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.Collections;
27 import java.util.HashMap;
28 import java.util.HashSet;
29 import java.util.Iterator;
30 import java.util.LinkedHashMap;
31 import java.util.List;
32 import java.util.Locale;
34 import java.util.Map.Entry;
35 import java.util.NoSuchElementException;
37 import java.util.SortedMap;
38 import java.util.TreeMap;
40 import jalview.bin.Console;
41 import jalview.commands.RemoveGapColCommand;
42 import jalview.datamodel.AlignedCodon;
43 import jalview.datamodel.AlignedCodonFrame;
44 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
45 import jalview.datamodel.Alignment;
46 import jalview.datamodel.AlignmentAnnotation;
47 import jalview.datamodel.AlignmentI;
48 import jalview.datamodel.ContactMatrixI;
49 import jalview.datamodel.DBRefEntry;
50 import jalview.datamodel.GeneLociI;
51 import jalview.datamodel.IncompleteCodonException;
52 import jalview.datamodel.Mapping;
53 import jalview.datamodel.Sequence;
54 import jalview.datamodel.SequenceFeature;
55 import jalview.datamodel.SequenceGroup;
56 import jalview.datamodel.SequenceI;
57 import jalview.datamodel.features.SequenceFeatures;
58 import jalview.io.gff.SequenceOntologyI;
59 import jalview.schemes.ResidueProperties;
60 import jalview.util.Comparison;
61 import jalview.util.DBRefUtils;
62 import jalview.util.IntRangeComparator;
63 import jalview.util.MapList;
64 import jalview.util.MappingUtils;
67 * grab bag of useful alignment manipulation operations Expect these to be
68 * refactored elsewhere at some point.
73 public class AlignmentUtils
75 private static final int CODON_LENGTH = 3;
77 private static final String SEQUENCE_VARIANT = "sequence_variant:";
80 * the 'id' attribute is provided for variant features fetched from
81 * Ensembl using its REST service with JSON format
83 public static final String VARIANT_ID = "id";
86 * A data model to hold the 'normal' base value at a position, and an optional
87 * sequence variant feature
89 static final class DnaVariant
93 SequenceFeature variant;
95 DnaVariant(String nuc)
101 DnaVariant(String nuc, SequenceFeature var)
107 public String getSource()
109 return variant == null ? null : variant.getFeatureGroup();
113 * toString for aid in the debugger only
116 public String toString()
118 return base + ":" + (variant == null ? "" : variant.getDescription());
123 * given an existing alignment, create a new alignment including all, or up to
124 * flankSize additional symbols from each sequence's dataset sequence
130 public static AlignmentI expandContext(AlignmentI core, int flankSize)
132 List<SequenceI> sq = new ArrayList<>();
134 for (SequenceI s : core.getSequences())
136 SequenceI newSeq = s.deriveSequence();
137 final int newSeqStart = newSeq.getStart() - 1;
138 if (newSeqStart > maxoffset
139 && newSeq.getDatasetSequence().getStart() < s.getStart())
141 maxoffset = newSeqStart;
147 maxoffset = Math.min(maxoffset, flankSize);
151 * now add offset left and right to create an expanded alignment
153 for (SequenceI s : sq)
156 while (ds.getDatasetSequence() != null)
158 ds = ds.getDatasetSequence();
160 int s_end = s.findPosition(s.getStart() + s.getLength());
161 // find available flanking residues for sequence
162 int ustream_ds = s.getStart() - ds.getStart();
163 int dstream_ds = ds.getEnd() - s_end;
165 // build new flanked sequence
167 // compute gap padding to start of flanking sequence
168 int offset = maxoffset - ustream_ds;
170 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
173 if (flankSize < ustream_ds)
175 // take up to flankSize residues
176 offset = maxoffset - flankSize;
177 ustream_ds = flankSize;
179 if (flankSize <= dstream_ds)
181 dstream_ds = flankSize - 1;
184 // TODO use Character.toLowerCase to avoid creating String objects?
185 char[] upstream = new String(ds
186 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
187 .toLowerCase(Locale.ROOT).toCharArray();
188 char[] downstream = new String(
189 ds.getSequence(s_end - 1, s_end + dstream_ds))
190 .toLowerCase(Locale.ROOT).toCharArray();
191 char[] coreseq = s.getSequence();
192 char[] nseq = new char[offset + upstream.length + downstream.length
194 char c = core.getGapCharacter();
197 for (; p < offset; p++)
202 System.arraycopy(upstream, 0, nseq, p, upstream.length);
203 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
205 System.arraycopy(downstream, 0, nseq,
206 p + coreseq.length + upstream.length, downstream.length);
207 s.setSequence(new String(nseq));
208 s.setStart(s.getStart() - ustream_ds);
209 s.setEnd(s_end + downstream.length);
211 AlignmentI newAl = new jalview.datamodel.Alignment(
212 sq.toArray(new SequenceI[0]));
213 for (SequenceI s : sq)
215 if (s.getAnnotation() != null)
217 for (AlignmentAnnotation aa : s.getAnnotation())
219 aa.adjustForAlignment(); // JAL-1712 fix
220 newAl.addAnnotation(aa);
224 newAl.setDataset(core.getDataset());
229 * Returns the index (zero-based position) of a sequence in an alignment, or
236 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
240 for (SequenceI alSeq : al.getSequences())
253 * Returns a map of lists of sequences in the alignment, keyed by sequence
254 * name. For use in mapping between different alignment views of the same
257 * @see jalview.datamodel.AlignmentI#getSequencesByName()
259 public static Map<String, List<SequenceI>> getSequencesByName(
262 Map<String, List<SequenceI>> theMap = new LinkedHashMap<>();
263 for (SequenceI seq : al.getSequences())
265 String name = seq.getName();
268 List<SequenceI> seqs = theMap.get(name);
271 seqs = new ArrayList<>();
272 theMap.put(name, seqs);
281 * Build mapping of protein to cDNA alignment. Mappings are made between
282 * sequences where the cDNA translates to the protein sequence. Any new
283 * mappings are added to the protein alignment. Returns true if any mappings
284 * either already exist or were added, else false.
286 * @param proteinAlignment
287 * @param cdnaAlignment
290 public static boolean mapProteinAlignmentToCdna(
291 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
293 if (proteinAlignment == null || cdnaAlignment == null)
298 Set<SequenceI> mappedDna = new HashSet<>();
299 Set<SequenceI> mappedProtein = new HashSet<>();
302 * First pass - map sequences where cross-references exist. This include
303 * 1-to-many mappings to support, for example, variant cDNA.
305 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
306 cdnaAlignment, mappedDna, mappedProtein, true);
309 * Second pass - map sequences where no cross-references exist. This only
310 * does 1-to-1 mappings and assumes corresponding sequences are in the same
311 * order in the alignments.
313 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
314 mappedDna, mappedProtein, false);
315 return mappingPerformed;
319 * Make mappings between compatible sequences (where the cDNA translation
320 * matches the protein).
322 * @param proteinAlignment
323 * @param cdnaAlignment
325 * a set of mapped DNA sequences (to add to)
326 * @param mappedProtein
327 * a set of mapped Protein sequences (to add to)
329 * if true, only map sequences where xrefs exist
332 protected static boolean mapProteinToCdna(
333 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
334 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
337 boolean mappingExistsOrAdded = false;
338 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
339 for (SequenceI aaSeq : thisSeqs)
341 boolean proteinMapped = false;
342 AlignedCodonFrame acf = new AlignedCodonFrame();
344 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
347 * Always try to map if sequences have xref to each other; this supports
348 * variant cDNA or alternative splicing for a protein sequence.
350 * If no xrefs, try to map progressively, assuming that alignments have
351 * mappable sequences in corresponding order. These are not
352 * many-to-many, as that would risk mixing species with similar cDNA
355 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
361 * Don't map non-xrefd sequences more than once each. This heuristic
362 * allows us to pair up similar sequences in ordered alignments.
364 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
365 || mappedDna.contains(cdnaSeq)))
369 if (mappingExists(proteinAlignment.getCodonFrames(),
370 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
372 mappingExistsOrAdded = true;
376 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
379 acf.addMap(cdnaSeq, aaSeq, map);
380 mappingExistsOrAdded = true;
381 proteinMapped = true;
382 mappedDna.add(cdnaSeq);
383 mappedProtein.add(aaSeq);
389 proteinAlignment.addCodonFrame(acf);
392 return mappingExistsOrAdded;
396 * Answers true if the mappings include one between the given (dataset)
399 protected static boolean mappingExists(List<AlignedCodonFrame> mappings,
400 SequenceI aaSeq, SequenceI cdnaSeq)
402 if (mappings != null)
404 for (AlignedCodonFrame acf : mappings)
406 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
416 * Builds a mapping (if possible) of a cDNA to a protein sequence.
418 * <li>first checks if the cdna translates exactly to the protein
420 * <li>else checks for translation after removing a STOP codon</li>
421 * <li>else checks for translation after removing a START codon</li>
422 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
424 * Returns null if no mapping is determined.
427 * the aligned protein sequence
429 * the aligned cdna sequence
432 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
436 * Here we handle either dataset sequence set (desktop) or absent (applet).
437 * Use only the char[] form of the sequence to avoid creating possibly large
440 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
441 char[] aaSeqChars = proteinDataset != null
442 ? proteinDataset.getSequence()
443 : proteinSeq.getSequence();
444 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
445 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
446 : cdnaSeq.getSequence();
447 if (aaSeqChars == null || cdnaSeqChars == null)
453 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
455 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
456 int cdnaLength = cdnaSeqChars.length;
457 int cdnaStart = cdnaSeq.getStart();
458 int cdnaEnd = cdnaSeq.getEnd();
459 final int proteinStart = proteinSeq.getStart();
460 final int proteinEnd = proteinSeq.getEnd();
463 * If lengths don't match, try ignoring stop codon (if present)
465 if (cdnaLength != mappedLength && cdnaLength > 2)
467 String lastCodon = String.valueOf(cdnaSeqChars,
468 cdnaLength - CODON_LENGTH, CODON_LENGTH)
469 .toUpperCase(Locale.ROOT);
470 for (String stop : ResidueProperties.STOP_CODONS)
472 if (lastCodon.equals(stop))
474 cdnaEnd -= CODON_LENGTH;
475 cdnaLength -= CODON_LENGTH;
482 * If lengths still don't match, try ignoring start codon.
485 if (cdnaLength != mappedLength && cdnaLength > 2
486 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH)
487 .toUpperCase(Locale.ROOT)
488 .equals(ResidueProperties.START))
490 startOffset += CODON_LENGTH;
491 cdnaStart += CODON_LENGTH;
492 cdnaLength -= CODON_LENGTH;
495 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
498 * protein is translation of dna (+/- start/stop codons)
500 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
502 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
507 * translation failed - try mapping CDS annotated regions of dna
509 return mapCdsToProtein(cdnaSeq, proteinSeq);
513 * Test whether the given cdna sequence, starting at the given offset,
514 * translates to the given amino acid sequence, using the standard translation
515 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
517 * @param cdnaSeqChars
522 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
525 if (cdnaSeqChars == null || aaSeqChars == null)
531 int dnaPos = cdnaStart;
532 for (; dnaPos < cdnaSeqChars.length - 2
533 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
535 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
536 final String translated = ResidueProperties.codonTranslate(codon);
539 * allow * in protein to match untranslatable in dna
541 final char aaRes = aaSeqChars[aaPos];
542 if ((translated == null || ResidueProperties.STOP.equals(translated))
547 if (translated == null || !(aaRes == translated.charAt(0)))
550 // jalview.bin.Console.outPrintln(("Mismatch at " + i + "/" + aaResidue
552 // + codon + "(" + translated + ") != " + aaRes));
558 * check we matched all of the protein sequence
560 if (aaPos != aaSeqChars.length)
566 * check we matched all of the dna except
567 * for optional trailing STOP codon
569 if (dnaPos == cdnaSeqChars.length)
573 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
575 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
576 if (ResidueProperties.STOP
577 .equals(ResidueProperties.codonTranslate(codon)))
586 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
587 * currently assumes that we are aligning cDNA to match protein.
590 * the sequence to be realigned
592 * the alignment whose sequence alignment is to be 'copied'
594 * character string represent a gap in the realigned sequence
595 * @param preserveUnmappedGaps
596 * @param preserveMappedGaps
597 * @return true if the sequence was realigned, false if it could not be
599 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
600 String gap, boolean preserveMappedGaps,
601 boolean preserveUnmappedGaps)
604 * Get any mappings from the source alignment to the target (dataset)
607 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
608 // all mappings. Would it help to constrain this?
609 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
610 if (mappings == null || mappings.isEmpty())
616 * Locate the aligned source sequence whose dataset sequence is mapped. We
617 * just take the first match here (as we can't align like more than one
620 SequenceI alignFrom = null;
621 AlignedCodonFrame mapping = null;
622 for (AlignedCodonFrame mp : mappings)
624 alignFrom = mp.findAlignedSequence(seq, al);
625 if (alignFrom != null)
632 if (alignFrom == null)
636 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
637 preserveMappedGaps, preserveUnmappedGaps);
642 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
643 * match residues and codons. Flags control whether existing gaps in unmapped
644 * (intron) and mapped (exon) regions are preserved or not. Gaps between
645 * intron and exon are only retained if both flags are set.
652 * @param preserveUnmappedGaps
653 * @param preserveMappedGaps
655 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
656 AlignedCodonFrame mapping, String myGap, char sourceGap,
657 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
659 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
661 // aligned and dataset sequence positions, all base zero
665 int basesWritten = 0;
666 char myGapChar = myGap.charAt(0);
667 int ratio = myGap.length();
669 int fromOffset = alignFrom.getStart() - 1;
670 int toOffset = alignTo.getStart() - 1;
671 int sourceGapMappedLength = 0;
672 boolean inExon = false;
673 final int toLength = alignTo.getLength();
674 final int fromLength = alignFrom.getLength();
675 StringBuilder thisAligned = new StringBuilder(2 * toLength);
678 * Traverse the 'model' aligned sequence
680 for (int i = 0; i < fromLength; i++)
682 char sourceChar = alignFrom.getCharAt(i);
683 if (sourceChar == sourceGap)
685 sourceGapMappedLength += ratio;
690 * Found a non-gap character. Locate its mapped region if any.
693 // Note mapping positions are base 1, our sequence positions base 0
694 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
695 sourceDsPos + fromOffset);
696 if (mappedPos == null)
699 * unmapped position; treat like a gap
701 sourceGapMappedLength += ratio;
702 // jalview.bin.Console.errPrintln("Can't align: no codon mapping to
704 // + sourceDsPos + "(" + sourceChar + ")");
709 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
710 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
711 StringBuilder trailingCopiedGap = new StringBuilder();
714 * Copy dna sequence up to and including this codon. Optionally, include
715 * gaps before the codon starts (in introns) and/or after the codon starts
718 * Note this only works for 'linear' splicing, not reverse or interleaved.
719 * But then 'align dna as protein' doesn't make much sense otherwise.
721 int intronLength = 0;
722 while (basesWritten + toOffset < mappedCodonEnd
723 && thisSeqPos < toLength)
725 final char c = alignTo.getCharAt(thisSeqPos++);
729 int sourcePosition = basesWritten + toOffset;
730 if (sourcePosition < mappedCodonStart)
733 * Found an unmapped (intron) base. First add in any preceding gaps
736 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
738 thisAligned.append(trailingCopiedGap.toString());
739 intronLength += trailingCopiedGap.length();
740 trailingCopiedGap = new StringBuilder();
747 final boolean startOfCodon = sourcePosition == mappedCodonStart;
748 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
749 preserveUnmappedGaps, sourceGapMappedLength, inExon,
750 trailingCopiedGap.length(), intronLength, startOfCodon);
751 for (int k = 0; k < gapsToAdd; k++)
753 thisAligned.append(myGapChar);
755 sourceGapMappedLength = 0;
758 thisAligned.append(c);
759 trailingCopiedGap = new StringBuilder();
763 if (inExon && preserveMappedGaps)
765 trailingCopiedGap.append(myGapChar);
767 else if (!inExon && preserveUnmappedGaps)
769 trailingCopiedGap.append(myGapChar);
776 * At end of model aligned sequence. Copy any remaining target sequence, optionally
777 * including (intron) gaps.
779 while (thisSeqPos < toLength)
781 final char c = alignTo.getCharAt(thisSeqPos++);
782 if (c != myGapChar || preserveUnmappedGaps)
784 thisAligned.append(c);
786 sourceGapMappedLength--;
790 * finally add gaps to pad for any trailing source gaps or
791 * unmapped characters
793 if (preserveUnmappedGaps)
795 while (sourceGapMappedLength > 0)
797 thisAligned.append(myGapChar);
798 sourceGapMappedLength--;
803 * All done aligning, set the aligned sequence.
805 alignTo.setSequence(new String(thisAligned));
809 * Helper method to work out how many gaps to insert when realigning.
811 * @param preserveMappedGaps
812 * @param preserveUnmappedGaps
813 * @param sourceGapMappedLength
815 * @param trailingCopiedGap
816 * @param intronLength
817 * @param startOfCodon
820 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
821 boolean preserveUnmappedGaps, int sourceGapMappedLength,
822 boolean inExon, int trailingGapLength, int intronLength,
823 final boolean startOfCodon)
829 * Reached start of codon. Ignore trailing gaps in intron unless we are
830 * preserving gaps in both exon and intron. Ignore them anyway if the
831 * protein alignment introduces a gap at least as large as the intronic
834 if (inExon && !preserveMappedGaps)
836 trailingGapLength = 0;
838 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
840 trailingGapLength = 0;
844 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
848 if (intronLength + trailingGapLength <= sourceGapMappedLength)
850 gapsToAdd = sourceGapMappedLength - intronLength;
854 gapsToAdd = Math.min(
855 intronLength + trailingGapLength - sourceGapMappedLength,
863 * second or third base of codon; check for any gaps in dna
865 if (!preserveMappedGaps)
867 trailingGapLength = 0;
869 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
875 * Realigns the given protein to match the alignment of the dna, using codon
876 * mappings to translate aligned codon positions to protein residues.
879 * the alignment whose sequences are realigned by this method
881 * the dna alignment whose alignment we are 'copying'
882 * @return the number of sequences that were realigned
884 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
886 if (protein.isNucleotide() || !dna.isNucleotide())
889 .errPrintln("Wrong alignment type in alignProteinAsDna");
892 List<SequenceI> unmappedProtein = new ArrayList<>();
893 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
894 protein, dna, unmappedProtein);
895 return alignProteinAs(protein, alignedCodons, unmappedProtein);
899 * Realigns the given dna to match the alignment of the protein, using codon
900 * mappings to translate aligned peptide positions to codons.
902 * Always produces a padded CDS alignment.
905 * the alignment whose sequences are realigned by this method
907 * the protein alignment whose alignment we are 'copying'
908 * @return the number of sequences that were realigned
910 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
912 if (protein.isNucleotide() || !dna.isNucleotide())
915 .errPrintln("Wrong alignment type in alignProteinAsDna");
918 // todo: implement this
919 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
920 int alignedCount = 0;
921 int width = 0; // alignment width for padding CDS
922 for (SequenceI dnaSeq : dna.getSequences())
924 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
925 dna.getGapCharacter()))
929 width = Math.max(dnaSeq.getLength(), width);
933 for (SequenceI dnaSeq : dna.getSequences())
935 oldwidth = dnaSeq.getLength();
936 diff = width - oldwidth;
939 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
946 * Helper method to align (if possible) the dna sequence to match the
947 * alignment of a mapped protein sequence. This is currently limited to
948 * handling coding sequence only.
956 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
957 AlignmentI protein, List<AlignedCodonFrame> mappings,
960 SequenceI cdsDss = cdsSeq.getDatasetSequence();
964 .println("alignCdsSequenceAsProtein needs aligned sequence!");
968 List<AlignedCodonFrame> dnaMappings = MappingUtils
969 .findMappingsForSequence(cdsSeq, mappings);
970 for (AlignedCodonFrame mapping : dnaMappings)
972 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
975 final int peptideLength = peptide.getLength();
976 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
979 MapList mapList = map.getMap();
980 if (map.getTo() == peptide.getDatasetSequence())
982 mapList = mapList.getInverse();
984 final int cdsLength = cdsDss.getLength();
985 int mappedFromLength = MappingUtils
986 .getLength(mapList.getFromRanges());
987 int mappedToLength = MappingUtils
988 .getLength(mapList.getToRanges());
989 boolean addStopCodon = (cdsLength == mappedFromLength
990 * CODON_LENGTH + CODON_LENGTH)
991 || (peptide.getDatasetSequence()
992 .getLength() == mappedFromLength - 1);
993 if (cdsLength != mappedToLength && !addStopCodon)
995 jalview.bin.Console.errPrintln(String.format(
996 "Can't align cds as protein (length mismatch %d/%d): %s",
997 cdsLength, mappedToLength, cdsSeq.getName()));
1001 * pre-fill the aligned cds sequence with gaps
1003 char[] alignedCds = new char[peptideLength * CODON_LENGTH
1004 + (addStopCodon ? CODON_LENGTH : 0)];
1005 Arrays.fill(alignedCds, gapChar);
1008 * walk over the aligned peptide sequence and insert mapped
1009 * codons for residues in the aligned cds sequence
1011 int copiedBases = 0;
1012 int cdsStart = cdsDss.getStart();
1013 int proteinPos = peptide.getStart() - 1;
1016 for (int col = 0; col < peptideLength; col++)
1018 char residue = peptide.getCharAt(col);
1020 if (Comparison.isGap(residue))
1022 cdsCol += CODON_LENGTH;
1027 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1030 // e.g. incomplete start codon, X in peptide
1031 cdsCol += CODON_LENGTH;
1035 for (int j = codon[0]; j <= codon[1]; j++)
1037 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1038 alignedCds[cdsCol++] = mappedBase;
1046 * append stop codon if not mapped from protein,
1047 * closing it up to the end of the mapped sequence
1049 if (copiedBases == cdsLength - CODON_LENGTH)
1051 for (int i = alignedCds.length - 1; i >= 0; i--)
1053 if (!Comparison.isGap(alignedCds[i]))
1055 cdsCol = i + 1; // gap just after end of sequence
1059 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1061 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1064 cdsSeq.setSequence(new String(alignedCds));
1073 * Builds a map whose key is an aligned codon position (3 alignment column
1074 * numbers base 0), and whose value is a map from protein sequence to each
1075 * protein's peptide residue for that codon. The map generates an ordering of
1076 * the codons, and allows us to read off the peptides at each position in
1077 * order to assemble 'aligned' protein sequences.
1080 * the protein alignment
1082 * the coding dna alignment
1083 * @param unmappedProtein
1084 * any unmapped proteins are added to this list
1087 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1088 AlignmentI protein, AlignmentI dna,
1089 List<SequenceI> unmappedProtein)
1092 * maintain a list of any proteins with no mappings - these will be
1093 * rendered 'as is' in the protein alignment as we can't align them
1095 unmappedProtein.addAll(protein.getSequences());
1097 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1100 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1101 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1102 * comparator keeps the codon positions ordered.
1104 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<>(
1105 new CodonComparator());
1107 for (SequenceI dnaSeq : dna.getSequences())
1109 for (AlignedCodonFrame mapping : mappings)
1111 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1114 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1115 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1117 unmappedProtein.remove(prot);
1123 * Finally add any unmapped peptide start residues (e.g. for incomplete
1124 * codons) as if at the codon position before the second residue
1126 // TODO resolve JAL-2022 so this fudge can be removed
1127 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1128 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1130 return alignedCodons;
1134 * Scans for any protein mapped from position 2 (meaning unmapped start
1135 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1136 * preceding position in the alignment
1138 * @param alignedCodons
1139 * the codon-to-peptide map
1140 * @param mappedSequenceCount
1141 * the number of distinct sequences in the map
1143 protected static void addUnmappedPeptideStarts(
1144 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1145 int mappedSequenceCount)
1147 // TODO delete this ugly hack once JAL-2022 is resolved
1148 // i.e. we can model startPhase > 0 (incomplete start codon)
1150 List<SequenceI> sequencesChecked = new ArrayList<>();
1151 AlignedCodon lastCodon = null;
1152 Map<SequenceI, AlignedCodon> toAdd = new HashMap<>();
1154 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1157 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1160 SequenceI seq = sequenceCodon.getKey();
1161 if (sequencesChecked.contains(seq))
1165 sequencesChecked.add(seq);
1166 AlignedCodon codon = sequenceCodon.getValue();
1167 if (codon.peptideCol > 1)
1169 jalview.bin.Console.errPrintln(
1170 "Problem mapping protein with >1 unmapped start positions: "
1173 else if (codon.peptideCol == 1)
1176 * first position (peptideCol == 0) was unmapped - add it
1178 if (lastCodon != null)
1180 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1181 lastCodon.pos2, lastCodon.pos3,
1182 String.valueOf(seq.getCharAt(0)), 0);
1183 toAdd.put(seq, firstPeptide);
1188 * unmapped residue at start of alignment (no prior column) -
1189 * 'insert' at nominal codon [0, 0, 0]
1191 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1192 String.valueOf(seq.getCharAt(0)), 0);
1193 toAdd.put(seq, firstPeptide);
1196 if (sequencesChecked.size() == mappedSequenceCount)
1198 // no need to check past first mapped position in all sequences
1202 lastCodon = entry.getKey();
1206 * add any new codons safely after iterating over the map
1208 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1210 addCodonToMap(alignedCodons, startCodon.getValue(),
1211 startCodon.getKey());
1216 * Update the aligned protein sequences to match the codon alignments given in
1220 * @param alignedCodons
1221 * an ordered map of codon positions (columns), with sequence/peptide
1222 * values present in each column
1223 * @param unmappedProtein
1226 protected static int alignProteinAs(AlignmentI protein,
1227 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1228 List<SequenceI> unmappedProtein)
1231 * prefill peptide sequences with gaps
1233 int alignedWidth = alignedCodons.size();
1234 char[] gaps = new char[alignedWidth];
1235 Arrays.fill(gaps, protein.getGapCharacter());
1236 Map<SequenceI, char[]> peptides = new HashMap<>();
1237 for (SequenceI seq : protein.getSequences())
1239 if (!unmappedProtein.contains(seq))
1241 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1246 * Traverse the codons left to right (as defined by CodonComparator)
1247 * and insert peptides in each column where the sequence is mapped.
1248 * This gives a peptide 'alignment' where residues are aligned if their
1249 * corresponding codons occupy the same columns in the cdna alignment.
1252 for (AlignedCodon codon : alignedCodons.keySet())
1254 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1256 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1258 char residue = entry.getValue().product.charAt(0);
1259 peptides.get(entry.getKey())[column] = residue;
1265 * and finally set the constructed sequences
1267 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1269 entry.getKey().setSequence(new String(entry.getValue()));
1276 * Populate the map of aligned codons by traversing the given sequence
1277 * mapping, locating the aligned positions of mapped codons, and adding those
1278 * positions and their translation products to the map.
1281 * the aligned sequence we are mapping from
1283 * the sequence to be aligned to the codons
1285 * the gap character in the dna sequence
1287 * a mapping to a sequence translation
1288 * @param alignedCodons
1289 * the map we are building up
1291 static void addCodonPositions(SequenceI dna, SequenceI protein,
1292 char gapChar, Mapping seqMap,
1293 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1295 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1298 * add codon positions, and their peptide translations, to the alignment
1299 * map, while remembering the first codon mapped
1301 while (codons.hasNext())
1305 AlignedCodon codon = codons.next();
1306 addCodonToMap(alignedCodons, codon, protein);
1307 } catch (IncompleteCodonException e)
1309 // possible incomplete trailing codon - ignore
1310 } catch (NoSuchElementException e)
1312 // possibly peptide lacking STOP
1318 * Helper method to add a codon-to-peptide entry to the aligned codons map
1320 * @param alignedCodons
1324 protected static void addCodonToMap(
1325 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1326 AlignedCodon codon, SequenceI protein)
1328 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1329 if (seqProduct == null)
1331 seqProduct = new HashMap<>();
1332 alignedCodons.put(codon, seqProduct);
1334 seqProduct.put(protein, codon);
1338 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1339 * between at least one pair of sequences in the two alignments. Currently,
1342 * <li>One alignment must be nucleotide, and the other protein</li>
1343 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1344 * <li>Mappable means the nucleotide translation matches the protein
1346 * <li>The translation may ignore start and stop codons if present in the
1354 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1356 if (al1 == null || al2 == null || al1 == al2)
1362 * Require one nucleotide and one protein
1364 if (al1.isNucleotide() == al2.isNucleotide())
1366 // or if protein - check if alternate coding
1367 if (al1.isNucleotide())
1371 return check3diPeptideMapping(al1,al2);
1373 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1374 AlignmentI protein = dna == al1 ? al2 : al1;
1375 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1376 for (SequenceI dnaSeq : dna.getSequences())
1378 for (SequenceI proteinSeq : protein.getSequences())
1380 if (isMappable(dnaSeq, proteinSeq, mappings))
1388 public static boolean check3diPeptideMapping(AlignmentI al1, AlignmentI al2)
1390 if (al1.getHeight()!=al2.getHeight())
1394 for (SequenceI al1seq: al1.getSequences())
1396 boolean foundMappable = false;
1397 for (SequenceI al2seq:al2.getSequences())
1399 if (canBuild3diMapping(al1seq,al2seq))
1401 foundMappable = true;
1410 if (mappable == al1.getHeight())
1418 * exact name, start-end, and identical length non-gap sequences
1423 public static boolean canBuild3diMapping(SequenceI al1seq,
1426 if (!al1seq.getDisplayId(true)
1427 .equalsIgnoreCase(al2seq.getDisplayId(true))) {
1430 String s1 = AlignSeq
1431 .extractGaps(jalview.util.Comparison.GapChars,
1432 al1seq.getSequenceAsString());
1433 String s2 = AlignSeq.extractGaps(
1434 jalview.util.Comparison.GapChars,
1435 al2seq.getSequenceAsString());
1436 return s1.length()==s2.length();
1439 public static boolean map3diPeptideToProteinAligment(
1440 AlignmentI proteinAlignment, AlignmentI tdiAlignment)
1442 if (proteinAlignment==null || tdiAlignment==null)
1446 Set<SequenceI> mappedDna = new HashSet<>();
1447 Set<SequenceI> mappedProtein = new HashSet<>();
1450 * First pass - map sequences where cross-references exist. This include
1451 * 1-to-many mappings to support, for example, variant cDNA.
1453 boolean mappingPerformed = mapProteinToTdiAlignment(proteinAlignment,
1454 tdiAlignment, mappedDna, mappedProtein);
1455 return mappingPerformed;
1462 * Make mappings between compatible sequences (ids are identical, length of seqs are identical).
1464 * @param proteinAlignment
1465 * @param tdiAlignment
1466 * @param mappedTdiSeq
1467 * a set of mapped DNA sequences (to add to)
1468 * @param mappedProtein
1469 * a set of mapped Protein sequences (to add to)
1471 * if true, only map sequences where xrefs exist
1474 protected static boolean mapProteinToTdiAlignment(
1475 final AlignmentI proteinAlignment, final AlignmentI tdiAlignment,
1476 Set<SequenceI> mappedTdiSeq, Set<SequenceI> mappedProtein)
1478 boolean mappingExistsOrAdded = false;
1479 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
1480 for (SequenceI aaSeq : thisSeqs)
1482 boolean proteinMapped = false;
1483 AlignedCodonFrame acf = new AlignedCodonFrame();
1485 for (SequenceI cdnaSeq : tdiAlignment.getSequences())
1488 if (mappingExists(proteinAlignment.getCodonFrames(),
1489 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
1491 mappingExistsOrAdded = true;
1495 if (canBuild3diMapping(aaSeq, cdnaSeq))
1497 MapList map = new MapList(new int[] { aaSeq.getStart(),aaSeq.getEnd()},new int[] { cdnaSeq.getStart(),cdnaSeq.getEnd()},1,1);
1498 acf.addMap(cdnaSeq, aaSeq, map);
1499 mappingExistsOrAdded = true;
1500 proteinMapped = true;
1501 mappedTdiSeq.add(cdnaSeq);
1502 mappedProtein.add(aaSeq);
1508 proteinAlignment.addCodonFrame(acf);
1511 return mappingExistsOrAdded;
1516 * Returns true if the dna sequence is mapped, or could be mapped, to the
1524 protected static boolean isMappable(SequenceI dnaSeq,
1525 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1527 if (dnaSeq == null || proteinSeq == null)
1532 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1533 : dnaSeq.getDatasetSequence();
1534 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1536 : proteinSeq.getDatasetSequence();
1538 for (AlignedCodonFrame mapping : mappings)
1540 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1550 * Just try to make a mapping (it is not yet stored), test whether
1553 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1557 * Finds any reference annotations associated with the sequences in
1558 * sequenceScope, that are not already added to the alignment, and adds them
1559 * to the 'candidates' map. Also populates a lookup table of annotation
1560 * labels, keyed by calcId, for use in constructing tooltips or the like.
1562 * @param sequenceScope
1563 * the sequences to scan for reference annotations
1564 * @param labelForCalcId
1565 * (optional) map to populate with label for calcId
1567 * map to populate with annotations for sequence
1569 * the alignment to check for presence of annotations
1571 public static void findAddableReferenceAnnotations(
1572 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1573 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1576 if (sequenceScope == null)
1582 * For each sequence in scope, make a list of any annotations on the
1583 * underlying dataset sequence which are not already on the alignment.
1585 * Add to a map of { alignmentSequence, <List of annotations to add> }
1587 for (SequenceI seq : sequenceScope)
1589 SequenceI dataset = seq.getDatasetSequence();
1590 if (dataset == null)
1594 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1595 if (datasetAnnotations == null)
1599 final List<AlignmentAnnotation> result = new ArrayList<>();
1600 for (AlignmentAnnotation dsann : datasetAnnotations)
1603 * Find matching annotations on the alignment. If none is found, then
1604 * add this annotation to the list of 'addable' annotations for this
1607 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1608 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1609 boolean found = false;
1610 if (matchedAlignmentAnnotations != null)
1612 for (AlignmentAnnotation matched : matchedAlignmentAnnotations)
1614 if (dsann.description.equals(matched.description))
1624 if (labelForCalcId != null)
1626 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1631 * Save any addable annotations for this sequence
1633 if (!result.isEmpty())
1635 candidates.put(seq, result);
1641 * Adds annotations to the top of the alignment annotations, in the same order
1642 * as their related sequences. If you already have an annotation and want to
1643 * add it to a sequence in an alignment use {@code addReferenceAnnotationTo}
1645 * @param annotations
1646 * the annotations to add
1648 * the alignment to add them to
1649 * @param selectionGroup
1650 * current selection group - may be null, if provided then any added
1651 * annotation will be trimmed to just those columns in the selection
1654 public static void addReferenceAnnotations(
1655 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1656 final AlignmentI alignment, final SequenceGroup selectionGroup)
1658 for (SequenceI seq : annotations.keySet())
1660 for (AlignmentAnnotation ann : annotations.get(seq))
1662 addReferenceAnnotationTo(alignment, seq, ann, selectionGroup);
1668 * Make a copy of a reference annotation {@code ann} and add it to an
1669 * alignment sequence {@code seq} in {@code alignment}, optionally limited to
1670 * the extent of {@code selectionGroup}
1675 * @param selectionGroup
1676 * current selection group - may be null, if provided then any added
1677 * annotation will be trimmed to just those columns in the selection
1679 * @return annotation added to {@code seq and {@code alignment}
1681 public static AlignmentAnnotation addReferenceAnnotationTo(
1682 final AlignmentI alignment, final SequenceI seq,
1683 final AlignmentAnnotation ann, final SequenceGroup selectionGroup)
1685 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1687 int endRes = ann.annotations.length;
1688 if (selectionGroup != null)
1690 startRes = -1 + Math.min(seq.getEnd(), Math.max(seq.getStart(),
1691 seq.findPosition(selectionGroup.getStartRes())));
1692 endRes = -1 + Math.min(seq.getEnd(),
1693 seq.findPosition(selectionGroup.getEndRes()));
1696 copyAnn.restrict(startRes, endRes + 0);
1699 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1700 * original annotation is already on the sequence.
1702 if (!seq.hasAnnotation(ann))
1704 ContactMatrixI cm = seq.getDatasetSequence().getContactMatrixFor(ann);
1707 seq.addContactListFor(copyAnn, cm);
1709 seq.addAlignmentAnnotation(copyAnn);
1712 copyAnn.adjustForAlignment();
1713 // add to the alignment and set visible
1714 alignment.addAnnotation(copyAnn);
1715 copyAnn.visible = true;
1721 * Set visibility of alignment annotations of specified types (labels), for
1722 * specified sequences. This supports controls like "Show all secondary
1723 * structure", "Hide all Temp factor", etc.
1725 * @al the alignment to scan for annotations
1727 * the types (labels) of annotations to be updated
1728 * @param forSequences
1729 * if not null, only annotations linked to one of these sequences are
1730 * in scope for update; if null, acts on all sequence annotations
1732 * if this flag is true, 'types' is ignored (label not checked)
1734 * if true, set visibility on, else set off
1736 public static void showOrHideSequenceAnnotations(AlignmentI al,
1737 Collection<String> types, List<SequenceI> forSequences,
1738 boolean anyType, boolean doShow)
1740 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1743 for (AlignmentAnnotation aa : anns)
1745 if (anyType || types.contains(aa.label))
1747 if ((aa.sequenceRef != null) && (forSequences == null
1748 || forSequences.contains(aa.sequenceRef)))
1750 aa.visible = doShow;
1757 public static AlignmentAnnotation getFirstSequenceAnnotationOfType(
1758 AlignmentI al, int graphType)
1760 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1763 for (AlignmentAnnotation aa : anns)
1765 if (aa.sequenceRef != null && aa.graph == graphType)
1773 * Returns true if either sequence has a cross-reference to the other
1779 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1781 // Note: moved here from class CrossRef as the latter class has dependencies
1782 // not availability to the applet's classpath
1783 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1787 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1788 * that sequence name is structured as Source|AccessionId.
1794 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1796 if (seq1 == null || seq2 == null)
1800 String name = seq2.getName();
1801 final List<DBRefEntry> xrefs = seq1.getDBRefs();
1804 for (int ix = 0, nx = xrefs.size(); ix < nx; ix++)
1806 DBRefEntry xref = xrefs.get(ix);
1807 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1808 // case-insensitive test, consistent with DBRefEntry.equalRef()
1809 if (xrefName.equalsIgnoreCase(name))
1819 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1820 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1821 * added to the original alignment's dataset, which is shared by the new
1822 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1823 * added to the alignment dataset.
1826 * aligned nucleotide (dna or cds) sequences
1828 * the alignment dataset the sequences belong to
1830 * (optional) to restrict results to CDS that map to specified
1832 * @return an alignment whose sequences are the cds-only parts of the dna
1833 * sequences (or null if no mappings are found)
1835 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1836 AlignmentI dataset, SequenceI[] products)
1838 if (dataset == null || dataset.getDataset() != null)
1840 throw new IllegalArgumentException(
1841 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1843 List<SequenceI> foundSeqs = new ArrayList<>();
1844 List<SequenceI> cdsSeqs = new ArrayList<>();
1845 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1846 HashSet<SequenceI> productSeqs = null;
1847 if (products != null)
1849 productSeqs = new HashSet<>();
1850 for (SequenceI seq : products)
1852 productSeqs.add(seq.getDatasetSequence() == null ? seq
1853 : seq.getDatasetSequence());
1858 * Construct CDS sequences from mappings on the alignment dataset.
1860 * - find the protein product(s) mapped to from each dna sequence
1861 * - if the mapping covers the whole dna sequence (give or take start/stop
1862 * codon), take the dna as the CDS sequence
1863 * - else search dataset mappings for a suitable dna sequence, i.e. one
1864 * whose whole sequence is mapped to the protein
1865 * - if no sequence found, construct one from the dna sequence and mapping
1866 * (and add it to dataset so it is found if this is repeated)
1868 for (SequenceI dnaSeq : dna)
1870 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1871 : dnaSeq.getDatasetSequence();
1873 List<AlignedCodonFrame> seqMappings = MappingUtils
1874 .findMappingsForSequence(dnaSeq, mappings);
1875 for (AlignedCodonFrame mapping : seqMappings)
1877 List<Mapping> mappingsFromSequence = mapping
1878 .getMappingsFromSequence(dnaSeq);
1880 for (Mapping aMapping : mappingsFromSequence)
1882 MapList mapList = aMapping.getMap();
1883 if (mapList.getFromRatio() == 1)
1886 * not a dna-to-protein mapping (likely dna-to-cds)
1892 * skip if mapping is not to one of the target set of proteins
1894 SequenceI proteinProduct = aMapping.getTo();
1895 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1901 * try to locate the CDS from the dataset mappings;
1902 * guard against duplicate results (for the case that protein has
1903 * dbrefs to both dna and cds sequences)
1905 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1906 seqMappings, aMapping);
1909 if (!foundSeqs.contains(cdsSeq))
1911 foundSeqs.add(cdsSeq);
1912 SequenceI derivedSequence = cdsSeq.deriveSequence();
1913 cdsSeqs.add(derivedSequence);
1914 if (!dataset.getSequences().contains(cdsSeq))
1916 dataset.addSequence(cdsSeq);
1923 * didn't find mapped CDS sequence - construct it and add
1924 * its dataset sequence to the dataset
1926 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1927 dataset).deriveSequence();
1928 // cdsSeq has a name constructed as CDS|<dbref>
1929 // <dbref> will be either the accession for the coding sequence,
1930 // marked in the /via/ dbref to the protein product accession
1931 // or it will be the original nucleotide accession.
1932 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1934 cdsSeqs.add(cdsSeq);
1937 * build the mapping from CDS to protein
1939 List<int[]> cdsRange = Collections
1940 .singletonList(new int[]
1941 { cdsSeq.getStart(),
1942 cdsSeq.getLength() + cdsSeq.getStart() - 1 });
1943 MapList cdsToProteinMap = new MapList(cdsRange,
1944 mapList.getToRanges(), mapList.getFromRatio(),
1945 mapList.getToRatio());
1947 if (!dataset.getSequences().contains(cdsSeqDss))
1950 * if this sequence is a newly created one, add it to the dataset
1951 * and made a CDS to protein mapping (if sequence already exists,
1952 * CDS-to-protein mapping _is_ the transcript-to-protein mapping)
1954 dataset.addSequence(cdsSeqDss);
1955 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1956 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1960 * guard against duplicating the mapping if repeating this action
1962 if (!mappings.contains(cdsToProteinMapping))
1964 mappings.add(cdsToProteinMapping);
1968 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1969 proteinProduct, aMapping);
1971 * add another mapping from original 'from' range to CDS
1973 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1974 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1976 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1978 if (!mappings.contains(dnaToCdsMapping))
1980 mappings.add(dnaToCdsMapping);
1984 * transfer dna chromosomal loci (if known) to the CDS
1985 * sequence (via the mapping)
1987 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1988 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1991 * add DBRef with mapping from protein to CDS
1992 * (this enables Get Cross-References from protein alignment)
1993 * This is tricky because we can't have two DBRefs with the
1994 * same source and accession, so need a different accession for
1995 * the CDS from the dna sequence
1998 // specific use case:
1999 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
2000 // ENSG02, ENSG03, with transcripts and products similarly named.
2001 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
2003 // JBPNote: ?? can't actually create an example that demonstrates we
2005 // synthesize an xref.
2007 List<DBRefEntry> primrefs = dnaDss.getPrimaryDBRefs();
2008 for (int ip = 0, np = primrefs.size(); ip < np; ip++)
2010 DBRefEntry primRef = primrefs.get(ip);
2012 * create a cross-reference from CDS to the source sequence's
2013 * primary reference and vice versa
2015 String source = primRef.getSource();
2016 String version = primRef.getVersion();
2017 DBRefEntry cdsCrossRef = new DBRefEntry(source,
2018 source + ":" + version, primRef.getAccessionId());
2020 .setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
2021 cdsSeqDss.addDBRef(cdsCrossRef);
2023 dnaSeq.addDBRef(new DBRefEntry(source, version,
2024 cdsSeq.getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
2025 // problem here is that the cross-reference is synthesized -
2026 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
2028 // assuming cds version same as dna ?!?
2030 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
2033 proteinToCdsRef.setMap(
2034 new Mapping(cdsSeqDss, cdsToProteinMap.getInverse()));
2035 proteinProduct.addDBRef(proteinToCdsRef);
2038 * transfer any features on dna that overlap the CDS
2040 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
2041 SequenceOntologyI.CDS);
2046 AlignmentI cds = new Alignment(
2047 cdsSeqs.toArray(new SequenceI[cdsSeqs.size()]));
2048 cds.setDataset(dataset);
2054 * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
2055 * toSeq, mediated by the given mapping between the sequences
2058 * @param targetToFrom
2062 protected static void transferGeneLoci(SequenceI fromSeq,
2063 MapList targetToFrom, SequenceI targetSeq)
2065 if (targetSeq.getGeneLoci() != null)
2067 // already have - don't override
2070 GeneLociI fromLoci = fromSeq.getGeneLoci();
2071 if (fromLoci == null)
2076 MapList newMap = targetToFrom.traverse(fromLoci.getMapping());
2080 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
2081 fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
2086 * A helper method that finds a CDS sequence in the alignment dataset that is
2087 * mapped to the given protein sequence, and either is, or has a mapping from,
2088 * the given dna sequence.
2091 * set of all mappings on the dataset
2093 * a dna (or cds) sequence we are searching from
2094 * @param seqMappings
2095 * the set of mappings involving dnaSeq
2097 * a transcript-to-peptide mapping
2100 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
2101 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
2105 * TODO a better dna-cds-protein mapping data representation to allow easy
2106 * navigation; until then this clunky looping around lists of mappings
2108 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
2109 : dnaSeq.getDatasetSequence();
2110 SequenceI proteinProduct = aMapping.getTo();
2113 * is this mapping from the whole dna sequence (i.e. CDS)?
2114 * allowing for possible stop codon on dna but not peptide
2116 int mappedFromLength = MappingUtils
2117 .getLength(aMapping.getMap().getFromRanges());
2118 int dnaLength = seqDss.getLength();
2119 if (mappedFromLength == dnaLength
2120 || mappedFromLength == dnaLength - CODON_LENGTH)
2123 * if sequence has CDS features, this is a transcript with no UTR
2124 * - do not take this as the CDS sequence! (JAL-2789)
2126 if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
2134 * looks like we found the dna-to-protein mapping; search for the
2135 * corresponding cds-to-protein mapping
2137 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
2138 .findMappingsForSequence(proteinProduct, mappings);
2139 for (AlignedCodonFrame acf : mappingsToPeptide)
2141 for (SequenceToSequenceMapping map : acf.getMappings())
2143 Mapping mapping = map.getMapping();
2144 if (mapping != aMapping
2145 && mapping.getMap().getFromRatio() == CODON_LENGTH
2146 && proteinProduct == mapping.getTo()
2147 && seqDss != map.getFromSeq())
2149 mappedFromLength = MappingUtils
2150 .getLength(mapping.getMap().getFromRanges());
2151 if (mappedFromLength == map.getFromSeq().getLength())
2154 * found a 3:1 mapping to the protein product which covers
2155 * the whole dna sequence i.e. is from CDS; finally check the CDS
2156 * is mapped from the given dna start sequence
2158 SequenceI cdsSeq = map.getFromSeq();
2159 // todo this test is weak if seqMappings contains multiple mappings;
2160 // we get away with it if transcript:cds relationship is 1:1
2161 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
2162 .findMappingsForSequence(cdsSeq, seqMappings);
2163 if (!dnaToCdsMaps.isEmpty())
2175 * Helper method that makes a CDS sequence as defined by the mappings from the
2176 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
2177 * forward or reverse strand).
2182 * - existing dataset. We check for sequences that look like the CDS
2183 * we are about to construct, if one exists already, then we will
2184 * just return that one.
2185 * @return CDS sequence (as a dataset sequence)
2187 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
2191 * construct CDS sequence name as "CDS|" with 'from id' held in the mapping
2192 * if set (e.g. EMBL protein_id), else sequence name appended
2194 String mapFromId = mapping.getMappedFromId();
2195 final String seqId = "CDS|"
2196 + (mapFromId != null ? mapFromId : seq.getName());
2198 SequenceI newSeq = null;
2201 * construct CDS sequence by splicing mapped from ranges
2203 char[] seqChars = seq.getSequence();
2204 List<int[]> fromRanges = mapping.getMap().getFromRanges();
2205 int cdsWidth = MappingUtils.getLength(fromRanges);
2206 char[] newSeqChars = new char[cdsWidth];
2209 for (int[] range : fromRanges)
2211 if (range[0] <= range[1])
2213 // forward strand mapping - just copy the range
2214 int length = range[1] - range[0] + 1;
2215 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
2221 // reverse strand mapping - copy and complement one by one
2222 for (int i = range[0]; i >= range[1]; i--)
2224 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
2228 newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2231 if (dataset != null)
2233 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2234 if (matches != null)
2236 boolean matched = false;
2237 for (SequenceI mtch : matches)
2239 if (mtch.getStart() != newSeq.getStart())
2243 if (mtch.getEnd() != newSeq.getEnd())
2247 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2259 "JAL-2154 regression: warning - found (and ignored) a duplicate CDS sequence:"
2265 // newSeq.setDescription(mapFromId);
2271 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2272 * the given mapping.
2276 * @param proteinProduct
2278 * @return list of DBRefEntrys added
2280 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2281 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2284 // gather direct refs from contig congruent with mapping
2285 List<DBRefEntry> direct = new ArrayList<>();
2286 HashSet<String> directSources = new HashSet<>();
2288 List<DBRefEntry> refs = contig.getDBRefs();
2291 for (int ib = 0, nb = refs.size(); ib < nb; ib++)
2293 DBRefEntry dbr = refs.get(ib);
2295 if (dbr.hasMap() && (map = dbr.getMap().getMap()).isTripletMap())
2297 // check if map is the CDS mapping
2298 if (mapping.getMap().equals(map))
2301 directSources.add(dbr.getSource());
2306 List<DBRefEntry> onSource = DBRefUtils.selectRefs(
2307 proteinProduct.getDBRefs(),
2308 directSources.toArray(new String[0]));
2309 List<DBRefEntry> propagated = new ArrayList<>();
2311 // and generate appropriate mappings
2312 for (int ic = 0, nc = direct.size(); ic < nc; ic++)
2314 DBRefEntry cdsref = direct.get(ic);
2315 Mapping m = cdsref.getMap();
2316 // clone maplist and mapping
2317 MapList cdsposmap = new MapList(
2318 Arrays.asList(new int[][]
2319 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2320 m.getMap().getToRanges(), 3, 1);
2321 Mapping cdsmap = new Mapping(m.getTo(), m.getMap());
2324 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2325 cdsref.getVersion(), cdsref.getAccessionId(),
2326 new Mapping(cdsmap.getTo(), cdsposmap));
2328 // and see if we can map to the protein product for this mapping.
2329 // onSource is the filtered set of accessions on protein that we are
2330 // tranferring, so we assume accession is the same.
2331 if (cdsmap.getTo() == null && onSource != null)
2333 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2334 cdsref.getAccessionId());
2335 if (sourceRefs != null)
2337 for (DBRefEntry srcref : sourceRefs)
2339 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2341 // we have found a complementary dbref on the protein product, so
2342 // update mapping's getTo
2343 newref.getMap().setTo(proteinProduct);
2348 cdsSeq.addDBRef(newref);
2349 propagated.add(newref);
2355 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2356 * feature start/end ranges, optionally omitting specified feature types.
2357 * Returns the number of features copied.
2362 * the mapping from 'fromSeq' to 'toSeq'
2364 * if not null, only features of this type are copied (including
2365 * subtypes in the Sequence Ontology)
2368 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2369 MapList mapping, String select, String... omitting)
2371 SequenceI copyTo = toSeq;
2372 while (copyTo.getDatasetSequence() != null)
2374 copyTo = copyTo.getDatasetSequence();
2376 if (fromSeq == copyTo || fromSeq.getDatasetSequence() == copyTo)
2378 return 0; // shared dataset sequence
2382 * get features, optionally restricted by an ontology term
2384 List<SequenceFeature> sfs = select == null
2385 ? fromSeq.getFeatures().getPositionalFeatures()
2386 : fromSeq.getFeatures().getFeaturesByOntology(select);
2389 for (SequenceFeature sf : sfs)
2391 String type = sf.getType();
2392 boolean omit = false;
2393 for (String toOmit : omitting)
2395 if (type.equals(toOmit))
2406 * locate the mapped range - null if either start or end is
2407 * not mapped (no partial overlaps are calculated)
2409 int start = sf.getBegin();
2410 int end = sf.getEnd();
2411 int[] mappedTo = mapping.locateInTo(start, end);
2413 * if whole exon range doesn't map, try interpreting it
2414 * as 5' or 3' exon overlapping the CDS range
2416 if (mappedTo == null)
2418 mappedTo = mapping.locateInTo(end, end);
2419 if (mappedTo != null)
2422 * end of exon is in CDS range - 5' overlap
2423 * to a range from the start of the peptide
2428 if (mappedTo == null)
2430 mappedTo = mapping.locateInTo(start, start);
2431 if (mappedTo != null)
2434 * start of exon is in CDS range - 3' overlap
2435 * to a range up to the end of the peptide
2437 mappedTo[1] = toSeq.getLength();
2440 if (mappedTo != null)
2442 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2443 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2444 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2445 sf.getFeatureGroup(), sf.getScore());
2446 copyTo.addSequenceFeature(copy);
2454 * Returns a mapping from dna to protein by inspecting sequence features of
2455 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2456 * length is 3 times the peptide length (optionally after dropping a trailing
2457 * stop codon). This method does not check whether the CDS nucleotide sequence
2458 * translates to the peptide sequence.
2464 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2465 SequenceI proteinSeq)
2467 List<int[]> ranges = findCdsPositions(dnaSeq);
2468 int mappedDnaLength = MappingUtils.getLength(ranges);
2471 * if not a whole number of codons, truncate mapping
2473 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2474 if (codonRemainder > 0)
2476 mappedDnaLength -= codonRemainder;
2477 MappingUtils.removeEndPositions(codonRemainder, ranges);
2480 int proteinLength = proteinSeq.getLength();
2481 int proteinStart = proteinSeq.getStart();
2482 int proteinEnd = proteinSeq.getEnd();
2485 * incomplete start codon may mean X at start of peptide
2486 * we ignore both for mapping purposes
2488 if (proteinSeq.getCharAt(0) == 'X')
2490 // todo JAL-2022 support startPhase > 0
2494 List<int[]> proteinRange = new ArrayList<>();
2497 * dna length should map to protein (or protein plus stop codon)
2499 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2500 if (codesForResidues == (proteinLength + 1))
2502 // assuming extra codon is for STOP and not in peptide
2503 // todo: check trailing codon is indeed a STOP codon
2505 mappedDnaLength -= CODON_LENGTH;
2506 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2509 if (codesForResidues == proteinLength)
2511 proteinRange.add(new int[] { proteinStart, proteinEnd });
2512 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2518 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2519 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2520 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2521 * position order, so this method is only valid for linear CDS in the same
2522 * sense as the protein product.
2527 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2529 List<int[]> result = new ArrayList<>();
2531 List<SequenceFeature> sfs = dnaSeq.getFeatures()
2532 .getFeaturesByOntology(SequenceOntologyI.CDS);
2537 SequenceFeatures.sortFeatures(sfs, true);
2539 for (SequenceFeature sf : sfs)
2544 String s = sf.getPhase();
2547 phase = Integer.parseInt(s);
2549 } catch (NumberFormatException e)
2554 * phase > 0 on first codon means 5' incomplete - skip to the start
2555 * of the next codon; example ENST00000496384
2557 int begin = sf.getBegin();
2558 int end = sf.getEnd();
2559 if (result.isEmpty() && phase > 0)
2564 // shouldn't happen!
2566 .println("Error: start phase extends beyond start CDS in "
2567 + dnaSeq.getName());
2570 result.add(new int[] { begin, end });
2574 * Finally sort ranges by start position. This avoids a dependency on
2575 * keeping features in order on the sequence (if they are in order anyway,
2576 * the sort will have almost no work to do). The implicit assumption is CDS
2577 * ranges are assembled in order. Other cases should not use this method,
2578 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2580 Collections.sort(result, IntRangeComparator.ASCENDING);
2585 * Makes an alignment with a copy of the given sequences, adding in any
2586 * non-redundant sequences which are mapped to by the cross-referenced
2592 * the alignment dataset shared by the new copy
2595 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2596 SequenceI[] xrefs, AlignmentI dataset)
2598 AlignmentI copy = new Alignment(new Alignment(seqs));
2599 copy.setDataset(dataset);
2600 boolean isProtein = !copy.isNucleotide();
2601 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2604 // BH 2019.01.25 recoded to remove iterators
2606 for (int ix = 0, nx = xrefs.length; ix < nx; ix++)
2608 SequenceI xref = xrefs[ix];
2609 List<DBRefEntry> dbrefs = xref.getDBRefs();
2612 for (int ir = 0, nir = dbrefs.size(); ir < nir; ir++)
2614 DBRefEntry dbref = dbrefs.get(ir);
2615 Mapping map = dbref.getMap();
2617 if (map == null || (mto = map.getTo()) == null
2618 || mto.isProtein() != isProtein)
2622 SequenceI mappedTo = mto;
2623 SequenceI match = matcher.findIdMatch(mappedTo);
2626 matcher.add(mappedTo);
2627 copy.addSequence(mappedTo);
2637 * Try to align sequences in 'unaligned' to match the alignment of their
2638 * mapped regions in 'aligned'. For example, could use this to align CDS
2639 * sequences which are mapped to their parent cDNA sequences.
2641 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2642 * dna-to-protein or protein-to-dna use alternative methods.
2645 * sequences to be aligned
2647 * holds aligned sequences and their mappings
2650 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2653 * easy case - aligning a copy of aligned sequences
2655 if (alignAsSameSequences(unaligned, aligned))
2657 return unaligned.getHeight();
2661 * fancy case - aligning via mappings between sequences
2663 List<SequenceI> unmapped = new ArrayList<>();
2664 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2665 unaligned, aligned, unmapped);
2666 int width = columnMap.size();
2667 char gap = unaligned.getGapCharacter();
2668 int realignedCount = 0;
2669 // TODO: verify this loop scales sensibly for very wide/high alignments
2671 for (SequenceI seq : unaligned.getSequences())
2673 if (!unmapped.contains(seq))
2675 char[] newSeq = new char[width];
2676 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2677 // Integer iteration below
2682 * traverse the map to find columns populated
2685 for (Integer column : columnMap.keySet())
2687 Character c = columnMap.get(column).get(seq);
2691 * sequence has a character at this position
2701 * trim trailing gaps
2703 if (lastCol < width)
2705 char[] tmp = new char[lastCol + 1];
2706 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2709 // TODO: optimise SequenceI to avoid char[]->String->char[]
2710 seq.setSequence(String.valueOf(newSeq));
2714 return realignedCount;
2718 * If unaligned and aligned sequences share the same dataset sequences, then
2719 * simply copies the aligned sequences to the unaligned sequences and returns
2720 * true; else returns false
2723 * - sequences to be aligned based on aligned
2725 * - 'guide' alignment containing sequences derived from same dataset
2729 static boolean alignAsSameSequences(AlignmentI unaligned,
2732 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2734 return false; // should only pass alignments with datasets here
2737 // map from dataset sequence to alignment sequence(s)
2738 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2739 for (SequenceI seq : aligned.getSequences())
2741 SequenceI ds = seq.getDatasetSequence();
2742 if (alignedDatasets.get(ds) == null)
2744 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2746 alignedDatasets.get(ds).add(seq);
2750 * first pass - check whether all sequences to be aligned share a
2751 * dataset sequence with an aligned sequence; also note the leftmost
2752 * ungapped column from which to copy
2754 int leftmost = Integer.MAX_VALUE;
2755 for (SequenceI seq : unaligned.getSequences())
2757 final SequenceI ds = seq.getDatasetSequence();
2758 if (!alignedDatasets.containsKey(ds))
2762 SequenceI alignedSeq = alignedDatasets.get(ds).get(0);
2763 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2764 leftmost = Math.min(leftmost, startCol);
2768 * second pass - copy aligned sequences;
2769 * heuristic rule: pair off sequences in order for the case where
2770 * more than one shares the same dataset sequence
2772 final char gapCharacter = aligned.getGapCharacter();
2773 for (SequenceI seq : unaligned.getSequences())
2775 List<SequenceI> alignedSequences = alignedDatasets
2776 .get(seq.getDatasetSequence());
2777 if (alignedSequences.isEmpty())
2780 * defensive check - shouldn't happen! (JAL-3536)
2784 SequenceI alignedSeq = alignedSequences.get(0);
2787 * gap fill for leading (5') UTR if any
2789 // TODO this copies intron columns - wrong!
2790 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2791 int endCol = alignedSeq.findIndex(seq.getEnd());
2792 char[] seqchars = new char[endCol - leftmost + 1];
2793 Arrays.fill(seqchars, gapCharacter);
2794 char[] toCopy = alignedSeq.getSequence(startCol - 1, endCol);
2795 System.arraycopy(toCopy, 0, seqchars, startCol - leftmost,
2797 seq.setSequence(String.valueOf(seqchars));
2798 if (alignedSequences.size() > 0)
2800 // pop off aligned sequences (except the last one)
2801 alignedSequences.remove(0);
2806 * finally remove gapped columns (e.g. introns)
2808 new RemoveGapColCommand("", unaligned.getSequencesArray(), 0,
2809 unaligned.getWidth() - 1, unaligned);
2815 * Returns a map whose key is alignment column number (base 1), and whose
2816 * values are a map of sequence characters in that column.
2823 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2824 AlignmentI unaligned, AlignmentI aligned,
2825 List<SequenceI> unmapped)
2828 * Map will hold, for each aligned column position, a map of
2829 * {unalignedSequence, characterPerSequence} at that position.
2830 * TreeMap keeps the entries in ascending column order.
2832 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2835 * record any sequences that have no mapping so can't be realigned
2837 unmapped.addAll(unaligned.getSequences());
2839 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2841 for (SequenceI seq : unaligned.getSequences())
2843 for (AlignedCodonFrame mapping : mappings)
2845 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2846 if (fromSeq != null)
2848 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2849 if (addMappedPositions(seq, fromSeq, seqMap, map))
2851 unmapped.remove(seq);
2860 * Helper method that adds to a map the mapped column positions of a sequence.
2862 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2863 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2867 * the sequence whose column positions we are recording
2869 * a sequence that is mapped to the first sequence
2871 * the mapping from 'fromSeq' to 'seq'
2873 * a map to add the column positions (in fromSeq) of the mapped
2877 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2878 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2886 * invert mapping if it is from unaligned to aligned sequence
2888 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2890 seqMap = new Mapping(seq.getDatasetSequence(),
2891 seqMap.getMap().getInverse());
2894 int toStart = seq.getStart();
2897 * traverse [start, end, start, end...] ranges in fromSeq
2899 for (int[] fromRange : seqMap.getMap().getFromRanges())
2901 for (int i = 0; i < fromRange.length - 1; i += 2)
2903 boolean forward = fromRange[i + 1] >= fromRange[i];
2906 * find the range mapped to (sequence positions base 1)
2908 int[] range = seqMap.locateMappedRange(fromRange[i],
2912 jalview.bin.Console.errPrintln("Error in mapping " + seqMap
2913 + " from " + fromSeq.getName());
2916 int fromCol = fromSeq.findIndex(fromRange[i]);
2917 int mappedCharPos = range[0];
2920 * walk over the 'from' aligned sequence in forward or reverse
2921 * direction; when a non-gap is found, record the column position
2922 * of the next character of the mapped-to sequence; stop when all
2923 * the characters of the range have been counted
2925 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
2928 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
2931 * mapped from sequence has a character in this column
2932 * record the column position for the mapped to character
2934 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2935 if (seqsMap == null)
2937 seqsMap = new HashMap<>();
2938 map.put(fromCol, seqsMap);
2940 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
2943 fromCol += (forward ? 1 : -1);
2950 // strictly temporary hack until proper criteria for aligning protein to cds
2951 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2952 public static boolean looksLikeEnsembl(AlignmentI alignment)
2954 for (SequenceI seq : alignment.getSequences())
2956 String name = seq.getName();
2957 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))