2 * Jalview - A Sequence Alignment Editor and Viewer (Version 2.8.2)
3 * Copyright (C) 2014 The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.analysis;
23 import jalview.datamodel.AlignedCodon;
24 import jalview.datamodel.AlignedCodonFrame;
25 import jalview.datamodel.AlignmentAnnotation;
26 import jalview.datamodel.AlignmentI;
27 import jalview.datamodel.Mapping;
28 import jalview.datamodel.SearchResults;
29 import jalview.datamodel.Sequence;
30 import jalview.datamodel.SequenceI;
31 import jalview.schemes.ResidueProperties;
32 import jalview.util.MapList;
34 import java.util.ArrayList;
35 import java.util.Arrays;
36 import java.util.HashMap;
37 import java.util.Iterator;
38 import java.util.LinkedHashMap;
39 import java.util.List;
41 import java.util.Map.Entry;
43 import java.util.TreeMap;
46 * grab bag of useful alignment manipulation operations Expect these to be
47 * refactored elsewhere at some point.
52 public class AlignmentUtils
56 * Represents the 3 possible results of trying to map one alignment to
59 public enum MappingResult
61 Mapped, NotMapped, AlreadyMapped
65 * given an existing alignment, create a new alignment including all, or up to
66 * flankSize additional symbols from each sequence's dataset sequence
72 public static AlignmentI expandContext(AlignmentI core, int flankSize)
74 List<SequenceI> sq = new ArrayList<SequenceI>();
76 for (SequenceI s : core.getSequences())
78 SequenceI newSeq = s.deriveSequence();
79 if (newSeq.getStart() > maxoffset
80 && newSeq.getDatasetSequence().getStart() < s.getStart())
82 maxoffset = newSeq.getStart();
88 maxoffset = flankSize;
90 // now add offset to create a new expanded alignment
91 for (SequenceI s : sq)
94 while (ds.getDatasetSequence() != null)
96 ds = ds.getDatasetSequence();
98 int s_end = s.findPosition(s.getStart() + s.getLength());
99 // find available flanking residues for sequence
100 int ustream_ds = s.getStart() - ds.getStart(), dstream_ds = ds
103 // build new flanked sequence
105 // compute gap padding to start of flanking sequence
106 int offset = maxoffset - ustream_ds;
108 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
111 if (flankSize < ustream_ds)
113 // take up to flankSize residues
114 offset = maxoffset - flankSize;
115 ustream_ds = flankSize;
117 if (flankSize < dstream_ds)
119 dstream_ds = flankSize;
122 char[] upstream = new String(ds.getSequence(s.getStart() - 1
123 - ustream_ds, s.getStart() - 1)).toLowerCase().toCharArray();
124 char[] downstream = new String(ds.getSequence(s_end - 1, s_end + 1
125 + dstream_ds)).toLowerCase().toCharArray();
126 char[] coreseq = s.getSequence();
127 char[] nseq = new char[offset + upstream.length + downstream.length
129 char c = core.getGapCharacter();
130 // TODO could lowercase the flanking regions
132 for (; p < offset; p++)
136 // s.setSequence(new String(upstream).toLowerCase()+new String(coreseq) +
137 // new String(downstream).toLowerCase());
138 System.arraycopy(upstream, 0, nseq, p, upstream.length);
139 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
141 System.arraycopy(downstream, 0, nseq, p + coreseq.length
142 + upstream.length, downstream.length);
143 s.setSequence(new String(nseq));
144 s.setStart(s.getStart() - ustream_ds);
145 s.setEnd(s_end + downstream.length);
147 AlignmentI newAl = new jalview.datamodel.Alignment(
148 sq.toArray(new SequenceI[0]));
149 for (SequenceI s : sq)
151 if (s.getAnnotation() != null)
153 for (AlignmentAnnotation aa : s.getAnnotation())
155 newAl.addAnnotation(aa);
159 newAl.setDataset(core.getDataset());
164 * Returns the index (zero-based position) of a sequence in an alignment, or
171 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
175 for (SequenceI alSeq : al.getSequences())
188 * Returns a map of lists of sequences in the alignment, keyed by sequence
189 * name. For use in mapping between different alignment views of the same
192 * @see jalview.datamodel.AlignmentI#getSequencesByName()
194 public static Map<String, List<SequenceI>> getSequencesByName(
197 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
198 for (SequenceI seq : al.getSequences())
200 String name = seq.getName();
203 List<SequenceI> seqs = theMap.get(name);
206 seqs = new ArrayList<SequenceI>();
207 theMap.put(name, seqs);
216 * Build mapping of protein to cDNA alignment. Mappings are made between
217 * sequences which have the same name and compatible lengths. Any new mappings
218 * are added to the protein alignment. Has a 3-valued result: either Mapped
219 * (at least one sequence mapping was created), AlreadyMapped (all possible
220 * sequence mappings already exist), or NotMapped (no possible sequence
223 * @param proteinAlignment
224 * @param cdnaAlignment
227 public static MappingResult mapProteinToCdna(
228 final AlignmentI proteinAlignment,
229 final AlignmentI cdnaAlignment)
231 if (proteinAlignment == null || cdnaAlignment == null)
233 return MappingResult.NotMapped;
236 boolean mappingPossible = false;
237 boolean mappingPerformed = false;
239 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
242 * Build a look-up of cDNA sequences by name, for matching purposes.
244 Map<String, List<SequenceI>> cdnaSeqs = cdnaAlignment
245 .getSequencesByName();
247 for (SequenceI aaSeq : thisSeqs)
249 AlignedCodonFrame acf = new AlignedCodonFrame();
250 List<SequenceI> candidates = cdnaSeqs.get(aaSeq.getName());
251 if (candidates == null)
254 * No cDNA sequence with matching name, so no mapping possible for this
259 mappingPossible = true;
260 for (SequenceI cdnaSeq : candidates)
262 if (!mappingExists(proteinAlignment.getCodonFrames(),
263 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
265 MapList map = mapProteinToCdna(aaSeq, cdnaSeq);
268 acf.addMap(cdnaSeq, aaSeq, map);
269 mappingPerformed = true;
273 proteinAlignment.addCodonFrame(acf);
277 * If at least one mapping was possible but none was done, then the
278 * alignments are already as mapped as they can be.
280 if (mappingPossible && !mappingPerformed)
282 return MappingResult.AlreadyMapped;
286 return mappingPerformed ? MappingResult.Mapped
287 : MappingResult.NotMapped;
292 * Answers true if the mappings include one between the given (dataset)
295 public static boolean mappingExists(Set<AlignedCodonFrame> set,
296 SequenceI aaSeq, SequenceI cdnaSeq)
300 for (AlignedCodonFrame acf : set)
302 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
312 * Build a mapping (if possible) of a protein to a cDNA sequence. The cDNA
313 * must be three times the length of the protein, possibly after ignoring
314 * start and/or stop codons. Returns null if no mapping is determined.
320 public static MapList mapProteinToCdna(SequenceI proteinSeq,
324 * Here we handle either dataset sequence set (desktop) or absent (applet)
326 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
327 String aaSeqString = proteinDataset != null ? proteinDataset
328 .getSequenceAsString() : proteinSeq.getSequenceAsString();
329 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
330 String cdnaSeqString = cdnaDataset != null ? cdnaDataset
331 .getSequenceAsString() : cdnaSeq.getSequenceAsString();
332 if (aaSeqString == null || cdnaSeqString == null)
337 final int mappedLength = 3 * aaSeqString.length();
338 int cdnaLength = cdnaSeqString.length();
340 int cdnaEnd = cdnaLength;
341 final int proteinStart = 1;
342 final int proteinEnd = aaSeqString.length();
345 * If lengths don't match, try ignoring stop codon.
347 if (cdnaLength != mappedLength)
349 for (Object stop : ResidueProperties.STOP)
351 if (cdnaSeqString.toUpperCase().endsWith((String) stop))
361 * If lengths still don't match, try ignoring start codon.
363 if (cdnaLength != mappedLength
364 && cdnaSeqString.toUpperCase().startsWith(
365 ResidueProperties.START))
371 if (cdnaLength == mappedLength)
373 MapList map = new MapList(new int[]
374 { cdnaStart, cdnaEnd }, new int[]
375 { proteinStart, proteinEnd }, 3, 1);
385 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
386 * currently assumes that we are aligning cDNA to match protein.
389 * the sequence to be realigned
391 * the alignment whose sequence alignment is to be 'copied'
393 * character string represent a gap in the realigned sequence
394 * @param preserveUnmappedGaps
395 * @param preserveMappedGaps
396 * @return true if the sequence was realigned, false if it could not be
398 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
399 String gap, boolean preserveMappedGaps,
400 boolean preserveUnmappedGaps)
403 * Get any mappings from the source alignment to the target (dataset) sequence.
405 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
406 // all mappings. Would it help to constrain this?
407 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
408 if (mappings == null || mappings.isEmpty())
414 * Locate the aligned source sequence whose dataset sequence is mapped. We
415 * just take the first match here (as we can't align cDNA like more than one
418 SequenceI alignFrom = null;
419 AlignedCodonFrame mapping = null;
420 for (AlignedCodonFrame mp : mappings)
422 alignFrom = mp.findAlignedSequence(seq.getDatasetSequence(), al);
423 if (alignFrom != null)
430 if (alignFrom == null)
434 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
435 preserveMappedGaps, preserveUnmappedGaps);
440 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
441 * match residues and codons. Flags control whether existing gaps in unmapped
442 * (intron) and mapped (exon) regions are preserved or not. Gaps linking intro
443 * and exon are only retained if both flags are set.
450 * @param preserveUnmappedGaps
451 * @param preserveMappedGaps
453 public static void alignSequenceAs(SequenceI alignTo,
455 AlignedCodonFrame mapping, String myGap, char sourceGap,
456 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
458 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
459 final char[] thisSeq = alignTo.getSequence();
460 final char[] thatAligned = alignFrom.getSequence();
461 StringBuilder thisAligned = new StringBuilder(2 * thisSeq.length);
463 // aligned and dataset sequence positions, all base zero
467 int basesWritten = 0;
468 char myGapChar = myGap.charAt(0);
469 int ratio = myGap.length();
472 * Traverse the aligned protein sequence.
474 int sourceGapMappedLength = 0;
475 boolean inExon = false;
476 for (char sourceChar : thatAligned)
478 if (sourceChar == sourceGap)
480 sourceGapMappedLength += ratio;
485 * Found a residue. Locate its mapped codon (start) position.
488 // Note mapping positions are base 1, our sequence positions base 0
489 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
491 if (mappedPos == null)
494 * Abort realignment if unmapped protein. Or could ignore it??
496 System.err.println("Can't align: no codon mapping to residue "
497 + sourceDsPos + "(" + sourceChar + ")");
501 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
502 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
503 StringBuilder trailingCopiedGap = new StringBuilder();
506 * Copy dna sequence up to and including this codon. Optionally, include
507 * gaps before the codon starts (in introns) and/or after the codon starts
510 * Note this only works for 'linear' splicing, not reverse or interleaved.
511 * But then 'align dna as protein' doesn't make much sense otherwise.
513 int intronLength = 0;
514 while (basesWritten < mappedCodonEnd && thisSeqPos < thisSeq.length)
516 final char c = thisSeq[thisSeqPos++];
521 if (basesWritten < mappedCodonStart)
524 * Found an unmapped (intron) base. First add in any preceding gaps
527 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
529 thisAligned.append(trailingCopiedGap.toString());
530 intronLength += trailingCopiedGap.length();
531 trailingCopiedGap = new StringBuilder();
538 final boolean startOfCodon = basesWritten == mappedCodonStart;
539 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
540 preserveUnmappedGaps, sourceGapMappedLength, inExon,
541 trailingCopiedGap.length(), intronLength, startOfCodon);
542 for (int i = 0; i < gapsToAdd; i++)
544 thisAligned.append(myGapChar);
546 sourceGapMappedLength = 0;
549 thisAligned.append(c);
550 trailingCopiedGap = new StringBuilder();
554 if (inExon && preserveMappedGaps)
556 trailingCopiedGap.append(myGapChar);
558 else if (!inExon && preserveUnmappedGaps)
560 trailingCopiedGap.append(myGapChar);
567 * At end of protein sequence. Copy any remaining dna sequence, optionally
568 * including (intron) gaps. We do not copy trailing gaps in protein.
570 while (thisSeqPos < thisSeq.length)
572 final char c = thisSeq[thisSeqPos++];
573 if (c != myGapChar || preserveUnmappedGaps)
575 thisAligned.append(c);
580 * All done aligning, set the aligned sequence.
582 alignTo.setSequence(new String(thisAligned));
586 * Helper method to work out how many gaps to insert when realigning.
588 * @param preserveMappedGaps
589 * @param preserveUnmappedGaps
590 * @param sourceGapMappedLength
592 * @param trailingCopiedGap
593 * @param intronLength
594 * @param startOfCodon
597 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
598 boolean preserveUnmappedGaps, int sourceGapMappedLength,
599 boolean inExon, int trailingGapLength,
600 int intronLength, final boolean startOfCodon)
606 * Reached start of codon. Ignore trailing gaps in intron unless we are
607 * preserving gaps in both exon and intron. Ignore them anyway if the
608 * protein alignment introduces a gap at least as large as the intronic
611 if (inExon && !preserveMappedGaps)
613 trailingGapLength = 0;
615 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
617 trailingGapLength = 0;
621 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
625 if (intronLength + trailingGapLength <= sourceGapMappedLength)
627 gapsToAdd = sourceGapMappedLength - intronLength;
631 gapsToAdd = Math.min(intronLength + trailingGapLength
632 - sourceGapMappedLength, trailingGapLength);
639 * second or third base of codon; check for any gaps in dna
641 if (!preserveMappedGaps)
643 trailingGapLength = 0;
645 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
651 * Returns a list of sequences mapped from the given sequences and aligned
652 * (gapped) in the same way. For example, the cDNA for aligned protein, where
653 * a single gap in protein generates three gaps in cDNA.
656 * @param gapCharacter
660 public static List<SequenceI> getAlignedTranslation(
661 List<SequenceI> sequences, char gapCharacter,
662 Set<AlignedCodonFrame> mappings)
664 List<SequenceI> alignedSeqs = new ArrayList<SequenceI>();
666 for (SequenceI seq : sequences)
668 List<SequenceI> mapped = getAlignedTranslation(seq, gapCharacter,
670 alignedSeqs.addAll(mapped);
676 * Returns sequences aligned 'like' the source sequence, as mapped by the
677 * given mappings. Normally we expect zero or one 'mapped' sequences, but this
678 * will support 1-to-many as well.
681 * @param gapCharacter
685 protected static List<SequenceI> getAlignedTranslation(SequenceI seq,
686 char gapCharacter, Set<AlignedCodonFrame> mappings)
688 List<SequenceI> result = new ArrayList<SequenceI>();
689 for (AlignedCodonFrame mapping : mappings)
691 if (mapping.involvesSequence(seq))
693 SequenceI mapped = getAlignedTranslation(seq, gapCharacter, mapping);
704 * Returns the translation of 'seq' (as held in the mapping) with
705 * corresponding alignment (gaps).
708 * @param gapCharacter
712 protected static SequenceI getAlignedTranslation(SequenceI seq,
713 char gapCharacter, AlignedCodonFrame mapping)
715 String gap = String.valueOf(gapCharacter);
716 boolean toDna = false;
718 SequenceI mapTo = mapping.getDnaForAaSeq(seq);
721 // mapping is from protein to nucleotide
723 // should ideally get gap count ratio from mapping
724 gap = String.valueOf(new char[]
725 { gapCharacter, gapCharacter, gapCharacter });
729 // mapping is from nucleotide to protein
730 mapTo = mapping.getAaForDnaSeq(seq);
733 StringBuilder newseq = new StringBuilder(seq.getLength()
736 int residueNo = 0; // in seq, base 1
737 int[] phrase = new int[fromRatio];
738 int phraseOffset = 0;
740 boolean first = true;
741 final Sequence alignedSeq = new Sequence("", "");
743 for (char c : seq.getSequence())
745 if (c == gapCharacter)
748 if (gapWidth >= fromRatio)
756 phrase[phraseOffset++] = residueNo + 1;
757 if (phraseOffset == fromRatio)
760 * Have read a whole codon (or protein residue), now translate: map
761 * source phrase to positions in target sequence add characters at
762 * these positions to newseq Note mapping positions are base 1, our
763 * sequence positions base 0.
765 SearchResults sr = new SearchResults();
766 for (int pos : phrase)
768 mapping.markMappedRegion(seq, pos, sr);
770 newseq.append(sr.toString());
774 // Hack: Copy sequence dataset, name and description from
775 // SearchResults.match[0].sequence
776 // TODO? carry over sequence names from original 'complement'
778 SequenceI mappedTo = sr.getResultSequence(0);
779 alignedSeq.setName(mappedTo.getName());
780 alignedSeq.setDescription(mappedTo.getDescription());
781 alignedSeq.setDatasetSequence(mappedTo);
788 alignedSeq.setSequence(newseq.toString());
793 * Realigns the given protein to match the alignment of the dna, using codon
794 * mappings to translate aligned codon positions to protein residues.
797 * the alignment whose sequences are realigned by this method
799 * the dna alignment whose alignment we are 'copying'
800 * @return the number of sequences that were realigned
802 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
804 Set<AlignedCodonFrame> mappings = protein.getCodonFrames();
807 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
808 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
809 * comparator keeps the codon positions ordered.
811 Map<AlignedCodon, Map<SequenceI, String>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, String>>(
812 new CodonComparator());
813 for (SequenceI dnaSeq : dna.getSequences())
815 for (AlignedCodonFrame mapping : mappings)
817 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
818 SequenceI prot = mapping.findAlignedSequence(
819 dnaSeq.getDatasetSequence(), protein);
822 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
823 seqMap, alignedCodons);
827 return alignProteinAs(protein, alignedCodons);
831 * Update the aligned protein sequences to match the codon alignments given in
835 * @param alignedCodons
836 * an ordered map of codon positions (columns), with sequence/peptide
837 * values present in each column
840 protected static int alignProteinAs(AlignmentI protein,
841 Map<AlignedCodon, Map<SequenceI, String>> alignedCodons)
844 * Prefill aligned sequences with gaps before inserting aligned protein
847 int alignedWidth = alignedCodons.size();
848 char[] gaps = new char[alignedWidth];
849 Arrays.fill(gaps, protein.getGapCharacter());
850 String allGaps = String.valueOf(gaps);
851 for (SequenceI seq : protein.getSequences())
853 seq.setSequence(allGaps);
857 for (AlignedCodon codon : alignedCodons.keySet())
859 final Map<SequenceI, String> columnResidues = alignedCodons.get(codon);
860 for (Entry<SequenceI, String> entry : columnResidues
863 // place translated codon at its column position in sequence
864 entry.getKey().getSequence()[column] = entry.getValue().charAt(0);
872 * Populate the map of aligned codons by traversing the given sequence
873 * mapping, locating the aligned positions of mapped codons, and adding those
874 * positions and their translation products to the map.
877 * the aligned sequence we are mapping from
879 * the sequence to be aligned to the codons
881 * the gap character in the dna sequence
883 * a mapping to a sequence translation
884 * @param alignedCodons
885 * the map we are building up
887 static void addCodonPositions(SequenceI dna, SequenceI protein,
890 Map<AlignedCodon, Map<SequenceI, String>> alignedCodons)
892 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
893 while (codons.hasNext())
895 AlignedCodon codon = codons.next();
896 Map<SequenceI, String> seqProduct = alignedCodons.get(codon);
897 if (seqProduct == null)
899 seqProduct = new HashMap<SequenceI, String>();
900 alignedCodons.put(codon, seqProduct);
902 seqProduct.put(protein, codon.product);