2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.analysis;
23 import static jalview.io.gff.GffConstants.CLINICAL_SIGNIFICANCE;
25 import jalview.datamodel.AlignedCodon;
26 import jalview.datamodel.AlignedCodonFrame;
27 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
28 import jalview.datamodel.Alignment;
29 import jalview.datamodel.AlignmentAnnotation;
30 import jalview.datamodel.AlignmentI;
31 import jalview.datamodel.DBRefEntry;
32 import jalview.datamodel.GeneLociI;
33 import jalview.datamodel.IncompleteCodonException;
34 import jalview.datamodel.Mapping;
35 import jalview.datamodel.Sequence;
36 import jalview.datamodel.SequenceFeature;
37 import jalview.datamodel.SequenceGroup;
38 import jalview.datamodel.SequenceI;
39 import jalview.datamodel.features.SequenceFeatures;
40 import jalview.io.gff.Gff3Helper;
41 import jalview.io.gff.SequenceOntologyI;
42 import jalview.schemes.ResidueProperties;
43 import jalview.util.Comparison;
44 import jalview.util.DBRefUtils;
45 import jalview.util.IntRangeComparator;
46 import jalview.util.MapList;
47 import jalview.util.MappingUtils;
48 import jalview.util.StringUtils;
50 import java.io.UnsupportedEncodingException;
51 import java.net.URLEncoder;
52 import java.util.ArrayList;
53 import java.util.Arrays;
54 import java.util.Collection;
55 import java.util.Collections;
56 import java.util.HashMap;
57 import java.util.HashSet;
58 import java.util.Iterator;
59 import java.util.LinkedHashMap;
60 import java.util.List;
62 import java.util.Map.Entry;
63 import java.util.NoSuchElementException;
65 import java.util.SortedMap;
66 import java.util.TreeMap;
69 * grab bag of useful alignment manipulation operations Expect these to be
70 * refactored elsewhere at some point.
75 public class AlignmentUtils
77 private static final int CODON_LENGTH = 3;
79 private static final String SEQUENCE_VARIANT = "sequence_variant:";
82 * the 'id' attribute is provided for variant features fetched from
83 * Ensembl using its REST service with JSON format
85 public static final String VARIANT_ID = "id";
88 * A data model to hold the 'normal' base value at a position, and an optional
89 * sequence variant feature
91 static final class DnaVariant
95 SequenceFeature variant;
97 DnaVariant(String nuc)
103 DnaVariant(String nuc, SequenceFeature var)
109 public String getSource()
111 return variant == null ? null : variant.getFeatureGroup();
115 * toString for aid in the debugger only
118 public String toString()
120 return base + ":" + (variant == null ? "" : variant.getDescription());
125 * given an existing alignment, create a new alignment including all, or up to
126 * flankSize additional symbols from each sequence's dataset sequence
132 public static AlignmentI expandContext(AlignmentI core, int flankSize)
134 List<SequenceI> sq = new ArrayList<>();
136 for (SequenceI s : core.getSequences())
138 SequenceI newSeq = s.deriveSequence();
139 final int newSeqStart = newSeq.getStart() - 1;
140 if (newSeqStart > maxoffset
141 && newSeq.getDatasetSequence().getStart() < s.getStart())
143 maxoffset = newSeqStart;
149 maxoffset = Math.min(maxoffset, flankSize);
153 * now add offset left and right to create an expanded alignment
155 for (SequenceI s : sq)
158 while (ds.getDatasetSequence() != null)
160 ds = ds.getDatasetSequence();
162 int s_end = s.findPosition(s.getStart() + s.getLength());
163 // find available flanking residues for sequence
164 int ustream_ds = s.getStart() - ds.getStart();
165 int dstream_ds = ds.getEnd() - s_end;
167 // build new flanked sequence
169 // compute gap padding to start of flanking sequence
170 int offset = maxoffset - ustream_ds;
172 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
175 if (flankSize < ustream_ds)
177 // take up to flankSize residues
178 offset = maxoffset - flankSize;
179 ustream_ds = flankSize;
181 if (flankSize <= dstream_ds)
183 dstream_ds = flankSize - 1;
186 // TODO use Character.toLowerCase to avoid creating String objects?
187 char[] upstream = new String(ds
188 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
189 .toLowerCase().toCharArray();
190 char[] downstream = new String(
191 ds.getSequence(s_end - 1, s_end + dstream_ds)).toLowerCase()
193 char[] coreseq = s.getSequence();
194 char[] nseq = new char[offset + upstream.length + downstream.length
196 char c = core.getGapCharacter();
199 for (; p < offset; p++)
204 System.arraycopy(upstream, 0, nseq, p, upstream.length);
205 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
207 System.arraycopy(downstream, 0, nseq,
208 p + coreseq.length + upstream.length, downstream.length);
209 s.setSequence(new String(nseq));
210 s.setStart(s.getStart() - ustream_ds);
211 s.setEnd(s_end + downstream.length);
213 AlignmentI newAl = new jalview.datamodel.Alignment(
214 sq.toArray(new SequenceI[0]));
215 for (SequenceI s : sq)
217 if (s.getAnnotation() != null)
219 for (AlignmentAnnotation aa : s.getAnnotation())
221 aa.adjustForAlignment(); // JAL-1712 fix
222 newAl.addAnnotation(aa);
226 newAl.setDataset(core.getDataset());
231 * Returns the index (zero-based position) of a sequence in an alignment, or
238 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
242 for (SequenceI alSeq : al.getSequences())
255 * Returns a map of lists of sequences in the alignment, keyed by sequence
256 * name. For use in mapping between different alignment views of the same
259 * @see jalview.datamodel.AlignmentI#getSequencesByName()
261 public static Map<String, List<SequenceI>> getSequencesByName(
264 Map<String, List<SequenceI>> theMap = new LinkedHashMap<>();
265 for (SequenceI seq : al.getSequences())
267 String name = seq.getName();
270 List<SequenceI> seqs = theMap.get(name);
273 seqs = new ArrayList<>();
274 theMap.put(name, seqs);
283 * Build mapping of protein to cDNA alignment. Mappings are made between
284 * sequences where the cDNA translates to the protein sequence. Any new
285 * mappings are added to the protein alignment. Returns true if any mappings
286 * either already exist or were added, else false.
288 * @param proteinAlignment
289 * @param cdnaAlignment
292 public static boolean mapProteinAlignmentToCdna(
293 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
295 if (proteinAlignment == null || cdnaAlignment == null)
300 Set<SequenceI> mappedDna = new HashSet<>();
301 Set<SequenceI> mappedProtein = new HashSet<>();
304 * First pass - map sequences where cross-references exist. This include
305 * 1-to-many mappings to support, for example, variant cDNA.
307 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
308 cdnaAlignment, mappedDna, mappedProtein, true);
311 * Second pass - map sequences where no cross-references exist. This only
312 * does 1-to-1 mappings and assumes corresponding sequences are in the same
313 * order in the alignments.
315 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
316 mappedDna, mappedProtein, false);
317 return mappingPerformed;
321 * Make mappings between compatible sequences (where the cDNA translation
322 * matches the protein).
324 * @param proteinAlignment
325 * @param cdnaAlignment
327 * a set of mapped DNA sequences (to add to)
328 * @param mappedProtein
329 * a set of mapped Protein sequences (to add to)
331 * if true, only map sequences where xrefs exist
334 protected static boolean mapProteinToCdna(
335 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
336 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
339 boolean mappingExistsOrAdded = false;
340 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
341 for (SequenceI aaSeq : thisSeqs)
343 boolean proteinMapped = false;
344 AlignedCodonFrame acf = new AlignedCodonFrame();
346 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
349 * Always try to map if sequences have xref to each other; this supports
350 * variant cDNA or alternative splicing for a protein sequence.
352 * If no xrefs, try to map progressively, assuming that alignments have
353 * mappable sequences in corresponding order. These are not
354 * many-to-many, as that would risk mixing species with similar cDNA
357 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
363 * Don't map non-xrefd sequences more than once each. This heuristic
364 * allows us to pair up similar sequences in ordered alignments.
366 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
367 || mappedDna.contains(cdnaSeq)))
371 if (mappingExists(proteinAlignment.getCodonFrames(),
372 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
374 mappingExistsOrAdded = true;
378 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
381 acf.addMap(cdnaSeq, aaSeq, map);
382 mappingExistsOrAdded = true;
383 proteinMapped = true;
384 mappedDna.add(cdnaSeq);
385 mappedProtein.add(aaSeq);
391 proteinAlignment.addCodonFrame(acf);
394 return mappingExistsOrAdded;
398 * Answers true if the mappings include one between the given (dataset)
401 protected static boolean mappingExists(List<AlignedCodonFrame> mappings,
402 SequenceI aaSeq, SequenceI cdnaSeq)
404 if (mappings != null)
406 for (AlignedCodonFrame acf : mappings)
408 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
418 * Builds a mapping (if possible) of a cDNA to a protein sequence.
420 * <li>first checks if the cdna translates exactly to the protein
422 * <li>else checks for translation after removing a STOP codon</li>
423 * <li>else checks for translation after removing a START codon</li>
424 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
426 * Returns null if no mapping is determined.
429 * the aligned protein sequence
431 * the aligned cdna sequence
434 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
438 * Here we handle either dataset sequence set (desktop) or absent (applet).
439 * Use only the char[] form of the sequence to avoid creating possibly large
442 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
443 char[] aaSeqChars = proteinDataset != null
444 ? proteinDataset.getSequence()
445 : proteinSeq.getSequence();
446 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
447 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
448 : cdnaSeq.getSequence();
449 if (aaSeqChars == null || cdnaSeqChars == null)
455 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
457 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
458 int cdnaLength = cdnaSeqChars.length;
459 int cdnaStart = cdnaSeq.getStart();
460 int cdnaEnd = cdnaSeq.getEnd();
461 final int proteinStart = proteinSeq.getStart();
462 final int proteinEnd = proteinSeq.getEnd();
465 * If lengths don't match, try ignoring stop codon (if present)
467 if (cdnaLength != mappedLength && cdnaLength > 2)
469 String lastCodon = String.valueOf(cdnaSeqChars,
470 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
471 for (String stop : ResidueProperties.STOP_CODONS)
473 if (lastCodon.equals(stop))
475 cdnaEnd -= CODON_LENGTH;
476 cdnaLength -= CODON_LENGTH;
483 * If lengths still don't match, try ignoring start codon.
486 if (cdnaLength != mappedLength && cdnaLength > 2
487 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
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.getLength(mapList
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 if (!matchedAlignmentAnnotations.iterator().hasNext())
1477 if (labelForCalcId != null)
1479 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1484 * Save any addable annotations for this sequence
1486 if (!result.isEmpty())
1488 candidates.put(seq, result);
1494 * Adds annotations to the top of the alignment annotations, in the same order
1495 * as their related sequences.
1497 * @param annotations
1498 * the annotations to add
1500 * the alignment to add them to
1501 * @param selectionGroup
1502 * current selection group (or null if none)
1504 public static void addReferenceAnnotations(
1505 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1506 final AlignmentI alignment, final SequenceGroup selectionGroup)
1508 for (SequenceI seq : annotations.keySet())
1510 for (AlignmentAnnotation ann : annotations.get(seq))
1512 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1514 int endRes = ann.annotations.length;
1515 if (selectionGroup != null)
1517 startRes = selectionGroup.getStartRes();
1518 endRes = selectionGroup.getEndRes();
1520 copyAnn.restrict(startRes, endRes);
1523 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1524 * original annotation is already on the sequence.
1526 if (!seq.hasAnnotation(ann))
1528 seq.addAlignmentAnnotation(copyAnn);
1531 copyAnn.adjustForAlignment();
1532 // add to the alignment and set visible
1533 alignment.addAnnotation(copyAnn);
1534 copyAnn.visible = true;
1540 * Set visibility of alignment annotations of specified types (labels), for
1541 * specified sequences. This supports controls like "Show all secondary
1542 * structure", "Hide all Temp factor", etc.
1544 * @al the alignment to scan for annotations
1546 * the types (labels) of annotations to be updated
1547 * @param forSequences
1548 * if not null, only annotations linked to one of these sequences are
1549 * in scope for update; if null, acts on all sequence annotations
1551 * if this flag is true, 'types' is ignored (label not checked)
1553 * if true, set visibility on, else set off
1555 public static void showOrHideSequenceAnnotations(AlignmentI al,
1556 Collection<String> types, List<SequenceI> forSequences,
1557 boolean anyType, boolean doShow)
1559 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1562 for (AlignmentAnnotation aa : anns)
1564 if (anyType || types.contains(aa.label))
1566 if ((aa.sequenceRef != null) && (forSequences == null
1567 || forSequences.contains(aa.sequenceRef)))
1569 aa.visible = doShow;
1577 * Returns true if either sequence has a cross-reference to the other
1583 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1585 // Note: moved here from class CrossRef as the latter class has dependencies
1586 // not availability to the applet's classpath
1587 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1591 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1592 * that sequence name is structured as Source|AccessionId.
1598 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1600 if (seq1 == null || seq2 == null)
1604 String name = seq2.getName();
1605 final DBRefEntry[] xrefs = seq1.getDBRefs();
1608 for (DBRefEntry xref : xrefs)
1610 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1611 // case-insensitive test, consistent with DBRefEntry.equalRef()
1612 if (xrefName.equalsIgnoreCase(name))
1622 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1623 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1624 * added to the original alignment's dataset, which is shared by the new
1625 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1626 * added to the alignment dataset.
1629 * aligned nucleotide (dna or cds) sequences
1631 * the alignment dataset the sequences belong to
1633 * (optional) to restrict results to CDS that map to specified
1635 * @return an alignment whose sequences are the cds-only parts of the dna
1636 * sequences (or null if no mappings are found)
1638 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1639 AlignmentI dataset, SequenceI[] products)
1641 if (dataset == null || dataset.getDataset() != null)
1643 throw new IllegalArgumentException(
1644 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1646 List<SequenceI> foundSeqs = new ArrayList<>();
1647 List<SequenceI> cdsSeqs = new ArrayList<>();
1648 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1649 HashSet<SequenceI> productSeqs = null;
1650 if (products != null)
1652 productSeqs = new HashSet<>();
1653 for (SequenceI seq : products)
1655 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1656 .getDatasetSequence());
1661 * Construct CDS sequences from mappings on the alignment dataset.
1663 * - find the protein product(s) mapped to from each dna sequence
1664 * - if the mapping covers the whole dna sequence (give or take start/stop
1665 * codon), take the dna as the CDS sequence
1666 * - else search dataset mappings for a suitable dna sequence, i.e. one
1667 * whose whole sequence is mapped to the protein
1668 * - if no sequence found, construct one from the dna sequence and mapping
1669 * (and add it to dataset so it is found if this is repeated)
1671 for (SequenceI dnaSeq : dna)
1673 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1674 : dnaSeq.getDatasetSequence();
1676 List<AlignedCodonFrame> seqMappings = MappingUtils
1677 .findMappingsForSequence(dnaSeq, mappings);
1678 for (AlignedCodonFrame mapping : seqMappings)
1680 List<Mapping> mappingsFromSequence = mapping
1681 .getMappingsFromSequence(dnaSeq);
1683 for (Mapping aMapping : mappingsFromSequence)
1685 MapList mapList = aMapping.getMap();
1686 if (mapList.getFromRatio() == 1)
1689 * not a dna-to-protein mapping (likely dna-to-cds)
1695 * skip if mapping is not to one of the target set of proteins
1697 SequenceI proteinProduct = aMapping.getTo();
1698 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1704 * try to locate the CDS from the dataset mappings;
1705 * guard against duplicate results (for the case that protein has
1706 * dbrefs to both dna and cds sequences)
1708 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1709 seqMappings, aMapping);
1712 if (!foundSeqs.contains(cdsSeq))
1714 foundSeqs.add(cdsSeq);
1715 SequenceI derivedSequence = cdsSeq.deriveSequence();
1716 cdsSeqs.add(derivedSequence);
1717 if (!dataset.getSequences().contains(cdsSeq))
1719 dataset.addSequence(cdsSeq);
1726 * didn't find mapped CDS sequence - construct it and add
1727 * its dataset sequence to the dataset
1729 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1730 dataset).deriveSequence();
1731 // cdsSeq has a name constructed as CDS|<dbref>
1732 // <dbref> will be either the accession for the coding sequence,
1733 // marked in the /via/ dbref to the protein product accession
1734 // or it will be the original nucleotide accession.
1735 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1737 cdsSeqs.add(cdsSeq);
1739 if (!dataset.getSequences().contains(cdsSeqDss))
1741 // check if this sequence is a newly created one
1742 // so needs adding to the dataset
1743 dataset.addSequence(cdsSeqDss);
1747 * add a mapping from CDS to the (unchanged) mapped to range
1749 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1750 cdsSeq.getLength() });
1751 MapList cdsToProteinMap = new MapList(cdsRange,
1752 mapList.getToRanges(), mapList.getFromRatio(),
1753 mapList.getToRatio());
1754 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1755 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1759 * guard against duplicating the mapping if repeating this action
1761 if (!mappings.contains(cdsToProteinMapping))
1763 mappings.add(cdsToProteinMapping);
1766 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1767 proteinProduct, aMapping);
1769 * add another mapping from original 'from' range to CDS
1771 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1772 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1774 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1776 if (!mappings.contains(dnaToCdsMapping))
1778 mappings.add(dnaToCdsMapping);
1782 * transfer dna chromosomal loci (if known) to the CDS
1783 * sequence (via the mapping)
1785 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1786 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1789 * add DBRef with mapping from protein to CDS
1790 * (this enables Get Cross-References from protein alignment)
1791 * This is tricky because we can't have two DBRefs with the
1792 * same source and accession, so need a different accession for
1793 * the CDS from the dna sequence
1796 // specific use case:
1797 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1798 // ENSG02, ENSG03, with transcripts and products similarly named.
1799 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1801 // JBPNote: ?? can't actually create an example that demonstrates we
1803 // synthesize an xref.
1805 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1808 * create a cross-reference from CDS to the source sequence's
1809 * primary reference and vice versa
1811 String source = primRef.getSource();
1812 String version = primRef.getVersion();
1813 DBRefEntry cdsCrossRef = new DBRefEntry(source, source + ":"
1814 + version, primRef.getAccessionId());
1815 cdsCrossRef.setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
1816 cdsSeqDss.addDBRef(cdsCrossRef);
1818 dnaSeq.addDBRef(new DBRefEntry(source, version, cdsSeq
1819 .getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
1821 // problem here is that the cross-reference is synthesized -
1822 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1824 // assuming cds version same as dna ?!?
1826 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
1829 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1831 proteinProduct.addDBRef(proteinToCdsRef);
1835 * transfer any features on dna that overlap the CDS
1837 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1838 SequenceOntologyI.CDS);
1843 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1845 cds.setDataset(dataset);
1851 * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
1852 * toSeq, mediated by the given mapping between the sequences
1855 * @param targetToFrom
1859 protected static void transferGeneLoci(SequenceI fromSeq,
1860 MapList targetToFrom, SequenceI targetSeq)
1862 if (targetSeq.getGeneLoci() != null)
1864 // already have - don't override
1867 GeneLociI fromLoci = fromSeq.getGeneLoci();
1868 if (fromLoci == null)
1873 MapList newMap = targetToFrom.traverse(fromLoci.getMapping());
1877 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
1878 fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
1883 * A helper method that finds a CDS sequence in the alignment dataset that is
1884 * mapped to the given protein sequence, and either is, or has a mapping from,
1885 * the given dna sequence.
1888 * set of all mappings on the dataset
1890 * a dna (or cds) sequence we are searching from
1891 * @param seqMappings
1892 * the set of mappings involving dnaSeq
1894 * a transcript-to-peptide mapping
1897 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1898 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1902 * TODO a better dna-cds-protein mapping data representation to allow easy
1903 * navigation; until then this clunky looping around lists of mappings
1905 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1906 : dnaSeq.getDatasetSequence();
1907 SequenceI proteinProduct = aMapping.getTo();
1910 * is this mapping from the whole dna sequence (i.e. CDS)?
1911 * allowing for possible stop codon on dna but not peptide
1913 int mappedFromLength = MappingUtils
1914 .getLength(aMapping.getMap().getFromRanges());
1915 int dnaLength = seqDss.getLength();
1916 if (mappedFromLength == dnaLength
1917 || mappedFromLength == dnaLength - CODON_LENGTH)
1920 * if sequence has CDS features, this is a transcript with no UTR
1921 * - do not take this as the CDS sequence! (JAL-2789)
1923 if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
1931 * looks like we found the dna-to-protein mapping; search for the
1932 * corresponding cds-to-protein mapping
1934 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1935 .findMappingsForSequence(proteinProduct, mappings);
1936 for (AlignedCodonFrame acf : mappingsToPeptide)
1938 for (SequenceToSequenceMapping map : acf.getMappings())
1940 Mapping mapping = map.getMapping();
1941 if (mapping != aMapping
1942 && mapping.getMap().getFromRatio() == CODON_LENGTH
1943 && proteinProduct == mapping.getTo()
1944 && seqDss != map.getFromSeq())
1946 mappedFromLength = MappingUtils
1947 .getLength(mapping.getMap().getFromRanges());
1948 if (mappedFromLength == map.getFromSeq().getLength())
1951 * found a 3:1 mapping to the protein product which covers
1952 * the whole dna sequence i.e. is from CDS; finally check the CDS
1953 * is mapped from the given dna start sequence
1955 SequenceI cdsSeq = map.getFromSeq();
1956 // todo this test is weak if seqMappings contains multiple mappings;
1957 // we get away with it if transcript:cds relationship is 1:1
1958 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1959 .findMappingsForSequence(cdsSeq, seqMappings);
1960 if (!dnaToCdsMaps.isEmpty())
1972 * Helper method that makes a CDS sequence as defined by the mappings from the
1973 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1974 * forward or reverse strand).
1979 * - existing dataset. We check for sequences that look like the CDS
1980 * we are about to construct, if one exists already, then we will
1981 * just return that one.
1982 * @return CDS sequence (as a dataset sequence)
1984 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1987 char[] seqChars = seq.getSequence();
1988 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1989 int cdsWidth = MappingUtils.getLength(fromRanges);
1990 char[] newSeqChars = new char[cdsWidth];
1993 for (int[] range : fromRanges)
1995 if (range[0] <= range[1])
1997 // forward strand mapping - just copy the range
1998 int length = range[1] - range[0] + 1;
1999 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
2005 // reverse strand mapping - copy and complement one by one
2006 for (int i = range[0]; i >= range[1]; i--)
2008 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
2014 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
2015 * else generate a sequence name
2017 String mapFromId = mapping.getMappedFromId();
2018 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
2019 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2020 if (dataset != null)
2022 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2023 if (matches != null)
2025 boolean matched = false;
2026 for (SequenceI mtch : matches)
2028 if (mtch.getStart() != newSeq.getStart())
2032 if (mtch.getEnd() != newSeq.getEnd())
2036 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2048 "JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
2054 // newSeq.setDescription(mapFromId);
2060 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2061 * the given mapping.
2065 * @param proteinProduct
2067 * @return list of DBRefEntrys added
2069 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2070 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2073 // gather direct refs from contig congruent with mapping
2074 List<DBRefEntry> direct = new ArrayList<>();
2075 HashSet<String> directSources = new HashSet<>();
2077 if (contig.getDBRefs() != null)
2079 for (DBRefEntry dbr : contig.getDBRefs())
2081 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
2083 MapList map = dbr.getMap().getMap();
2084 // check if map is the CDS mapping
2085 if (mapping.getMap().equals(map))
2088 directSources.add(dbr.getSource());
2093 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2094 proteinProduct.getDBRefs(),
2095 directSources.toArray(new String[0]));
2096 List<DBRefEntry> propagated = new ArrayList<>();
2098 // and generate appropriate mappings
2099 for (DBRefEntry cdsref : direct)
2101 // clone maplist and mapping
2102 MapList cdsposmap = new MapList(
2103 Arrays.asList(new int[][]
2104 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2105 cdsref.getMap().getMap().getToRanges(), 3, 1);
2106 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(),
2107 cdsref.getMap().getMap());
2110 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2111 cdsref.getVersion(), cdsref.getAccessionId(),
2112 new Mapping(cdsmap.getTo(), cdsposmap));
2114 // and see if we can map to the protein product for this mapping.
2115 // onSource is the filtered set of accessions on protein that we are
2116 // tranferring, so we assume accession is the same.
2117 if (cdsmap.getTo() == null && onSource != null)
2119 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2120 cdsref.getAccessionId());
2121 if (sourceRefs != null)
2123 for (DBRefEntry srcref : sourceRefs)
2125 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2127 // we have found a complementary dbref on the protein product, so
2128 // update mapping's getTo
2129 newref.getMap().setTo(proteinProduct);
2134 cdsSeq.addDBRef(newref);
2135 propagated.add(newref);
2141 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2142 * feature start/end ranges, optionally omitting specified feature types.
2143 * Returns the number of features copied.
2148 * the mapping from 'fromSeq' to 'toSeq'
2150 * if not null, only features of this type are copied (including
2151 * subtypes in the Sequence Ontology)
2154 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2155 MapList mapping, String select, String... omitting)
2157 SequenceI copyTo = toSeq;
2158 while (copyTo.getDatasetSequence() != null)
2160 copyTo = copyTo.getDatasetSequence();
2164 * get features, optionally restricted by an ontology term
2166 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2167 .getPositionalFeatures() : fromSeq.getFeatures()
2168 .getFeaturesByOntology(select);
2171 for (SequenceFeature sf : sfs)
2173 String type = sf.getType();
2174 boolean omit = false;
2175 for (String toOmit : omitting)
2177 if (type.equals(toOmit))
2188 * locate the mapped range - null if either start or end is
2189 * not mapped (no partial overlaps are calculated)
2191 int start = sf.getBegin();
2192 int end = sf.getEnd();
2193 int[] mappedTo = mapping.locateInTo(start, end);
2195 * if whole exon range doesn't map, try interpreting it
2196 * as 5' or 3' exon overlapping the CDS range
2198 if (mappedTo == null)
2200 mappedTo = mapping.locateInTo(end, end);
2201 if (mappedTo != null)
2204 * end of exon is in CDS range - 5' overlap
2205 * to a range from the start of the peptide
2210 if (mappedTo == null)
2212 mappedTo = mapping.locateInTo(start, start);
2213 if (mappedTo != null)
2216 * start of exon is in CDS range - 3' overlap
2217 * to a range up to the end of the peptide
2219 mappedTo[1] = toSeq.getLength();
2222 if (mappedTo != null)
2224 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2225 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2226 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2227 sf.getFeatureGroup(), sf.getScore());
2228 copyTo.addSequenceFeature(copy);
2236 * Returns a mapping from dna to protein by inspecting sequence features of
2237 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2238 * length is 3 times the peptide length (optionally after dropping a trailing
2239 * stop codon). This method does not check whether the CDS nucleotide sequence
2240 * translates to the peptide sequence.
2246 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2247 SequenceI proteinSeq)
2249 List<int[]> ranges = findCdsPositions(dnaSeq);
2250 int mappedDnaLength = MappingUtils.getLength(ranges);
2253 * if not a whole number of codons, truncate mapping
2255 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2256 if (codonRemainder > 0)
2258 mappedDnaLength -= codonRemainder;
2259 MappingUtils.removeEndPositions(codonRemainder, ranges);
2262 int proteinLength = proteinSeq.getLength();
2263 int proteinStart = proteinSeq.getStart();
2264 int proteinEnd = proteinSeq.getEnd();
2267 * incomplete start codon may mean X at start of peptide
2268 * we ignore both for mapping purposes
2270 if (proteinSeq.getCharAt(0) == 'X')
2272 // todo JAL-2022 support startPhase > 0
2276 List<int[]> proteinRange = new ArrayList<>();
2279 * dna length should map to protein (or protein plus stop codon)
2281 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2282 if (codesForResidues == (proteinLength + 1))
2284 // assuming extra codon is for STOP and not in peptide
2285 // todo: check trailing codon is indeed a STOP codon
2287 mappedDnaLength -= CODON_LENGTH;
2288 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2291 if (codesForResidues == proteinLength)
2293 proteinRange.add(new int[] { proteinStart, proteinEnd });
2294 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2300 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2301 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2302 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2303 * position order, so this method is only valid for linear CDS in the same
2304 * sense as the protein product.
2309 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2311 List<int[]> result = new ArrayList<>();
2313 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2314 SequenceOntologyI.CDS);
2319 SequenceFeatures.sortFeatures(sfs, true);
2321 for (SequenceFeature sf : sfs)
2326 phase = Integer.parseInt(sf.getPhase());
2327 } catch (NumberFormatException e)
2332 * phase > 0 on first codon means 5' incomplete - skip to the start
2333 * of the next codon; example ENST00000496384
2335 int begin = sf.getBegin();
2336 int end = sf.getEnd();
2337 if (result.isEmpty() && phase > 0)
2342 // shouldn't happen!
2344 .println("Error: start phase extends beyond start CDS in "
2345 + dnaSeq.getName());
2348 result.add(new int[] { begin, end });
2352 * Finally sort ranges by start position. This avoids a dependency on
2353 * keeping features in order on the sequence (if they are in order anyway,
2354 * the sort will have almost no work to do). The implicit assumption is CDS
2355 * ranges are assembled in order. Other cases should not use this method,
2356 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2358 Collections.sort(result, IntRangeComparator.ASCENDING);
2363 * Maps exon features from dna to protein, and computes variants in peptide
2364 * product generated by variants in dna, and adds them as sequence_variant
2365 * features on the protein sequence. Returns the number of variant features
2370 * @param dnaToProtein
2372 public static int computeProteinFeatures(SequenceI dnaSeq,
2373 SequenceI peptide, MapList dnaToProtein)
2375 while (dnaSeq.getDatasetSequence() != null)
2377 dnaSeq = dnaSeq.getDatasetSequence();
2379 while (peptide.getDatasetSequence() != null)
2381 peptide = peptide.getDatasetSequence();
2384 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2387 * compute protein variants from dna variants and codon mappings;
2388 * NB - alternatively we could retrieve this using the REST service e.g.
2389 * http://rest.ensembl.org/overlap/translation
2390 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2391 * which would be a bit slower but possibly more reliable
2395 * build a map with codon variations for each potentially varying peptide
2397 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2398 dnaSeq, dnaToProtein);
2401 * scan codon variations, compute peptide variants and add to peptide sequence
2404 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2406 int peptidePos = variant.getKey();
2407 List<DnaVariant>[] codonVariants = variant.getValue();
2408 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2415 * Computes non-synonymous peptide variants from codon variants and adds them
2416 * as sequence_variant features on the protein sequence (one feature per
2417 * allele variant). Selected attributes (variant id, clinical significance)
2418 * are copied over to the new features.
2421 * the protein sequence
2423 * the position to compute peptide variants for
2424 * @param codonVariants
2425 * a list of dna variants per codon position
2426 * @return the number of features added
2428 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2429 List<DnaVariant>[] codonVariants)
2431 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2433 String base1 = codonVariants[0].get(0).base;
2434 String base2 = codonVariants[1].get(0).base;
2435 String base3 = codonVariants[2].get(0).base;
2438 * variants in first codon base
2440 for (DnaVariant var : codonVariants[0])
2442 if (var.variant != null)
2444 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2445 if (alleles != null)
2447 for (String base : alleles.split(","))
2449 if (!base1.equalsIgnoreCase(base))
2451 String codon = base.toUpperCase() + base2.toLowerCase()
2452 + base3.toLowerCase();
2453 String canonical = base1.toUpperCase() + base2.toLowerCase()
2454 + base3.toLowerCase();
2455 if (addPeptideVariant(peptide, peptidePos, residue, var,
2467 * variants in second codon base
2469 for (DnaVariant var : codonVariants[1])
2471 if (var.variant != null)
2473 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2474 if (alleles != null)
2476 for (String base : alleles.split(","))
2478 if (!base2.equalsIgnoreCase(base))
2480 String codon = base1.toLowerCase() + base.toUpperCase()
2481 + base3.toLowerCase();
2482 String canonical = base1.toLowerCase() + base2.toUpperCase()
2483 + base3.toLowerCase();
2484 if (addPeptideVariant(peptide, peptidePos, residue, var,
2496 * variants in third codon base
2498 for (DnaVariant var : codonVariants[2])
2500 if (var.variant != null)
2502 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2503 if (alleles != null)
2505 for (String base : alleles.split(","))
2507 if (!base3.equalsIgnoreCase(base))
2509 String codon = base1.toLowerCase() + base2.toLowerCase()
2510 + base.toUpperCase();
2511 String canonical = base1.toLowerCase() + base2.toLowerCase()
2512 + base3.toUpperCase();
2513 if (addPeptideVariant(peptide, peptidePos, residue, var,
2528 * Helper method that adds a peptide variant feature. ID and
2529 * clinical_significance attributes of the dna variant (if present) are copied
2530 * to the new feature.
2537 * the variant codon e.g. aCg
2539 * the 'normal' codon e.g. aTg
2540 * @return true if a feature was added, else false
2542 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2543 String residue, DnaVariant var, String codon, String canonical)
2546 * get peptide translation of codon e.g. GAT -> D
2547 * note that variants which are not single alleles,
2548 * e.g. multibase variants or HGMD_MUTATION etc
2549 * are currently ignored here
2551 String trans = codon.contains("-") ? null
2552 : (codon.length() > CODON_LENGTH ? null
2553 : ResidueProperties.codonTranslate(codon));
2558 String desc = canonical + "/" + codon;
2559 String featureType = "";
2560 if (trans.equals(residue))
2562 featureType = SequenceOntologyI.SYNONYMOUS_VARIANT;
2564 else if (ResidueProperties.STOP.equals(trans))
2566 featureType = SequenceOntologyI.STOP_GAINED;
2570 String residue3Char = StringUtils
2571 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2572 String trans3Char = StringUtils
2573 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2574 desc = "p." + residue3Char + peptidePos + trans3Char;
2575 featureType = SequenceOntologyI.NONSYNONYMOUS_VARIANT;
2577 SequenceFeature sf = new SequenceFeature(featureType, desc, peptidePos,
2578 peptidePos, var.getSource());
2580 StringBuilder attributes = new StringBuilder(32);
2581 String id = (String) var.variant.getValue(VARIANT_ID);
2584 if (id.startsWith(SEQUENCE_VARIANT))
2586 id = id.substring(SEQUENCE_VARIANT.length());
2588 sf.setValue(VARIANT_ID, id);
2589 attributes.append(VARIANT_ID).append("=").append(id);
2590 // TODO handle other species variants JAL-2064
2591 StringBuilder link = new StringBuilder(32);
2594 link.append(desc).append(" ").append(id).append(
2595 "|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2596 .append(URLEncoder.encode(id, "UTF-8"));
2597 sf.addLink(link.toString());
2598 } catch (UnsupportedEncodingException e)
2603 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2604 if (clinSig != null)
2606 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2607 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2610 peptide.addSequenceFeature(sf);
2611 if (attributes.length() > 0)
2613 sf.setAttributes(attributes.toString());
2619 * Builds a map whose key is position in the protein sequence, and value is a
2620 * list of the base and all variants for each corresponding codon position.
2622 * This depends on dna variants being held as a comma-separated list as
2623 * property "alleles" on variant features.
2626 * @param dnaToProtein
2629 @SuppressWarnings("unchecked")
2630 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2631 SequenceI dnaSeq, MapList dnaToProtein)
2634 * map from peptide position to all variants of the codon which codes for it
2635 * LinkedHashMap ensures we keep the peptide features in sequence order
2637 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<>();
2639 List<SequenceFeature> dnaFeatures = dnaSeq.getFeatures()
2640 .getFeaturesByOntology(SequenceOntologyI.SEQUENCE_VARIANT);
2641 if (dnaFeatures.isEmpty())
2646 int dnaStart = dnaSeq.getStart();
2647 int[] lastCodon = null;
2648 int lastPeptidePostion = 0;
2651 * build a map of codon variations for peptides
2653 for (SequenceFeature sf : dnaFeatures)
2655 int dnaCol = sf.getBegin();
2656 if (dnaCol != sf.getEnd())
2658 // not handling multi-locus variant features
2663 * ignore variant if not a SNP
2665 String alls = (String) sf.getValue(Gff3Helper.ALLELES);
2668 continue; // non-SNP VCF variant perhaps - can't process this
2671 String[] alleles = alls.toUpperCase().split(",");
2672 boolean isSnp = true;
2673 for (String allele : alleles)
2675 if (allele.trim().length() > 1)
2685 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2688 // feature doesn't lie within coding region
2691 int peptidePosition = mapsTo[0];
2692 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2693 if (codonVariants == null)
2695 codonVariants = new ArrayList[CODON_LENGTH];
2696 codonVariants[0] = new ArrayList<>();
2697 codonVariants[1] = new ArrayList<>();
2698 codonVariants[2] = new ArrayList<>();
2699 variants.put(peptidePosition, codonVariants);
2703 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2705 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2706 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2707 peptidePosition, peptidePosition));
2708 lastPeptidePostion = peptidePosition;
2712 * save nucleotide (and any variant) for each codon position
2714 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2716 String nucleotide = String.valueOf(
2717 dnaSeq.getCharAt(codon[codonPos] - dnaStart)).toUpperCase();
2718 List<DnaVariant> codonVariant = codonVariants[codonPos];
2719 if (codon[codonPos] == dnaCol)
2721 if (!codonVariant.isEmpty()
2722 && codonVariant.get(0).variant == null)
2725 * already recorded base value, add this variant
2727 codonVariant.get(0).variant = sf;
2732 * add variant with base value
2734 codonVariant.add(new DnaVariant(nucleotide, sf));
2737 else if (codonVariant.isEmpty())
2740 * record (possibly non-varying) base value
2742 codonVariant.add(new DnaVariant(nucleotide));
2750 * Makes an alignment with a copy of the given sequences, adding in any
2751 * non-redundant sequences which are mapped to by the cross-referenced
2757 * the alignment dataset shared by the new copy
2760 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2761 SequenceI[] xrefs, AlignmentI dataset)
2763 AlignmentI copy = new Alignment(new Alignment(seqs));
2764 copy.setDataset(dataset);
2765 boolean isProtein = !copy.isNucleotide();
2766 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2769 for (SequenceI xref : xrefs)
2771 DBRefEntry[] dbrefs = xref.getDBRefs();
2774 for (DBRefEntry dbref : dbrefs)
2776 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2777 || dbref.getMap().getTo().isProtein() != isProtein)
2781 SequenceI mappedTo = dbref.getMap().getTo();
2782 SequenceI match = matcher.findIdMatch(mappedTo);
2785 matcher.add(mappedTo);
2786 copy.addSequence(mappedTo);
2796 * Try to align sequences in 'unaligned' to match the alignment of their
2797 * mapped regions in 'aligned'. For example, could use this to align CDS
2798 * sequences which are mapped to their parent cDNA sequences.
2800 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2801 * dna-to-protein or protein-to-dna use alternative methods.
2804 * sequences to be aligned
2806 * holds aligned sequences and their mappings
2809 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2812 * easy case - aligning a copy of aligned sequences
2814 if (alignAsSameSequences(unaligned, aligned))
2816 return unaligned.getHeight();
2820 * fancy case - aligning via mappings between sequences
2822 List<SequenceI> unmapped = new ArrayList<>();
2823 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2824 unaligned, aligned, unmapped);
2825 int width = columnMap.size();
2826 char gap = unaligned.getGapCharacter();
2827 int realignedCount = 0;
2828 // TODO: verify this loop scales sensibly for very wide/high alignments
2830 for (SequenceI seq : unaligned.getSequences())
2832 if (!unmapped.contains(seq))
2834 char[] newSeq = new char[width];
2835 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2836 // Integer iteration below
2841 * traverse the map to find columns populated
2844 for (Integer column : columnMap.keySet())
2846 Character c = columnMap.get(column).get(seq);
2850 * sequence has a character at this position
2860 * trim trailing gaps
2862 if (lastCol < width)
2864 char[] tmp = new char[lastCol + 1];
2865 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2868 // TODO: optimise SequenceI to avoid char[]->String->char[]
2869 seq.setSequence(String.valueOf(newSeq));
2873 return realignedCount;
2877 * If unaligned and aligned sequences share the same dataset sequences, then
2878 * simply copies the aligned sequences to the unaligned sequences and returns
2879 * true; else returns false
2882 * - sequences to be aligned based on aligned
2884 * - 'guide' alignment containing sequences derived from same dataset
2888 static boolean alignAsSameSequences(AlignmentI unaligned,
2891 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2893 return false; // should only pass alignments with datasets here
2896 // map from dataset sequence to alignment sequence(s)
2897 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2898 for (SequenceI seq : aligned.getSequences())
2900 SequenceI ds = seq.getDatasetSequence();
2901 if (alignedDatasets.get(ds) == null)
2903 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2905 alignedDatasets.get(ds).add(seq);
2909 * first pass - check whether all sequences to be aligned share a dataset
2910 * sequence with an aligned sequence
2912 for (SequenceI seq : unaligned.getSequences())
2914 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2921 * second pass - copy aligned sequences;
2922 * heuristic rule: pair off sequences in order for the case where
2923 * more than one shares the same dataset sequence
2925 for (SequenceI seq : unaligned.getSequences())
2927 List<SequenceI> alignedSequences = alignedDatasets
2928 .get(seq.getDatasetSequence());
2929 // TODO: getSequenceAsString() will be deprecated in the future
2930 // TODO: need to leave to SequenceI implementor to update gaps
2931 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2932 if (alignedSequences.size() > 0)
2934 // pop off aligned sequences (except the last one)
2935 alignedSequences.remove(0);
2943 * Returns a map whose key is alignment column number (base 1), and whose
2944 * values are a map of sequence characters in that column.
2951 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2952 AlignmentI unaligned, AlignmentI aligned,
2953 List<SequenceI> unmapped)
2956 * Map will hold, for each aligned column position, a map of
2957 * {unalignedSequence, characterPerSequence} at that position.
2958 * TreeMap keeps the entries in ascending column order.
2960 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2963 * record any sequences that have no mapping so can't be realigned
2965 unmapped.addAll(unaligned.getSequences());
2967 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2969 for (SequenceI seq : unaligned.getSequences())
2971 for (AlignedCodonFrame mapping : mappings)
2973 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2974 if (fromSeq != null)
2976 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2977 if (addMappedPositions(seq, fromSeq, seqMap, map))
2979 unmapped.remove(seq);
2988 * Helper method that adds to a map the mapped column positions of a sequence.
2990 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2991 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2995 * the sequence whose column positions we are recording
2997 * a sequence that is mapped to the first sequence
2999 * the mapping from 'fromSeq' to 'seq'
3001 * a map to add the column positions (in fromSeq) of the mapped
3005 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
3006 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
3014 * invert mapping if it is from unaligned to aligned sequence
3016 if (seqMap.getTo() == fromSeq.getDatasetSequence())
3018 seqMap = new Mapping(seq.getDatasetSequence(),
3019 seqMap.getMap().getInverse());
3022 int toStart = seq.getStart();
3025 * traverse [start, end, start, end...] ranges in fromSeq
3027 for (int[] fromRange : seqMap.getMap().getFromRanges())
3029 for (int i = 0; i < fromRange.length - 1; i += 2)
3031 boolean forward = fromRange[i + 1] >= fromRange[i];
3034 * find the range mapped to (sequence positions base 1)
3036 int[] range = seqMap.locateMappedRange(fromRange[i],
3040 System.err.println("Error in mapping " + seqMap + " from "
3041 + fromSeq.getName());
3044 int fromCol = fromSeq.findIndex(fromRange[i]);
3045 int mappedCharPos = range[0];
3048 * walk over the 'from' aligned sequence in forward or reverse
3049 * direction; when a non-gap is found, record the column position
3050 * of the next character of the mapped-to sequence; stop when all
3051 * the characters of the range have been counted
3053 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
3056 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
3059 * mapped from sequence has a character in this column
3060 * record the column position for the mapped to character
3062 Map<SequenceI, Character> seqsMap = map.get(fromCol);
3063 if (seqsMap == null)
3065 seqsMap = new HashMap<>();
3066 map.put(fromCol, seqsMap);
3068 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
3071 fromCol += (forward ? 1 : -1);
3078 // strictly temporary hack until proper criteria for aligning protein to cds
3079 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
3080 public static boolean looksLikeEnsembl(AlignmentI alignment)
3082 for (SequenceI seq : alignment.getSequences())
3084 String name = seq.getName();
3085 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))