }
+ }
+ return map;
+ }
+
+ /**
+ * Helper method that adds to a map the mapped column positions of a sequence. <br>
+ * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
+ * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
+ * sequence.
+ *
+ * @param seq
+ * the sequence whose column positions we are recording
+ * @param fromSeq
+ * a sequence that is mapped to the first sequence
+ * @param seqMap
+ * the mapping from 'fromSeq' to 'seq'
+ * @param map
+ * a map to add the column positions (in fromSeq) of the mapped
+ * positions of seq
+ * @return
+ */
+ static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
+ Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
+ {
+ if (seqMap == null)
+ {
+ return false;
+ }
+
+ char[] fromChars = fromSeq.getSequence();
+ int toStart = seq.getStart();
+ char[] toChars = seq.getSequence();
+
+ /*
+ * traverse [start, end, start, end...] ranges in fromSeq
+ */
+ for (int[] fromRange : seqMap.getMap().getFromRanges())
+ {
+ for (int i = 0; i < fromRange.length - 1; i += 2)
+ {
+ boolean forward = fromRange[i + 1] >= fromRange[i];
+
+ /*
+ * find the range mapped to (sequence positions base 1)
+ */
+ int[] range = seqMap.locateMappedRange(fromRange[i],
+ fromRange[i + 1]);
+ if (range == null)
+ {
+ System.err.println("Error in mapping " + seqMap + " from "
+ + fromSeq.getName());
+ return false;
+ }
+ int fromCol = fromSeq.findIndex(fromRange[i]);
+ int mappedCharPos = range[0];
+
+ /*
+ * walk over the 'from' aligned sequence in forward or reverse
+ * direction; when a non-gap is found, record the column position
+ * of the next character of the mapped-to sequence; stop when all
+ * the characters of the range have been counted
+ */
+ while (mappedCharPos <= range[1])
+ {
+ if (!Comparison.isGap(fromChars[fromCol - 1]))
+ {
+ /*
+ * mapped from sequence has a character in this column
+ * record the column position for the mapped to character
+ */
+ Map<SequenceI, Character> seqsMap = map.get(fromCol);
+ if (seqsMap == null)
+ {
+ seqsMap = new HashMap<SequenceI, Character>();
+ map.put(fromCol, seqsMap);
+ }
+ seqsMap.put(seq, toChars[mappedCharPos - toStart]);
+ mappedCharPos++;
+ }
+ fromCol += (forward ? 1 : -1);
+ }
+ }
+ }
+ return true;
+ }
+
+ // strictly temporary hack until proper criteria for aligning protein to cds
+ // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
+ public static boolean looksLikeEnsembl(AlignmentI alignment)
+ {
+ for (SequenceI seq : alignment.getSequences())
+ {
+ String name = seq.getName();
+ if (!name.startsWith("ENSG") && !name.startsWith("ENST"))
+ {
+ return false;
+ }
+ }
+ return true;
}
}