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.api.AlignCalcWorkerI;
41 import jalview.bin.Console;
42 import jalview.commands.RemoveGapColCommand;
43 import jalview.datamodel.AlignedCodon;
44 import jalview.datamodel.AlignedCodonFrame;
45 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
46 import jalview.datamodel.Alignment;
47 import jalview.datamodel.AlignmentAnnotation;
48 import jalview.datamodel.AlignmentI;
49 import jalview.datamodel.ContactMatrixI;
50 import jalview.datamodel.DBRefEntry;
51 import jalview.datamodel.GeneLociI;
52 import jalview.datamodel.IncompleteCodonException;
53 import jalview.datamodel.Mapping;
54 import jalview.datamodel.Sequence;
55 import jalview.datamodel.SequenceFeature;
56 import jalview.datamodel.SequenceGroup;
57 import jalview.datamodel.SequenceI;
58 import jalview.datamodel.features.SequenceFeatures;
59 import jalview.gui.AlignmentPanel;
60 import jalview.io.gff.SequenceOntologyI;
61 import jalview.schemes.ResidueProperties;
62 import jalview.util.Comparison;
63 import jalview.util.DBRefUtils;
64 import jalview.util.IntRangeComparator;
65 import jalview.util.MapList;
66 import jalview.util.MappingUtils;
67 import jalview.workers.SecondaryStructureConsensusThread;
70 * grab bag of useful alignment manipulation operations Expect these to be
71 * refactored elsewhere at some point.
76 public class AlignmentUtils
78 private static final int CODON_LENGTH = 3;
80 private static final String SEQUENCE_VARIANT = "sequence_variant:";
83 private static final Map<String, String> SECONDARY_STRUCTURE_LABELS = new HashMap<>();
86 SECONDARY_STRUCTURE_LABELS.put("Secondary Structure", "3D Structures");
87 SECONDARY_STRUCTURE_LABELS.put("jnetpred", "JPred");
88 // Add other secondary structure labels here if needed
91 private static final String SS_ANNOTATION_LABEL = "Secondary Structure";
94 * the 'id' attribute is provided for variant features fetched from
95 * Ensembl using its REST service with JSON format
97 public static final String VARIANT_ID = "id";
100 * A data model to hold the 'normal' base value at a position, and an optional
101 * sequence variant feature
103 static final class DnaVariant
107 SequenceFeature variant;
109 DnaVariant(String nuc)
115 DnaVariant(String nuc, SequenceFeature var)
121 public String getSource()
123 return variant == null ? null : variant.getFeatureGroup();
127 * toString for aid in the debugger only
130 public String toString()
132 return base + ":" + (variant == null ? "" : variant.getDescription());
137 * given an existing alignment, create a new alignment including all, or up to
138 * flankSize additional symbols from each sequence's dataset sequence
144 public static AlignmentI expandContext(AlignmentI core, int flankSize)
146 List<SequenceI> sq = new ArrayList<>();
148 for (SequenceI s : core.getSequences())
150 SequenceI newSeq = s.deriveSequence();
151 final int newSeqStart = newSeq.getStart() - 1;
152 if (newSeqStart > maxoffset
153 && newSeq.getDatasetSequence().getStart() < s.getStart())
155 maxoffset = newSeqStart;
161 maxoffset = Math.min(maxoffset, flankSize);
165 * now add offset left and right to create an expanded alignment
167 for (SequenceI s : sq)
170 while (ds.getDatasetSequence() != null)
172 ds = ds.getDatasetSequence();
174 int s_end = s.findPosition(s.getStart() + s.getLength());
175 // find available flanking residues for sequence
176 int ustream_ds = s.getStart() - ds.getStart();
177 int dstream_ds = ds.getEnd() - s_end;
179 // build new flanked sequence
181 // compute gap padding to start of flanking sequence
182 int offset = maxoffset - ustream_ds;
184 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
187 if (flankSize < ustream_ds)
189 // take up to flankSize residues
190 offset = maxoffset - flankSize;
191 ustream_ds = flankSize;
193 if (flankSize <= dstream_ds)
195 dstream_ds = flankSize - 1;
198 // TODO use Character.toLowerCase to avoid creating String objects?
199 char[] upstream = new String(ds
200 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
201 .toLowerCase(Locale.ROOT).toCharArray();
202 char[] downstream = new String(
203 ds.getSequence(s_end - 1, s_end + dstream_ds))
204 .toLowerCase(Locale.ROOT).toCharArray();
205 char[] coreseq = s.getSequence();
206 char[] nseq = new char[offset + upstream.length + downstream.length
208 char c = core.getGapCharacter();
211 for (; p < offset; p++)
216 System.arraycopy(upstream, 0, nseq, p, upstream.length);
217 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
219 System.arraycopy(downstream, 0, nseq,
220 p + coreseq.length + upstream.length, downstream.length);
221 s.setSequence(new String(nseq));
222 s.setStart(s.getStart() - ustream_ds);
223 s.setEnd(s_end + downstream.length);
225 AlignmentI newAl = new jalview.datamodel.Alignment(
226 sq.toArray(new SequenceI[0]));
227 for (SequenceI s : sq)
229 if (s.getAnnotation() != null)
231 for (AlignmentAnnotation aa : s.getAnnotation())
233 aa.adjustForAlignment(); // JAL-1712 fix
234 newAl.addAnnotation(aa);
238 newAl.setDataset(core.getDataset());
243 * Returns the index (zero-based position) of a sequence in an alignment, or
250 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
254 for (SequenceI alSeq : al.getSequences())
267 * Returns a map of lists of sequences in the alignment, keyed by sequence
268 * name. For use in mapping between different alignment views of the same
271 * @see jalview.datamodel.AlignmentI#getSequencesByName()
273 public static Map<String, List<SequenceI>> getSequencesByName(
276 Map<String, List<SequenceI>> theMap = new LinkedHashMap<>();
277 for (SequenceI seq : al.getSequences())
279 String name = seq.getName();
282 List<SequenceI> seqs = theMap.get(name);
285 seqs = new ArrayList<>();
286 theMap.put(name, seqs);
295 * Build mapping of protein to cDNA alignment. Mappings are made between
296 * sequences where the cDNA translates to the protein sequence. Any new
297 * mappings are added to the protein alignment. Returns true if any mappings
298 * either already exist or were added, else false.
300 * @param proteinAlignment
301 * @param cdnaAlignment
304 public static boolean mapProteinAlignmentToCdna(
305 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
307 if (proteinAlignment == null || cdnaAlignment == null)
312 Set<SequenceI> mappedDna = new HashSet<>();
313 Set<SequenceI> mappedProtein = new HashSet<>();
316 * First pass - map sequences where cross-references exist. This include
317 * 1-to-many mappings to support, for example, variant cDNA.
319 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
320 cdnaAlignment, mappedDna, mappedProtein, true);
323 * Second pass - map sequences where no cross-references exist. This only
324 * does 1-to-1 mappings and assumes corresponding sequences are in the same
325 * order in the alignments.
327 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
328 mappedDna, mappedProtein, false);
329 return mappingPerformed;
333 * Make mappings between compatible sequences (where the cDNA translation
334 * matches the protein).
336 * @param proteinAlignment
337 * @param cdnaAlignment
339 * a set of mapped DNA sequences (to add to)
340 * @param mappedProtein
341 * a set of mapped Protein sequences (to add to)
343 * if true, only map sequences where xrefs exist
346 protected static boolean mapProteinToCdna(
347 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
348 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
351 boolean mappingExistsOrAdded = false;
352 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
353 for (SequenceI aaSeq : thisSeqs)
355 boolean proteinMapped = false;
356 AlignedCodonFrame acf = new AlignedCodonFrame();
358 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
361 * Always try to map if sequences have xref to each other; this supports
362 * variant cDNA or alternative splicing for a protein sequence.
364 * If no xrefs, try to map progressively, assuming that alignments have
365 * mappable sequences in corresponding order. These are not
366 * many-to-many, as that would risk mixing species with similar cDNA
369 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
375 * Don't map non-xrefd sequences more than once each. This heuristic
376 * allows us to pair up similar sequences in ordered alignments.
378 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
379 || mappedDna.contains(cdnaSeq)))
383 if (mappingExists(proteinAlignment.getCodonFrames(),
384 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
386 mappingExistsOrAdded = true;
390 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
393 acf.addMap(cdnaSeq, aaSeq, map);
394 mappingExistsOrAdded = true;
395 proteinMapped = true;
396 mappedDna.add(cdnaSeq);
397 mappedProtein.add(aaSeq);
403 proteinAlignment.addCodonFrame(acf);
406 return mappingExistsOrAdded;
410 * Answers true if the mappings include one between the given (dataset)
413 protected static boolean mappingExists(List<AlignedCodonFrame> mappings,
414 SequenceI aaSeq, SequenceI cdnaSeq)
416 if (mappings != null)
418 for (AlignedCodonFrame acf : mappings)
420 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
430 * Builds a mapping (if possible) of a cDNA to a protein sequence.
432 * <li>first checks if the cdna translates exactly to the protein
434 * <li>else checks for translation after removing a STOP codon</li>
435 * <li>else checks for translation after removing a START codon</li>
436 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
438 * Returns null if no mapping is determined.
441 * the aligned protein sequence
443 * the aligned cdna sequence
446 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
450 * Here we handle either dataset sequence set (desktop) or absent (applet).
451 * Use only the char[] form of the sequence to avoid creating possibly large
454 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
455 char[] aaSeqChars = proteinDataset != null
456 ? proteinDataset.getSequence()
457 : proteinSeq.getSequence();
458 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
459 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
460 : cdnaSeq.getSequence();
461 if (aaSeqChars == null || cdnaSeqChars == null)
467 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
469 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
470 int cdnaLength = cdnaSeqChars.length;
471 int cdnaStart = cdnaSeq.getStart();
472 int cdnaEnd = cdnaSeq.getEnd();
473 final int proteinStart = proteinSeq.getStart();
474 final int proteinEnd = proteinSeq.getEnd();
477 * If lengths don't match, try ignoring stop codon (if present)
479 if (cdnaLength != mappedLength && cdnaLength > 2)
481 String lastCodon = String.valueOf(cdnaSeqChars,
482 cdnaLength - CODON_LENGTH, CODON_LENGTH)
483 .toUpperCase(Locale.ROOT);
484 for (String stop : ResidueProperties.STOP_CODONS)
486 if (lastCodon.equals(stop))
488 cdnaEnd -= CODON_LENGTH;
489 cdnaLength -= CODON_LENGTH;
496 * If lengths still don't match, try ignoring start codon.
499 if (cdnaLength != mappedLength && cdnaLength > 2
500 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH)
501 .toUpperCase(Locale.ROOT)
502 .equals(ResidueProperties.START))
504 startOffset += CODON_LENGTH;
505 cdnaStart += CODON_LENGTH;
506 cdnaLength -= CODON_LENGTH;
509 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
512 * protein is translation of dna (+/- start/stop codons)
514 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
516 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
521 * translation failed - try mapping CDS annotated regions of dna
523 return mapCdsToProtein(cdnaSeq, proteinSeq);
527 * Test whether the given cdna sequence, starting at the given offset,
528 * translates to the given amino acid sequence, using the standard translation
529 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
531 * @param cdnaSeqChars
536 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
539 if (cdnaSeqChars == null || aaSeqChars == null)
545 int dnaPos = cdnaStart;
546 for (; dnaPos < cdnaSeqChars.length - 2
547 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
549 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
550 final String translated = ResidueProperties.codonTranslate(codon);
553 * allow * in protein to match untranslatable in dna
555 final char aaRes = aaSeqChars[aaPos];
556 if ((translated == null || ResidueProperties.STOP.equals(translated))
561 if (translated == null || !(aaRes == translated.charAt(0)))
564 // jalview.bin.Console.outPrintln(("Mismatch at " + i + "/" + aaResidue
566 // + codon + "(" + translated + ") != " + aaRes));
572 * check we matched all of the protein sequence
574 if (aaPos != aaSeqChars.length)
580 * check we matched all of the dna except
581 * for optional trailing STOP codon
583 if (dnaPos == cdnaSeqChars.length)
587 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
589 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
590 if (ResidueProperties.STOP
591 .equals(ResidueProperties.codonTranslate(codon)))
600 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
601 * currently assumes that we are aligning cDNA to match protein.
604 * the sequence to be realigned
606 * the alignment whose sequence alignment is to be 'copied'
608 * character string represent a gap in the realigned sequence
609 * @param preserveUnmappedGaps
610 * @param preserveMappedGaps
611 * @return true if the sequence was realigned, false if it could not be
613 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
614 String gap, boolean preserveMappedGaps,
615 boolean preserveUnmappedGaps)
618 * Get any mappings from the source alignment to the target (dataset)
621 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
622 // all mappings. Would it help to constrain this?
623 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
624 if (mappings == null || mappings.isEmpty())
630 * Locate the aligned source sequence whose dataset sequence is mapped. We
631 * just take the first match here (as we can't align like more than one
634 SequenceI alignFrom = null;
635 AlignedCodonFrame mapping = null;
636 for (AlignedCodonFrame mp : mappings)
638 alignFrom = mp.findAlignedSequence(seq, al);
639 if (alignFrom != null)
646 if (alignFrom == null)
650 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
651 preserveMappedGaps, preserveUnmappedGaps);
656 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
657 * match residues and codons. Flags control whether existing gaps in unmapped
658 * (intron) and mapped (exon) regions are preserved or not. Gaps between
659 * intron and exon are only retained if both flags are set.
666 * @param preserveUnmappedGaps
667 * @param preserveMappedGaps
669 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
670 AlignedCodonFrame mapping, String myGap, char sourceGap,
671 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
673 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
675 // aligned and dataset sequence positions, all base zero
679 int basesWritten = 0;
680 char myGapChar = myGap.charAt(0);
681 int ratio = myGap.length();
683 int fromOffset = alignFrom.getStart() - 1;
684 int toOffset = alignTo.getStart() - 1;
685 int sourceGapMappedLength = 0;
686 boolean inExon = false;
687 final int toLength = alignTo.getLength();
688 final int fromLength = alignFrom.getLength();
689 StringBuilder thisAligned = new StringBuilder(2 * toLength);
692 * Traverse the 'model' aligned sequence
694 for (int i = 0; i < fromLength; i++)
696 char sourceChar = alignFrom.getCharAt(i);
697 if (sourceChar == sourceGap)
699 sourceGapMappedLength += ratio;
704 * Found a non-gap character. Locate its mapped region if any.
707 // Note mapping positions are base 1, our sequence positions base 0
708 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
709 sourceDsPos + fromOffset);
710 if (mappedPos == null)
713 * unmapped position; treat like a gap
715 sourceGapMappedLength += ratio;
716 // jalview.bin.Console.errPrintln("Can't align: no codon mapping to
718 // + sourceDsPos + "(" + sourceChar + ")");
723 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
724 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
725 StringBuilder trailingCopiedGap = new StringBuilder();
728 * Copy dna sequence up to and including this codon. Optionally, include
729 * gaps before the codon starts (in introns) and/or after the codon starts
732 * Note this only works for 'linear' splicing, not reverse or interleaved.
733 * But then 'align dna as protein' doesn't make much sense otherwise.
735 int intronLength = 0;
736 while (basesWritten + toOffset < mappedCodonEnd
737 && thisSeqPos < toLength)
739 final char c = alignTo.getCharAt(thisSeqPos++);
743 int sourcePosition = basesWritten + toOffset;
744 if (sourcePosition < mappedCodonStart)
747 * Found an unmapped (intron) base. First add in any preceding gaps
750 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
752 thisAligned.append(trailingCopiedGap.toString());
753 intronLength += trailingCopiedGap.length();
754 trailingCopiedGap = new StringBuilder();
761 final boolean startOfCodon = sourcePosition == mappedCodonStart;
762 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
763 preserveUnmappedGaps, sourceGapMappedLength, inExon,
764 trailingCopiedGap.length(), intronLength, startOfCodon);
765 for (int k = 0; k < gapsToAdd; k++)
767 thisAligned.append(myGapChar);
769 sourceGapMappedLength = 0;
772 thisAligned.append(c);
773 trailingCopiedGap = new StringBuilder();
777 if (inExon && preserveMappedGaps)
779 trailingCopiedGap.append(myGapChar);
781 else if (!inExon && preserveUnmappedGaps)
783 trailingCopiedGap.append(myGapChar);
790 * At end of model aligned sequence. Copy any remaining target sequence, optionally
791 * including (intron) gaps.
793 while (thisSeqPos < toLength)
795 final char c = alignTo.getCharAt(thisSeqPos++);
796 if (c != myGapChar || preserveUnmappedGaps)
798 thisAligned.append(c);
800 sourceGapMappedLength--;
804 * finally add gaps to pad for any trailing source gaps or
805 * unmapped characters
807 if (preserveUnmappedGaps)
809 while (sourceGapMappedLength > 0)
811 thisAligned.append(myGapChar);
812 sourceGapMappedLength--;
817 * All done aligning, set the aligned sequence.
819 alignTo.setSequence(new String(thisAligned));
823 * Helper method to work out how many gaps to insert when realigning.
825 * @param preserveMappedGaps
826 * @param preserveUnmappedGaps
827 * @param sourceGapMappedLength
829 * @param trailingCopiedGap
830 * @param intronLength
831 * @param startOfCodon
834 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
835 boolean preserveUnmappedGaps, int sourceGapMappedLength,
836 boolean inExon, int trailingGapLength, int intronLength,
837 final boolean startOfCodon)
843 * Reached start of codon. Ignore trailing gaps in intron unless we are
844 * preserving gaps in both exon and intron. Ignore them anyway if the
845 * protein alignment introduces a gap at least as large as the intronic
848 if (inExon && !preserveMappedGaps)
850 trailingGapLength = 0;
852 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
854 trailingGapLength = 0;
858 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
862 if (intronLength + trailingGapLength <= sourceGapMappedLength)
864 gapsToAdd = sourceGapMappedLength - intronLength;
868 gapsToAdd = Math.min(
869 intronLength + trailingGapLength - sourceGapMappedLength,
877 * second or third base of codon; check for any gaps in dna
879 if (!preserveMappedGaps)
881 trailingGapLength = 0;
883 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
889 * Realigns the given protein to match the alignment of the dna, using codon
890 * mappings to translate aligned codon positions to protein residues.
893 * the alignment whose sequences are realigned by this method
895 * the dna alignment whose alignment we are 'copying'
896 * @return the number of sequences that were realigned
898 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
900 if (protein.isNucleotide() || !dna.isNucleotide())
903 .errPrintln("Wrong alignment type in alignProteinAsDna");
906 List<SequenceI> unmappedProtein = new ArrayList<>();
907 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
908 protein, dna, unmappedProtein);
909 return alignProteinAs(protein, alignedCodons, unmappedProtein);
913 * Realigns the given dna to match the alignment of the protein, using codon
914 * mappings to translate aligned peptide positions to codons.
916 * Always produces a padded CDS alignment.
919 * the alignment whose sequences are realigned by this method
921 * the protein alignment whose alignment we are 'copying'
922 * @return the number of sequences that were realigned
924 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
926 if (protein.isNucleotide() || !dna.isNucleotide())
929 .errPrintln("Wrong alignment type in alignProteinAsDna");
932 // todo: implement this
933 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
934 int alignedCount = 0;
935 int width = 0; // alignment width for padding CDS
936 for (SequenceI dnaSeq : dna.getSequences())
938 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
939 dna.getGapCharacter()))
943 width = Math.max(dnaSeq.getLength(), width);
947 for (SequenceI dnaSeq : dna.getSequences())
949 oldwidth = dnaSeq.getLength();
950 diff = width - oldwidth;
953 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
960 * Helper method to align (if possible) the dna sequence to match the
961 * alignment of a mapped protein sequence. This is currently limited to
962 * handling coding sequence only.
970 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
971 AlignmentI protein, List<AlignedCodonFrame> mappings,
974 SequenceI cdsDss = cdsSeq.getDatasetSequence();
978 .println("alignCdsSequenceAsProtein needs aligned sequence!");
982 List<AlignedCodonFrame> dnaMappings = MappingUtils
983 .findMappingsForSequence(cdsSeq, mappings);
984 for (AlignedCodonFrame mapping : dnaMappings)
986 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
989 final int peptideLength = peptide.getLength();
990 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
993 MapList mapList = map.getMap();
994 if (map.getTo() == peptide.getDatasetSequence())
996 mapList = mapList.getInverse();
998 final int cdsLength = cdsDss.getLength();
999 int mappedFromLength = MappingUtils
1000 .getLength(mapList.getFromRanges());
1001 int mappedToLength = MappingUtils
1002 .getLength(mapList.getToRanges());
1003 boolean addStopCodon = (cdsLength == mappedFromLength
1004 * CODON_LENGTH + CODON_LENGTH)
1005 || (peptide.getDatasetSequence()
1006 .getLength() == mappedFromLength - 1);
1007 if (cdsLength != mappedToLength && !addStopCodon)
1009 jalview.bin.Console.errPrintln(String.format(
1010 "Can't align cds as protein (length mismatch %d/%d): %s",
1011 cdsLength, mappedToLength, cdsSeq.getName()));
1015 * pre-fill the aligned cds sequence with gaps
1017 char[] alignedCds = new char[peptideLength * CODON_LENGTH
1018 + (addStopCodon ? CODON_LENGTH : 0)];
1019 Arrays.fill(alignedCds, gapChar);
1022 * walk over the aligned peptide sequence and insert mapped
1023 * codons for residues in the aligned cds sequence
1025 int copiedBases = 0;
1026 int cdsStart = cdsDss.getStart();
1027 int proteinPos = peptide.getStart() - 1;
1030 for (int col = 0; col < peptideLength; col++)
1032 char residue = peptide.getCharAt(col);
1034 if (Comparison.isGap(residue))
1036 cdsCol += CODON_LENGTH;
1041 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1044 // e.g. incomplete start codon, X in peptide
1045 cdsCol += CODON_LENGTH;
1049 for (int j = codon[0]; j <= codon[1]; j++)
1051 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1052 alignedCds[cdsCol++] = mappedBase;
1060 * append stop codon if not mapped from protein,
1061 * closing it up to the end of the mapped sequence
1063 if (copiedBases == cdsLength - CODON_LENGTH)
1065 for (int i = alignedCds.length - 1; i >= 0; i--)
1067 if (!Comparison.isGap(alignedCds[i]))
1069 cdsCol = i + 1; // gap just after end of sequence
1073 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1075 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1078 cdsSeq.setSequence(new String(alignedCds));
1087 * Builds a map whose key is an aligned codon position (3 alignment column
1088 * numbers base 0), and whose value is a map from protein sequence to each
1089 * protein's peptide residue for that codon. The map generates an ordering of
1090 * the codons, and allows us to read off the peptides at each position in
1091 * order to assemble 'aligned' protein sequences.
1094 * the protein alignment
1096 * the coding dna alignment
1097 * @param unmappedProtein
1098 * any unmapped proteins are added to this list
1101 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1102 AlignmentI protein, AlignmentI dna,
1103 List<SequenceI> unmappedProtein)
1106 * maintain a list of any proteins with no mappings - these will be
1107 * rendered 'as is' in the protein alignment as we can't align them
1109 unmappedProtein.addAll(protein.getSequences());
1111 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1114 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1115 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1116 * comparator keeps the codon positions ordered.
1118 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<>(
1119 new CodonComparator());
1121 for (SequenceI dnaSeq : dna.getSequences())
1123 for (AlignedCodonFrame mapping : mappings)
1125 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1128 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1129 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1131 unmappedProtein.remove(prot);
1137 * Finally add any unmapped peptide start residues (e.g. for incomplete
1138 * codons) as if at the codon position before the second residue
1140 // TODO resolve JAL-2022 so this fudge can be removed
1141 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1142 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1144 return alignedCodons;
1148 * Scans for any protein mapped from position 2 (meaning unmapped start
1149 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1150 * preceding position in the alignment
1152 * @param alignedCodons
1153 * the codon-to-peptide map
1154 * @param mappedSequenceCount
1155 * the number of distinct sequences in the map
1157 protected static void addUnmappedPeptideStarts(
1158 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1159 int mappedSequenceCount)
1161 // TODO delete this ugly hack once JAL-2022 is resolved
1162 // i.e. we can model startPhase > 0 (incomplete start codon)
1164 List<SequenceI> sequencesChecked = new ArrayList<>();
1165 AlignedCodon lastCodon = null;
1166 Map<SequenceI, AlignedCodon> toAdd = new HashMap<>();
1168 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1171 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1174 SequenceI seq = sequenceCodon.getKey();
1175 if (sequencesChecked.contains(seq))
1179 sequencesChecked.add(seq);
1180 AlignedCodon codon = sequenceCodon.getValue();
1181 if (codon.peptideCol > 1)
1183 jalview.bin.Console.errPrintln(
1184 "Problem mapping protein with >1 unmapped start positions: "
1187 else if (codon.peptideCol == 1)
1190 * first position (peptideCol == 0) was unmapped - add it
1192 if (lastCodon != null)
1194 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1195 lastCodon.pos2, lastCodon.pos3,
1196 String.valueOf(seq.getCharAt(0)), 0);
1197 toAdd.put(seq, firstPeptide);
1202 * unmapped residue at start of alignment (no prior column) -
1203 * 'insert' at nominal codon [0, 0, 0]
1205 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1206 String.valueOf(seq.getCharAt(0)), 0);
1207 toAdd.put(seq, firstPeptide);
1210 if (sequencesChecked.size() == mappedSequenceCount)
1212 // no need to check past first mapped position in all sequences
1216 lastCodon = entry.getKey();
1220 * add any new codons safely after iterating over the map
1222 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1224 addCodonToMap(alignedCodons, startCodon.getValue(),
1225 startCodon.getKey());
1230 * Update the aligned protein sequences to match the codon alignments given in
1234 * @param alignedCodons
1235 * an ordered map of codon positions (columns), with sequence/peptide
1236 * values present in each column
1237 * @param unmappedProtein
1240 protected static int alignProteinAs(AlignmentI protein,
1241 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1242 List<SequenceI> unmappedProtein)
1245 * prefill peptide sequences with gaps
1247 int alignedWidth = alignedCodons.size();
1248 char[] gaps = new char[alignedWidth];
1249 Arrays.fill(gaps, protein.getGapCharacter());
1250 Map<SequenceI, char[]> peptides = new HashMap<>();
1251 for (SequenceI seq : protein.getSequences())
1253 if (!unmappedProtein.contains(seq))
1255 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1260 * Traverse the codons left to right (as defined by CodonComparator)
1261 * and insert peptides in each column where the sequence is mapped.
1262 * This gives a peptide 'alignment' where residues are aligned if their
1263 * corresponding codons occupy the same columns in the cdna alignment.
1266 for (AlignedCodon codon : alignedCodons.keySet())
1268 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1270 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1272 char residue = entry.getValue().product.charAt(0);
1273 peptides.get(entry.getKey())[column] = residue;
1279 * and finally set the constructed sequences
1281 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1283 entry.getKey().setSequence(new String(entry.getValue()));
1290 * Populate the map of aligned codons by traversing the given sequence
1291 * mapping, locating the aligned positions of mapped codons, and adding those
1292 * positions and their translation products to the map.
1295 * the aligned sequence we are mapping from
1297 * the sequence to be aligned to the codons
1299 * the gap character in the dna sequence
1301 * a mapping to a sequence translation
1302 * @param alignedCodons
1303 * the map we are building up
1305 static void addCodonPositions(SequenceI dna, SequenceI protein,
1306 char gapChar, Mapping seqMap,
1307 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1309 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1312 * add codon positions, and their peptide translations, to the alignment
1313 * map, while remembering the first codon mapped
1315 while (codons.hasNext())
1319 AlignedCodon codon = codons.next();
1320 addCodonToMap(alignedCodons, codon, protein);
1321 } catch (IncompleteCodonException e)
1323 // possible incomplete trailing codon - ignore
1324 } catch (NoSuchElementException e)
1326 // possibly peptide lacking STOP
1332 * Helper method to add a codon-to-peptide entry to the aligned codons map
1334 * @param alignedCodons
1338 protected static void addCodonToMap(
1339 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1340 AlignedCodon codon, SequenceI protein)
1342 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1343 if (seqProduct == null)
1345 seqProduct = new HashMap<>();
1346 alignedCodons.put(codon, seqProduct);
1348 seqProduct.put(protein, codon);
1352 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1353 * between at least one pair of sequences in the two alignments. Currently,
1356 * <li>One alignment must be nucleotide, and the other protein</li>
1357 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1358 * <li>Mappable means the nucleotide translation matches the protein
1360 * <li>The translation may ignore start and stop codons if present in the
1368 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1370 if (al1 == null || al2 == null)
1376 * Require one nucleotide and one protein
1378 if (al1.isNucleotide() == al2.isNucleotide())
1382 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1383 AlignmentI protein = dna == al1 ? al2 : al1;
1384 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1385 for (SequenceI dnaSeq : dna.getSequences())
1387 for (SequenceI proteinSeq : protein.getSequences())
1389 if (isMappable(dnaSeq, proteinSeq, mappings))
1399 * Returns true if the dna sequence is mapped, or could be mapped, to the
1407 protected static boolean isMappable(SequenceI dnaSeq,
1408 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1410 if (dnaSeq == null || proteinSeq == null)
1415 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1416 : dnaSeq.getDatasetSequence();
1417 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1419 : proteinSeq.getDatasetSequence();
1421 for (AlignedCodonFrame mapping : mappings)
1423 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1433 * Just try to make a mapping (it is not yet stored), test whether
1436 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1440 * Finds any reference annotations associated with the sequences in
1441 * sequenceScope, that are not already added to the alignment, and adds them
1442 * to the 'candidates' map. Also populates a lookup table of annotation
1443 * labels, keyed by calcId, for use in constructing tooltips or the like.
1445 * @param sequenceScope
1446 * the sequences to scan for reference annotations
1447 * @param labelForCalcId
1448 * (optional) map to populate with label for calcId
1450 * map to populate with annotations for sequence
1452 * the alignment to check for presence of annotations
1454 public static void findAddableReferenceAnnotations(
1455 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1456 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1459 if (sequenceScope == null)
1465 * For each sequence in scope, make a list of any annotations on the
1466 * underlying dataset sequence which are not already on the alignment.
1468 * Add to a map of { alignmentSequence, <List of annotations to add> }
1470 for (SequenceI seq : sequenceScope)
1472 SequenceI dataset = seq.getDatasetSequence();
1473 if (dataset == null)
1477 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1478 if (datasetAnnotations == null)
1482 final List<AlignmentAnnotation> result = new ArrayList<>();
1483 for (AlignmentAnnotation dsann : datasetAnnotations)
1486 * Find matching annotations on the alignment. If none is found, then
1487 * add this annotation to the list of 'addable' annotations for this
1490 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1491 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1492 boolean found = false;
1493 if (matchedAlignmentAnnotations != null)
1495 for (AlignmentAnnotation matched : matchedAlignmentAnnotations)
1497 if (dsann.description.equals(matched.description))
1507 if (labelForCalcId != null)
1509 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1514 * Save any addable annotations for this sequence
1516 if (!result.isEmpty())
1518 candidates.put(seq, result);
1524 * Adds annotations to the top of the alignment annotations, in the same order
1525 * as their related sequences. If you already have an annotation and want to
1526 * add it to a sequence in an alignment use {@code addReferenceAnnotationTo}
1528 * @param annotations
1529 * the annotations to add
1531 * the alignment to add them to
1532 * @param selectionGroup
1533 * current selection group - may be null, if provided then any added
1534 * annotation will be trimmed to just those columns in the selection
1537 public static void addReferenceAnnotations(
1538 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1539 final AlignmentI alignment, final SequenceGroup selectionGroup)
1541 for (SequenceI seq : annotations.keySet())
1543 for (AlignmentAnnotation ann : annotations.get(seq))
1545 addReferenceAnnotationTo(alignment, seq, ann, selectionGroup);
1551 public static boolean isSSAnnotationPresent( Map<SequenceI, List<AlignmentAnnotation>> annotations) {
1553 for (SequenceI seq : annotations.keySet())
1555 for (AlignmentAnnotation ann : annotations.get(seq))
1557 if(ann.getDescription(false).startsWith(SS_ANNOTATION_LABEL)) {
1566 * Make a copy of a reference annotation {@code ann} and add it to an
1567 * alignment sequence {@code seq} in {@code alignment}, optionally limited to
1568 * the extent of {@code selectionGroup}
1573 * @param selectionGroup
1574 * current selection group - may be null, if provided then any added
1575 * annotation will be trimmed to just those columns in the selection
1577 * @return annotation added to {@code seq and {@code alignment}
1579 public static AlignmentAnnotation addReferenceAnnotationTo(
1580 final AlignmentI alignment, final SequenceI seq,
1581 final AlignmentAnnotation ann, final SequenceGroup selectionGroup)
1583 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1585 int endRes = ann.annotations.length;
1586 if (selectionGroup != null)
1588 startRes = -1 + Math.min(seq.getEnd(), Math.max(seq.getStart(),
1589 seq.findPosition(selectionGroup.getStartRes())));
1590 endRes = -1 + Math.min(seq.getEnd(),
1591 seq.findPosition(selectionGroup.getEndRes()));
1594 copyAnn.restrict(startRes, endRes + 0);
1597 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1598 * original annotation is already on the sequence.
1600 if (!seq.hasAnnotation(ann))
1602 ContactMatrixI cm = seq.getDatasetSequence().getContactMatrixFor(ann);
1605 seq.addContactListFor(copyAnn, cm);
1607 seq.addAlignmentAnnotation(copyAnn);
1610 copyAnn.adjustForAlignment();
1611 // add to the alignment and set visible
1612 alignment.addAnnotation(copyAnn);
1613 copyAnn.visible = true;
1619 * Set visibility of alignment annotations of specified types (labels), for
1620 * specified sequences. This supports controls like "Show all secondary
1621 * structure", "Hide all Temp factor", etc.
1623 * @al the alignment to scan for annotations
1625 * the types (labels) of annotations to be updated
1626 * @param forSequences
1627 * if not null, only annotations linked to one of these sequences are
1628 * in scope for update; if null, acts on all sequence annotations
1630 * if this flag is true, 'types' is ignored (label not checked)
1632 * if true, set visibility on, else set off
1634 public static void showOrHideSequenceAnnotations(AlignmentI al,
1635 Collection<String> types, List<SequenceI> forSequences,
1636 boolean anyType, boolean doShow)
1638 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1641 for (AlignmentAnnotation aa : anns)
1643 if (anyType || types.contains(aa.label))
1645 if ((aa.sequenceRef != null) && (forSequences == null
1646 || forSequences.contains(aa.sequenceRef)))
1648 aa.visible = doShow;
1655 public static AlignmentAnnotation getFirstSequenceAnnotationOfType(
1656 AlignmentI al, int graphType)
1658 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1661 for (AlignmentAnnotation aa : anns)
1663 if (aa.sequenceRef != null && aa.graph == graphType)
1671 * Returns true if either sequence has a cross-reference to the other
1677 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1679 // Note: moved here from class CrossRef as the latter class has dependencies
1680 // not availability to the applet's classpath
1681 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1685 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1686 * that sequence name is structured as Source|AccessionId.
1692 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1694 if (seq1 == null || seq2 == null)
1698 String name = seq2.getName();
1699 final List<DBRefEntry> xrefs = seq1.getDBRefs();
1702 for (int ix = 0, nx = xrefs.size(); ix < nx; ix++)
1704 DBRefEntry xref = xrefs.get(ix);
1705 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1706 // case-insensitive test, consistent with DBRefEntry.equalRef()
1707 if (xrefName.equalsIgnoreCase(name))
1717 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1718 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1719 * added to the original alignment's dataset, which is shared by the new
1720 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1721 * added to the alignment dataset.
1724 * aligned nucleotide (dna or cds) sequences
1726 * the alignment dataset the sequences belong to
1728 * (optional) to restrict results to CDS that map to specified
1730 * @return an alignment whose sequences are the cds-only parts of the dna
1731 * sequences (or null if no mappings are found)
1733 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1734 AlignmentI dataset, SequenceI[] products)
1736 if (dataset == null || dataset.getDataset() != null)
1738 throw new IllegalArgumentException(
1739 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1741 List<SequenceI> foundSeqs = new ArrayList<>();
1742 List<SequenceI> cdsSeqs = new ArrayList<>();
1743 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1744 HashSet<SequenceI> productSeqs = null;
1745 if (products != null)
1747 productSeqs = new HashSet<>();
1748 for (SequenceI seq : products)
1750 productSeqs.add(seq.getDatasetSequence() == null ? seq
1751 : seq.getDatasetSequence());
1756 * Construct CDS sequences from mappings on the alignment dataset.
1758 * - find the protein product(s) mapped to from each dna sequence
1759 * - if the mapping covers the whole dna sequence (give or take start/stop
1760 * codon), take the dna as the CDS sequence
1761 * - else search dataset mappings for a suitable dna sequence, i.e. one
1762 * whose whole sequence is mapped to the protein
1763 * - if no sequence found, construct one from the dna sequence and mapping
1764 * (and add it to dataset so it is found if this is repeated)
1766 for (SequenceI dnaSeq : dna)
1768 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1769 : dnaSeq.getDatasetSequence();
1771 List<AlignedCodonFrame> seqMappings = MappingUtils
1772 .findMappingsForSequence(dnaSeq, mappings);
1773 for (AlignedCodonFrame mapping : seqMappings)
1775 List<Mapping> mappingsFromSequence = mapping
1776 .getMappingsFromSequence(dnaSeq);
1778 for (Mapping aMapping : mappingsFromSequence)
1780 MapList mapList = aMapping.getMap();
1781 if (mapList.getFromRatio() == 1)
1784 * not a dna-to-protein mapping (likely dna-to-cds)
1790 * skip if mapping is not to one of the target set of proteins
1792 SequenceI proteinProduct = aMapping.getTo();
1793 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1799 * try to locate the CDS from the dataset mappings;
1800 * guard against duplicate results (for the case that protein has
1801 * dbrefs to both dna and cds sequences)
1803 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1804 seqMappings, aMapping);
1807 if (!foundSeqs.contains(cdsSeq))
1809 foundSeqs.add(cdsSeq);
1810 SequenceI derivedSequence = cdsSeq.deriveSequence();
1811 cdsSeqs.add(derivedSequence);
1812 if (!dataset.getSequences().contains(cdsSeq))
1814 dataset.addSequence(cdsSeq);
1821 * didn't find mapped CDS sequence - construct it and add
1822 * its dataset sequence to the dataset
1824 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1825 dataset).deriveSequence();
1826 // cdsSeq has a name constructed as CDS|<dbref>
1827 // <dbref> will be either the accession for the coding sequence,
1828 // marked in the /via/ dbref to the protein product accession
1829 // or it will be the original nucleotide accession.
1830 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1832 cdsSeqs.add(cdsSeq);
1835 * build the mapping from CDS to protein
1837 List<int[]> cdsRange = Collections
1838 .singletonList(new int[]
1839 { cdsSeq.getStart(),
1840 cdsSeq.getLength() + cdsSeq.getStart() - 1 });
1841 MapList cdsToProteinMap = new MapList(cdsRange,
1842 mapList.getToRanges(), mapList.getFromRatio(),
1843 mapList.getToRatio());
1845 if (!dataset.getSequences().contains(cdsSeqDss))
1848 * if this sequence is a newly created one, add it to the dataset
1849 * and made a CDS to protein mapping (if sequence already exists,
1850 * CDS-to-protein mapping _is_ the transcript-to-protein mapping)
1852 dataset.addSequence(cdsSeqDss);
1853 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1854 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1858 * guard against duplicating the mapping if repeating this action
1860 if (!mappings.contains(cdsToProteinMapping))
1862 mappings.add(cdsToProteinMapping);
1866 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1867 proteinProduct, aMapping);
1869 * add another mapping from original 'from' range to CDS
1871 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1872 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1874 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1876 if (!mappings.contains(dnaToCdsMapping))
1878 mappings.add(dnaToCdsMapping);
1882 * transfer dna chromosomal loci (if known) to the CDS
1883 * sequence (via the mapping)
1885 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1886 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1889 * add DBRef with mapping from protein to CDS
1890 * (this enables Get Cross-References from protein alignment)
1891 * This is tricky because we can't have two DBRefs with the
1892 * same source and accession, so need a different accession for
1893 * the CDS from the dna sequence
1896 // specific use case:
1897 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1898 // ENSG02, ENSG03, with transcripts and products similarly named.
1899 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1901 // JBPNote: ?? can't actually create an example that demonstrates we
1903 // synthesize an xref.
1905 List<DBRefEntry> primrefs = dnaDss.getPrimaryDBRefs();
1906 for (int ip = 0, np = primrefs.size(); ip < np; ip++)
1908 DBRefEntry primRef = primrefs.get(ip);
1910 * create a cross-reference from CDS to the source sequence's
1911 * primary reference and vice versa
1913 String source = primRef.getSource();
1914 String version = primRef.getVersion();
1915 DBRefEntry cdsCrossRef = new DBRefEntry(source,
1916 source + ":" + version, primRef.getAccessionId());
1918 .setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
1919 cdsSeqDss.addDBRef(cdsCrossRef);
1921 dnaSeq.addDBRef(new DBRefEntry(source, version,
1922 cdsSeq.getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
1923 // problem here is that the cross-reference is synthesized -
1924 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1926 // assuming cds version same as dna ?!?
1928 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
1931 proteinToCdsRef.setMap(
1932 new Mapping(cdsSeqDss, cdsToProteinMap.getInverse()));
1933 proteinProduct.addDBRef(proteinToCdsRef);
1936 * transfer any features on dna that overlap the CDS
1938 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1939 SequenceOntologyI.CDS);
1944 AlignmentI cds = new Alignment(
1945 cdsSeqs.toArray(new SequenceI[cdsSeqs.size()]));
1946 cds.setDataset(dataset);
1952 * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
1953 * toSeq, mediated by the given mapping between the sequences
1956 * @param targetToFrom
1960 protected static void transferGeneLoci(SequenceI fromSeq,
1961 MapList targetToFrom, SequenceI targetSeq)
1963 if (targetSeq.getGeneLoci() != null)
1965 // already have - don't override
1968 GeneLociI fromLoci = fromSeq.getGeneLoci();
1969 if (fromLoci == null)
1974 MapList newMap = targetToFrom.traverse(fromLoci.getMapping());
1978 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
1979 fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
1984 * A helper method that finds a CDS sequence in the alignment dataset that is
1985 * mapped to the given protein sequence, and either is, or has a mapping from,
1986 * the given dna sequence.
1989 * set of all mappings on the dataset
1991 * a dna (or cds) sequence we are searching from
1992 * @param seqMappings
1993 * the set of mappings involving dnaSeq
1995 * a transcript-to-peptide mapping
1998 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1999 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
2003 * TODO a better dna-cds-protein mapping data representation to allow easy
2004 * navigation; until then this clunky looping around lists of mappings
2006 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
2007 : dnaSeq.getDatasetSequence();
2008 SequenceI proteinProduct = aMapping.getTo();
2011 * is this mapping from the whole dna sequence (i.e. CDS)?
2012 * allowing for possible stop codon on dna but not peptide
2014 int mappedFromLength = MappingUtils
2015 .getLength(aMapping.getMap().getFromRanges());
2016 int dnaLength = seqDss.getLength();
2017 if (mappedFromLength == dnaLength
2018 || mappedFromLength == dnaLength - CODON_LENGTH)
2021 * if sequence has CDS features, this is a transcript with no UTR
2022 * - do not take this as the CDS sequence! (JAL-2789)
2024 if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
2032 * looks like we found the dna-to-protein mapping; search for the
2033 * corresponding cds-to-protein mapping
2035 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
2036 .findMappingsForSequence(proteinProduct, mappings);
2037 for (AlignedCodonFrame acf : mappingsToPeptide)
2039 for (SequenceToSequenceMapping map : acf.getMappings())
2041 Mapping mapping = map.getMapping();
2042 if (mapping != aMapping
2043 && mapping.getMap().getFromRatio() == CODON_LENGTH
2044 && proteinProduct == mapping.getTo()
2045 && seqDss != map.getFromSeq())
2047 mappedFromLength = MappingUtils
2048 .getLength(mapping.getMap().getFromRanges());
2049 if (mappedFromLength == map.getFromSeq().getLength())
2052 * found a 3:1 mapping to the protein product which covers
2053 * the whole dna sequence i.e. is from CDS; finally check the CDS
2054 * is mapped from the given dna start sequence
2056 SequenceI cdsSeq = map.getFromSeq();
2057 // todo this test is weak if seqMappings contains multiple mappings;
2058 // we get away with it if transcript:cds relationship is 1:1
2059 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
2060 .findMappingsForSequence(cdsSeq, seqMappings);
2061 if (!dnaToCdsMaps.isEmpty())
2073 * Helper method that makes a CDS sequence as defined by the mappings from the
2074 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
2075 * forward or reverse strand).
2080 * - existing dataset. We check for sequences that look like the CDS
2081 * we are about to construct, if one exists already, then we will
2082 * just return that one.
2083 * @return CDS sequence (as a dataset sequence)
2085 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
2089 * construct CDS sequence name as "CDS|" with 'from id' held in the mapping
2090 * if set (e.g. EMBL protein_id), else sequence name appended
2092 String mapFromId = mapping.getMappedFromId();
2093 final String seqId = "CDS|"
2094 + (mapFromId != null ? mapFromId : seq.getName());
2096 SequenceI newSeq = null;
2099 * construct CDS sequence by splicing mapped from ranges
2101 char[] seqChars = seq.getSequence();
2102 List<int[]> fromRanges = mapping.getMap().getFromRanges();
2103 int cdsWidth = MappingUtils.getLength(fromRanges);
2104 char[] newSeqChars = new char[cdsWidth];
2107 for (int[] range : fromRanges)
2109 if (range[0] <= range[1])
2111 // forward strand mapping - just copy the range
2112 int length = range[1] - range[0] + 1;
2113 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
2119 // reverse strand mapping - copy and complement one by one
2120 for (int i = range[0]; i >= range[1]; i--)
2122 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
2126 newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2129 if (dataset != null)
2131 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2132 if (matches != null)
2134 boolean matched = false;
2135 for (SequenceI mtch : matches)
2137 if (mtch.getStart() != newSeq.getStart())
2141 if (mtch.getEnd() != newSeq.getEnd())
2145 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2157 "JAL-2154 regression: warning - found (and ignored) a duplicate CDS sequence:"
2163 // newSeq.setDescription(mapFromId);
2169 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2170 * the given mapping.
2174 * @param proteinProduct
2176 * @return list of DBRefEntrys added
2178 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2179 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2182 // gather direct refs from contig congruent with mapping
2183 List<DBRefEntry> direct = new ArrayList<>();
2184 HashSet<String> directSources = new HashSet<>();
2186 List<DBRefEntry> refs = contig.getDBRefs();
2189 for (int ib = 0, nb = refs.size(); ib < nb; ib++)
2191 DBRefEntry dbr = refs.get(ib);
2193 if (dbr.hasMap() && (map = dbr.getMap().getMap()).isTripletMap())
2195 // check if map is the CDS mapping
2196 if (mapping.getMap().equals(map))
2199 directSources.add(dbr.getSource());
2204 List<DBRefEntry> onSource = DBRefUtils.selectRefs(
2205 proteinProduct.getDBRefs(),
2206 directSources.toArray(new String[0]));
2207 List<DBRefEntry> propagated = new ArrayList<>();
2209 // and generate appropriate mappings
2210 for (int ic = 0, nc = direct.size(); ic < nc; ic++)
2212 DBRefEntry cdsref = direct.get(ic);
2213 Mapping m = cdsref.getMap();
2214 // clone maplist and mapping
2215 MapList cdsposmap = new MapList(
2216 Arrays.asList(new int[][]
2217 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2218 m.getMap().getToRanges(), 3, 1);
2219 Mapping cdsmap = new Mapping(m.getTo(), m.getMap());
2222 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2223 cdsref.getVersion(), cdsref.getAccessionId(),
2224 new Mapping(cdsmap.getTo(), cdsposmap));
2226 // and see if we can map to the protein product for this mapping.
2227 // onSource is the filtered set of accessions on protein that we are
2228 // tranferring, so we assume accession is the same.
2229 if (cdsmap.getTo() == null && onSource != null)
2231 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2232 cdsref.getAccessionId());
2233 if (sourceRefs != null)
2235 for (DBRefEntry srcref : sourceRefs)
2237 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2239 // we have found a complementary dbref on the protein product, so
2240 // update mapping's getTo
2241 newref.getMap().setTo(proteinProduct);
2246 cdsSeq.addDBRef(newref);
2247 propagated.add(newref);
2253 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2254 * feature start/end ranges, optionally omitting specified feature types.
2255 * Returns the number of features copied.
2260 * the mapping from 'fromSeq' to 'toSeq'
2262 * if not null, only features of this type are copied (including
2263 * subtypes in the Sequence Ontology)
2266 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2267 MapList mapping, String select, String... omitting)
2269 SequenceI copyTo = toSeq;
2270 while (copyTo.getDatasetSequence() != null)
2272 copyTo = copyTo.getDatasetSequence();
2274 if (fromSeq == copyTo || fromSeq.getDatasetSequence() == copyTo)
2276 return 0; // shared dataset sequence
2280 * get features, optionally restricted by an ontology term
2282 List<SequenceFeature> sfs = select == null
2283 ? fromSeq.getFeatures().getPositionalFeatures()
2284 : fromSeq.getFeatures().getFeaturesByOntology(select);
2287 for (SequenceFeature sf : sfs)
2289 String type = sf.getType();
2290 boolean omit = false;
2291 for (String toOmit : omitting)
2293 if (type.equals(toOmit))
2304 * locate the mapped range - null if either start or end is
2305 * not mapped (no partial overlaps are calculated)
2307 int start = sf.getBegin();
2308 int end = sf.getEnd();
2309 int[] mappedTo = mapping.locateInTo(start, end);
2311 * if whole exon range doesn't map, try interpreting it
2312 * as 5' or 3' exon overlapping the CDS range
2314 if (mappedTo == null)
2316 mappedTo = mapping.locateInTo(end, end);
2317 if (mappedTo != null)
2320 * end of exon is in CDS range - 5' overlap
2321 * to a range from the start of the peptide
2326 if (mappedTo == null)
2328 mappedTo = mapping.locateInTo(start, start);
2329 if (mappedTo != null)
2332 * start of exon is in CDS range - 3' overlap
2333 * to a range up to the end of the peptide
2335 mappedTo[1] = toSeq.getLength();
2338 if (mappedTo != null)
2340 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2341 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2342 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2343 sf.getFeatureGroup(), sf.getScore());
2344 copyTo.addSequenceFeature(copy);
2352 * Returns a mapping from dna to protein by inspecting sequence features of
2353 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2354 * length is 3 times the peptide length (optionally after dropping a trailing
2355 * stop codon). This method does not check whether the CDS nucleotide sequence
2356 * translates to the peptide sequence.
2362 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2363 SequenceI proteinSeq)
2365 List<int[]> ranges = findCdsPositions(dnaSeq);
2366 int mappedDnaLength = MappingUtils.getLength(ranges);
2369 * if not a whole number of codons, truncate mapping
2371 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2372 if (codonRemainder > 0)
2374 mappedDnaLength -= codonRemainder;
2375 MappingUtils.removeEndPositions(codonRemainder, ranges);
2378 int proteinLength = proteinSeq.getLength();
2379 int proteinStart = proteinSeq.getStart();
2380 int proteinEnd = proteinSeq.getEnd();
2383 * incomplete start codon may mean X at start of peptide
2384 * we ignore both for mapping purposes
2386 if (proteinSeq.getCharAt(0) == 'X')
2388 // todo JAL-2022 support startPhase > 0
2392 List<int[]> proteinRange = new ArrayList<>();
2395 * dna length should map to protein (or protein plus stop codon)
2397 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2398 if (codesForResidues == (proteinLength + 1))
2400 // assuming extra codon is for STOP and not in peptide
2401 // todo: check trailing codon is indeed a STOP codon
2403 mappedDnaLength -= CODON_LENGTH;
2404 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2407 if (codesForResidues == proteinLength)
2409 proteinRange.add(new int[] { proteinStart, proteinEnd });
2410 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2416 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2417 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2418 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2419 * position order, so this method is only valid for linear CDS in the same
2420 * sense as the protein product.
2425 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2427 List<int[]> result = new ArrayList<>();
2429 List<SequenceFeature> sfs = dnaSeq.getFeatures()
2430 .getFeaturesByOntology(SequenceOntologyI.CDS);
2435 SequenceFeatures.sortFeatures(sfs, true);
2437 for (SequenceFeature sf : sfs)
2442 String s = sf.getPhase();
2445 phase = Integer.parseInt(s);
2447 } catch (NumberFormatException e)
2452 * phase > 0 on first codon means 5' incomplete - skip to the start
2453 * of the next codon; example ENST00000496384
2455 int begin = sf.getBegin();
2456 int end = sf.getEnd();
2457 if (result.isEmpty() && phase > 0)
2462 // shouldn't happen!
2464 .println("Error: start phase extends beyond start CDS in "
2465 + dnaSeq.getName());
2468 result.add(new int[] { begin, end });
2472 * Finally sort ranges by start position. This avoids a dependency on
2473 * keeping features in order on the sequence (if they are in order anyway,
2474 * the sort will have almost no work to do). The implicit assumption is CDS
2475 * ranges are assembled in order. Other cases should not use this method,
2476 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2478 Collections.sort(result, IntRangeComparator.ASCENDING);
2483 * Makes an alignment with a copy of the given sequences, adding in any
2484 * non-redundant sequences which are mapped to by the cross-referenced
2490 * the alignment dataset shared by the new copy
2493 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2494 SequenceI[] xrefs, AlignmentI dataset)
2496 AlignmentI copy = new Alignment(new Alignment(seqs));
2497 copy.setDataset(dataset);
2498 boolean isProtein = !copy.isNucleotide();
2499 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2502 // BH 2019.01.25 recoded to remove iterators
2504 for (int ix = 0, nx = xrefs.length; ix < nx; ix++)
2506 SequenceI xref = xrefs[ix];
2507 List<DBRefEntry> dbrefs = xref.getDBRefs();
2510 for (int ir = 0, nir = dbrefs.size(); ir < nir; ir++)
2512 DBRefEntry dbref = dbrefs.get(ir);
2513 Mapping map = dbref.getMap();
2515 if (map == null || (mto = map.getTo()) == null
2516 || mto.isProtein() != isProtein)
2520 SequenceI mappedTo = mto;
2521 SequenceI match = matcher.findIdMatch(mappedTo);
2524 matcher.add(mappedTo);
2525 copy.addSequence(mappedTo);
2535 * Try to align sequences in 'unaligned' to match the alignment of their
2536 * mapped regions in 'aligned'. For example, could use this to align CDS
2537 * sequences which are mapped to their parent cDNA sequences.
2539 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2540 * dna-to-protein or protein-to-dna use alternative methods.
2543 * sequences to be aligned
2545 * holds aligned sequences and their mappings
2548 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2551 * easy case - aligning a copy of aligned sequences
2553 if (alignAsSameSequences(unaligned, aligned))
2555 return unaligned.getHeight();
2559 * fancy case - aligning via mappings between sequences
2561 List<SequenceI> unmapped = new ArrayList<>();
2562 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2563 unaligned, aligned, unmapped);
2564 int width = columnMap.size();
2565 char gap = unaligned.getGapCharacter();
2566 int realignedCount = 0;
2567 // TODO: verify this loop scales sensibly for very wide/high alignments
2569 for (SequenceI seq : unaligned.getSequences())
2571 if (!unmapped.contains(seq))
2573 char[] newSeq = new char[width];
2574 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2575 // Integer iteration below
2580 * traverse the map to find columns populated
2583 for (Integer column : columnMap.keySet())
2585 Character c = columnMap.get(column).get(seq);
2589 * sequence has a character at this position
2599 * trim trailing gaps
2601 if (lastCol < width)
2603 char[] tmp = new char[lastCol + 1];
2604 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2607 // TODO: optimise SequenceI to avoid char[]->String->char[]
2608 seq.setSequence(String.valueOf(newSeq));
2612 return realignedCount;
2616 * If unaligned and aligned sequences share the same dataset sequences, then
2617 * simply copies the aligned sequences to the unaligned sequences and returns
2618 * true; else returns false
2621 * - sequences to be aligned based on aligned
2623 * - 'guide' alignment containing sequences derived from same dataset
2627 static boolean alignAsSameSequences(AlignmentI unaligned,
2630 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2632 return false; // should only pass alignments with datasets here
2635 // map from dataset sequence to alignment sequence(s)
2636 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2637 for (SequenceI seq : aligned.getSequences())
2639 SequenceI ds = seq.getDatasetSequence();
2640 if (alignedDatasets.get(ds) == null)
2642 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2644 alignedDatasets.get(ds).add(seq);
2648 * first pass - check whether all sequences to be aligned share a
2649 * dataset sequence with an aligned sequence; also note the leftmost
2650 * ungapped column from which to copy
2652 int leftmost = Integer.MAX_VALUE;
2653 for (SequenceI seq : unaligned.getSequences())
2655 final SequenceI ds = seq.getDatasetSequence();
2656 if (!alignedDatasets.containsKey(ds))
2660 SequenceI alignedSeq = alignedDatasets.get(ds).get(0);
2661 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2662 leftmost = Math.min(leftmost, startCol);
2666 * second pass - copy aligned sequences;
2667 * heuristic rule: pair off sequences in order for the case where
2668 * more than one shares the same dataset sequence
2670 final char gapCharacter = aligned.getGapCharacter();
2671 for (SequenceI seq : unaligned.getSequences())
2673 List<SequenceI> alignedSequences = alignedDatasets
2674 .get(seq.getDatasetSequence());
2675 if (alignedSequences.isEmpty())
2678 * defensive check - shouldn't happen! (JAL-3536)
2682 SequenceI alignedSeq = alignedSequences.get(0);
2685 * gap fill for leading (5') UTR if any
2687 // TODO this copies intron columns - wrong!
2688 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2689 int endCol = alignedSeq.findIndex(seq.getEnd());
2690 char[] seqchars = new char[endCol - leftmost + 1];
2691 Arrays.fill(seqchars, gapCharacter);
2692 char[] toCopy = alignedSeq.getSequence(startCol - 1, endCol);
2693 System.arraycopy(toCopy, 0, seqchars, startCol - leftmost,
2695 seq.setSequence(String.valueOf(seqchars));
2696 if (alignedSequences.size() > 0)
2698 // pop off aligned sequences (except the last one)
2699 alignedSequences.remove(0);
2704 * finally remove gapped columns (e.g. introns)
2706 new RemoveGapColCommand("", unaligned.getSequencesArray(), 0,
2707 unaligned.getWidth() - 1, unaligned);
2713 * Returns a map whose key is alignment column number (base 1), and whose
2714 * values are a map of sequence characters in that column.
2721 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2722 AlignmentI unaligned, AlignmentI aligned,
2723 List<SequenceI> unmapped)
2726 * Map will hold, for each aligned column position, a map of
2727 * {unalignedSequence, characterPerSequence} at that position.
2728 * TreeMap keeps the entries in ascending column order.
2730 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2733 * record any sequences that have no mapping so can't be realigned
2735 unmapped.addAll(unaligned.getSequences());
2737 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2739 for (SequenceI seq : unaligned.getSequences())
2741 for (AlignedCodonFrame mapping : mappings)
2743 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2744 if (fromSeq != null)
2746 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2747 if (addMappedPositions(seq, fromSeq, seqMap, map))
2749 unmapped.remove(seq);
2758 * Helper method that adds to a map the mapped column positions of a sequence.
2760 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2761 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2765 * the sequence whose column positions we are recording
2767 * a sequence that is mapped to the first sequence
2769 * the mapping from 'fromSeq' to 'seq'
2771 * a map to add the column positions (in fromSeq) of the mapped
2775 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2776 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2784 * invert mapping if it is from unaligned to aligned sequence
2786 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2788 seqMap = new Mapping(seq.getDatasetSequence(),
2789 seqMap.getMap().getInverse());
2792 int toStart = seq.getStart();
2795 * traverse [start, end, start, end...] ranges in fromSeq
2797 for (int[] fromRange : seqMap.getMap().getFromRanges())
2799 for (int i = 0; i < fromRange.length - 1; i += 2)
2801 boolean forward = fromRange[i + 1] >= fromRange[i];
2804 * find the range mapped to (sequence positions base 1)
2806 int[] range = seqMap.locateMappedRange(fromRange[i],
2810 jalview.bin.Console.errPrintln("Error in mapping " + seqMap
2811 + " from " + fromSeq.getName());
2814 int fromCol = fromSeq.findIndex(fromRange[i]);
2815 int mappedCharPos = range[0];
2818 * walk over the 'from' aligned sequence in forward or reverse
2819 * direction; when a non-gap is found, record the column position
2820 * of the next character of the mapped-to sequence; stop when all
2821 * the characters of the range have been counted
2823 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
2826 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
2829 * mapped from sequence has a character in this column
2830 * record the column position for the mapped to character
2832 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2833 if (seqsMap == null)
2835 seqsMap = new HashMap<>();
2836 map.put(fromCol, seqsMap);
2838 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
2841 fromCol += (forward ? 1 : -1);
2848 // strictly temporary hack until proper criteria for aligning protein to cds
2849 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2850 public static boolean looksLikeEnsembl(AlignmentI alignment)
2852 for (SequenceI seq : alignment.getSequences())
2854 String name = seq.getName();
2855 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))
2864 public static List<String> getSecondaryStructureSources(AlignmentAnnotation[] annotations) {
2866 List<String> ssSources = new ArrayList<>();
2867 Set<String> addedLabels = new HashSet<>(); // to keep track of added labels
2869 for (AlignmentAnnotation annotation : annotations) {
2870 String label = annotation.label;
2871 if (SECONDARY_STRUCTURE_LABELS.containsKey(label) && !addedLabels.contains(label)) {
2872 ssSources.add(SECONDARY_STRUCTURE_LABELS.get(label));
2873 addedLabels.add(label); // Add the label to the set
2880 public static boolean isSecondaryStructurePresent(AlignmentAnnotation[] annotations)
2882 boolean ssPresent = false;
2884 for (AlignmentAnnotation aa : annotations)
2890 if (SECONDARY_STRUCTURE_LABELS.containsKey(aa.label)) {