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 java.util.ArrayList;
24 import java.util.Arrays;
25 import java.util.Collection;
26 import java.util.HashMap;
27 import java.util.HashSet;
28 import java.util.Iterator;
29 import java.util.LinkedHashMap;
30 import java.util.LinkedHashSet;
31 import java.util.List;
33 import java.util.Map.Entry;
35 import java.util.TreeMap;
37 import jalview.datamodel.AlignedCodon;
38 import jalview.datamodel.AlignedCodonFrame;
39 import jalview.datamodel.Alignment;
40 import jalview.datamodel.AlignmentAnnotation;
41 import jalview.datamodel.AlignmentI;
42 import jalview.datamodel.DBRefEntry;
43 import jalview.datamodel.DBRefSource;
44 import jalview.datamodel.FeatureProperties;
45 import jalview.datamodel.Mapping;
46 import jalview.datamodel.SearchResults;
47 import jalview.datamodel.Sequence;
48 import jalview.datamodel.SequenceGroup;
49 import jalview.datamodel.SequenceI;
50 import jalview.schemes.ResidueProperties;
51 import jalview.util.DBRefUtils;
52 import jalview.util.MapList;
53 import jalview.util.MappingUtils;
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 mapProteinToCdna(
232 final AlignmentI proteinAlignment,
233 final AlignmentI cdnaAlignment)
235 if (proteinAlignment == null || cdnaAlignment == null)
240 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
241 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
244 * First pass - map sequences where cross-references exist. This include
245 * 1-to-many mappings to support, for example, variant cDNA.
247 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
248 cdnaAlignment, mappedDna, mappedProtein, true);
251 * Second pass - map sequences where no cross-references exist. This only
252 * does 1-to-1 mappings and assumes corresponding sequences are in the same
253 * order in the alignments.
255 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
256 mappedDna, mappedProtein, false);
257 return mappingPerformed;
261 * Make mappings between compatible sequences (where the cDNA translation
262 * matches the protein).
264 * @param proteinAlignment
265 * @param cdnaAlignment
267 * a set of mapped DNA sequences (to add to)
268 * @param mappedProtein
269 * a set of mapped Protein sequences (to add to)
271 * if true, only map sequences where xrefs exist
274 protected static boolean mapProteinToCdna(
275 final AlignmentI proteinAlignment,
276 final AlignmentI cdnaAlignment, Set<SequenceI> mappedDna,
277 Set<SequenceI> mappedProtein, boolean xrefsOnly)
279 boolean mappingPerformed = false;
280 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
281 for (SequenceI aaSeq : thisSeqs)
283 boolean proteinMapped = false;
284 AlignedCodonFrame acf = new AlignedCodonFrame();
286 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
289 * Always try to map if sequences have xref to each other; this supports
290 * variant cDNA or alternative splicing for a protein sequence.
292 * If no xrefs, try to map progressively, assuming that alignments have
293 * mappable sequences in corresponding order. These are not
294 * many-to-many, as that would risk mixing species with similar cDNA
297 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
303 * Don't map non-xrefd sequences more than once each. This heuristic
304 * allows us to pair up similar sequences in ordered alignments.
307 && (mappedProtein.contains(aaSeq) || mappedDna
312 if (!mappingExists(proteinAlignment.getCodonFrames(),
313 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
315 MapList map = mapProteinToCdna(aaSeq, cdnaSeq);
318 acf.addMap(cdnaSeq, aaSeq, map);
319 mappingPerformed = true;
320 proteinMapped = true;
321 mappedDna.add(cdnaSeq);
322 mappedProtein.add(aaSeq);
328 proteinAlignment.addCodonFrame(acf);
331 return mappingPerformed;
335 * Answers true if the mappings include one between the given (dataset)
338 public static boolean mappingExists(Set<AlignedCodonFrame> set,
339 SequenceI aaSeq, SequenceI cdnaSeq)
343 for (AlignedCodonFrame acf : set)
345 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
355 * Build a mapping (if possible) of a protein to a cDNA sequence. The cDNA
356 * must be three times the length of the protein, possibly after ignoring
357 * start and/or stop codons, and must translate to the protein. Returns null
358 * if no mapping is determined.
364 public static MapList mapProteinToCdna(SequenceI proteinSeq,
368 * Here we handle either dataset sequence set (desktop) or absent (applet).
369 * Use only the char[] form of the sequence to avoid creating possibly large
372 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
373 char[] aaSeqChars = proteinDataset != null ? proteinDataset
374 .getSequence() : proteinSeq.getSequence();
375 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
376 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
377 : cdnaSeq.getSequence();
378 if (aaSeqChars == null || cdnaSeqChars == null)
384 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
386 final int mappedLength = 3 * aaSeqChars.length;
387 int cdnaLength = cdnaSeqChars.length;
389 int cdnaEnd = cdnaLength;
390 final int proteinStart = 1;
391 final int proteinEnd = aaSeqChars.length;
394 * If lengths don't match, try ignoring stop codon.
396 if (cdnaLength != mappedLength && cdnaLength > 2)
398 String lastCodon = String.valueOf(cdnaSeqChars, cdnaLength - 3, 3)
400 for (String stop : ResidueProperties.STOP)
402 if (lastCodon.equals(stop))
412 * If lengths still don't match, try ignoring start codon.
414 if (cdnaLength != mappedLength
416 && String.valueOf(cdnaSeqChars, 0, 3).toUpperCase()
418 ResidueProperties.START))
424 if (cdnaLength != mappedLength)
428 if (!translatesAs(cdnaSeqChars, cdnaStart - 1, aaSeqChars))
432 MapList map = new MapList(new int[]
433 { cdnaStart, cdnaEnd }, new int[]
434 { proteinStart, proteinEnd }, 3, 1);
439 * Test whether the given cdna sequence, starting at the given offset,
440 * translates to the given amino acid sequence, using the standard translation
441 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
443 * @param cdnaSeqChars
448 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
452 for (int i = cdnaStart; i < cdnaSeqChars.length - 2
453 && aaResidue < aaSeqChars.length; i += 3, aaResidue++)
455 String codon = String.valueOf(cdnaSeqChars, i, 3);
456 final String translated = ResidueProperties.codonTranslate(
459 * ? allow X in protein to match untranslatable in dna ?
461 final char aaRes = aaSeqChars[aaResidue];
462 if ((translated == null || "STOP".equals(translated)) && aaRes == 'X')
466 if (translated == null
467 || !(aaRes == translated.charAt(0)))
470 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
471 // + codon + "(" + translated + ") != " + aaRes));
475 // fail if we didn't match all of the aa sequence
476 return (aaResidue == aaSeqChars.length);
480 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
481 * currently assumes that we are aligning cDNA to match protein.
484 * the sequence to be realigned
486 * the alignment whose sequence alignment is to be 'copied'
488 * character string represent a gap in the realigned sequence
489 * @param preserveUnmappedGaps
490 * @param preserveMappedGaps
491 * @return true if the sequence was realigned, false if it could not be
493 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
494 String gap, boolean preserveMappedGaps,
495 boolean preserveUnmappedGaps)
498 * Get any mappings from the source alignment to the target (dataset) sequence.
500 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
501 // all mappings. Would it help to constrain this?
502 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
503 if (mappings == null || mappings.isEmpty())
509 * Locate the aligned source sequence whose dataset sequence is mapped. We
510 * just take the first match here (as we can't align cDNA like more than one
513 SequenceI alignFrom = null;
514 AlignedCodonFrame mapping = null;
515 for (AlignedCodonFrame mp : mappings)
517 alignFrom = mp.findAlignedSequence(seq.getDatasetSequence(), al);
518 if (alignFrom != null)
525 if (alignFrom == null)
529 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
530 preserveMappedGaps, preserveUnmappedGaps);
535 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
536 * match residues and codons. Flags control whether existing gaps in unmapped
537 * (intron) and mapped (exon) regions are preserved or not. Gaps linking intro
538 * and exon are only retained if both flags are set.
545 * @param preserveUnmappedGaps
546 * @param preserveMappedGaps
548 public static void alignSequenceAs(SequenceI alignTo,
550 AlignedCodonFrame mapping, String myGap, char sourceGap,
551 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
553 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
554 final char[] thisSeq = alignTo.getSequence();
555 final char[] thatAligned = alignFrom.getSequence();
556 StringBuilder thisAligned = new StringBuilder(2 * thisSeq.length);
558 // aligned and dataset sequence positions, all base zero
562 int basesWritten = 0;
563 char myGapChar = myGap.charAt(0);
564 int ratio = myGap.length();
567 * Traverse the aligned protein sequence.
569 int sourceGapMappedLength = 0;
570 boolean inExon = false;
571 for (char sourceChar : thatAligned)
573 if (sourceChar == sourceGap)
575 sourceGapMappedLength += ratio;
580 * Found a residue. Locate its mapped codon (start) position.
583 // Note mapping positions are base 1, our sequence positions base 0
584 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
586 if (mappedPos == null)
589 * Abort realignment if unmapped protein. Or could ignore it??
591 System.err.println("Can't align: no codon mapping to residue "
592 + sourceDsPos + "(" + sourceChar + ")");
596 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
597 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
598 StringBuilder trailingCopiedGap = new StringBuilder();
601 * Copy dna sequence up to and including this codon. Optionally, include
602 * gaps before the codon starts (in introns) and/or after the codon starts
605 * Note this only works for 'linear' splicing, not reverse or interleaved.
606 * But then 'align dna as protein' doesn't make much sense otherwise.
608 int intronLength = 0;
609 while (basesWritten < mappedCodonEnd && thisSeqPos < thisSeq.length)
611 final char c = thisSeq[thisSeqPos++];
616 if (basesWritten < mappedCodonStart)
619 * Found an unmapped (intron) base. First add in any preceding gaps
622 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
624 thisAligned.append(trailingCopiedGap.toString());
625 intronLength += trailingCopiedGap.length();
626 trailingCopiedGap = new StringBuilder();
633 final boolean startOfCodon = basesWritten == mappedCodonStart;
634 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
635 preserveUnmappedGaps, sourceGapMappedLength, inExon,
636 trailingCopiedGap.length(), intronLength, startOfCodon);
637 for (int i = 0; i < gapsToAdd; i++)
639 thisAligned.append(myGapChar);
641 sourceGapMappedLength = 0;
644 thisAligned.append(c);
645 trailingCopiedGap = new StringBuilder();
649 if (inExon && preserveMappedGaps)
651 trailingCopiedGap.append(myGapChar);
653 else if (!inExon && preserveUnmappedGaps)
655 trailingCopiedGap.append(myGapChar);
662 * At end of protein sequence. Copy any remaining dna sequence, optionally
663 * including (intron) gaps. We do not copy trailing gaps in protein.
665 while (thisSeqPos < thisSeq.length)
667 final char c = thisSeq[thisSeqPos++];
668 if (c != myGapChar || preserveUnmappedGaps)
670 thisAligned.append(c);
675 * All done aligning, set the aligned sequence.
677 alignTo.setSequence(new String(thisAligned));
681 * Helper method to work out how many gaps to insert when realigning.
683 * @param preserveMappedGaps
684 * @param preserveUnmappedGaps
685 * @param sourceGapMappedLength
687 * @param trailingCopiedGap
688 * @param intronLength
689 * @param startOfCodon
692 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
693 boolean preserveUnmappedGaps, int sourceGapMappedLength,
694 boolean inExon, int trailingGapLength,
695 int intronLength, final boolean startOfCodon)
701 * Reached start of codon. Ignore trailing gaps in intron unless we are
702 * preserving gaps in both exon and intron. Ignore them anyway if the
703 * protein alignment introduces a gap at least as large as the intronic
706 if (inExon && !preserveMappedGaps)
708 trailingGapLength = 0;
710 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
712 trailingGapLength = 0;
716 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
720 if (intronLength + trailingGapLength <= sourceGapMappedLength)
722 gapsToAdd = sourceGapMappedLength - intronLength;
726 gapsToAdd = Math.min(intronLength + trailingGapLength
727 - sourceGapMappedLength, trailingGapLength);
734 * second or third base of codon; check for any gaps in dna
736 if (!preserveMappedGaps)
738 trailingGapLength = 0;
740 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
746 * Returns a list of sequences mapped from the given sequences and aligned
747 * (gapped) in the same way. For example, the cDNA for aligned protein, where
748 * a single gap in protein generates three gaps in cDNA.
751 * @param gapCharacter
755 public static List<SequenceI> getAlignedTranslation(
756 List<SequenceI> sequences, char gapCharacter,
757 Set<AlignedCodonFrame> mappings)
759 List<SequenceI> alignedSeqs = new ArrayList<SequenceI>();
761 for (SequenceI seq : sequences)
763 List<SequenceI> mapped = getAlignedTranslation(seq, gapCharacter,
765 alignedSeqs.addAll(mapped);
771 * Returns sequences aligned 'like' the source sequence, as mapped by the
772 * given mappings. Normally we expect zero or one 'mapped' sequences, but this
773 * will support 1-to-many as well.
776 * @param gapCharacter
780 protected static List<SequenceI> getAlignedTranslation(SequenceI seq,
781 char gapCharacter, Set<AlignedCodonFrame> mappings)
783 List<SequenceI> result = new ArrayList<SequenceI>();
784 for (AlignedCodonFrame mapping : mappings)
786 if (mapping.involvesSequence(seq))
788 SequenceI mapped = getAlignedTranslation(seq, gapCharacter, mapping);
799 * Returns the translation of 'seq' (as held in the mapping) with
800 * corresponding alignment (gaps).
803 * @param gapCharacter
807 protected static SequenceI getAlignedTranslation(SequenceI seq,
808 char gapCharacter, AlignedCodonFrame mapping)
810 String gap = String.valueOf(gapCharacter);
811 boolean toDna = false;
813 SequenceI mapTo = mapping.getDnaForAaSeq(seq);
816 // mapping is from protein to nucleotide
818 // should ideally get gap count ratio from mapping
819 gap = String.valueOf(new char[]
820 { gapCharacter, gapCharacter, gapCharacter });
824 // mapping is from nucleotide to protein
825 mapTo = mapping.getAaForDnaSeq(seq);
828 StringBuilder newseq = new StringBuilder(seq.getLength()
831 int residueNo = 0; // in seq, base 1
832 int[] phrase = new int[fromRatio];
833 int phraseOffset = 0;
835 boolean first = true;
836 final Sequence alignedSeq = new Sequence("", "");
838 for (char c : seq.getSequence())
840 if (c == gapCharacter)
843 if (gapWidth >= fromRatio)
851 phrase[phraseOffset++] = residueNo + 1;
852 if (phraseOffset == fromRatio)
855 * Have read a whole codon (or protein residue), now translate: map
856 * source phrase to positions in target sequence add characters at
857 * these positions to newseq Note mapping positions are base 1, our
858 * sequence positions base 0.
860 SearchResults sr = new SearchResults();
861 for (int pos : phrase)
863 mapping.markMappedRegion(seq, pos, sr);
865 newseq.append(sr.toString());
869 // Hack: Copy sequence dataset, name and description from
870 // SearchResults.match[0].sequence
871 // TODO? carry over sequence names from original 'complement'
873 SequenceI mappedTo = sr.getResultSequence(0);
874 alignedSeq.setName(mappedTo.getName());
875 alignedSeq.setDescription(mappedTo.getDescription());
876 alignedSeq.setDatasetSequence(mappedTo);
883 alignedSeq.setSequence(newseq.toString());
888 * Realigns the given protein to match the alignment of the dna, using codon
889 * mappings to translate aligned codon positions to protein residues.
892 * the alignment whose sequences are realigned by this method
894 * the dna alignment whose alignment we are 'copying'
895 * @return the number of sequences that were realigned
897 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
899 Set<AlignedCodonFrame> mappings = protein.getCodonFrames();
902 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
903 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
904 * comparator keeps the codon positions ordered.
906 Map<AlignedCodon, Map<SequenceI, String>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, String>>(
907 new CodonComparator());
908 for (SequenceI dnaSeq : dna.getSequences())
910 for (AlignedCodonFrame mapping : mappings)
912 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
913 SequenceI prot = mapping.findAlignedSequence(
914 dnaSeq.getDatasetSequence(), protein);
917 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
918 seqMap, alignedCodons);
922 return alignProteinAs(protein, alignedCodons);
926 * Update the aligned protein sequences to match the codon alignments given in
930 * @param alignedCodons
931 * an ordered map of codon positions (columns), with sequence/peptide
932 * values present in each column
935 protected static int alignProteinAs(AlignmentI protein,
936 Map<AlignedCodon, Map<SequenceI, String>> alignedCodons)
939 * Prefill aligned sequences with gaps before inserting aligned protein
942 int alignedWidth = alignedCodons.size();
943 char[] gaps = new char[alignedWidth];
944 Arrays.fill(gaps, protein.getGapCharacter());
945 String allGaps = String.valueOf(gaps);
946 for (SequenceI seq : protein.getSequences())
948 seq.setSequence(allGaps);
952 for (AlignedCodon codon : alignedCodons.keySet())
954 final Map<SequenceI, String> columnResidues = alignedCodons.get(codon);
955 for (Entry<SequenceI, String> entry : columnResidues
958 // place translated codon at its column position in sequence
959 entry.getKey().getSequence()[column] = entry.getValue().charAt(0);
967 * Populate the map of aligned codons by traversing the given sequence
968 * mapping, locating the aligned positions of mapped codons, and adding those
969 * positions and their translation products to the map.
972 * the aligned sequence we are mapping from
974 * the sequence to be aligned to the codons
976 * the gap character in the dna sequence
978 * a mapping to a sequence translation
979 * @param alignedCodons
980 * the map we are building up
982 static void addCodonPositions(SequenceI dna, SequenceI protein,
985 Map<AlignedCodon, Map<SequenceI, String>> alignedCodons)
987 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
988 while (codons.hasNext())
990 AlignedCodon codon = codons.next();
991 Map<SequenceI, String> seqProduct = alignedCodons.get(codon);
992 if (seqProduct == null)
994 seqProduct = new HashMap<SequenceI, String>();
995 alignedCodons.put(codon, seqProduct);
997 seqProduct.put(protein, codon.product);
1002 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1003 * between at least one pair of sequences in the two alignments. Currently,
1006 * <li>One alignment must be nucleotide, and the other protein</li>
1007 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1008 * <li>Mappable means the nucleotide translation matches the protein sequence</li>
1009 * <li>The translation may ignore start and stop codons if present in the
1017 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1020 * Require one nucleotide and one protein
1022 if (al1.isNucleotide() == al2.isNucleotide())
1026 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1027 AlignmentI protein = dna == al1 ? al2 : al1;
1028 Set<AlignedCodonFrame> mappings = protein.getCodonFrames();
1029 for (SequenceI dnaSeq : dna.getSequences())
1031 for (SequenceI proteinSeq : protein.getSequences())
1033 if (isMappable(dnaSeq, proteinSeq, mappings))
1043 * Returns true if the dna sequence is mapped, or could be mapped, to the
1051 public static boolean isMappable(SequenceI dnaSeq, SequenceI proteinSeq,
1052 Set<AlignedCodonFrame> mappings)
1054 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq : dnaSeq.getDatasetSequence();
1055 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null ? proteinSeq
1056 : proteinSeq.getDatasetSequence();
1061 for (AlignedCodonFrame mapping : mappings) {
1062 if ( proteinDs == mapping.getAaForDnaSeq(dnaDs)) {
1068 * Just try to make a mapping (it is not yet stored), test whether
1071 return mapProteinToCdna(proteinDs, dnaDs) != null;
1075 * Finds any reference annotations associated with the sequences in
1076 * sequenceScope, that are not already added to the alignment, and adds them
1077 * to the 'candidates' map. Also populates a lookup table of annotation
1078 * labels, keyed by calcId, for use in constructing tooltips or the like.
1080 * @param sequenceScope
1081 * the sequences to scan for reference annotations
1082 * @param labelForCalcId
1083 * (optional) map to populate with label for calcId
1085 * map to populate with annotations for sequence
1087 * the alignment to check for presence of annotations
1089 public static void findAddableReferenceAnnotations(
1090 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1091 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1094 if (sequenceScope == null)
1100 * For each sequence in scope, make a list of any annotations on the
1101 * underlying dataset sequence which are not already on the alignment.
1103 * Add to a map of { alignmentSequence, <List of annotations to add> }
1105 for (SequenceI seq : sequenceScope)
1107 SequenceI dataset = seq.getDatasetSequence();
1108 if (dataset == null)
1112 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1113 if (datasetAnnotations == null)
1117 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1118 for (AlignmentAnnotation dsann : datasetAnnotations)
1121 * Find matching annotations on the alignment. If none is found, then
1122 * add this annotation to the list of 'addable' annotations for this
1125 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1126 .findAnnotations(seq, dsann.getCalcId(),
1128 if (!matchedAlignmentAnnotations.iterator().hasNext())
1131 if (labelForCalcId != null)
1133 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1138 * Save any addable annotations for this sequence
1140 if (!result.isEmpty())
1142 candidates.put(seq, result);
1148 * Adds annotations to the top of the alignment annotations, in the same order
1149 * as their related sequences.
1151 * @param annotations
1152 * the annotations to add
1154 * the alignment to add them to
1155 * @param selectionGroup
1156 * current selection group (or null if none)
1158 public static void addReferenceAnnotations(
1159 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1160 final AlignmentI alignment, final SequenceGroup selectionGroup)
1162 for (SequenceI seq : annotations.keySet())
1164 for (AlignmentAnnotation ann : annotations.get(seq))
1166 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1168 int endRes = ann.annotations.length;
1169 if (selectionGroup != null)
1171 startRes = selectionGroup.getStartRes();
1172 endRes = selectionGroup.getEndRes();
1174 copyAnn.restrict(startRes, endRes);
1177 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1178 * original annotation is already on the sequence.
1180 if (!seq.hasAnnotation(ann))
1182 seq.addAlignmentAnnotation(copyAnn);
1185 copyAnn.adjustForAlignment();
1186 // add to the alignment and set visible
1187 alignment.addAnnotation(copyAnn);
1188 copyAnn.visible = true;
1194 * Set visibility of alignment annotations of specified types (labels), for
1195 * specified sequences. This supports controls like
1196 * "Show all secondary structure", "Hide all Temp factor", etc.
1198 * @al the alignment to scan for annotations
1200 * the types (labels) of annotations to be updated
1201 * @param forSequences
1202 * if not null, only annotations linked to one of these sequences are
1203 * in scope for update; if null, acts on all sequence annotations
1205 * if this flag is true, 'types' is ignored (label not checked)
1207 * if true, set visibility on, else set off
1209 public static void showOrHideSequenceAnnotations(AlignmentI al,
1210 Collection<String> types, List<SequenceI> forSequences,
1211 boolean anyType, boolean doShow)
1213 for (AlignmentAnnotation aa : al
1214 .getAlignmentAnnotation())
1216 if (anyType || types.contains(aa.label))
1218 if ((aa.sequenceRef != null)
1219 && (forSequences == null || forSequences
1220 .contains(aa.sequenceRef)))
1222 aa.visible = doShow;
1229 * Returns true if either sequence has a cross-reference to the other
1235 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1237 // Note: moved here from class CrossRef as the latter class has dependencies
1238 // not availability to the applet's classpath
1239 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1243 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1244 * that sequence name is structured as Source|AccessId.
1250 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1252 if (seq1 == null || seq2 == null)
1256 String name = seq2.getName();
1257 final DBRefEntry[] xrefs = seq1.getDBRef();
1260 for (DBRefEntry xref : xrefs)
1262 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1263 // case-insensitive test, consistent with DBRefEntry.equalRef()
1264 if (xrefName.equalsIgnoreCase(name))
1274 * Constructs an alignment consisting of the mapped exon regions in the given
1275 * nucleotide sequences, and updates mappings to match.
1278 * aligned dna sequences
1280 * from dna to protein; these are replaced with new mappings
1281 * @return an alignment whose sequences are the exon-only parts of the dna
1282 * sequences (or null if no exons are found)
1284 public static AlignmentI makeExonAlignment(SequenceI[] dna,
1285 Set<AlignedCodonFrame> mappings)
1287 Set<AlignedCodonFrame> newMappings = new LinkedHashSet<AlignedCodonFrame>();
1288 List<SequenceI> exonSequences = new ArrayList<SequenceI>();
1290 for (SequenceI dnaSeq : dna)
1292 final SequenceI ds = dnaSeq.getDatasetSequence();
1293 List<AlignedCodonFrame> seqMappings = MappingUtils
1294 .findMappingsForSequence(ds, mappings);
1295 for (AlignedCodonFrame acf : seqMappings)
1297 AlignedCodonFrame newMapping = new AlignedCodonFrame();
1298 final List<SequenceI> mappedExons = makeExonSequences(ds, acf,
1300 if (!mappedExons.isEmpty())
1302 exonSequences.addAll(mappedExons);
1303 newMappings.add(newMapping);
1307 AlignmentI al = new Alignment(
1308 exonSequences.toArray(new SequenceI[exonSequences.size()]));
1309 al.setDataset(null);
1312 * Replace the old mappings with the new ones
1315 mappings.addAll(newMappings);
1321 * Helper method to make exon-only sequences and populate their mappings to
1324 * For example, if ggCCaTTcGAg has mappings [3, 4, 6, 7, 9, 10] to protein
1325 * then generate a sequence CCTTGA with mapping [1, 6] to the same protein
1328 * Typically eukaryotic dna will include exons encoding for a single peptide
1329 * sequence i.e. return a single result. Bacterial dna may have overlapping
1330 * exon mappings coding for multiple peptides so return multiple results
1331 * (example EMBL KF591215).
1334 * a dna dataset sequence
1336 * containing one or more mappings of the sequence to protein
1338 * the new mapping to populate, from the exon-only sequences to their
1339 * mapped protein sequences
1342 protected static List<SequenceI> makeExonSequences(SequenceI dnaSeq,
1343 AlignedCodonFrame mapping, AlignedCodonFrame newMapping)
1345 List<SequenceI> exonSequences = new ArrayList<SequenceI>();
1346 List<Mapping> seqMappings = mapping.getMappingsForSequence(dnaSeq);
1347 final char[] dna = dnaSeq.getSequence();
1348 for (Mapping seqMapping : seqMappings)
1350 StringBuilder newSequence = new StringBuilder(dnaSeq.getLength());
1353 * Get the codon regions as { [2, 5], [7, 12], [14, 14] etc }
1355 final List<int[]> dnaExonRanges = seqMapping.getMap().getFromRanges();
1356 for (int[] range : dnaExonRanges)
1358 for (int pos = range[0]; pos <= range[1]; pos++)
1360 newSequence.append(dna[pos - 1]);
1364 SequenceI exon = new Sequence(dnaSeq.getName(),
1365 newSequence.toString());
1368 * Locate any xrefs to CDS database on the protein product and attach to
1369 * the CDS sequence. Also add as a sub-token of the sequence name.
1371 // default to "CDS" if we can't locate an actual gene id
1372 String cdsAccId = FeatureProperties
1373 .getCodingFeature(DBRefSource.EMBL);
1374 DBRefEntry[] cdsRefs = DBRefUtils.selectRefs(seqMapping.getTo()
1375 .getDBRef(), DBRefSource.CODINGDBS);
1376 if (cdsRefs != null)
1378 for (DBRefEntry cdsRef : cdsRefs)
1380 exon.addDBRef(new DBRefEntry(cdsRef));
1381 cdsAccId = cdsRef.getAccessionId();
1384 exon.setName(exon.getName() + "|" + cdsAccId);
1385 exon.createDatasetSequence();
1388 * Build new mappings - from the same protein regions, but now to
1391 List<int[]> exonRange = new ArrayList<int[]>();
1392 exonRange.add(new int[]
1393 { 1, newSequence.length() });
1394 MapList map = new MapList(exonRange, seqMapping.getMap()
1397 newMapping.addMap(exon.getDatasetSequence(), seqMapping.getTo(), map);
1398 MapList cdsToDnaMap = new MapList(dnaExonRanges, exonRange, 1, 1);
1399 newMapping.addMap(dnaSeq, exon.getDatasetSequence(), cdsToDnaMap);
1401 exonSequences.add(exon);
1403 return exonSequences;