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 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
551 // + codon + "(" + translated + ") != " + aaRes));
557 * check we matched all of the protein sequence
559 if (aaPos != aaSeqChars.length)
565 * check we matched all of the dna except
566 * for optional trailing STOP codon
568 if (dnaPos == cdnaSeqChars.length)
572 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
574 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
575 if (ResidueProperties.STOP
576 .equals(ResidueProperties.codonTranslate(codon)))
585 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
586 * currently assumes that we are aligning cDNA to match protein.
589 * the sequence to be realigned
591 * the alignment whose sequence alignment is to be 'copied'
593 * character string represent a gap in the realigned sequence
594 * @param preserveUnmappedGaps
595 * @param preserveMappedGaps
596 * @return true if the sequence was realigned, false if it could not be
598 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
599 String gap, boolean preserveMappedGaps,
600 boolean preserveUnmappedGaps)
603 * Get any mappings from the source alignment to the target (dataset)
606 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
607 // all mappings. Would it help to constrain this?
608 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
609 if (mappings == null || mappings.isEmpty())
615 * Locate the aligned source sequence whose dataset sequence is mapped. We
616 * just take the first match here (as we can't align like more than one
619 SequenceI alignFrom = null;
620 AlignedCodonFrame mapping = null;
621 for (AlignedCodonFrame mp : mappings)
623 alignFrom = mp.findAlignedSequence(seq, al);
624 if (alignFrom != null)
631 if (alignFrom == null)
635 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
636 preserveMappedGaps, preserveUnmappedGaps);
641 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
642 * match residues and codons. Flags control whether existing gaps in unmapped
643 * (intron) and mapped (exon) regions are preserved or not. Gaps between
644 * intron and exon are only retained if both flags are set.
651 * @param preserveUnmappedGaps
652 * @param preserveMappedGaps
654 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
655 AlignedCodonFrame mapping, String myGap, char sourceGap,
656 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
658 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
660 // aligned and dataset sequence positions, all base zero
664 int basesWritten = 0;
665 char myGapChar = myGap.charAt(0);
666 int ratio = myGap.length();
668 int fromOffset = alignFrom.getStart() - 1;
669 int toOffset = alignTo.getStart() - 1;
670 int sourceGapMappedLength = 0;
671 boolean inExon = false;
672 final int toLength = alignTo.getLength();
673 final int fromLength = alignFrom.getLength();
674 StringBuilder thisAligned = new StringBuilder(2 * toLength);
677 * Traverse the 'model' aligned sequence
679 for (int i = 0; i < fromLength; i++)
681 char sourceChar = alignFrom.getCharAt(i);
682 if (sourceChar == sourceGap)
684 sourceGapMappedLength += ratio;
689 * Found a non-gap character. Locate its mapped region if any.
692 // Note mapping positions are base 1, our sequence positions base 0
693 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
694 sourceDsPos + fromOffset);
695 if (mappedPos == null)
698 * unmapped position; treat like a gap
700 sourceGapMappedLength += ratio;
701 // System.err.println("Can't align: no codon mapping to residue "
702 // + sourceDsPos + "(" + sourceChar + ")");
707 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
708 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
709 StringBuilder trailingCopiedGap = new StringBuilder();
712 * Copy dna sequence up to and including this codon. Optionally, include
713 * gaps before the codon starts (in introns) and/or after the codon starts
716 * Note this only works for 'linear' splicing, not reverse or interleaved.
717 * But then 'align dna as protein' doesn't make much sense otherwise.
719 int intronLength = 0;
720 while (basesWritten + toOffset < mappedCodonEnd
721 && thisSeqPos < toLength)
723 final char c = alignTo.getCharAt(thisSeqPos++);
727 int sourcePosition = basesWritten + toOffset;
728 if (sourcePosition < mappedCodonStart)
731 * Found an unmapped (intron) base. First add in any preceding gaps
734 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
736 thisAligned.append(trailingCopiedGap.toString());
737 intronLength += trailingCopiedGap.length();
738 trailingCopiedGap = new StringBuilder();
745 final boolean startOfCodon = sourcePosition == mappedCodonStart;
746 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
747 preserveUnmappedGaps, sourceGapMappedLength, inExon,
748 trailingCopiedGap.length(), intronLength, startOfCodon);
749 for (int k = 0; k < gapsToAdd; k++)
751 thisAligned.append(myGapChar);
753 sourceGapMappedLength = 0;
756 thisAligned.append(c);
757 trailingCopiedGap = new StringBuilder();
761 if (inExon && preserveMappedGaps)
763 trailingCopiedGap.append(myGapChar);
765 else if (!inExon && preserveUnmappedGaps)
767 trailingCopiedGap.append(myGapChar);
774 * At end of model aligned sequence. Copy any remaining target sequence, optionally
775 * including (intron) gaps.
777 while (thisSeqPos < toLength)
779 final char c = alignTo.getCharAt(thisSeqPos++);
780 if (c != myGapChar || preserveUnmappedGaps)
782 thisAligned.append(c);
784 sourceGapMappedLength--;
788 * finally add gaps to pad for any trailing source gaps or
789 * unmapped characters
791 if (preserveUnmappedGaps)
793 while (sourceGapMappedLength > 0)
795 thisAligned.append(myGapChar);
796 sourceGapMappedLength--;
801 * All done aligning, set the aligned sequence.
803 alignTo.setSequence(new String(thisAligned));
807 * Helper method to work out how many gaps to insert when realigning.
809 * @param preserveMappedGaps
810 * @param preserveUnmappedGaps
811 * @param sourceGapMappedLength
813 * @param trailingCopiedGap
814 * @param intronLength
815 * @param startOfCodon
818 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
819 boolean preserveUnmappedGaps, int sourceGapMappedLength,
820 boolean inExon, int trailingGapLength, int intronLength,
821 final boolean startOfCodon)
827 * Reached start of codon. Ignore trailing gaps in intron unless we are
828 * preserving gaps in both exon and intron. Ignore them anyway if the
829 * protein alignment introduces a gap at least as large as the intronic
832 if (inExon && !preserveMappedGaps)
834 trailingGapLength = 0;
836 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
838 trailingGapLength = 0;
842 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
846 if (intronLength + trailingGapLength <= sourceGapMappedLength)
848 gapsToAdd = sourceGapMappedLength - intronLength;
852 gapsToAdd = Math.min(
853 intronLength + trailingGapLength - sourceGapMappedLength,
861 * second or third base of codon; check for any gaps in dna
863 if (!preserveMappedGaps)
865 trailingGapLength = 0;
867 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
873 * Realigns the given protein to match the alignment of the dna, using codon
874 * mappings to translate aligned codon positions to protein residues.
877 * the alignment whose sequences are realigned by this method
879 * the dna alignment whose alignment we are 'copying'
880 * @return the number of sequences that were realigned
882 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
884 if (protein.isNucleotide() || !dna.isNucleotide())
886 System.err.println("Wrong alignment type in alignProteinAsDna");
889 List<SequenceI> unmappedProtein = new ArrayList<>();
890 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
891 protein, dna, unmappedProtein);
892 return alignProteinAs(protein, alignedCodons, unmappedProtein);
896 * Realigns the given dna to match the alignment of the protein, using codon
897 * mappings to translate aligned peptide positions to codons.
899 * Always produces a padded CDS alignment.
902 * the alignment whose sequences are realigned by this method
904 * the protein alignment whose alignment we are 'copying'
905 * @return the number of sequences that were realigned
907 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
909 if (protein.isNucleotide() || !dna.isNucleotide())
911 System.err.println("Wrong alignment type in alignProteinAsDna");
914 // todo: implement this
915 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
916 int alignedCount = 0;
917 int width = 0; // alignment width for padding CDS
918 for (SequenceI dnaSeq : dna.getSequences())
920 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
921 dna.getGapCharacter()))
925 width = Math.max(dnaSeq.getLength(), width);
929 for (SequenceI dnaSeq : dna.getSequences())
931 oldwidth = dnaSeq.getLength();
932 diff = width - oldwidth;
935 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
942 * Helper method to align (if possible) the dna sequence to match the
943 * alignment of a mapped protein sequence. This is currently limited to
944 * handling coding sequence only.
952 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
953 AlignmentI protein, List<AlignedCodonFrame> mappings,
956 SequenceI cdsDss = cdsSeq.getDatasetSequence();
960 .println("alignCdsSequenceAsProtein needs aligned sequence!");
964 List<AlignedCodonFrame> dnaMappings = MappingUtils
965 .findMappingsForSequence(cdsSeq, mappings);
966 for (AlignedCodonFrame mapping : dnaMappings)
968 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
971 final int peptideLength = peptide.getLength();
972 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
975 MapList mapList = map.getMap();
976 if (map.getTo() == peptide.getDatasetSequence())
978 mapList = mapList.getInverse();
980 final int cdsLength = cdsDss.getLength();
981 int mappedFromLength = MappingUtils
982 .getLength(mapList.getFromRanges());
983 int mappedToLength = MappingUtils
984 .getLength(mapList.getToRanges());
985 boolean addStopCodon = (cdsLength == mappedFromLength
986 * CODON_LENGTH + CODON_LENGTH)
987 || (peptide.getDatasetSequence()
988 .getLength() == mappedFromLength - 1);
989 if (cdsLength != mappedToLength && !addStopCodon)
991 System.err.println(String.format(
992 "Can't align cds as protein (length mismatch %d/%d): %s",
993 cdsLength, mappedToLength, cdsSeq.getName()));
997 * pre-fill the aligned cds sequence with gaps
999 char[] alignedCds = new char[peptideLength * CODON_LENGTH
1000 + (addStopCodon ? CODON_LENGTH : 0)];
1001 Arrays.fill(alignedCds, gapChar);
1004 * walk over the aligned peptide sequence and insert mapped
1005 * codons for residues in the aligned cds sequence
1007 int copiedBases = 0;
1008 int cdsStart = cdsDss.getStart();
1009 int proteinPos = peptide.getStart() - 1;
1012 for (int col = 0; col < peptideLength; col++)
1014 char residue = peptide.getCharAt(col);
1016 if (Comparison.isGap(residue))
1018 cdsCol += CODON_LENGTH;
1023 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1026 // e.g. incomplete start codon, X in peptide
1027 cdsCol += CODON_LENGTH;
1031 for (int j = codon[0]; j <= codon[1]; j++)
1033 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1034 alignedCds[cdsCol++] = mappedBase;
1042 * append stop codon if not mapped from protein,
1043 * closing it up to the end of the mapped sequence
1045 if (copiedBases == cdsLength - CODON_LENGTH)
1047 for (int i = alignedCds.length - 1; i >= 0; i--)
1049 if (!Comparison.isGap(alignedCds[i]))
1051 cdsCol = i + 1; // gap just after end of sequence
1055 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1057 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1060 cdsSeq.setSequence(new String(alignedCds));
1069 * Builds a map whose key is an aligned codon position (3 alignment column
1070 * numbers base 0), and whose value is a map from protein sequence to each
1071 * protein's peptide residue for that codon. The map generates an ordering of
1072 * the codons, and allows us to read off the peptides at each position in
1073 * order to assemble 'aligned' protein sequences.
1076 * the protein alignment
1078 * the coding dna alignment
1079 * @param unmappedProtein
1080 * any unmapped proteins are added to this list
1083 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1084 AlignmentI protein, AlignmentI dna,
1085 List<SequenceI> unmappedProtein)
1088 * maintain a list of any proteins with no mappings - these will be
1089 * rendered 'as is' in the protein alignment as we can't align them
1091 unmappedProtein.addAll(protein.getSequences());
1093 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1096 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1097 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1098 * comparator keeps the codon positions ordered.
1100 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<>(
1101 new CodonComparator());
1103 for (SequenceI dnaSeq : dna.getSequences())
1105 for (AlignedCodonFrame mapping : mappings)
1107 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1110 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1111 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1113 unmappedProtein.remove(prot);
1119 * Finally add any unmapped peptide start residues (e.g. for incomplete
1120 * codons) as if at the codon position before the second residue
1122 // TODO resolve JAL-2022 so this fudge can be removed
1123 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1124 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1126 return alignedCodons;
1130 * Scans for any protein mapped from position 2 (meaning unmapped start
1131 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1132 * preceding position in the alignment
1134 * @param alignedCodons
1135 * the codon-to-peptide map
1136 * @param mappedSequenceCount
1137 * the number of distinct sequences in the map
1139 protected static void addUnmappedPeptideStarts(
1140 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1141 int mappedSequenceCount)
1143 // TODO delete this ugly hack once JAL-2022 is resolved
1144 // i.e. we can model startPhase > 0 (incomplete start codon)
1146 List<SequenceI> sequencesChecked = new ArrayList<>();
1147 AlignedCodon lastCodon = null;
1148 Map<SequenceI, AlignedCodon> toAdd = new HashMap<>();
1150 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1153 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1156 SequenceI seq = sequenceCodon.getKey();
1157 if (sequencesChecked.contains(seq))
1161 sequencesChecked.add(seq);
1162 AlignedCodon codon = sequenceCodon.getValue();
1163 if (codon.peptideCol > 1)
1166 "Problem mapping protein with >1 unmapped start positions: "
1169 else if (codon.peptideCol == 1)
1172 * first position (peptideCol == 0) was unmapped - add it
1174 if (lastCodon != null)
1176 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1177 lastCodon.pos2, lastCodon.pos3,
1178 String.valueOf(seq.getCharAt(0)), 0);
1179 toAdd.put(seq, firstPeptide);
1184 * unmapped residue at start of alignment (no prior column) -
1185 * 'insert' at nominal codon [0, 0, 0]
1187 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1188 String.valueOf(seq.getCharAt(0)), 0);
1189 toAdd.put(seq, firstPeptide);
1192 if (sequencesChecked.size() == mappedSequenceCount)
1194 // no need to check past first mapped position in all sequences
1198 lastCodon = entry.getKey();
1202 * add any new codons safely after iterating over the map
1204 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1206 addCodonToMap(alignedCodons, startCodon.getValue(),
1207 startCodon.getKey());
1212 * Update the aligned protein sequences to match the codon alignments given in
1216 * @param alignedCodons
1217 * an ordered map of codon positions (columns), with sequence/peptide
1218 * values present in each column
1219 * @param unmappedProtein
1222 protected static int alignProteinAs(AlignmentI protein,
1223 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1224 List<SequenceI> unmappedProtein)
1227 * prefill peptide sequences with gaps
1229 int alignedWidth = alignedCodons.size();
1230 char[] gaps = new char[alignedWidth];
1231 Arrays.fill(gaps, protein.getGapCharacter());
1232 Map<SequenceI, char[]> peptides = new HashMap<>();
1233 for (SequenceI seq : protein.getSequences())
1235 if (!unmappedProtein.contains(seq))
1237 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1242 * Traverse the codons left to right (as defined by CodonComparator)
1243 * and insert peptides in each column where the sequence is mapped.
1244 * This gives a peptide 'alignment' where residues are aligned if their
1245 * corresponding codons occupy the same columns in the cdna alignment.
1248 for (AlignedCodon codon : alignedCodons.keySet())
1250 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1252 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1254 char residue = entry.getValue().product.charAt(0);
1255 peptides.get(entry.getKey())[column] = residue;
1261 * and finally set the constructed sequences
1263 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1265 entry.getKey().setSequence(new String(entry.getValue()));
1272 * Populate the map of aligned codons by traversing the given sequence
1273 * mapping, locating the aligned positions of mapped codons, and adding those
1274 * positions and their translation products to the map.
1277 * the aligned sequence we are mapping from
1279 * the sequence to be aligned to the codons
1281 * the gap character in the dna sequence
1283 * a mapping to a sequence translation
1284 * @param alignedCodons
1285 * the map we are building up
1287 static void addCodonPositions(SequenceI dna, SequenceI protein,
1288 char gapChar, Mapping seqMap,
1289 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1291 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1294 * add codon positions, and their peptide translations, to the alignment
1295 * map, while remembering the first codon mapped
1297 while (codons.hasNext())
1301 AlignedCodon codon = codons.next();
1302 addCodonToMap(alignedCodons, codon, protein);
1303 } catch (IncompleteCodonException e)
1305 // possible incomplete trailing codon - ignore
1306 } catch (NoSuchElementException e)
1308 // possibly peptide lacking STOP
1314 * Helper method to add a codon-to-peptide entry to the aligned codons map
1316 * @param alignedCodons
1320 protected static void addCodonToMap(
1321 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1322 AlignedCodon codon, SequenceI protein)
1324 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1325 if (seqProduct == null)
1327 seqProduct = new HashMap<>();
1328 alignedCodons.put(codon, seqProduct);
1330 seqProduct.put(protein, codon);
1334 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1335 * between at least one pair of sequences in the two alignments. Currently,
1338 * <li>One alignment must be nucleotide, and the other protein</li>
1339 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1340 * <li>Mappable means the nucleotide translation matches the protein
1342 * <li>The translation may ignore start and stop codons if present in the
1350 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1352 if (al1 == null || al2 == null)
1358 * Require one nucleotide and one protein
1360 if (al1.isNucleotide() == al2.isNucleotide())
1364 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1365 AlignmentI protein = dna == al1 ? al2 : al1;
1366 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1367 for (SequenceI dnaSeq : dna.getSequences())
1369 for (SequenceI proteinSeq : protein.getSequences())
1371 if (isMappable(dnaSeq, proteinSeq, mappings))
1381 * Returns true if the dna sequence is mapped, or could be mapped, to the
1389 protected static boolean isMappable(SequenceI dnaSeq,
1390 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1392 if (dnaSeq == null || proteinSeq == null)
1397 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1398 : dnaSeq.getDatasetSequence();
1399 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1401 : proteinSeq.getDatasetSequence();
1403 for (AlignedCodonFrame mapping : mappings)
1405 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1415 * Just try to make a mapping (it is not yet stored), test whether
1418 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1422 * Finds any reference annotations associated with the sequences in
1423 * sequenceScope, that are not already added to the alignment, and adds them
1424 * to the 'candidates' map. Also populates a lookup table of annotation
1425 * labels, keyed by calcId, for use in constructing tooltips or the like.
1427 * @param sequenceScope
1428 * the sequences to scan for reference annotations
1429 * @param labelForCalcId
1430 * (optional) map to populate with label for calcId
1432 * map to populate with annotations for sequence
1434 * the alignment to check for presence of annotations
1436 public static void findAddableReferenceAnnotations(
1437 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1438 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1441 if (sequenceScope == null)
1447 * For each sequence in scope, make a list of any annotations on the
1448 * underlying dataset sequence which are not already on the alignment.
1450 * Add to a map of { alignmentSequence, <List of annotations to add> }
1452 for (SequenceI seq : sequenceScope)
1454 SequenceI dataset = seq.getDatasetSequence();
1455 if (dataset == null)
1459 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1460 if (datasetAnnotations == null)
1464 final List<AlignmentAnnotation> result = new ArrayList<>();
1465 for (AlignmentAnnotation dsann : datasetAnnotations)
1468 * Find matching annotations on the alignment. If none is found, then
1469 * add this annotation to the list of 'addable' annotations for this
1472 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1473 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1474 boolean found = false;
1475 if (matchedAlignmentAnnotations != null)
1477 for (AlignmentAnnotation matched : matchedAlignmentAnnotations)
1479 if (dsann.description.equals(matched.description))
1489 if (labelForCalcId != null)
1491 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1496 * Save any addable annotations for this sequence
1498 if (!result.isEmpty())
1500 candidates.put(seq, result);
1506 * Adds annotations to the top of the alignment annotations, in the same order
1507 * as their related sequences.
1509 * @param annotations
1510 * the annotations to add
1512 * the alignment to add them to
1513 * @param selectionGroup
1514 * current selection group (or null if none)
1516 public static void addReferenceAnnotations(
1517 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1518 final AlignmentI alignment, final SequenceGroup selectionGroup)
1520 for (SequenceI seq : annotations.keySet())
1522 for (AlignmentAnnotation ann : annotations.get(seq))
1524 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1526 int endRes = ann.annotations.length;
1527 if (selectionGroup != null)
1529 startRes = selectionGroup.getStartRes();
1530 endRes = selectionGroup.getEndRes();
1532 copyAnn.restrict(startRes, endRes + 0);
1535 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1536 * original annotation is already on the sequence.
1538 if (!seq.hasAnnotation(ann))
1540 ContactMatrixI cm = seq.getDatasetSequence()
1541 .getContactMatrixFor(ann);
1544 seq.addContactListFor(copyAnn, cm);
1546 seq.addAlignmentAnnotation(copyAnn);
1549 copyAnn.adjustForAlignment();
1550 // add to the alignment and set visible
1551 alignment.addAnnotation(copyAnn);
1552 copyAnn.visible = true;
1559 * Set visibility of alignment annotations of specified types (labels), for
1560 * specified sequences. This supports controls like "Show all secondary
1561 * structure", "Hide all Temp factor", etc.
1563 * @al the alignment to scan for annotations
1565 * the types (labels) of annotations to be updated
1566 * @param forSequences
1567 * if not null, only annotations linked to one of these sequences are
1568 * in scope for update; if null, acts on all sequence annotations
1570 * if this flag is true, 'types' is ignored (label not checked)
1572 * if true, set visibility on, else set off
1574 public static void showOrHideSequenceAnnotations(AlignmentI al,
1575 Collection<String> types, List<SequenceI> forSequences,
1576 boolean anyType, boolean doShow)
1578 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1581 for (AlignmentAnnotation aa : anns)
1583 if (anyType || types.contains(aa.label))
1585 if ((aa.sequenceRef != null) && (forSequences == null
1586 || forSequences.contains(aa.sequenceRef)))
1588 aa.visible = doShow;
1595 public static AlignmentAnnotation getFirstSequenceAnnotationOfType(
1596 AlignmentI al, int graphType)
1598 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1601 for (AlignmentAnnotation aa : anns)
1603 if (aa.sequenceRef != null && aa.graph == graphType)
1611 * Returns true if either sequence has a cross-reference to the other
1617 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1619 // Note: moved here from class CrossRef as the latter class has dependencies
1620 // not availability to the applet's classpath
1621 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1625 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1626 * that sequence name is structured as Source|AccessionId.
1632 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1634 if (seq1 == null || seq2 == null)
1638 String name = seq2.getName();
1639 final List<DBRefEntry> xrefs = seq1.getDBRefs();
1642 for (int ix = 0, nx = xrefs.size(); ix < nx; ix++)
1644 DBRefEntry xref = xrefs.get(ix);
1645 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1646 // case-insensitive test, consistent with DBRefEntry.equalRef()
1647 if (xrefName.equalsIgnoreCase(name))
1657 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1658 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1659 * added to the original alignment's dataset, which is shared by the new
1660 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1661 * added to the alignment dataset.
1664 * aligned nucleotide (dna or cds) sequences
1666 * the alignment dataset the sequences belong to
1668 * (optional) to restrict results to CDS that map to specified
1670 * @return an alignment whose sequences are the cds-only parts of the dna
1671 * sequences (or null if no mappings are found)
1673 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1674 AlignmentI dataset, SequenceI[] products)
1676 if (dataset == null || dataset.getDataset() != null)
1678 throw new IllegalArgumentException(
1679 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1681 List<SequenceI> foundSeqs = new ArrayList<>();
1682 List<SequenceI> cdsSeqs = new ArrayList<>();
1683 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1684 HashSet<SequenceI> productSeqs = null;
1685 if (products != null)
1687 productSeqs = new HashSet<>();
1688 for (SequenceI seq : products)
1690 productSeqs.add(seq.getDatasetSequence() == null ? seq
1691 : seq.getDatasetSequence());
1696 * Construct CDS sequences from mappings on the alignment dataset.
1698 * - find the protein product(s) mapped to from each dna sequence
1699 * - if the mapping covers the whole dna sequence (give or take start/stop
1700 * codon), take the dna as the CDS sequence
1701 * - else search dataset mappings for a suitable dna sequence, i.e. one
1702 * whose whole sequence is mapped to the protein
1703 * - if no sequence found, construct one from the dna sequence and mapping
1704 * (and add it to dataset so it is found if this is repeated)
1706 for (SequenceI dnaSeq : dna)
1708 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1709 : dnaSeq.getDatasetSequence();
1711 List<AlignedCodonFrame> seqMappings = MappingUtils
1712 .findMappingsForSequence(dnaSeq, mappings);
1713 for (AlignedCodonFrame mapping : seqMappings)
1715 List<Mapping> mappingsFromSequence = mapping
1716 .getMappingsFromSequence(dnaSeq);
1718 for (Mapping aMapping : mappingsFromSequence)
1720 MapList mapList = aMapping.getMap();
1721 if (mapList.getFromRatio() == 1)
1724 * not a dna-to-protein mapping (likely dna-to-cds)
1730 * skip if mapping is not to one of the target set of proteins
1732 SequenceI proteinProduct = aMapping.getTo();
1733 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1739 * try to locate the CDS from the dataset mappings;
1740 * guard against duplicate results (for the case that protein has
1741 * dbrefs to both dna and cds sequences)
1743 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1744 seqMappings, aMapping);
1747 if (!foundSeqs.contains(cdsSeq))
1749 foundSeqs.add(cdsSeq);
1750 SequenceI derivedSequence = cdsSeq.deriveSequence();
1751 cdsSeqs.add(derivedSequence);
1752 if (!dataset.getSequences().contains(cdsSeq))
1754 dataset.addSequence(cdsSeq);
1761 * didn't find mapped CDS sequence - construct it and add
1762 * its dataset sequence to the dataset
1764 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1765 dataset).deriveSequence();
1766 // cdsSeq has a name constructed as CDS|<dbref>
1767 // <dbref> will be either the accession for the coding sequence,
1768 // marked in the /via/ dbref to the protein product accession
1769 // or it will be the original nucleotide accession.
1770 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1772 cdsSeqs.add(cdsSeq);
1775 * build the mapping from CDS to protein
1777 List<int[]> cdsRange = Collections
1778 .singletonList(new int[]
1779 { cdsSeq.getStart(),
1780 cdsSeq.getLength() + cdsSeq.getStart() - 1 });
1781 MapList cdsToProteinMap = new MapList(cdsRange,
1782 mapList.getToRanges(), mapList.getFromRatio(),
1783 mapList.getToRatio());
1785 if (!dataset.getSequences().contains(cdsSeqDss))
1788 * if this sequence is a newly created one, add it to the dataset
1789 * and made a CDS to protein mapping (if sequence already exists,
1790 * CDS-to-protein mapping _is_ the transcript-to-protein mapping)
1792 dataset.addSequence(cdsSeqDss);
1793 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1794 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1798 * guard against duplicating the mapping if repeating this action
1800 if (!mappings.contains(cdsToProteinMapping))
1802 mappings.add(cdsToProteinMapping);
1806 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1807 proteinProduct, aMapping);
1809 * add another mapping from original 'from' range to CDS
1811 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1812 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1814 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1816 if (!mappings.contains(dnaToCdsMapping))
1818 mappings.add(dnaToCdsMapping);
1822 * transfer dna chromosomal loci (if known) to the CDS
1823 * sequence (via the mapping)
1825 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1826 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1829 * add DBRef with mapping from protein to CDS
1830 * (this enables Get Cross-References from protein alignment)
1831 * This is tricky because we can't have two DBRefs with the
1832 * same source and accession, so need a different accession for
1833 * the CDS from the dna sequence
1836 // specific use case:
1837 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1838 // ENSG02, ENSG03, with transcripts and products similarly named.
1839 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1841 // JBPNote: ?? can't actually create an example that demonstrates we
1843 // synthesize an xref.
1845 List<DBRefEntry> primrefs = dnaDss.getPrimaryDBRefs();
1846 for (int ip = 0, np = primrefs.size(); ip < np; ip++)
1848 DBRefEntry primRef = primrefs.get(ip);
1850 * create a cross-reference from CDS to the source sequence's
1851 * primary reference and vice versa
1853 String source = primRef.getSource();
1854 String version = primRef.getVersion();
1855 DBRefEntry cdsCrossRef = new DBRefEntry(source,
1856 source + ":" + version, primRef.getAccessionId());
1858 .setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
1859 cdsSeqDss.addDBRef(cdsCrossRef);
1861 dnaSeq.addDBRef(new DBRefEntry(source, version,
1862 cdsSeq.getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
1863 // problem here is that the cross-reference is synthesized -
1864 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1866 // assuming cds version same as dna ?!?
1868 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
1871 proteinToCdsRef.setMap(
1872 new Mapping(cdsSeqDss, cdsToProteinMap.getInverse()));
1873 proteinProduct.addDBRef(proteinToCdsRef);
1876 * transfer any features on dna that overlap the CDS
1878 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1879 SequenceOntologyI.CDS);
1884 AlignmentI cds = new Alignment(
1885 cdsSeqs.toArray(new SequenceI[cdsSeqs.size()]));
1886 cds.setDataset(dataset);
1892 * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
1893 * toSeq, mediated by the given mapping between the sequences
1896 * @param targetToFrom
1900 protected static void transferGeneLoci(SequenceI fromSeq,
1901 MapList targetToFrom, SequenceI targetSeq)
1903 if (targetSeq.getGeneLoci() != null)
1905 // already have - don't override
1908 GeneLociI fromLoci = fromSeq.getGeneLoci();
1909 if (fromLoci == null)
1914 MapList newMap = targetToFrom.traverse(fromLoci.getMapping());
1918 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
1919 fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
1924 * A helper method that finds a CDS sequence in the alignment dataset that is
1925 * mapped to the given protein sequence, and either is, or has a mapping from,
1926 * the given dna sequence.
1929 * set of all mappings on the dataset
1931 * a dna (or cds) sequence we are searching from
1932 * @param seqMappings
1933 * the set of mappings involving dnaSeq
1935 * a transcript-to-peptide mapping
1938 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1939 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1943 * TODO a better dna-cds-protein mapping data representation to allow easy
1944 * navigation; until then this clunky looping around lists of mappings
1946 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1947 : dnaSeq.getDatasetSequence();
1948 SequenceI proteinProduct = aMapping.getTo();
1951 * is this mapping from the whole dna sequence (i.e. CDS)?
1952 * allowing for possible stop codon on dna but not peptide
1954 int mappedFromLength = MappingUtils
1955 .getLength(aMapping.getMap().getFromRanges());
1956 int dnaLength = seqDss.getLength();
1957 if (mappedFromLength == dnaLength
1958 || mappedFromLength == dnaLength - CODON_LENGTH)
1961 * if sequence has CDS features, this is a transcript with no UTR
1962 * - do not take this as the CDS sequence! (JAL-2789)
1964 if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
1972 * looks like we found the dna-to-protein mapping; search for the
1973 * corresponding cds-to-protein mapping
1975 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1976 .findMappingsForSequence(proteinProduct, mappings);
1977 for (AlignedCodonFrame acf : mappingsToPeptide)
1979 for (SequenceToSequenceMapping map : acf.getMappings())
1981 Mapping mapping = map.getMapping();
1982 if (mapping != aMapping
1983 && mapping.getMap().getFromRatio() == CODON_LENGTH
1984 && proteinProduct == mapping.getTo()
1985 && seqDss != map.getFromSeq())
1987 mappedFromLength = MappingUtils
1988 .getLength(mapping.getMap().getFromRanges());
1989 if (mappedFromLength == map.getFromSeq().getLength())
1992 * found a 3:1 mapping to the protein product which covers
1993 * the whole dna sequence i.e. is from CDS; finally check the CDS
1994 * is mapped from the given dna start sequence
1996 SequenceI cdsSeq = map.getFromSeq();
1997 // todo this test is weak if seqMappings contains multiple mappings;
1998 // we get away with it if transcript:cds relationship is 1:1
1999 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
2000 .findMappingsForSequence(cdsSeq, seqMappings);
2001 if (!dnaToCdsMaps.isEmpty())
2013 * Helper method that makes a CDS sequence as defined by the mappings from the
2014 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
2015 * forward or reverse strand).
2020 * - existing dataset. We check for sequences that look like the CDS
2021 * we are about to construct, if one exists already, then we will
2022 * just return that one.
2023 * @return CDS sequence (as a dataset sequence)
2025 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
2029 * construct CDS sequence name as "CDS|" with 'from id' held in the mapping
2030 * if set (e.g. EMBL protein_id), else sequence name appended
2032 String mapFromId = mapping.getMappedFromId();
2033 final String seqId = "CDS|"
2034 + (mapFromId != null ? mapFromId : seq.getName());
2036 SequenceI newSeq = null;
2039 * construct CDS sequence by splicing mapped from ranges
2041 char[] seqChars = seq.getSequence();
2042 List<int[]> fromRanges = mapping.getMap().getFromRanges();
2043 int cdsWidth = MappingUtils.getLength(fromRanges);
2044 char[] newSeqChars = new char[cdsWidth];
2047 for (int[] range : fromRanges)
2049 if (range[0] <= range[1])
2051 // forward strand mapping - just copy the range
2052 int length = range[1] - range[0] + 1;
2053 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
2059 // reverse strand mapping - copy and complement one by one
2060 for (int i = range[0]; i >= range[1]; i--)
2062 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
2066 newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2069 if (dataset != null)
2071 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2072 if (matches != null)
2074 boolean matched = false;
2075 for (SequenceI mtch : matches)
2077 if (mtch.getStart() != newSeq.getStart())
2081 if (mtch.getEnd() != newSeq.getEnd())
2085 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2097 "JAL-2154 regression: warning - found (and ignored) a duplicate CDS sequence:"
2103 // newSeq.setDescription(mapFromId);
2109 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2110 * the given mapping.
2114 * @param proteinProduct
2116 * @return list of DBRefEntrys added
2118 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2119 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2122 // gather direct refs from contig congruent with mapping
2123 List<DBRefEntry> direct = new ArrayList<>();
2124 HashSet<String> directSources = new HashSet<>();
2126 List<DBRefEntry> refs = contig.getDBRefs();
2129 for (int ib = 0, nb = refs.size(); ib < nb; ib++)
2131 DBRefEntry dbr = refs.get(ib);
2133 if (dbr.hasMap() && (map = dbr.getMap().getMap()).isTripletMap())
2135 // check if map is the CDS mapping
2136 if (mapping.getMap().equals(map))
2139 directSources.add(dbr.getSource());
2144 List<DBRefEntry> onSource = DBRefUtils.selectRefs(
2145 proteinProduct.getDBRefs(),
2146 directSources.toArray(new String[0]));
2147 List<DBRefEntry> propagated = new ArrayList<>();
2149 // and generate appropriate mappings
2150 for (int ic = 0, nc = direct.size(); ic < nc; ic++)
2152 DBRefEntry cdsref = direct.get(ic);
2153 Mapping m = cdsref.getMap();
2154 // clone maplist and mapping
2155 MapList cdsposmap = new MapList(
2156 Arrays.asList(new int[][]
2157 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2158 m.getMap().getToRanges(), 3, 1);
2159 Mapping cdsmap = new Mapping(m.getTo(), m.getMap());
2162 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2163 cdsref.getVersion(), cdsref.getAccessionId(),
2164 new Mapping(cdsmap.getTo(), cdsposmap));
2166 // and see if we can map to the protein product for this mapping.
2167 // onSource is the filtered set of accessions on protein that we are
2168 // tranferring, so we assume accession is the same.
2169 if (cdsmap.getTo() == null && onSource != null)
2171 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2172 cdsref.getAccessionId());
2173 if (sourceRefs != null)
2175 for (DBRefEntry srcref : sourceRefs)
2177 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2179 // we have found a complementary dbref on the protein product, so
2180 // update mapping's getTo
2181 newref.getMap().setTo(proteinProduct);
2186 cdsSeq.addDBRef(newref);
2187 propagated.add(newref);
2193 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2194 * feature start/end ranges, optionally omitting specified feature types.
2195 * Returns the number of features copied.
2200 * the mapping from 'fromSeq' to 'toSeq'
2202 * if not null, only features of this type are copied (including
2203 * subtypes in the Sequence Ontology)
2206 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2207 MapList mapping, String select, String... omitting)
2209 SequenceI copyTo = toSeq;
2210 while (copyTo.getDatasetSequence() != null)
2212 copyTo = copyTo.getDatasetSequence();
2214 if (fromSeq == copyTo || fromSeq.getDatasetSequence() == copyTo)
2216 return 0; // shared dataset sequence
2220 * get features, optionally restricted by an ontology term
2222 List<SequenceFeature> sfs = select == null
2223 ? fromSeq.getFeatures().getPositionalFeatures()
2224 : fromSeq.getFeatures().getFeaturesByOntology(select);
2227 for (SequenceFeature sf : sfs)
2229 String type = sf.getType();
2230 boolean omit = false;
2231 for (String toOmit : omitting)
2233 if (type.equals(toOmit))
2244 * locate the mapped range - null if either start or end is
2245 * not mapped (no partial overlaps are calculated)
2247 int start = sf.getBegin();
2248 int end = sf.getEnd();
2249 int[] mappedTo = mapping.locateInTo(start, end);
2251 * if whole exon range doesn't map, try interpreting it
2252 * as 5' or 3' exon overlapping the CDS range
2254 if (mappedTo == null)
2256 mappedTo = mapping.locateInTo(end, end);
2257 if (mappedTo != null)
2260 * end of exon is in CDS range - 5' overlap
2261 * to a range from the start of the peptide
2266 if (mappedTo == null)
2268 mappedTo = mapping.locateInTo(start, start);
2269 if (mappedTo != null)
2272 * start of exon is in CDS range - 3' overlap
2273 * to a range up to the end of the peptide
2275 mappedTo[1] = toSeq.getLength();
2278 if (mappedTo != null)
2280 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2281 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2282 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2283 sf.getFeatureGroup(), sf.getScore());
2284 copyTo.addSequenceFeature(copy);
2292 * Returns a mapping from dna to protein by inspecting sequence features of
2293 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2294 * length is 3 times the peptide length (optionally after dropping a trailing
2295 * stop codon). This method does not check whether the CDS nucleotide sequence
2296 * translates to the peptide sequence.
2302 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2303 SequenceI proteinSeq)
2305 List<int[]> ranges = findCdsPositions(dnaSeq);
2306 int mappedDnaLength = MappingUtils.getLength(ranges);
2309 * if not a whole number of codons, truncate mapping
2311 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2312 if (codonRemainder > 0)
2314 mappedDnaLength -= codonRemainder;
2315 MappingUtils.removeEndPositions(codonRemainder, ranges);
2318 int proteinLength = proteinSeq.getLength();
2319 int proteinStart = proteinSeq.getStart();
2320 int proteinEnd = proteinSeq.getEnd();
2323 * incomplete start codon may mean X at start of peptide
2324 * we ignore both for mapping purposes
2326 if (proteinSeq.getCharAt(0) == 'X')
2328 // todo JAL-2022 support startPhase > 0
2332 List<int[]> proteinRange = new ArrayList<>();
2335 * dna length should map to protein (or protein plus stop codon)
2337 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2338 if (codesForResidues == (proteinLength + 1))
2340 // assuming extra codon is for STOP and not in peptide
2341 // todo: check trailing codon is indeed a STOP codon
2343 mappedDnaLength -= CODON_LENGTH;
2344 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2347 if (codesForResidues == proteinLength)
2349 proteinRange.add(new int[] { proteinStart, proteinEnd });
2350 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2356 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2357 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2358 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2359 * position order, so this method is only valid for linear CDS in the same
2360 * sense as the protein product.
2365 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2367 List<int[]> result = new ArrayList<>();
2369 List<SequenceFeature> sfs = dnaSeq.getFeatures()
2370 .getFeaturesByOntology(SequenceOntologyI.CDS);
2375 SequenceFeatures.sortFeatures(sfs, true);
2377 for (SequenceFeature sf : sfs)
2382 String s = sf.getPhase();
2385 phase = Integer.parseInt(s);
2387 } catch (NumberFormatException e)
2392 * phase > 0 on first codon means 5' incomplete - skip to the start
2393 * of the next codon; example ENST00000496384
2395 int begin = sf.getBegin();
2396 int end = sf.getEnd();
2397 if (result.isEmpty() && phase > 0)
2402 // shouldn't happen!
2404 .println("Error: start phase extends beyond start CDS in "
2405 + dnaSeq.getName());
2408 result.add(new int[] { begin, end });
2412 * Finally sort ranges by start position. This avoids a dependency on
2413 * keeping features in order on the sequence (if they are in order anyway,
2414 * the sort will have almost no work to do). The implicit assumption is CDS
2415 * ranges are assembled in order. Other cases should not use this method,
2416 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2418 Collections.sort(result, IntRangeComparator.ASCENDING);
2423 * Makes an alignment with a copy of the given sequences, adding in any
2424 * non-redundant sequences which are mapped to by the cross-referenced
2430 * the alignment dataset shared by the new copy
2433 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2434 SequenceI[] xrefs, AlignmentI dataset)
2436 AlignmentI copy = new Alignment(new Alignment(seqs));
2437 copy.setDataset(dataset);
2438 boolean isProtein = !copy.isNucleotide();
2439 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2442 // BH 2019.01.25 recoded to remove iterators
2444 for (int ix = 0, nx = xrefs.length; ix < nx; ix++)
2446 SequenceI xref = xrefs[ix];
2447 List<DBRefEntry> dbrefs = xref.getDBRefs();
2450 for (int ir = 0, nir = dbrefs.size(); ir < nir; ir++)
2452 DBRefEntry dbref = dbrefs.get(ir);
2453 Mapping map = dbref.getMap();
2455 if (map == null || (mto = map.getTo()) == null
2456 || mto.isProtein() != isProtein)
2460 SequenceI mappedTo = mto;
2461 SequenceI match = matcher.findIdMatch(mappedTo);
2464 matcher.add(mappedTo);
2465 copy.addSequence(mappedTo);
2475 * Try to align sequences in 'unaligned' to match the alignment of their
2476 * mapped regions in 'aligned'. For example, could use this to align CDS
2477 * sequences which are mapped to their parent cDNA sequences.
2479 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2480 * dna-to-protein or protein-to-dna use alternative methods.
2483 * sequences to be aligned
2485 * holds aligned sequences and their mappings
2488 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2491 * easy case - aligning a copy of aligned sequences
2493 if (alignAsSameSequences(unaligned, aligned))
2495 return unaligned.getHeight();
2499 * fancy case - aligning via mappings between sequences
2501 List<SequenceI> unmapped = new ArrayList<>();
2502 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2503 unaligned, aligned, unmapped);
2504 int width = columnMap.size();
2505 char gap = unaligned.getGapCharacter();
2506 int realignedCount = 0;
2507 // TODO: verify this loop scales sensibly for very wide/high alignments
2509 for (SequenceI seq : unaligned.getSequences())
2511 if (!unmapped.contains(seq))
2513 char[] newSeq = new char[width];
2514 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2515 // Integer iteration below
2520 * traverse the map to find columns populated
2523 for (Integer column : columnMap.keySet())
2525 Character c = columnMap.get(column).get(seq);
2529 * sequence has a character at this position
2539 * trim trailing gaps
2541 if (lastCol < width)
2543 char[] tmp = new char[lastCol + 1];
2544 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2547 // TODO: optimise SequenceI to avoid char[]->String->char[]
2548 seq.setSequence(String.valueOf(newSeq));
2552 return realignedCount;
2556 * If unaligned and aligned sequences share the same dataset sequences, then
2557 * simply copies the aligned sequences to the unaligned sequences and returns
2558 * true; else returns false
2561 * - sequences to be aligned based on aligned
2563 * - 'guide' alignment containing sequences derived from same dataset
2567 static boolean alignAsSameSequences(AlignmentI unaligned,
2570 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2572 return false; // should only pass alignments with datasets here
2575 // map from dataset sequence to alignment sequence(s)
2576 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2577 for (SequenceI seq : aligned.getSequences())
2579 SequenceI ds = seq.getDatasetSequence();
2580 if (alignedDatasets.get(ds) == null)
2582 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2584 alignedDatasets.get(ds).add(seq);
2588 * first pass - check whether all sequences to be aligned share a
2589 * dataset sequence with an aligned sequence; also note the leftmost
2590 * ungapped column from which to copy
2592 int leftmost = Integer.MAX_VALUE;
2593 for (SequenceI seq : unaligned.getSequences())
2595 final SequenceI ds = seq.getDatasetSequence();
2596 if (!alignedDatasets.containsKey(ds))
2600 SequenceI alignedSeq = alignedDatasets.get(ds).get(0);
2601 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2602 leftmost = Math.min(leftmost, startCol);
2606 * second pass - copy aligned sequences;
2607 * heuristic rule: pair off sequences in order for the case where
2608 * more than one shares the same dataset sequence
2610 final char gapCharacter = aligned.getGapCharacter();
2611 for (SequenceI seq : unaligned.getSequences())
2613 List<SequenceI> alignedSequences = alignedDatasets
2614 .get(seq.getDatasetSequence());
2615 if (alignedSequences.isEmpty())
2618 * defensive check - shouldn't happen! (JAL-3536)
2622 SequenceI alignedSeq = alignedSequences.get(0);
2625 * gap fill for leading (5') UTR if any
2627 // TODO this copies intron columns - wrong!
2628 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2629 int endCol = alignedSeq.findIndex(seq.getEnd());
2630 char[] seqchars = new char[endCol - leftmost + 1];
2631 Arrays.fill(seqchars, gapCharacter);
2632 char[] toCopy = alignedSeq.getSequence(startCol - 1, endCol);
2633 System.arraycopy(toCopy, 0, seqchars, startCol - leftmost,
2635 seq.setSequence(String.valueOf(seqchars));
2636 if (alignedSequences.size() > 0)
2638 // pop off aligned sequences (except the last one)
2639 alignedSequences.remove(0);
2644 * finally remove gapped columns (e.g. introns)
2646 new RemoveGapColCommand("", unaligned.getSequencesArray(), 0,
2647 unaligned.getWidth() - 1, unaligned);
2653 * Returns a map whose key is alignment column number (base 1), and whose
2654 * values are a map of sequence characters in that column.
2661 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2662 AlignmentI unaligned, AlignmentI aligned,
2663 List<SequenceI> unmapped)
2666 * Map will hold, for each aligned column position, a map of
2667 * {unalignedSequence, characterPerSequence} at that position.
2668 * TreeMap keeps the entries in ascending column order.
2670 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2673 * record any sequences that have no mapping so can't be realigned
2675 unmapped.addAll(unaligned.getSequences());
2677 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2679 for (SequenceI seq : unaligned.getSequences())
2681 for (AlignedCodonFrame mapping : mappings)
2683 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2684 if (fromSeq != null)
2686 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2687 if (addMappedPositions(seq, fromSeq, seqMap, map))
2689 unmapped.remove(seq);
2698 * Helper method that adds to a map the mapped column positions of a sequence.
2700 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2701 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2705 * the sequence whose column positions we are recording
2707 * a sequence that is mapped to the first sequence
2709 * the mapping from 'fromSeq' to 'seq'
2711 * a map to add the column positions (in fromSeq) of the mapped
2715 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2716 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2724 * invert mapping if it is from unaligned to aligned sequence
2726 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2728 seqMap = new Mapping(seq.getDatasetSequence(),
2729 seqMap.getMap().getInverse());
2732 int toStart = seq.getStart();
2735 * traverse [start, end, start, end...] ranges in fromSeq
2737 for (int[] fromRange : seqMap.getMap().getFromRanges())
2739 for (int i = 0; i < fromRange.length - 1; i += 2)
2741 boolean forward = fromRange[i + 1] >= fromRange[i];
2744 * find the range mapped to (sequence positions base 1)
2746 int[] range = seqMap.locateMappedRange(fromRange[i],
2750 System.err.println("Error in mapping " + seqMap + " from "
2751 + fromSeq.getName());
2754 int fromCol = fromSeq.findIndex(fromRange[i]);
2755 int mappedCharPos = range[0];
2758 * walk over the 'from' aligned sequence in forward or reverse
2759 * direction; when a non-gap is found, record the column position
2760 * of the next character of the mapped-to sequence; stop when all
2761 * the characters of the range have been counted
2763 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
2766 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
2769 * mapped from sequence has a character in this column
2770 * record the column position for the mapped to character
2772 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2773 if (seqsMap == null)
2775 seqsMap = new HashMap<>();
2776 map.put(fromCol, seqsMap);
2778 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
2781 fromCol += (forward ? 1 : -1);
2788 // strictly temporary hack until proper criteria for aligning protein to cds
2789 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2790 public static boolean looksLikeEnsembl(AlignmentI alignment)
2792 for (SequenceI seq : alignment.getSequences())
2794 String name = seq.getName();
2795 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))