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 static jalview.io.gff.GffConstants.CLINICAL_SIGNIFICANCE;
25 import jalview.api.DBRefEntryI;
26 import jalview.datamodel.AlignedCodon;
27 import jalview.datamodel.AlignedCodonFrame;
28 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
29 import jalview.datamodel.Alignment;
30 import jalview.datamodel.AlignmentAnnotation;
31 import jalview.datamodel.AlignmentI;
32 import jalview.datamodel.DBRefEntry;
33 import jalview.datamodel.IncompleteCodonException;
34 import jalview.datamodel.Mapping;
35 import jalview.datamodel.Sequence;
36 import jalview.datamodel.SequenceFeature;
37 import jalview.datamodel.SequenceGroup;
38 import jalview.datamodel.SequenceI;
39 import jalview.io.gff.SequenceOntologyFactory;
40 import jalview.io.gff.SequenceOntologyI;
41 import jalview.schemes.ResidueProperties;
42 import jalview.util.Comparison;
43 import jalview.util.DBRefUtils;
44 import jalview.util.MapList;
45 import jalview.util.MappingUtils;
46 import jalview.util.StringUtils;
48 import java.io.UnsupportedEncodingException;
49 import java.net.URLEncoder;
50 import java.util.ArrayList;
51 import java.util.Arrays;
52 import java.util.Collection;
53 import java.util.Collections;
54 import java.util.Comparator;
55 import java.util.HashMap;
56 import java.util.HashSet;
57 import java.util.Iterator;
58 import java.util.LinkedHashMap;
59 import java.util.List;
61 import java.util.Map.Entry;
62 import java.util.NoSuchElementException;
64 import java.util.TreeMap;
67 * grab bag of useful alignment manipulation operations Expect these to be
68 * refactored elsewhere at some point.
73 public class AlignmentUtils
76 private static final String SEQUENCE_VARIANT = "sequence_variant:";
77 private static final String ID = "ID";
80 * A data model to hold the 'normal' base value at a position, and an optional
81 * sequence variant feature
83 static class DnaVariant
87 SequenceFeature variant;
89 DnaVariant(String nuc)
94 DnaVariant(String nuc, SequenceFeature var)
102 * given an existing alignment, create a new alignment including all, or up to
103 * flankSize additional symbols from each sequence's dataset sequence
109 public static AlignmentI expandContext(AlignmentI core, int flankSize)
111 List<SequenceI> sq = new ArrayList<SequenceI>();
113 for (SequenceI s : core.getSequences())
115 SequenceI newSeq = s.deriveSequence();
116 final int newSeqStart = newSeq.getStart() - 1;
117 if (newSeqStart > maxoffset
118 && newSeq.getDatasetSequence().getStart() < s.getStart())
120 maxoffset = newSeqStart;
126 maxoffset = Math.min(maxoffset, flankSize);
130 * now add offset left and right to create an expanded alignment
132 for (SequenceI s : sq)
135 while (ds.getDatasetSequence() != null)
137 ds = ds.getDatasetSequence();
139 int s_end = s.findPosition(s.getStart() + s.getLength());
140 // find available flanking residues for sequence
141 int ustream_ds = s.getStart() - ds.getStart();
142 int dstream_ds = ds.getEnd() - s_end;
144 // build new flanked sequence
146 // compute gap padding to start of flanking sequence
147 int offset = maxoffset - ustream_ds;
149 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
152 if (flankSize < ustream_ds)
154 // take up to flankSize residues
155 offset = maxoffset - flankSize;
156 ustream_ds = flankSize;
158 if (flankSize <= dstream_ds)
160 dstream_ds = flankSize - 1;
163 // TODO use Character.toLowerCase to avoid creating String objects?
164 char[] upstream = new String(ds.getSequence(s.getStart() - 1
165 - ustream_ds, s.getStart() - 1)).toLowerCase().toCharArray();
166 char[] downstream = new String(ds.getSequence(s_end - 1, s_end
167 + dstream_ds)).toLowerCase().toCharArray();
168 char[] coreseq = s.getSequence();
169 char[] nseq = new char[offset + upstream.length + downstream.length
171 char c = core.getGapCharacter();
174 for (; p < offset; p++)
179 System.arraycopy(upstream, 0, nseq, p, upstream.length);
180 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
182 System.arraycopy(downstream, 0, nseq, p + coreseq.length
183 + upstream.length, downstream.length);
184 s.setSequence(new String(nseq));
185 s.setStart(s.getStart() - ustream_ds);
186 s.setEnd(s_end + downstream.length);
188 AlignmentI newAl = new jalview.datamodel.Alignment(
189 sq.toArray(new SequenceI[0]));
190 for (SequenceI s : sq)
192 if (s.getAnnotation() != null)
194 for (AlignmentAnnotation aa : s.getAnnotation())
196 aa.adjustForAlignment(); // JAL-1712 fix
197 newAl.addAnnotation(aa);
201 newAl.setDataset(core.getDataset());
206 * Returns the index (zero-based position) of a sequence in an alignment, or
213 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
217 for (SequenceI alSeq : al.getSequences())
230 * Returns a map of lists of sequences in the alignment, keyed by sequence
231 * name. For use in mapping between different alignment views of the same
234 * @see jalview.datamodel.AlignmentI#getSequencesByName()
236 public static Map<String, List<SequenceI>> getSequencesByName(
239 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
240 for (SequenceI seq : al.getSequences())
242 String name = seq.getName();
245 List<SequenceI> seqs = theMap.get(name);
248 seqs = new ArrayList<SequenceI>();
249 theMap.put(name, seqs);
258 * Build mapping of protein to cDNA alignment. Mappings are made between
259 * sequences where the cDNA translates to the protein sequence. Any new
260 * mappings are added to the protein alignment. Returns true if any mappings
261 * either already exist or were added, else false.
263 * @param proteinAlignment
264 * @param cdnaAlignment
267 public static boolean mapProteinAlignmentToCdna(
268 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
270 if (proteinAlignment == null || cdnaAlignment == null)
275 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
276 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
279 * First pass - map sequences where cross-references exist. This include
280 * 1-to-many mappings to support, for example, variant cDNA.
282 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
283 cdnaAlignment, mappedDna, mappedProtein, true);
286 * Second pass - map sequences where no cross-references exist. This only
287 * does 1-to-1 mappings and assumes corresponding sequences are in the same
288 * order in the alignments.
290 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
291 mappedDna, mappedProtein, false);
292 return mappingPerformed;
296 * Make mappings between compatible sequences (where the cDNA translation
297 * matches the protein).
299 * @param proteinAlignment
300 * @param cdnaAlignment
302 * a set of mapped DNA sequences (to add to)
303 * @param mappedProtein
304 * a set of mapped Protein sequences (to add to)
306 * if true, only map sequences where xrefs exist
309 protected static boolean mapProteinToCdna(
310 final AlignmentI proteinAlignment,
311 final AlignmentI cdnaAlignment, Set<SequenceI> mappedDna,
312 Set<SequenceI> mappedProtein, boolean xrefsOnly)
314 boolean mappingExistsOrAdded = false;
315 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
316 for (SequenceI aaSeq : thisSeqs)
318 boolean proteinMapped = false;
319 AlignedCodonFrame acf = new AlignedCodonFrame();
321 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
324 * Always try to map if sequences have xref to each other; this supports
325 * variant cDNA or alternative splicing for a protein sequence.
327 * If no xrefs, try to map progressively, assuming that alignments have
328 * mappable sequences in corresponding order. These are not
329 * many-to-many, as that would risk mixing species with similar cDNA
332 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
338 * Don't map non-xrefd sequences more than once each. This heuristic
339 * allows us to pair up similar sequences in ordered alignments.
342 && (mappedProtein.contains(aaSeq) || mappedDna
347 if (mappingExists(proteinAlignment.getCodonFrames(),
348 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
350 mappingExistsOrAdded = true;
354 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
357 acf.addMap(cdnaSeq, aaSeq, map);
358 mappingExistsOrAdded = true;
359 proteinMapped = true;
360 mappedDna.add(cdnaSeq);
361 mappedProtein.add(aaSeq);
367 proteinAlignment.addCodonFrame(acf);
370 return mappingExistsOrAdded;
374 * Answers true if the mappings include one between the given (dataset)
377 public static boolean mappingExists(List<AlignedCodonFrame> mappings,
378 SequenceI aaSeq, SequenceI cdnaSeq)
380 if (mappings != null)
382 for (AlignedCodonFrame acf : mappings)
384 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
394 * Builds a mapping (if possible) of a cDNA to a protein sequence.
396 * <li>first checks if the cdna translates exactly to the protein sequence</li>
397 * <li>else checks for translation after removing a STOP codon</li>
398 * <li>else checks for translation after removing a START codon</li>
399 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
401 * Returns null if no mapping is determined.
404 * the aligned protein sequence
406 * the aligned cdna sequence
409 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
413 * Here we handle either dataset sequence set (desktop) or absent (applet).
414 * Use only the char[] form of the sequence to avoid creating possibly large
417 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
418 char[] aaSeqChars = proteinDataset != null ? proteinDataset
419 .getSequence() : proteinSeq.getSequence();
420 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
421 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
422 : cdnaSeq.getSequence();
423 if (aaSeqChars == null || cdnaSeqChars == null)
429 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
431 final int mappedLength = 3 * aaSeqChars.length;
432 int cdnaLength = cdnaSeqChars.length;
433 int cdnaStart = cdnaSeq.getStart();
434 int cdnaEnd = cdnaSeq.getEnd();
435 final int proteinStart = proteinSeq.getStart();
436 final int proteinEnd = proteinSeq.getEnd();
439 * If lengths don't match, try ignoring stop codon (if present)
441 if (cdnaLength != mappedLength && cdnaLength > 2)
443 String lastCodon = String.valueOf(cdnaSeqChars, cdnaLength - 3, 3)
445 for (String stop : ResidueProperties.STOP)
447 if (lastCodon.equals(stop))
457 * If lengths still don't match, try ignoring start codon.
460 if (cdnaLength != mappedLength
462 && String.valueOf(cdnaSeqChars, 0, 3).toUpperCase()
463 .equals(ResidueProperties.START))
470 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
473 * protein is translation of dna (+/- start/stop codons)
475 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd }, new int[]
476 { proteinStart, proteinEnd }, 3, 1);
481 * translation failed - try mapping CDS annotated regions of dna
483 return mapCdsToProtein(cdnaSeq, proteinSeq);
487 * Test whether the given cdna sequence, starting at the given offset,
488 * translates to the given amino acid sequence, using the standard translation
489 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
491 * @param cdnaSeqChars
496 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
499 if (cdnaSeqChars == null || aaSeqChars == null)
505 int dnaPos = cdnaStart;
506 for (; dnaPos < cdnaSeqChars.length - 2
507 && aaPos < aaSeqChars.length; dnaPos += 3, aaPos++)
509 String codon = String.valueOf(cdnaSeqChars, dnaPos, 3);
510 final String translated = ResidueProperties.codonTranslate(codon);
513 * allow * in protein to match untranslatable in dna
515 final char aaRes = aaSeqChars[aaPos];
516 if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
520 if (translated == null || !(aaRes == translated.charAt(0)))
523 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
524 // + codon + "(" + translated + ") != " + aaRes));
530 * check we matched all of the protein sequence
532 if (aaPos != aaSeqChars.length)
538 * check we matched all of the dna except
539 * for optional trailing STOP codon
541 if (dnaPos == cdnaSeqChars.length)
545 if (dnaPos == cdnaSeqChars.length - 3)
547 String codon = String.valueOf(cdnaSeqChars, dnaPos, 3);
548 if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
557 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
558 * currently assumes that we are aligning cDNA to match protein.
561 * the sequence to be realigned
563 * the alignment whose sequence alignment is to be 'copied'
565 * character string represent a gap in the realigned sequence
566 * @param preserveUnmappedGaps
567 * @param preserveMappedGaps
568 * @return true if the sequence was realigned, false if it could not be
570 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
571 String gap, boolean preserveMappedGaps,
572 boolean preserveUnmappedGaps)
575 * Get any mappings from the source alignment to the target (dataset)
578 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
579 // all mappings. Would it help to constrain this?
580 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
581 if (mappings == null || mappings.isEmpty())
587 * Locate the aligned source sequence whose dataset sequence is mapped. We
588 * just take the first match here (as we can't align like more than one
591 SequenceI alignFrom = null;
592 AlignedCodonFrame mapping = null;
593 for (AlignedCodonFrame mp : mappings)
595 alignFrom = mp.findAlignedSequence(seq, al);
596 if (alignFrom != null)
603 if (alignFrom == null)
607 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
608 preserveMappedGaps, preserveUnmappedGaps);
613 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
614 * match residues and codons. Flags control whether existing gaps in unmapped
615 * (intron) and mapped (exon) regions are preserved or not. Gaps between
616 * intron and exon are only retained if both flags are set.
623 * @param preserveUnmappedGaps
624 * @param preserveMappedGaps
626 public static void alignSequenceAs(SequenceI alignTo,
627 SequenceI alignFrom, AlignedCodonFrame mapping, String myGap,
628 char sourceGap, boolean preserveMappedGaps,
629 boolean preserveUnmappedGaps)
631 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
633 // aligned and dataset sequence positions, all base zero
637 int basesWritten = 0;
638 char myGapChar = myGap.charAt(0);
639 int ratio = myGap.length();
641 int fromOffset = alignFrom.getStart() - 1;
642 int toOffset = alignTo.getStart() - 1;
643 int sourceGapMappedLength = 0;
644 boolean inExon = false;
645 final char[] thisSeq = alignTo.getSequence();
646 final char[] thatAligned = alignFrom.getSequence();
647 StringBuilder thisAligned = new StringBuilder(2 * thisSeq.length);
650 * Traverse the 'model' aligned sequence
652 for (char sourceChar : thatAligned)
654 if (sourceChar == sourceGap)
656 sourceGapMappedLength += ratio;
661 * Found a non-gap character. Locate its mapped region if any.
664 // Note mapping positions are base 1, our sequence positions base 0
665 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
666 sourceDsPos + fromOffset);
667 if (mappedPos == null)
670 * unmapped position; treat like a gap
672 sourceGapMappedLength += ratio;
673 // System.err.println("Can't align: no codon mapping to residue "
674 // + sourceDsPos + "(" + sourceChar + ")");
679 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
680 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
681 StringBuilder trailingCopiedGap = new StringBuilder();
684 * Copy dna sequence up to and including this codon. Optionally, include
685 * gaps before the codon starts (in introns) and/or after the codon starts
688 * Note this only works for 'linear' splicing, not reverse or interleaved.
689 * But then 'align dna as protein' doesn't make much sense otherwise.
691 int intronLength = 0;
692 while (basesWritten + toOffset < mappedCodonEnd
693 && thisSeqPos < thisSeq.length)
695 final char c = thisSeq[thisSeqPos++];
699 int sourcePosition = basesWritten + toOffset;
700 if (sourcePosition < mappedCodonStart)
703 * Found an unmapped (intron) base. First add in any preceding gaps
706 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
708 thisAligned.append(trailingCopiedGap.toString());
709 intronLength += trailingCopiedGap.length();
710 trailingCopiedGap = new StringBuilder();
717 final boolean startOfCodon = sourcePosition == mappedCodonStart;
718 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
719 preserveUnmappedGaps, sourceGapMappedLength, inExon,
720 trailingCopiedGap.length(), intronLength, startOfCodon);
721 for (int i = 0; i < gapsToAdd; i++)
723 thisAligned.append(myGapChar);
725 sourceGapMappedLength = 0;
728 thisAligned.append(c);
729 trailingCopiedGap = new StringBuilder();
733 if (inExon && preserveMappedGaps)
735 trailingCopiedGap.append(myGapChar);
737 else if (!inExon && preserveUnmappedGaps)
739 trailingCopiedGap.append(myGapChar);
746 * At end of model aligned sequence. Copy any remaining target sequence, optionally
747 * including (intron) gaps.
749 while (thisSeqPos < thisSeq.length)
751 final char c = thisSeq[thisSeqPos++];
752 if (c != myGapChar || preserveUnmappedGaps)
754 thisAligned.append(c);
756 sourceGapMappedLength--;
760 * finally add gaps to pad for any trailing source gaps or
761 * unmapped characters
763 if (preserveUnmappedGaps)
765 while (sourceGapMappedLength > 0)
767 thisAligned.append(myGapChar);
768 sourceGapMappedLength--;
773 * All done aligning, set the aligned sequence.
775 alignTo.setSequence(new String(thisAligned));
779 * Helper method to work out how many gaps to insert when realigning.
781 * @param preserveMappedGaps
782 * @param preserveUnmappedGaps
783 * @param sourceGapMappedLength
785 * @param trailingCopiedGap
786 * @param intronLength
787 * @param startOfCodon
790 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
791 boolean preserveUnmappedGaps, int sourceGapMappedLength,
792 boolean inExon, int trailingGapLength, int intronLength,
793 final boolean startOfCodon)
799 * Reached start of codon. Ignore trailing gaps in intron unless we are
800 * preserving gaps in both exon and intron. Ignore them anyway if the
801 * protein alignment introduces a gap at least as large as the intronic
804 if (inExon && !preserveMappedGaps)
806 trailingGapLength = 0;
808 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
810 trailingGapLength = 0;
814 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
818 if (intronLength + trailingGapLength <= sourceGapMappedLength)
820 gapsToAdd = sourceGapMappedLength - intronLength;
824 gapsToAdd = Math.min(intronLength + trailingGapLength
825 - sourceGapMappedLength, trailingGapLength);
832 * second or third base of codon; check for any gaps in dna
834 if (!preserveMappedGaps)
836 trailingGapLength = 0;
838 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
844 * Realigns the given protein to match the alignment of the dna, using codon
845 * mappings to translate aligned codon positions to protein residues.
848 * the alignment whose sequences are realigned by this method
850 * the dna alignment whose alignment we are 'copying'
851 * @return the number of sequences that were realigned
853 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
855 if (protein.isNucleotide() || !dna.isNucleotide())
857 System.err.println("Wrong alignment type in alignProteinAsDna");
860 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
861 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
862 protein, dna, unmappedProtein);
863 return alignProteinAs(protein, alignedCodons, unmappedProtein);
867 * Realigns the given dna to match the alignment of the protein, using codon
868 * mappings to translate aligned peptide positions to codons.
871 * the alignment whose sequences are realigned by this method
873 * the protein alignment whose alignment we are 'copying'
874 * @return the number of sequences that were realigned
876 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
878 if (protein.isNucleotide() || !dna.isNucleotide())
880 System.err.println("Wrong alignment type in alignProteinAsDna");
883 // todo: implement this
884 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
885 int alignedCount = 0;
886 for (SequenceI dnaSeq : dna.getSequences())
888 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings))
897 * Helper method to align (if possible) the dna sequence (cds only) to match
898 * the alignment of a mapped protein sequence
905 static boolean alignCdsSequenceAsProtein(SequenceI dnaSeq,
906 AlignmentI protein, List<AlignedCodonFrame> mappings)
908 List<AlignedCodonFrame> dnaMappings = MappingUtils
909 .findMappingsForSequence(dnaSeq, mappings);
910 for (AlignedCodonFrame mapping : dnaMappings)
912 SequenceI peptide = mapping.findAlignedSequence(dnaSeq, protein);
915 Mapping map = mapping.getMappingBetween(dnaSeq, peptide);
919 * traverse peptide adding gaps or codons to new cds sequence
921 MapList mapList = map.getMap();
929 * Builds a map whose key is an aligned codon position (3 alignment column
930 * numbers base 0), and whose value is a map from protein sequence to each
931 * protein's peptide residue for that codon. The map generates an ordering of
932 * the codons, and allows us to read off the peptides at each position in
933 * order to assemble 'aligned' protein sequences.
936 * the protein alignment
938 * the coding dna alignment
939 * @param unmappedProtein
940 * any unmapped proteins are added to this list
943 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
944 AlignmentI protein, AlignmentI dna,
945 List<SequenceI> unmappedProtein)
948 * maintain a list of any proteins with no mappings - these will be
949 * rendered 'as is' in the protein alignment as we can't align them
951 unmappedProtein.addAll(protein.getSequences());
953 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
956 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
957 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
958 * comparator keeps the codon positions ordered.
960 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
961 new CodonComparator());
963 for (SequenceI dnaSeq : dna.getSequences())
965 for (AlignedCodonFrame mapping : mappings)
967 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
970 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
971 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
972 seqMap, alignedCodons);
973 unmappedProtein.remove(prot);
979 * Finally add any unmapped peptide start residues (e.g. for incomplete
980 * codons) as if at the codon position before the second residue
982 // TODO resolve JAL-2022 so this fudge can be removed
983 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
984 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
986 return alignedCodons;
990 * Scans for any protein mapped from position 2 (meaning unmapped start
991 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
992 * preceding position in the alignment
994 * @param alignedCodons
995 * the codon-to-peptide map
996 * @param mappedSequenceCount
997 * the number of distinct sequences in the map
999 protected static void addUnmappedPeptideStarts(
1000 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1001 int mappedSequenceCount)
1003 // TODO delete this ugly hack once JAL-2022 is resolved
1004 // i.e. we can model startPhase > 0 (incomplete start codon)
1006 List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
1007 AlignedCodon lastCodon = null;
1008 Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
1010 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1013 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1016 SequenceI seq = sequenceCodon.getKey();
1017 if (sequencesChecked.contains(seq))
1021 sequencesChecked.add(seq);
1022 AlignedCodon codon = sequenceCodon.getValue();
1023 if (codon.peptideCol > 1)
1026 .println("Problem mapping protein with >1 unmapped start positions: "
1029 else if (codon.peptideCol == 1)
1032 * first position (peptideCol == 0) was unmapped - add it
1034 if (lastCodon != null)
1036 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1037 lastCodon.pos2, lastCodon.pos3, String.valueOf(seq
1039 toAdd.put(seq, firstPeptide);
1044 * unmapped residue at start of alignment (no prior column) -
1045 * 'insert' at nominal codon [0, 0, 0]
1047 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1048 String.valueOf(seq.getCharAt(0)), 0);
1049 toAdd.put(seq, firstPeptide);
1052 if (sequencesChecked.size() == mappedSequenceCount)
1054 // no need to check past first mapped position in all sequences
1058 lastCodon = entry.getKey();
1062 * add any new codons safely after iterating over the map
1064 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1066 addCodonToMap(alignedCodons, startCodon.getValue(),
1067 startCodon.getKey());
1072 * Update the aligned protein sequences to match the codon alignments given in
1076 * @param alignedCodons
1077 * an ordered map of codon positions (columns), with sequence/peptide
1078 * values present in each column
1079 * @param unmappedProtein
1082 protected static int alignProteinAs(AlignmentI protein,
1083 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1084 List<SequenceI> unmappedProtein)
1087 * Prefill aligned sequences with gaps before inserting aligned protein
1090 int alignedWidth = alignedCodons.size();
1091 char[] gaps = new char[alignedWidth];
1092 Arrays.fill(gaps, protein.getGapCharacter());
1093 String allGaps = String.valueOf(gaps);
1094 for (SequenceI seq : protein.getSequences())
1096 if (!unmappedProtein.contains(seq))
1098 seq.setSequence(allGaps);
1103 for (AlignedCodon codon : alignedCodons.keySet())
1105 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1107 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1109 // place translated codon at its column position in sequence
1110 entry.getKey().getSequence()[column] = entry.getValue().product
1119 * Populate the map of aligned codons by traversing the given sequence
1120 * mapping, locating the aligned positions of mapped codons, and adding those
1121 * positions and their translation products to the map.
1124 * the aligned sequence we are mapping from
1126 * the sequence to be aligned to the codons
1128 * the gap character in the dna sequence
1130 * a mapping to a sequence translation
1131 * @param alignedCodons
1132 * the map we are building up
1134 static void addCodonPositions(SequenceI dna, SequenceI protein,
1135 char gapChar, Mapping seqMap,
1136 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1138 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1141 * add codon positions, and their peptide translations, to the alignment
1142 * map, while remembering the first codon mapped
1144 while (codons.hasNext())
1148 AlignedCodon codon = codons.next();
1149 addCodonToMap(alignedCodons, codon, protein);
1150 } catch (IncompleteCodonException e)
1152 // possible incomplete trailing codon - ignore
1153 } catch (NoSuchElementException e)
1155 // possibly peptide lacking STOP
1161 * Helper method to add a codon-to-peptide entry to the aligned codons map
1163 * @param alignedCodons
1167 protected static void addCodonToMap(
1168 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1169 AlignedCodon codon, SequenceI protein)
1171 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1172 if (seqProduct == null)
1174 seqProduct = new HashMap<SequenceI, AlignedCodon>();
1175 alignedCodons.put(codon, seqProduct);
1177 seqProduct.put(protein, codon);
1181 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1182 * between at least one pair of sequences in the two alignments. Currently,
1185 * <li>One alignment must be nucleotide, and the other protein</li>
1186 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1187 * <li>Mappable means the nucleotide translation matches the protein sequence</li>
1188 * <li>The translation may ignore start and stop codons if present in the
1196 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1198 if (al1 == null || al2 == null)
1204 * Require one nucleotide and one protein
1206 if (al1.isNucleotide() == al2.isNucleotide())
1210 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1211 AlignmentI protein = dna == al1 ? al2 : al1;
1212 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1213 for (SequenceI dnaSeq : dna.getSequences())
1215 for (SequenceI proteinSeq : protein.getSequences())
1217 if (isMappable(dnaSeq, proteinSeq, mappings))
1227 * Returns true if the dna sequence is mapped, or could be mapped, to the
1235 protected static boolean isMappable(SequenceI dnaSeq,
1236 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1238 if (dnaSeq == null || proteinSeq == null)
1243 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq : dnaSeq
1244 .getDatasetSequence();
1245 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null ? proteinSeq
1246 : proteinSeq.getDatasetSequence();
1248 for (AlignedCodonFrame mapping : mappings)
1250 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1260 * Just try to make a mapping (it is not yet stored), test whether
1263 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1267 * Finds any reference annotations associated with the sequences in
1268 * sequenceScope, that are not already added to the alignment, and adds them
1269 * to the 'candidates' map. Also populates a lookup table of annotation
1270 * labels, keyed by calcId, for use in constructing tooltips or the like.
1272 * @param sequenceScope
1273 * the sequences to scan for reference annotations
1274 * @param labelForCalcId
1275 * (optional) map to populate with label for calcId
1277 * map to populate with annotations for sequence
1279 * the alignment to check for presence of annotations
1281 public static void findAddableReferenceAnnotations(
1282 List<SequenceI> sequenceScope,
1283 Map<String, String> labelForCalcId,
1284 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1287 if (sequenceScope == null)
1293 * For each sequence in scope, make a list of any annotations on the
1294 * underlying dataset sequence which are not already on the alignment.
1296 * Add to a map of { alignmentSequence, <List of annotations to add> }
1298 for (SequenceI seq : sequenceScope)
1300 SequenceI dataset = seq.getDatasetSequence();
1301 if (dataset == null)
1305 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1306 if (datasetAnnotations == null)
1310 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1311 for (AlignmentAnnotation dsann : datasetAnnotations)
1314 * Find matching annotations on the alignment. If none is found, then
1315 * add this annotation to the list of 'addable' annotations for this
1318 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1319 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1320 if (!matchedAlignmentAnnotations.iterator().hasNext())
1323 if (labelForCalcId != null)
1325 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1330 * Save any addable annotations for this sequence
1332 if (!result.isEmpty())
1334 candidates.put(seq, result);
1340 * Adds annotations to the top of the alignment annotations, in the same order
1341 * as their related sequences.
1343 * @param annotations
1344 * the annotations to add
1346 * the alignment to add them to
1347 * @param selectionGroup
1348 * current selection group (or null if none)
1350 public static void addReferenceAnnotations(
1351 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1352 final AlignmentI alignment, final SequenceGroup selectionGroup)
1354 for (SequenceI seq : annotations.keySet())
1356 for (AlignmentAnnotation ann : annotations.get(seq))
1358 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1360 int endRes = ann.annotations.length;
1361 if (selectionGroup != null)
1363 startRes = selectionGroup.getStartRes();
1364 endRes = selectionGroup.getEndRes();
1366 copyAnn.restrict(startRes, endRes);
1369 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1370 * original annotation is already on the sequence.
1372 if (!seq.hasAnnotation(ann))
1374 seq.addAlignmentAnnotation(copyAnn);
1377 copyAnn.adjustForAlignment();
1378 // add to the alignment and set visible
1379 alignment.addAnnotation(copyAnn);
1380 copyAnn.visible = true;
1386 * Set visibility of alignment annotations of specified types (labels), for
1387 * specified sequences. This supports controls like
1388 * "Show all secondary structure", "Hide all Temp factor", etc.
1390 * @al the alignment to scan for annotations
1392 * the types (labels) of annotations to be updated
1393 * @param forSequences
1394 * if not null, only annotations linked to one of these sequences are
1395 * in scope for update; if null, acts on all sequence annotations
1397 * if this flag is true, 'types' is ignored (label not checked)
1399 * if true, set visibility on, else set off
1401 public static void showOrHideSequenceAnnotations(AlignmentI al,
1402 Collection<String> types, List<SequenceI> forSequences,
1403 boolean anyType, boolean doShow)
1405 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1408 for (AlignmentAnnotation aa : anns)
1410 if (anyType || types.contains(aa.label))
1412 if ((aa.sequenceRef != null)
1413 && (forSequences == null || forSequences
1414 .contains(aa.sequenceRef)))
1416 aa.visible = doShow;
1424 * Returns true if either sequence has a cross-reference to the other
1430 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1432 // Note: moved here from class CrossRef as the latter class has dependencies
1433 // not availability to the applet's classpath
1434 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1438 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1439 * that sequence name is structured as Source|AccessionId.
1445 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1447 if (seq1 == null || seq2 == null)
1451 String name = seq2.getName();
1452 final DBRefEntry[] xrefs = seq1.getDBRefs();
1455 for (DBRefEntry xref : xrefs)
1457 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1458 // case-insensitive test, consistent with DBRefEntry.equalRef()
1459 if (xrefName.equalsIgnoreCase(name))
1469 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1470 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1471 * added to the original alignment's dataset, which is shared by the new
1472 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1473 * added to the alignment dataset.
1476 * aligned nucleotide (dna or cds) sequences
1478 * the alignment dataset the sequences belong to
1480 * (optional) to restrict results to CDS that map to specified
1482 * @return an alignment whose sequences are the cds-only parts of the dna
1483 * sequences (or null if no mappings are found)
1485 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1486 AlignmentI dataset, SequenceI[] products)
1488 if (dataset == null || dataset.getDataset() != null)
1490 throw new IllegalArgumentException(
1491 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1493 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
1494 List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
1495 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1496 HashSet<SequenceI> productSeqs = null;
1497 if (products != null)
1499 productSeqs = new HashSet<SequenceI>();
1500 for (SequenceI seq : products)
1502 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1503 .getDatasetSequence());
1508 * Construct CDS sequences from mappings on the alignment dataset.
1510 * - find the protein product(s) mapped to from each dna sequence
1511 * - if the mapping covers the whole dna sequence (give or take start/stop
1512 * codon), take the dna as the CDS sequence
1513 * - else search dataset mappings for a suitable dna sequence, i.e. one
1514 * whose whole sequence is mapped to the protein
1515 * - if no sequence found, construct one from the dna sequence and mapping
1516 * (and add it to dataset so it is found if this is repeated)
1518 for (SequenceI dnaSeq : dna)
1520 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1521 : dnaSeq.getDatasetSequence();
1523 List<AlignedCodonFrame> seqMappings = MappingUtils
1524 .findMappingsForSequence(dnaSeq, mappings);
1525 for (AlignedCodonFrame mapping : seqMappings)
1527 List<Mapping> mappingsFromSequence = mapping
1528 .getMappingsFromSequence(dnaSeq);
1530 for (Mapping aMapping : mappingsFromSequence)
1532 MapList mapList = aMapping.getMap();
1533 if (mapList.getFromRatio() == 1)
1536 * not a dna-to-protein mapping (likely dna-to-cds)
1542 * skip if mapping is not to one of the target set of proteins
1544 SequenceI proteinProduct = aMapping.getTo();
1545 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1551 * try to locate the CDS from the dataset mappings;
1552 * guard against duplicate results (for the case that protein has
1553 * dbrefs to both dna and cds sequences)
1555 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1556 seqMappings, aMapping);
1559 if (!foundSeqs.contains(cdsSeq))
1561 foundSeqs.add(cdsSeq);
1562 SequenceI derivedSequence = cdsSeq.deriveSequence();
1563 cdsSeqs.add(derivedSequence);
1564 if (!dataset.getSequences().contains(cdsSeq))
1566 dataset.addSequence(cdsSeq);
1573 * didn't find mapped CDS sequence - construct it and add
1574 * its dataset sequence to the dataset
1576 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping);
1577 SequenceI cdsSeqDss = cdsSeq.createDatasetSequence();
1578 cdsSeqs.add(cdsSeq);
1579 if (!dataset.getSequences().contains(cdsSeqDss))
1581 dataset.addSequence(cdsSeqDss);
1585 * add a mapping from CDS to the (unchanged) mapped to range
1587 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1588 cdsSeq.getLength() });
1589 MapList cdsToProteinMap = new MapList(cdsRange, mapList.getToRanges(),
1590 mapList.getFromRatio(), mapList.getToRatio());
1591 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1592 cdsToProteinMapping.addMap(cdsSeq, proteinProduct, cdsToProteinMap);
1595 * guard against duplicating the mapping if repeating this action
1597 if (!mappings.contains(cdsToProteinMapping))
1599 mappings.add(cdsToProteinMapping);
1603 * copy protein's dbrefs to CDS sequence
1604 * this enables Get Cross-References from CDS alignment
1606 DBRefEntry[] proteinRefs = DBRefUtils.selectDbRefs(false,
1607 proteinProduct.getDBRefs());
1608 if (proteinRefs != null)
1610 for (DBRefEntry ref : proteinRefs)
1612 DBRefEntry cdsToProteinRef = new DBRefEntry(ref);
1613 cdsToProteinRef.setMap(new Mapping(proteinProduct,
1615 cdsSeqDss.addDBRef(cdsToProteinRef);
1620 * add another mapping from original 'from' range to CDS
1622 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1623 MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1626 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeq,
1628 if (!mappings.contains(dnaToCdsMapping))
1630 mappings.add(dnaToCdsMapping);
1634 * add DBRef with mapping from protein to CDS
1635 * (this enables Get Cross-References from protein alignment)
1636 * This is tricky because we can't have two DBRefs with the
1637 * same source and accession, so need a different accession for
1638 * the CDS from the dna sequence
1640 DBRefEntryI dnaRef = dnaDss.getSourceDBRef();
1643 // assuming cds version same as dna ?!?
1644 DBRefEntry proteinToCdsRef = new DBRefEntry(dnaRef.getSource(),
1645 dnaRef.getVersion(), cdsSeq.getName());
1646 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1648 proteinProduct.addDBRef(proteinToCdsRef);
1652 * transfer any features on dna that overlap the CDS
1654 transferFeatures(dnaSeq, cdsSeq, cdsToProteinMap, null,
1655 SequenceOntologyI.CDS);
1660 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1662 cds.setDataset(dataset);
1668 * A helper method that finds a CDS sequence in the alignment dataset that is
1669 * mapped to the given protein sequence, and either is, or has a mapping from,
1670 * the given dna sequence.
1673 * set of all mappings on the dataset
1675 * a dna (or cds) sequence we are searching from
1676 * @param seqMappings
1677 * the set of mappings involving dnaSeq
1679 * an initial candidate from seqMappings
1682 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1683 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1687 * TODO a better dna-cds-protein mapping data representation to allow easy
1688 * navigation; until then this clunky looping around lists of mappings
1690 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1691 : dnaSeq.getDatasetSequence();
1692 SequenceI proteinProduct = aMapping.getTo();
1695 * is this mapping from the whole dna sequence (i.e. CDS)?
1696 * allowing for possible stop codon on dna but not peptide
1698 int mappedFromLength = MappingUtils.getLength(aMapping.getMap()
1700 int dnaLength = seqDss.getLength();
1701 if (mappedFromLength == dnaLength || mappedFromLength == dnaLength - 3)
1707 * looks like we found the dna-to-protein mapping; search for the
1708 * corresponding cds-to-protein mapping
1710 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1711 .findMappingsForSequence(proteinProduct, mappings);
1712 for (AlignedCodonFrame acf : mappingsToPeptide)
1714 for (SequenceToSequenceMapping map : acf.getMappings())
1716 Mapping mapping = map.getMapping();
1717 if (mapping != aMapping && mapping.getMap().getFromRatio() == 3
1718 && proteinProduct == mapping.getTo()
1719 && seqDss != map.getFromSeq())
1721 mappedFromLength = MappingUtils.getLength(mapping.getMap()
1723 if (mappedFromLength == map.getFromSeq().getLength())
1726 * found a 3:1 mapping to the protein product which covers
1727 * the whole dna sequence i.e. is from CDS; finally check it
1728 * is from the dna start sequence
1730 SequenceI cdsSeq = map.getFromSeq();
1731 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1732 .findMappingsForSequence(cdsSeq, seqMappings);
1733 if (!dnaToCdsMaps.isEmpty())
1745 * Helper method that makes a CDS sequence as defined by the mappings from the
1746 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1747 * forward or reverse strand).
1751 * @return CDS sequence (as a dataset sequence)
1753 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping)
1755 char[] seqChars = seq.getSequence();
1756 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1757 int cdsWidth = MappingUtils.getLength(fromRanges);
1758 char[] newSeqChars = new char[cdsWidth];
1761 for (int[] range : fromRanges)
1763 if (range[0] <= range[1])
1765 // forward strand mapping - just copy the range
1766 int length = range[1] - range[0] + 1;
1767 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1773 // reverse strand mapping - copy and complement one by one
1774 for (int i = range[0]; i >= range[1]; i--)
1776 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1782 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
1783 * else generate a sequence name
1785 String mapFromId = mapping.getMappedFromId();
1786 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
1787 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
1788 // newSeq.setDescription(mapFromId);
1794 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
1795 * feature start/end ranges, optionally omitting specified feature types.
1796 * Returns the number of features copied.
1801 * if not null, only features of this type are copied (including
1802 * subtypes in the Sequence Ontology)
1804 * the mapping from 'fromSeq' to 'toSeq'
1807 public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
1808 MapList mapping, String select, String... omitting)
1810 SequenceI copyTo = toSeq;
1811 while (copyTo.getDatasetSequence() != null)
1813 copyTo = copyTo.getDatasetSequence();
1816 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
1818 SequenceFeature[] sfs = fromSeq.getSequenceFeatures();
1821 for (SequenceFeature sf : sfs)
1823 String type = sf.getType();
1824 if (select != null && !so.isA(type, select))
1828 boolean omit = false;
1829 for (String toOmit : omitting)
1831 if (type.equals(toOmit))
1842 * locate the mapped range - null if either start or end is
1843 * not mapped (no partial overlaps are calculated)
1845 int start = sf.getBegin();
1846 int end = sf.getEnd();
1847 int[] mappedTo = mapping.locateInTo(start, end);
1849 * if whole exon range doesn't map, try interpreting it
1850 * as 5' or 3' exon overlapping the CDS range
1852 if (mappedTo == null)
1854 mappedTo = mapping.locateInTo(end, end);
1855 if (mappedTo != null)
1858 * end of exon is in CDS range - 5' overlap
1859 * to a range from the start of the peptide
1864 if (mappedTo == null)
1866 mappedTo = mapping.locateInTo(start, start);
1867 if (mappedTo != null)
1870 * start of exon is in CDS range - 3' overlap
1871 * to a range up to the end of the peptide
1873 mappedTo[1] = toSeq.getLength();
1876 if (mappedTo != null)
1878 SequenceFeature copy = new SequenceFeature(sf);
1879 copy.setBegin(Math.min(mappedTo[0], mappedTo[1]));
1880 copy.setEnd(Math.max(mappedTo[0], mappedTo[1]));
1881 copyTo.addSequenceFeature(copy);
1890 * Returns a mapping from dna to protein by inspecting sequence features of
1891 * type "CDS" on the dna.
1897 public static MapList mapCdsToProtein(SequenceI dnaSeq,
1898 SequenceI proteinSeq)
1900 List<int[]> ranges = findCdsPositions(dnaSeq);
1901 int mappedDnaLength = MappingUtils.getLength(ranges);
1903 int proteinLength = proteinSeq.getLength();
1904 int proteinStart = proteinSeq.getStart();
1905 int proteinEnd = proteinSeq.getEnd();
1908 * incomplete start codon may mean X at start of peptide
1909 * we ignore both for mapping purposes
1911 if (proteinSeq.getCharAt(0) == 'X')
1913 // todo JAL-2022 support startPhase > 0
1917 List<int[]> proteinRange = new ArrayList<int[]>();
1920 * dna length should map to protein (or protein plus stop codon)
1922 int codesForResidues = mappedDnaLength / 3;
1923 if (codesForResidues == (proteinLength + 1))
1925 // assuming extra codon is for STOP and not in peptide
1928 if (codesForResidues == proteinLength)
1930 proteinRange.add(new int[] { proteinStart, proteinEnd });
1931 return new MapList(ranges, proteinRange, 3, 1);
1937 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
1938 * start/end positions of sequence features of type "CDS" (or a sub-type of
1939 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
1940 * position order, so this method is only valid for linear CDS in the same
1941 * sense as the protein product.
1946 public static List<int[]> findCdsPositions(SequenceI dnaSeq)
1948 List<int[]> result = new ArrayList<int[]>();
1949 SequenceFeature[] sfs = dnaSeq.getSequenceFeatures();
1955 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
1958 for (SequenceFeature sf : sfs)
1961 * process a CDS feature (or a sub-type of CDS)
1963 if (so.isA(sf.getType(), SequenceOntologyI.CDS))
1968 phase = Integer.parseInt(sf.getPhase());
1969 } catch (NumberFormatException e)
1974 * phase > 0 on first codon means 5' incomplete - skip to the start
1975 * of the next codon; example ENST00000496384
1977 int begin = sf.getBegin();
1978 int end = sf.getEnd();
1979 if (result.isEmpty())
1984 // shouldn't happen!
1986 .println("Error: start phase extends beyond start CDS in "
1987 + dnaSeq.getName());
1990 result.add(new int[] { begin, end });
1995 * remove 'startPhase' positions (usually 0) from the first range
1996 * so we begin at the start of a complete codon
1998 if (!result.isEmpty())
2000 // TODO JAL-2022 correctly model start phase > 0
2001 result.get(0)[0] += startPhase;
2005 * Finally sort ranges by start position. This avoids a dependency on
2006 * keeping features in order on the sequence (if they are in order anyway,
2007 * the sort will have almost no work to do). The implicit assumption is CDS
2008 * ranges are assembled in order. Other cases should not use this method,
2009 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2011 Collections.sort(result, new Comparator<int[]>()
2014 public int compare(int[] o1, int[] o2)
2016 return Integer.compare(o1[0], o2[0]);
2023 * Maps exon features from dna to protein, and computes variants in peptide
2024 * product generated by variants in dna, and adds them as sequence_variant
2025 * features on the protein sequence. Returns the number of variant features
2030 * @param dnaToProtein
2032 public static int computeProteinFeatures(SequenceI dnaSeq,
2033 SequenceI peptide, MapList dnaToProtein)
2035 while (dnaSeq.getDatasetSequence() != null)
2037 dnaSeq = dnaSeq.getDatasetSequence();
2039 while (peptide.getDatasetSequence() != null)
2041 peptide = peptide.getDatasetSequence();
2044 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2047 * compute protein variants from dna variants and codon mappings;
2048 * NB - alternatively we could retrieve this using the REST service e.g.
2049 * http://rest.ensembl.org/overlap/translation
2050 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2051 * which would be a bit slower but possibly more reliable
2055 * build a map with codon variations for each potentially varying peptide
2057 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2058 dnaSeq, dnaToProtein);
2061 * scan codon variations, compute peptide variants and add to peptide sequence
2064 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2066 int peptidePos = variant.getKey();
2067 List<DnaVariant>[] codonVariants = variant.getValue();
2068 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2072 * sort to get sequence features in start position order
2073 * - would be better to store in Sequence as a TreeSet or NCList?
2075 if (peptide.getSequenceFeatures() != null)
2077 Arrays.sort(peptide.getSequenceFeatures(),
2078 new Comparator<SequenceFeature>()
2081 public int compare(SequenceFeature o1, SequenceFeature o2)
2083 int c = Integer.compare(o1.getBegin(), o2.getBegin());
2084 return c == 0 ? Integer.compare(o1.getEnd(), o2.getEnd())
2093 * Computes non-synonymous peptide variants from codon variants and adds them
2094 * as sequence_variant features on the protein sequence (one feature per
2095 * allele variant). Selected attributes (variant id, clinical significance)
2096 * are copied over to the new features.
2099 * the protein sequence
2101 * the position to compute peptide variants for
2102 * @param codonVariants
2103 * a list of dna variants per codon position
2104 * @return the number of features added
2106 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2107 List<DnaVariant>[] codonVariants)
2109 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2111 String base1 = codonVariants[0].get(0).base;
2112 String base2 = codonVariants[1].get(0).base;
2113 String base3 = codonVariants[2].get(0).base;
2116 * variants in first codon base
2118 for (DnaVariant var : codonVariants[0])
2120 if (var.variant != null)
2122 String alleles = (String) var.variant.getValue("alleles");
2123 if (alleles != null)
2125 for (String base : alleles.split(","))
2127 String codon = base + base2 + base3;
2128 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2138 * variants in second codon base
2140 for (DnaVariant var : codonVariants[1])
2142 if (var.variant != null)
2144 String alleles = (String) var.variant.getValue("alleles");
2145 if (alleles != null)
2147 for (String base : alleles.split(","))
2149 String codon = base1 + base + base3;
2150 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2160 * variants in third codon base
2162 for (DnaVariant var : codonVariants[2])
2164 if (var.variant != null)
2166 String alleles = (String) var.variant.getValue("alleles");
2167 if (alleles != null)
2169 for (String base : alleles.split(","))
2171 String codon = base1 + base2 + base;
2172 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2185 * Helper method that adds a peptide variant feature, provided the given codon
2186 * translates to a value different to the current residue (is a non-synonymous
2187 * variant). ID and clinical_significance attributes of the dna variant (if
2188 * present) are copied to the new feature.
2195 * @return true if a feature was added, else false
2197 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2198 String residue, DnaVariant var, String codon)
2201 * get peptide translation of codon e.g. GAT -> D
2202 * note that variants which are not single alleles,
2203 * e.g. multibase variants or HGMD_MUTATION etc
2204 * are currently ignored here
2206 String trans = codon.contains("-") ? "-"
2207 : (codon.length() > 3 ? null : ResidueProperties
2208 .codonTranslate(codon));
2209 if (trans != null && !trans.equals(residue))
2211 String residue3Char = StringUtils
2212 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2213 String trans3Char = StringUtils
2214 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2215 String desc = "p." + residue3Char + peptidePos + trans3Char;
2216 // set score to 0f so 'graduated colour' option is offered! JAL-2060
2217 SequenceFeature sf = new SequenceFeature(
2218 SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
2219 peptidePos, 0f, "Jalview");
2220 StringBuilder attributes = new StringBuilder(32);
2221 String id = (String) var.variant.getValue(ID);
2224 if (id.startsWith(SEQUENCE_VARIANT))
2226 id = id.substring(SEQUENCE_VARIANT.length());
2228 sf.setValue(ID, id);
2229 attributes.append(ID).append("=").append(id);
2230 // TODO handle other species variants
2231 StringBuilder link = new StringBuilder(32);
2234 link.append(desc).append(" ").append(id)
2235 .append("|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2236 .append(URLEncoder.encode(id, "UTF-8"));
2237 sf.addLink(link.toString());
2238 } catch (UnsupportedEncodingException e)
2243 String clinSig = (String) var.variant
2244 .getValue(CLINICAL_SIGNIFICANCE);
2245 if (clinSig != null)
2247 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2248 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2251 peptide.addSequenceFeature(sf);
2252 if (attributes.length() > 0)
2254 sf.setAttributes(attributes.toString());
2262 * Builds a map whose key is position in the protein sequence, and value is a
2263 * list of the base and all variants for each corresponding codon position
2266 * @param dnaToProtein
2269 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2270 SequenceI dnaSeq, MapList dnaToProtein)
2273 * map from peptide position to all variants of the codon which codes for it
2274 * LinkedHashMap ensures we keep the peptide features in sequence order
2276 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
2277 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2279 SequenceFeature[] dnaFeatures = dnaSeq.getSequenceFeatures();
2280 if (dnaFeatures == null)
2285 int dnaStart = dnaSeq.getStart();
2286 int[] lastCodon = null;
2287 int lastPeptidePostion = 0;
2290 * build a map of codon variations for peptides
2292 for (SequenceFeature sf : dnaFeatures)
2294 int dnaCol = sf.getBegin();
2295 if (dnaCol != sf.getEnd())
2297 // not handling multi-locus variant features
2300 if (so.isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
2302 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2305 // feature doesn't lie within coding region
2308 int peptidePosition = mapsTo[0];
2309 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2310 if (codonVariants == null)
2312 codonVariants = new ArrayList[3];
2313 codonVariants[0] = new ArrayList<DnaVariant>();
2314 codonVariants[1] = new ArrayList<DnaVariant>();
2315 codonVariants[2] = new ArrayList<DnaVariant>();
2316 variants.put(peptidePosition, codonVariants);
2320 * extract dna variants to a string array
2322 String alls = (String) sf.getValue("alleles");
2327 String[] alleles = alls.toUpperCase().split(",");
2329 for (String allele : alleles)
2331 alleles[i++] = allele.trim(); // lose any space characters "A, G"
2335 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2337 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2338 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2339 peptidePosition, peptidePosition));
2340 lastPeptidePostion = peptidePosition;
2344 * save nucleotide (and any variant) for each codon position
2346 for (int codonPos = 0; codonPos < 3; codonPos++)
2348 String nucleotide = String.valueOf(
2349 dnaSeq.getCharAt(codon[codonPos] - dnaStart))
2351 List<DnaVariant> codonVariant = codonVariants[codonPos];
2352 if (codon[codonPos] == dnaCol)
2354 if (!codonVariant.isEmpty()
2355 && codonVariant.get(0).variant == null)
2358 * already recorded base value, add this variant
2360 codonVariant.get(0).variant = sf;
2365 * add variant with base value
2367 codonVariant.add(new DnaVariant(nucleotide, sf));
2370 else if (codonVariant.isEmpty())
2373 * record (possibly non-varying) base value
2375 codonVariant.add(new DnaVariant(nucleotide));
2384 * Makes an alignment with a copy of the given sequences, adding in any
2385 * non-redundant sequences which are mapped to by the cross-referenced
2391 * the alignment dataset shared by the new copy
2394 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2395 SequenceI[] xrefs, AlignmentI dataset)
2397 AlignmentI copy = new Alignment(new Alignment(seqs));
2398 copy.setDataset(dataset);
2400 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2403 for (SequenceI xref : xrefs)
2405 DBRefEntry[] dbrefs = xref.getDBRefs();
2408 for (DBRefEntry dbref : dbrefs)
2410 if (dbref.getMap() == null || dbref.getMap().getTo() == null)
2414 SequenceI mappedTo = dbref.getMap().getTo();
2415 SequenceI match = matcher.findIdMatch(mappedTo);
2418 matcher.add(mappedTo);
2419 copy.addSequence(mappedTo);
2429 * Try to align sequences in 'unaligned' to match the alignment of their
2430 * mapped regions in 'aligned'. For example, could use this to align CDS
2431 * sequences which are mapped to their parent cDNA sequences.
2433 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2434 * dna-to-protein or protein-to-dna use alternative methods.
2437 * sequences to be aligned
2439 * holds aligned sequences and their mappings
2442 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2444 List<SequenceI> unmapped = new ArrayList<SequenceI>();
2445 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2446 unaligned, aligned, unmapped);
2447 int width = columnMap.size();
2448 char gap = unaligned.getGapCharacter();
2449 int realignedCount = 0;
2451 for (SequenceI seq : unaligned.getSequences())
2453 if (!unmapped.contains(seq))
2455 char[] newSeq = new char[width];
2456 Arrays.fill(newSeq, gap);
2461 * traverse the map to find columns populated
2464 for (Integer column : columnMap.keySet())
2466 Character c = columnMap.get(column).get(seq);
2470 * sequence has a character at this position
2480 * trim trailing gaps
2482 if (lastCol < width)
2484 char[] tmp = new char[lastCol + 1];
2485 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2488 seq.setSequence(String.valueOf(newSeq));
2492 return realignedCount;
2496 * Returns a map whose key is alignment column number (base 1), and whose
2497 * values are a map of sequence characters in that column.
2504 static Map<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2505 AlignmentI unaligned, AlignmentI aligned, List<SequenceI> unmapped)
2508 * Map will hold, for each aligned column position, a map of
2509 * {unalignedSequence, characterPerSequence} at that position.
2510 * TreeMap keeps the entries in ascending column order.
2512 Map<Integer, Map<SequenceI, Character>> map = new TreeMap<Integer, Map<SequenceI, Character>>();
2515 * record any sequences that have no mapping so can't be realigned
2517 unmapped.addAll(unaligned.getSequences());
2519 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2521 for (SequenceI seq : unaligned.getSequences())
2523 for (AlignedCodonFrame mapping : mappings)
2525 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2526 if (fromSeq != null)
2528 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2529 if (addMappedPositions(seq, fromSeq, seqMap, map))
2531 unmapped.remove(seq);
2540 * Helper method that adds to a map the mapped column positions of a sequence. <br>
2541 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2542 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2546 * the sequence whose column positions we are recording
2548 * a sequence that is mapped to the first sequence
2550 * the mapping from 'fromSeq' to 'seq'
2552 * a map to add the column positions (in fromSeq) of the mapped
2556 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2557 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2565 * invert mapping if it is from unaligned to aligned sequence
2567 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2569 seqMap = new Mapping(seq.getDatasetSequence(), seqMap.getMap()
2573 char[] fromChars = fromSeq.getSequence();
2574 int toStart = seq.getStart();
2575 char[] toChars = seq.getSequence();
2578 * traverse [start, end, start, end...] ranges in fromSeq
2580 for (int[] fromRange : seqMap.getMap().getFromRanges())
2582 for (int i = 0; i < fromRange.length - 1; i += 2)
2584 boolean forward = fromRange[i + 1] >= fromRange[i];
2587 * find the range mapped to (sequence positions base 1)
2589 int[] range = seqMap.locateMappedRange(fromRange[i],
2593 System.err.println("Error in mapping " + seqMap + " from "
2594 + fromSeq.getName());
2597 int fromCol = fromSeq.findIndex(fromRange[i]);
2598 int mappedCharPos = range[0];
2601 * walk over the 'from' aligned sequence in forward or reverse
2602 * direction; when a non-gap is found, record the column position
2603 * of the next character of the mapped-to sequence; stop when all
2604 * the characters of the range have been counted
2606 while (mappedCharPos <= range[1] && fromCol <= fromChars.length
2609 if (!Comparison.isGap(fromChars[fromCol - 1]))
2612 * mapped from sequence has a character in this column
2613 * record the column position for the mapped to character
2615 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2616 if (seqsMap == null)
2618 seqsMap = new HashMap<SequenceI, Character>();
2619 map.put(fromCol, seqsMap);
2621 seqsMap.put(seq, toChars[mappedCharPos - toStart]);
2624 fromCol += (forward ? 1 : -1);
2631 // strictly temporary hack until proper criteria for aligning protein to cds
2632 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2633 public static boolean looksLikeEnsembl(AlignmentI alignment)
2635 for (SequenceI seq : alignment.getSequences())
2637 String name = seq.getName();
2638 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))