2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ 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.Alignment;
26 import jalview.datamodel.AlignmentAnnotation;
27 import jalview.datamodel.AlignmentI;
28 import jalview.datamodel.DBRefEntry;
29 import jalview.datamodel.DBRefSource;
30 import jalview.datamodel.FeatureProperties;
31 import jalview.datamodel.Mapping;
32 import jalview.datamodel.SearchResults;
33 import jalview.datamodel.Sequence;
34 import jalview.datamodel.SequenceFeature;
35 import jalview.datamodel.SequenceGroup;
36 import jalview.datamodel.SequenceI;
37 import jalview.schemes.ResidueProperties;
38 import jalview.util.DBRefUtils;
39 import jalview.util.MapList;
40 import jalview.util.MappingUtils;
42 import java.util.ArrayList;
43 import java.util.Arrays;
44 import java.util.Collection;
45 import java.util.HashMap;
46 import java.util.HashSet;
47 import java.util.Iterator;
48 import java.util.LinkedHashMap;
49 import java.util.List;
51 import java.util.Map.Entry;
53 import java.util.TreeMap;
56 * grab bag of useful alignment manipulation operations Expect these to be
57 * refactored elsewhere at some point.
62 public class AlignmentUtils
66 * given an existing alignment, create a new alignment including all, or up to
67 * flankSize additional symbols from each sequence's dataset sequence
73 public static AlignmentI expandContext(AlignmentI core, int flankSize)
75 List<SequenceI> sq = new ArrayList<SequenceI>();
77 for (SequenceI s : core.getSequences())
79 SequenceI newSeq = s.deriveSequence();
80 final int newSeqStart = newSeq.getStart() - 1;
81 if (newSeqStart > maxoffset
82 && newSeq.getDatasetSequence().getStart() < s.getStart())
84 maxoffset = newSeqStart;
90 maxoffset = Math.min(maxoffset, flankSize);
94 * now add offset left and right to create an expanded alignment
96 for (SequenceI s : sq)
99 while (ds.getDatasetSequence() != null)
101 ds = ds.getDatasetSequence();
103 int s_end = s.findPosition(s.getStart() + s.getLength());
104 // find available flanking residues for sequence
105 int ustream_ds = s.getStart() - ds.getStart();
106 int dstream_ds = ds.getEnd() - s_end;
108 // build new flanked sequence
110 // compute gap padding to start of flanking sequence
111 int offset = maxoffset - ustream_ds;
113 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
116 if (flankSize < ustream_ds)
118 // take up to flankSize residues
119 offset = maxoffset - flankSize;
120 ustream_ds = flankSize;
122 if (flankSize <= dstream_ds)
124 dstream_ds = flankSize - 1;
127 // TODO use Character.toLowerCase to avoid creating String objects?
128 char[] upstream = new String(ds.getSequence(s.getStart() - 1
129 - ustream_ds, s.getStart() - 1)).toLowerCase().toCharArray();
130 char[] downstream = new String(ds.getSequence(s_end - 1, s_end
131 + dstream_ds)).toLowerCase().toCharArray();
132 char[] coreseq = s.getSequence();
133 char[] nseq = new char[offset + upstream.length + downstream.length
135 char c = core.getGapCharacter();
138 for (; p < offset; p++)
143 System.arraycopy(upstream, 0, nseq, p, upstream.length);
144 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
146 System.arraycopy(downstream, 0, nseq, p + coreseq.length
147 + upstream.length, downstream.length);
148 s.setSequence(new String(nseq));
149 s.setStart(s.getStart() - ustream_ds);
150 s.setEnd(s_end + downstream.length);
152 AlignmentI newAl = new jalview.datamodel.Alignment(
153 sq.toArray(new SequenceI[0]));
154 for (SequenceI s : sq)
156 if (s.getAnnotation() != null)
158 for (AlignmentAnnotation aa : s.getAnnotation())
160 aa.adjustForAlignment(); // JAL-1712 fix
161 newAl.addAnnotation(aa);
165 newAl.setDataset(core.getDataset());
170 * Returns the index (zero-based position) of a sequence in an alignment, or
177 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
181 for (SequenceI alSeq : al.getSequences())
194 * Returns a map of lists of sequences in the alignment, keyed by sequence
195 * name. For use in mapping between different alignment views of the same
198 * @see jalview.datamodel.AlignmentI#getSequencesByName()
200 public static Map<String, List<SequenceI>> getSequencesByName(
203 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
204 for (SequenceI seq : al.getSequences())
206 String name = seq.getName();
209 List<SequenceI> seqs = theMap.get(name);
212 seqs = new ArrayList<SequenceI>();
213 theMap.put(name, seqs);
222 * Build mapping of protein to cDNA alignment. Mappings are made between
223 * sequences where the cDNA translates to the protein sequence. Any new
224 * mappings are added to the protein alignment. Returns true if any mappings
225 * either already exist or were added, else false.
227 * @param proteinAlignment
228 * @param cdnaAlignment
231 public static boolean mapProteinAlignmentToCdna(
232 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
234 if (proteinAlignment == null || cdnaAlignment == null)
239 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
240 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
243 * First pass - map sequences where cross-references exist. This include
244 * 1-to-many mappings to support, for example, variant cDNA.
246 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
247 cdnaAlignment, mappedDna, mappedProtein, true);
250 * Second pass - map sequences where no cross-references exist. This only
251 * does 1-to-1 mappings and assumes corresponding sequences are in the same
252 * order in the alignments.
254 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
255 mappedDna, mappedProtein, false);
256 return mappingPerformed;
260 * Make mappings between compatible sequences (where the cDNA translation
261 * matches the protein).
263 * @param proteinAlignment
264 * @param cdnaAlignment
266 * a set of mapped DNA sequences (to add to)
267 * @param mappedProtein
268 * a set of mapped Protein sequences (to add to)
270 * if true, only map sequences where xrefs exist
273 protected static boolean mapProteinToCdna(
274 final AlignmentI proteinAlignment,
275 final AlignmentI cdnaAlignment, Set<SequenceI> mappedDna,
276 Set<SequenceI> mappedProtein, boolean xrefsOnly)
278 boolean mappingExistsOrAdded = false;
279 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
280 for (SequenceI aaSeq : thisSeqs)
282 boolean proteinMapped = false;
283 AlignedCodonFrame acf = new AlignedCodonFrame();
285 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
288 * Always try to map if sequences have xref to each other; this supports
289 * variant cDNA or alternative splicing for a protein sequence.
291 * If no xrefs, try to map progressively, assuming that alignments have
292 * mappable sequences in corresponding order. These are not
293 * many-to-many, as that would risk mixing species with similar cDNA
296 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
302 * Don't map non-xrefd sequences more than once each. This heuristic
303 * allows us to pair up similar sequences in ordered alignments.
306 && (mappedProtein.contains(aaSeq) || mappedDna
311 if (mappingExists(proteinAlignment.getCodonFrames(),
312 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
314 mappingExistsOrAdded = true;
318 MapList map = mapProteinSequenceToCdna(aaSeq, cdnaSeq);
321 acf.addMap(cdnaSeq, aaSeq, map);
322 mappingExistsOrAdded = true;
323 proteinMapped = true;
324 mappedDna.add(cdnaSeq);
325 mappedProtein.add(aaSeq);
331 proteinAlignment.addCodonFrame(acf);
334 return mappingExistsOrAdded;
338 * Answers true if the mappings include one between the given (dataset)
341 public static boolean mappingExists(List<AlignedCodonFrame> mappings,
342 SequenceI aaSeq, SequenceI cdnaSeq)
344 if (mappings != null)
346 for (AlignedCodonFrame acf : mappings)
348 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
358 * Build a mapping (if possible) of a protein to a cDNA sequence. The cDNA
359 * must be three times the length of the protein, possibly after ignoring
360 * start and/or stop codons, and must translate to the protein. Returns null
361 * if no mapping is determined.
367 public static MapList mapProteinSequenceToCdna(SequenceI proteinSeq,
371 * Here we handle either dataset sequence set (desktop) or absent (applet).
372 * Use only the char[] form of the sequence to avoid creating possibly large
375 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
376 char[] aaSeqChars = proteinDataset != null ? proteinDataset
377 .getSequence() : proteinSeq.getSequence();
378 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
379 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
380 : cdnaSeq.getSequence();
381 if (aaSeqChars == null || cdnaSeqChars == null)
387 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
389 final int mappedLength = 3 * aaSeqChars.length;
390 int cdnaLength = cdnaSeqChars.length;
391 int cdnaStart = cdnaSeq.getStart();
392 int cdnaEnd = cdnaSeq.getEnd();
393 final int proteinStart = proteinSeq.getStart();
394 final int proteinEnd = proteinSeq.getEnd();
397 * If lengths don't match, try ignoring stop codon.
399 if (cdnaLength != mappedLength && cdnaLength > 2)
401 String lastCodon = String.valueOf(cdnaSeqChars, cdnaLength - 3, 3)
403 for (String stop : ResidueProperties.STOP)
405 if (lastCodon.equals(stop))
415 * If lengths still don't match, try ignoring start codon.
418 if (cdnaLength != mappedLength
420 && String.valueOf(cdnaSeqChars, 0, 3).toUpperCase()
421 .equals(ResidueProperties.START))
428 if (cdnaLength != mappedLength)
432 if (!translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
436 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd }, new int[] {
437 proteinStart, proteinEnd }, 3, 1);
442 * Test whether the given cdna sequence, starting at the given offset,
443 * translates to the given amino acid sequence, using the standard translation
444 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
446 * @param cdnaSeqChars
451 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
454 if (cdnaSeqChars == null || aaSeqChars == null)
460 for (int i = cdnaStart; i < cdnaSeqChars.length - 2
461 && aaResidue < aaSeqChars.length; i += 3, aaResidue++)
463 String codon = String.valueOf(cdnaSeqChars, i, 3);
464 final String translated = ResidueProperties.codonTranslate(codon);
466 * allow * in protein to match untranslatable in dna
468 final char aaRes = aaSeqChars[aaResidue];
469 if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
473 if (translated == null || !(aaRes == translated.charAt(0)))
476 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
477 // + codon + "(" + translated + ") != " + aaRes));
481 // fail if we didn't match all of the aa sequence
482 return (aaResidue == aaSeqChars.length);
486 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
487 * currently assumes that we are aligning cDNA to match protein.
490 * the sequence to be realigned
492 * the alignment whose sequence alignment is to be 'copied'
494 * character string represent a gap in the realigned sequence
495 * @param preserveUnmappedGaps
496 * @param preserveMappedGaps
497 * @return true if the sequence was realigned, false if it could not be
499 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
500 String gap, boolean preserveMappedGaps,
501 boolean preserveUnmappedGaps)
504 * Get any mappings from the source alignment to the target (dataset)
507 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
508 // all mappings. Would it help to constrain this?
509 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
510 if (mappings == null || mappings.isEmpty())
516 * Locate the aligned source sequence whose dataset sequence is mapped. We
517 * just take the first match here (as we can't align like more than one
520 SequenceI alignFrom = null;
521 AlignedCodonFrame mapping = null;
522 for (AlignedCodonFrame mp : mappings)
524 alignFrom = mp.findAlignedSequence(seq.getDatasetSequence(), al);
525 if (alignFrom != null)
532 if (alignFrom == null)
536 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
537 preserveMappedGaps, preserveUnmappedGaps);
542 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
543 * match residues and codons. Flags control whether existing gaps in unmapped
544 * (intron) and mapped (exon) regions are preserved or not. Gaps between
545 * intron and exon are only retained if both flags are set.
552 * @param preserveUnmappedGaps
553 * @param preserveMappedGaps
555 public static void alignSequenceAs(SequenceI alignTo,
556 SequenceI alignFrom, AlignedCodonFrame mapping, String myGap,
557 char sourceGap, boolean preserveMappedGaps,
558 boolean preserveUnmappedGaps)
560 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
562 // aligned and dataset sequence positions, all base zero
566 int basesWritten = 0;
567 char myGapChar = myGap.charAt(0);
568 int ratio = myGap.length();
570 int fromOffset = alignFrom.getStart() - 1;
571 int toOffset = alignTo.getStart() - 1;
572 int sourceGapMappedLength = 0;
573 boolean inExon = false;
574 final char[] thisSeq = alignTo.getSequence();
575 final char[] thatAligned = alignFrom.getSequence();
576 StringBuilder thisAligned = new StringBuilder(2 * thisSeq.length);
579 * Traverse the 'model' aligned sequence
581 for (char sourceChar : thatAligned)
583 if (sourceChar == sourceGap)
585 sourceGapMappedLength += ratio;
590 * Found a non-gap character. Locate its mapped region if any.
593 // Note mapping positions are base 1, our sequence positions base 0
594 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
595 sourceDsPos + fromOffset);
596 if (mappedPos == null)
599 * unmapped position; treat like a gap
601 sourceGapMappedLength += ratio;
602 // System.err.println("Can't align: no codon mapping to residue "
603 // + sourceDsPos + "(" + sourceChar + ")");
608 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
609 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
610 StringBuilder trailingCopiedGap = new StringBuilder();
613 * Copy dna sequence up to and including this codon. Optionally, include
614 * gaps before the codon starts (in introns) and/or after the codon starts
617 * Note this only works for 'linear' splicing, not reverse or interleaved.
618 * But then 'align dna as protein' doesn't make much sense otherwise.
620 int intronLength = 0;
621 while (basesWritten + toOffset < mappedCodonEnd
622 && thisSeqPos < thisSeq.length)
624 final char c = thisSeq[thisSeqPos++];
628 int sourcePosition = basesWritten + toOffset;
629 if (sourcePosition < mappedCodonStart)
632 * Found an unmapped (intron) base. First add in any preceding gaps
635 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
637 thisAligned.append(trailingCopiedGap.toString());
638 intronLength += trailingCopiedGap.length();
639 trailingCopiedGap = new StringBuilder();
646 final boolean startOfCodon = sourcePosition == mappedCodonStart;
647 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
648 preserveUnmappedGaps, sourceGapMappedLength, inExon,
649 trailingCopiedGap.length(), intronLength, startOfCodon);
650 for (int i = 0; i < gapsToAdd; i++)
652 thisAligned.append(myGapChar);
654 sourceGapMappedLength = 0;
657 thisAligned.append(c);
658 trailingCopiedGap = new StringBuilder();
662 if (inExon && preserveMappedGaps)
664 trailingCopiedGap.append(myGapChar);
666 else if (!inExon && preserveUnmappedGaps)
668 trailingCopiedGap.append(myGapChar);
675 * At end of model aligned sequence. Copy any remaining target sequence, optionally
676 * including (intron) gaps.
678 while (thisSeqPos < thisSeq.length)
680 final char c = thisSeq[thisSeqPos++];
681 if (c != myGapChar || preserveUnmappedGaps)
683 thisAligned.append(c);
685 sourceGapMappedLength--;
689 * finally add gaps to pad for any trailing source gaps or
690 * unmapped characters
692 if (preserveUnmappedGaps)
694 while (sourceGapMappedLength > 0)
696 thisAligned.append(myGapChar);
697 sourceGapMappedLength--;
702 * All done aligning, set the aligned sequence.
704 alignTo.setSequence(new String(thisAligned));
708 * Helper method to work out how many gaps to insert when realigning.
710 * @param preserveMappedGaps
711 * @param preserveUnmappedGaps
712 * @param sourceGapMappedLength
714 * @param trailingCopiedGap
715 * @param intronLength
716 * @param startOfCodon
719 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
720 boolean preserveUnmappedGaps, int sourceGapMappedLength,
721 boolean inExon, int trailingGapLength, int intronLength,
722 final boolean startOfCodon)
728 * Reached start of codon. Ignore trailing gaps in intron unless we are
729 * preserving gaps in both exon and intron. Ignore them anyway if the
730 * protein alignment introduces a gap at least as large as the intronic
733 if (inExon && !preserveMappedGaps)
735 trailingGapLength = 0;
737 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
739 trailingGapLength = 0;
743 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
747 if (intronLength + trailingGapLength <= sourceGapMappedLength)
749 gapsToAdd = sourceGapMappedLength - intronLength;
753 gapsToAdd = Math.min(intronLength + trailingGapLength
754 - sourceGapMappedLength, trailingGapLength);
761 * second or third base of codon; check for any gaps in dna
763 if (!preserveMappedGaps)
765 trailingGapLength = 0;
767 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
773 * Returns a list of sequences mapped from the given sequences and aligned
774 * (gapped) in the same way. For example, the cDNA for aligned protein, where
775 * a single gap in protein generates three gaps in cDNA.
778 * @param gapCharacter
782 public static List<SequenceI> getAlignedTranslation(
783 List<SequenceI> sequences, char gapCharacter,
784 Set<AlignedCodonFrame> mappings)
786 List<SequenceI> alignedSeqs = new ArrayList<SequenceI>();
788 for (SequenceI seq : sequences)
790 List<SequenceI> mapped = getAlignedTranslation(seq, gapCharacter,
792 alignedSeqs.addAll(mapped);
798 * Returns sequences aligned 'like' the source sequence, as mapped by the
799 * given mappings. Normally we expect zero or one 'mapped' sequences, but this
800 * will support 1-to-many as well.
803 * @param gapCharacter
807 protected static List<SequenceI> getAlignedTranslation(SequenceI seq,
808 char gapCharacter, Set<AlignedCodonFrame> mappings)
810 List<SequenceI> result = new ArrayList<SequenceI>();
811 for (AlignedCodonFrame mapping : mappings)
813 if (mapping.involvesSequence(seq))
815 SequenceI mapped = getAlignedTranslation(seq, gapCharacter, mapping);
826 * Returns the translation of 'seq' (as held in the mapping) with
827 * corresponding alignment (gaps).
830 * @param gapCharacter
834 protected static SequenceI getAlignedTranslation(SequenceI seq,
835 char gapCharacter, AlignedCodonFrame mapping)
837 String gap = String.valueOf(gapCharacter);
838 boolean toDna = false;
840 SequenceI mapTo = mapping.getDnaForAaSeq(seq);
843 // mapping is from protein to nucleotide
845 // should ideally get gap count ratio from mapping
846 gap = String.valueOf(new char[] { gapCharacter, gapCharacter,
851 // mapping is from nucleotide to protein
852 mapTo = mapping.getAaForDnaSeq(seq);
855 StringBuilder newseq = new StringBuilder(seq.getLength()
858 int residueNo = 0; // in seq, base 1
859 int[] phrase = new int[fromRatio];
860 int phraseOffset = 0;
862 boolean first = true;
863 final Sequence alignedSeq = new Sequence("", "");
865 for (char c : seq.getSequence())
867 if (c == gapCharacter)
870 if (gapWidth >= fromRatio)
878 phrase[phraseOffset++] = residueNo + 1;
879 if (phraseOffset == fromRatio)
882 * Have read a whole codon (or protein residue), now translate: map
883 * source phrase to positions in target sequence add characters at
884 * these positions to newseq Note mapping positions are base 1, our
885 * sequence positions base 0.
887 SearchResults sr = new SearchResults();
888 for (int pos : phrase)
890 mapping.markMappedRegion(seq, pos, sr);
892 newseq.append(sr.getCharacters());
896 // Hack: Copy sequence dataset, name and description from
897 // SearchResults.match[0].sequence
898 // TODO? carry over sequence names from original 'complement'
900 SequenceI mappedTo = sr.getResultSequence(0);
901 alignedSeq.setName(mappedTo.getName());
902 alignedSeq.setDescription(mappedTo.getDescription());
903 alignedSeq.setDatasetSequence(mappedTo);
910 alignedSeq.setSequence(newseq.toString());
915 * Realigns the given protein to match the alignment of the dna, using codon
916 * mappings to translate aligned codon positions to protein residues.
919 * the alignment whose sequences are realigned by this method
921 * the dna alignment whose alignment we are 'copying'
922 * @return the number of sequences that were realigned
924 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
926 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
927 unmappedProtein.addAll(protein.getSequences());
929 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
932 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
933 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
934 * comparator keeps the codon positions ordered.
936 Map<AlignedCodon, Map<SequenceI, String>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, String>>(
937 new CodonComparator());
938 for (SequenceI dnaSeq : dna.getSequences())
940 for (AlignedCodonFrame mapping : mappings)
942 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
943 SequenceI prot = mapping.findAlignedSequence(
944 dnaSeq.getDatasetSequence(), protein);
947 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
948 seqMap, alignedCodons);
949 unmappedProtein.remove(prot);
953 return alignProteinAs(protein, alignedCodons, unmappedProtein);
957 * Update the aligned protein sequences to match the codon alignments given in
961 * @param alignedCodons
962 * an ordered map of codon positions (columns), with sequence/peptide
963 * values present in each column
964 * @param unmappedProtein
967 protected static int alignProteinAs(AlignmentI protein,
968 Map<AlignedCodon, Map<SequenceI, String>> alignedCodons,
969 List<SequenceI> unmappedProtein)
972 * Prefill aligned sequences with gaps before inserting aligned protein
975 int alignedWidth = alignedCodons.size();
976 char[] gaps = new char[alignedWidth];
977 Arrays.fill(gaps, protein.getGapCharacter());
978 String allGaps = String.valueOf(gaps);
979 for (SequenceI seq : protein.getSequences())
981 if (!unmappedProtein.contains(seq))
983 seq.setSequence(allGaps);
988 for (AlignedCodon codon : alignedCodons.keySet())
990 final Map<SequenceI, String> columnResidues = alignedCodons
992 for (Entry<SequenceI, String> entry : columnResidues.entrySet())
994 // place translated codon at its column position in sequence
995 entry.getKey().getSequence()[column] = entry.getValue().charAt(0);
1003 * Populate the map of aligned codons by traversing the given sequence
1004 * mapping, locating the aligned positions of mapped codons, and adding those
1005 * positions and their translation products to the map.
1008 * the aligned sequence we are mapping from
1010 * the sequence to be aligned to the codons
1012 * the gap character in the dna sequence
1014 * a mapping to a sequence translation
1015 * @param alignedCodons
1016 * the map we are building up
1018 static void addCodonPositions(SequenceI dna, SequenceI protein,
1019 char gapChar, Mapping seqMap,
1020 Map<AlignedCodon, Map<SequenceI, String>> alignedCodons)
1022 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1023 while (codons.hasNext())
1025 AlignedCodon codon = codons.next();
1026 Map<SequenceI, String> seqProduct = alignedCodons.get(codon);
1027 if (seqProduct == null)
1029 seqProduct = new HashMap<SequenceI, String>();
1030 alignedCodons.put(codon, seqProduct);
1032 seqProduct.put(protein, codon.product);
1037 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1038 * between at least one pair of sequences in the two alignments. Currently,
1041 * <li>One alignment must be nucleotide, and the other protein</li>
1042 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1043 * <li>Mappable means the nucleotide translation matches the protein sequence</li>
1044 * <li>The translation may ignore start and stop codons if present in the
1052 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1054 if (al1 == null || al2 == null)
1060 * Require one nucleotide and one protein
1062 if (al1.isNucleotide() == al2.isNucleotide())
1066 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1067 AlignmentI protein = dna == al1 ? al2 : al1;
1068 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1069 for (SequenceI dnaSeq : dna.getSequences())
1071 for (SequenceI proteinSeq : protein.getSequences())
1073 if (isMappable(dnaSeq, proteinSeq, mappings))
1083 * Returns true if the dna sequence is mapped, or could be mapped, to the
1091 protected static boolean isMappable(SequenceI dnaSeq,
1092 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1094 if (dnaSeq == null || proteinSeq == null)
1099 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq : dnaSeq
1100 .getDatasetSequence();
1101 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null ? proteinSeq
1102 : proteinSeq.getDatasetSequence();
1104 for (AlignedCodonFrame mapping : mappings)
1106 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1116 * Just try to make a mapping (it is not yet stored), test whether
1119 return mapProteinSequenceToCdna(proteinDs, dnaDs) != null;
1123 * Finds any reference annotations associated with the sequences in
1124 * sequenceScope, that are not already added to the alignment, and adds them
1125 * to the 'candidates' map. Also populates a lookup table of annotation
1126 * labels, keyed by calcId, for use in constructing tooltips or the like.
1128 * @param sequenceScope
1129 * the sequences to scan for reference annotations
1130 * @param labelForCalcId
1131 * (optional) map to populate with label for calcId
1133 * map to populate with annotations for sequence
1135 * the alignment to check for presence of annotations
1137 public static void findAddableReferenceAnnotations(
1138 List<SequenceI> sequenceScope,
1139 Map<String, String> labelForCalcId,
1140 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1143 if (sequenceScope == null)
1149 * For each sequence in scope, make a list of any annotations on the
1150 * underlying dataset sequence which are not already on the alignment.
1152 * Add to a map of { alignmentSequence, <List of annotations to add> }
1154 for (SequenceI seq : sequenceScope)
1156 SequenceI dataset = seq.getDatasetSequence();
1157 if (dataset == null)
1161 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1162 if (datasetAnnotations == null)
1166 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1167 for (AlignmentAnnotation dsann : datasetAnnotations)
1170 * Find matching annotations on the alignment. If none is found, then
1171 * add this annotation to the list of 'addable' annotations for this
1174 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1175 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1176 if (!matchedAlignmentAnnotations.iterator().hasNext())
1179 if (labelForCalcId != null)
1181 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1186 * Save any addable annotations for this sequence
1188 if (!result.isEmpty())
1190 candidates.put(seq, result);
1196 * Adds annotations to the top of the alignment annotations, in the same order
1197 * as their related sequences.
1199 * @param annotations
1200 * the annotations to add
1202 * the alignment to add them to
1203 * @param selectionGroup
1204 * current selection group (or null if none)
1206 public static void addReferenceAnnotations(
1207 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1208 final AlignmentI alignment, final SequenceGroup selectionGroup)
1210 for (SequenceI seq : annotations.keySet())
1212 for (AlignmentAnnotation ann : annotations.get(seq))
1214 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1216 int endRes = ann.annotations.length;
1217 if (selectionGroup != null)
1219 startRes = selectionGroup.getStartRes();
1220 endRes = selectionGroup.getEndRes();
1222 copyAnn.restrict(startRes, endRes);
1225 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1226 * original annotation is already on the sequence.
1228 if (!seq.hasAnnotation(ann))
1230 seq.addAlignmentAnnotation(copyAnn);
1233 copyAnn.adjustForAlignment();
1234 // add to the alignment and set visible
1235 alignment.addAnnotation(copyAnn);
1236 copyAnn.visible = true;
1242 * Set visibility of alignment annotations of specified types (labels), for
1243 * specified sequences. This supports controls like
1244 * "Show all secondary structure", "Hide all Temp factor", etc.
1246 * @al the alignment to scan for annotations
1248 * the types (labels) of annotations to be updated
1249 * @param forSequences
1250 * if not null, only annotations linked to one of these sequences are
1251 * in scope for update; if null, acts on all sequence annotations
1253 * if this flag is true, 'types' is ignored (label not checked)
1255 * if true, set visibility on, else set off
1257 public static void showOrHideSequenceAnnotations(AlignmentI al,
1258 Collection<String> types, List<SequenceI> forSequences,
1259 boolean anyType, boolean doShow)
1261 for (AlignmentAnnotation aa : al.getAlignmentAnnotation())
1263 if (anyType || types.contains(aa.label))
1265 if ((aa.sequenceRef != null)
1266 && (forSequences == null || forSequences
1267 .contains(aa.sequenceRef)))
1269 aa.visible = doShow;
1276 * Returns true if either sequence has a cross-reference to the other
1282 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1284 // Note: moved here from class CrossRef as the latter class has dependencies
1285 // not availability to the applet's classpath
1286 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1290 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1291 * that sequence name is structured as Source|AccessionId.
1297 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1299 if (seq1 == null || seq2 == null)
1303 String name = seq2.getName();
1304 final DBRefEntry[] xrefs = seq1.getDBRef();
1307 for (DBRefEntry xref : xrefs)
1309 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1310 // case-insensitive test, consistent with DBRefEntry.equalRef()
1311 if (xrefName.equalsIgnoreCase(name))
1321 * Constructs an alignment consisting of the mapped cds regions in the given
1322 * nucleotide sequences, and updates mappings to match.
1325 * aligned dna sequences
1327 * from dna to protein; these are replaced with new mappings
1328 * @return an alignment whose sequences are the cds-only parts of the dna
1329 * sequences (or null if no cds are found)
1331 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1332 List<AlignedCodonFrame> mappings)
1334 List<AlignedCodonFrame> newMappings = new ArrayList<AlignedCodonFrame>();
1335 List<SequenceI> cdsSequences = new ArrayList<SequenceI>();
1337 for (SequenceI dnaSeq : dna)
1339 final SequenceI ds = dnaSeq.getDatasetSequence();
1340 List<AlignedCodonFrame> seqMappings = MappingUtils
1341 .findMappingsForSequence(ds, mappings);
1342 for (AlignedCodonFrame acf : seqMappings)
1344 AlignedCodonFrame newMapping = new AlignedCodonFrame();
1345 final List<SequenceI> mappedCds = makeCdsSequences(ds, acf,
1347 if (!mappedCds.isEmpty())
1349 cdsSequences.addAll(mappedCds);
1350 newMappings.add(newMapping);
1354 AlignmentI al = new Alignment(
1355 cdsSequences.toArray(new SequenceI[cdsSequences.size()]));
1356 al.setDataset(null);
1359 * Replace the old mappings with the new ones
1362 mappings.addAll(newMappings);
1368 * Helper method to make cds-only sequences and populate their mappings to
1371 * For example, if ggCCaTTcGAg has mappings [3, 4, 6, 7, 9, 10] to protein
1372 * then generate a sequence CCTTGA with mapping [1, 6] to the same protein
1375 * Typically eukaryotic dna will include cds encoding for a single peptide
1376 * sequence i.e. return a single result. Bacterial dna may have overlapping
1377 * cds mappings coding for multiple peptides so return multiple results
1378 * (example EMBL KF591215).
1381 * a dna dataset sequence
1383 * containing one or more mappings of the sequence to protein
1384 * @param newMappings
1385 * the new mapping to populate, from the cds-only sequences to their
1386 * mapped protein sequences
1389 protected static List<SequenceI> makeCdsSequences(SequenceI dnaSeq,
1390 AlignedCodonFrame mapping, AlignedCodonFrame newMappings)
1392 List<SequenceI> cdsSequences = new ArrayList<SequenceI>();
1393 List<Mapping> seqMappings = mapping.getMappingsForSequence(dnaSeq);
1395 for (Mapping seqMapping : seqMappings)
1397 SequenceI cds = makeCdsSequence(dnaSeq, seqMapping, newMappings);
1398 cdsSequences.add(cds);
1401 * add new mappings, from dna to cds, and from cds to peptide
1403 MapList dnaToCds = addCdsMappings(dnaSeq, cds, seqMapping,
1407 * transfer any features on dna that overlap the CDS
1409 transferFeatures(dnaSeq, cds, dnaToCds, "CDS" /* SequenceOntology.CDS */);
1411 return cdsSequences;
1415 * Transfers any co-located features on 'fromSeq' to 'toSeq', adjusting the
1416 * feature start/end ranges, optionally omitting specified feature types.
1421 * the mapping from 'fromSeq' to 'toSeq'
1424 protected static void transferFeatures(SequenceI fromSeq,
1425 SequenceI toSeq, MapList mapping, String... omitting)
1427 SequenceFeature[] sfs = fromSeq.getSequenceFeatures();
1430 for (SequenceFeature sf : sfs)
1432 String type = sf.getType();
1433 boolean omit = false;
1434 for (String toOmit : omitting)
1436 if (type.equals(toOmit))
1447 * locate the mapped range - null if either start or end is
1448 * not mapped (no partial overlaps are calculated)
1450 int[] mappedTo = mapping.locateInTo(sf.getBegin(), sf.getEnd());
1451 if (mappedTo != null)
1453 SequenceFeature copy = new SequenceFeature(sf);
1454 copy.setBegin(Math.min(mappedTo[0], mappedTo[1]));
1455 copy.setEnd(Math.max(mappedTo[0], mappedTo[1]));
1456 toSeq.addSequenceFeature(copy);
1463 * Creates and adds mappings
1465 * <li>from cds to peptide</li>
1466 * <li>from dna to cds</li>
1468 * and returns the dna-to-cds mapping
1473 * @param newMappings
1476 protected static MapList addCdsMappings(SequenceI dnaSeq,
1478 Mapping dnaMapping, AlignedCodonFrame newMappings)
1480 cdsSeq.createDatasetSequence();
1483 * CDS to peptide is just a contiguous 3:1 mapping, with
1484 * the peptide ranges taken unchanged from the dna mapping
1486 List<int[]> cdsRanges = new ArrayList<int[]>();
1487 cdsRanges.add(new int[] { 1, cdsSeq.getLength() });
1488 MapList cdsToPeptide = new MapList(cdsRanges, dnaMapping.getMap()
1489 .getToRanges(), 3, 1);
1490 newMappings.addMap(cdsSeq.getDatasetSequence(), dnaMapping.getTo(),
1494 * dna 'from' ranges map 1:1 to the contiguous extracted CDS
1496 MapList dnaToCds = new MapList(
1497 dnaMapping.getMap().getFromRanges(), cdsRanges, 1, 1);
1498 newMappings.addMap(dnaSeq, cdsSeq.getDatasetSequence(), dnaToCds);
1503 * Makes and returns a CDS-only sequence, where the CDS regions are identified
1504 * as the 'from' ranges of the mapping on the dna. Any sequence features on
1505 * the dna which overlap the CDS regions are copied to the new sequence.
1508 * nucleotide sequence
1510 * mappings from CDS regions of nucleotide
1511 * @param exonMappings
1512 * CDS-to-peptide mapping (to add to)
1515 protected static SequenceI makeCdsSequence(SequenceI dnaSeq,
1516 Mapping seqMapping, AlignedCodonFrame exonMappings)
1518 StringBuilder newSequence = new StringBuilder(dnaSeq.getLength());
1519 final char[] dna = dnaSeq.getSequence();
1520 int offset = dnaSeq.getStart() - 1;
1523 * Get the codon regions as { [2, 5], [7, 12], [14, 14] etc }
1525 final List<int[]> dnaCdsRanges = seqMapping.getMap().getFromRanges();
1526 for (int[] range : dnaCdsRanges)
1528 // TODO handle reverse mapping as well (range[1] < range[0])
1529 for (int pos = range[0]; pos <= range[1]; pos++)
1531 newSequence.append(dna[pos - offset - 1]);
1535 SequenceI cds = new Sequence(dnaSeq.getName(),
1536 newSequence.toString());
1538 transferDbRefs(seqMapping.getTo(), cds);
1543 * Locate any xrefs to CDS databases on the protein product and attach to the
1544 * CDS sequence. Also add as a sub-token of the sequence name.
1549 protected static void transferDbRefs(SequenceI from, SequenceI to)
1551 String cdsAccId = FeatureProperties.getCodingFeature(DBRefSource.EMBL);
1552 DBRefEntry[] cdsRefs = DBRefUtils.selectRefs(from.getDBRef(),
1553 DBRefSource.CODINGDBS);
1554 if (cdsRefs != null)
1556 for (DBRefEntry cdsRef : cdsRefs)
1558 to.addDBRef(new DBRefEntry(cdsRef));
1559 cdsAccId = cdsRef.getAccessionId();
1562 if (!to.getName().contains(cdsAccId))
1564 to.setName(to.getName() + "|" + cdsAccId);