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)
1467 if (dsann.annotations != null) // ignore non-positional annotation
1470 * Find matching annotations on the alignment. If none is found, then
1471 * add this annotation to the list of 'addable' annotations for this
1474 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1475 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1476 if (!matchedAlignmentAnnotations.iterator().hasNext())
1479 if (labelForCalcId != null)
1481 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1486 * Save any addable annotations for this sequence
1488 if (!result.isEmpty())
1490 candidates.put(seq, result);
1497 * Adds annotations to the top of the alignment annotations, in the same order
1498 * as their related sequences.
1500 * @param annotations
1501 * the annotations to add
1503 * the alignment to add them to
1504 * @param selectionGroup
1505 * current selection group (or null if none)
1507 public static void addReferenceAnnotations(
1508 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1509 final AlignmentI alignment, final SequenceGroup selectionGroup)
1511 for (SequenceI seq : annotations.keySet())
1513 for (AlignmentAnnotation ann : annotations.get(seq))
1515 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1517 int endRes = ann.annotations.length;
1518 if (selectionGroup != null)
1520 startRes = selectionGroup.getStartRes();
1521 endRes = selectionGroup.getEndRes();
1523 copyAnn.restrict(startRes, endRes);
1526 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1527 * original annotation is already on the sequence.
1529 if (!seq.hasAnnotation(ann))
1531 seq.addAlignmentAnnotation(copyAnn);
1534 copyAnn.adjustForAlignment();
1535 // add to the alignment and set visible
1536 alignment.addAnnotation(copyAnn);
1537 copyAnn.visible = true;
1543 * Set visibility of alignment annotations of specified types (labels), for
1544 * specified sequences. This supports controls like "Show all secondary
1545 * structure", "Hide all Temp factor", etc.
1547 * @al the alignment to scan for annotations
1549 * the types (labels) of annotations to be updated
1550 * @param forSequences
1551 * if not null, only annotations linked to one of these sequences are
1552 * in scope for update; if null, acts on all sequence annotations
1554 * if this flag is true, 'types' is ignored (label not checked)
1556 * if true, set visibility on, else set off
1558 public static void showOrHideSequenceAnnotations(AlignmentI al,
1559 Collection<String> types, List<SequenceI> forSequences,
1560 boolean anyType, boolean doShow)
1562 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1565 for (AlignmentAnnotation aa : anns)
1567 if (anyType || types.contains(aa.label))
1569 if ((aa.sequenceRef != null) && (forSequences == null
1570 || forSequences.contains(aa.sequenceRef)))
1572 aa.visible = doShow;
1580 * Returns true if either sequence has a cross-reference to the other
1586 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1588 // Note: moved here from class CrossRef as the latter class has dependencies
1589 // not availability to the applet's classpath
1590 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1594 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1595 * that sequence name is structured as Source|AccessionId.
1601 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1603 if (seq1 == null || seq2 == null)
1607 String name = seq2.getName();
1608 final DBRefEntry[] xrefs = seq1.getDBRefs();
1611 for (DBRefEntry xref : xrefs)
1613 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1614 // case-insensitive test, consistent with DBRefEntry.equalRef()
1615 if (xrefName.equalsIgnoreCase(name))
1625 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1626 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1627 * added to the original alignment's dataset, which is shared by the new
1628 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1629 * added to the alignment dataset.
1632 * aligned nucleotide (dna or cds) sequences
1634 * the alignment dataset the sequences belong to
1636 * (optional) to restrict results to CDS that map to specified
1638 * @return an alignment whose sequences are the cds-only parts of the dna
1639 * sequences (or null if no mappings are found)
1641 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1642 AlignmentI dataset, SequenceI[] products)
1644 if (dataset == null || dataset.getDataset() != null)
1646 throw new IllegalArgumentException(
1647 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1649 List<SequenceI> foundSeqs = new ArrayList<>();
1650 List<SequenceI> cdsSeqs = new ArrayList<>();
1651 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1652 HashSet<SequenceI> productSeqs = null;
1653 if (products != null)
1655 productSeqs = new HashSet<>();
1656 for (SequenceI seq : products)
1658 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1659 .getDatasetSequence());
1664 * Construct CDS sequences from mappings on the alignment dataset.
1666 * - find the protein product(s) mapped to from each dna sequence
1667 * - if the mapping covers the whole dna sequence (give or take start/stop
1668 * codon), take the dna as the CDS sequence
1669 * - else search dataset mappings for a suitable dna sequence, i.e. one
1670 * whose whole sequence is mapped to the protein
1671 * - if no sequence found, construct one from the dna sequence and mapping
1672 * (and add it to dataset so it is found if this is repeated)
1674 for (SequenceI dnaSeq : dna)
1676 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1677 : dnaSeq.getDatasetSequence();
1679 List<AlignedCodonFrame> seqMappings = MappingUtils
1680 .findMappingsForSequence(dnaSeq, mappings);
1681 for (AlignedCodonFrame mapping : seqMappings)
1683 List<Mapping> mappingsFromSequence = mapping
1684 .getMappingsFromSequence(dnaSeq);
1686 for (Mapping aMapping : mappingsFromSequence)
1688 MapList mapList = aMapping.getMap();
1689 if (mapList.getFromRatio() == 1)
1692 * not a dna-to-protein mapping (likely dna-to-cds)
1698 * skip if mapping is not to one of the target set of proteins
1700 SequenceI proteinProduct = aMapping.getTo();
1701 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1707 * try to locate the CDS from the dataset mappings;
1708 * guard against duplicate results (for the case that protein has
1709 * dbrefs to both dna and cds sequences)
1711 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1712 seqMappings, aMapping);
1715 if (!foundSeqs.contains(cdsSeq))
1717 foundSeqs.add(cdsSeq);
1718 SequenceI derivedSequence = cdsSeq.deriveSequence();
1719 cdsSeqs.add(derivedSequence);
1720 if (!dataset.getSequences().contains(cdsSeq))
1722 dataset.addSequence(cdsSeq);
1729 * didn't find mapped CDS sequence - construct it and add
1730 * its dataset sequence to the dataset
1732 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1733 dataset).deriveSequence();
1734 // cdsSeq has a name constructed as CDS|<dbref>
1735 // <dbref> will be either the accession for the coding sequence,
1736 // marked in the /via/ dbref to the protein product accession
1737 // or it will be the original nucleotide accession.
1738 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1740 cdsSeqs.add(cdsSeq);
1742 if (!dataset.getSequences().contains(cdsSeqDss))
1744 // check if this sequence is a newly created one
1745 // so needs adding to the dataset
1746 dataset.addSequence(cdsSeqDss);
1750 * add a mapping from CDS to the (unchanged) mapped to range
1752 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1753 cdsSeq.getLength() });
1754 MapList cdsToProteinMap = new MapList(cdsRange,
1755 mapList.getToRanges(), mapList.getFromRatio(),
1756 mapList.getToRatio());
1757 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1758 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1762 * guard against duplicating the mapping if repeating this action
1764 if (!mappings.contains(cdsToProteinMapping))
1766 mappings.add(cdsToProteinMapping);
1769 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1770 proteinProduct, aMapping);
1772 * add another mapping from original 'from' range to CDS
1774 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1775 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1777 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1779 if (!mappings.contains(dnaToCdsMapping))
1781 mappings.add(dnaToCdsMapping);
1785 * transfer dna chromosomal loci (if known) to the CDS
1786 * sequence (via the mapping)
1788 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1789 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1792 * add DBRef with mapping from protein to CDS
1793 * (this enables Get Cross-References from protein alignment)
1794 * This is tricky because we can't have two DBRefs with the
1795 * same source and accession, so need a different accession for
1796 * the CDS from the dna sequence
1799 // specific use case:
1800 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1801 // ENSG02, ENSG03, with transcripts and products similarly named.
1802 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1804 // JBPNote: ?? can't actually create an example that demonstrates we
1806 // synthesize an xref.
1808 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1811 * create a cross-reference from CDS to the source sequence's
1812 * primary reference and vice versa
1814 String source = primRef.getSource();
1815 String version = primRef.getVersion();
1816 DBRefEntry cdsCrossRef = new DBRefEntry(source, source + ":"
1817 + version, primRef.getAccessionId());
1818 cdsCrossRef.setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
1819 cdsSeqDss.addDBRef(cdsCrossRef);
1821 dnaSeq.addDBRef(new DBRefEntry(source, version, cdsSeq
1822 .getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
1824 // problem here is that the cross-reference is synthesized -
1825 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1827 // assuming cds version same as dna ?!?
1829 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
1832 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1834 proteinProduct.addDBRef(proteinToCdsRef);
1838 * transfer any features on dna that overlap the CDS
1840 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1841 SequenceOntologyI.CDS);
1846 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1848 cds.setDataset(dataset);
1854 * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
1855 * toSeq, mediated by the given mapping between the sequences
1858 * @param targetToFrom
1862 protected static void transferGeneLoci(SequenceI fromSeq,
1863 MapList targetToFrom, SequenceI targetSeq)
1865 if (targetSeq.getGeneLoci() != null)
1867 // already have - don't override
1870 GeneLociI fromLoci = fromSeq.getGeneLoci();
1871 if (fromLoci == null)
1876 MapList newMap = targetToFrom.traverse(fromLoci.getMap());
1880 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
1881 fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
1886 * A helper method that finds a CDS sequence in the alignment dataset that is
1887 * mapped to the given protein sequence, and either is, or has a mapping from,
1888 * the given dna sequence.
1891 * set of all mappings on the dataset
1893 * a dna (or cds) sequence we are searching from
1894 * @param seqMappings
1895 * the set of mappings involving dnaSeq
1897 * a transcript-to-peptide mapping
1900 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1901 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1905 * TODO a better dna-cds-protein mapping data representation to allow easy
1906 * navigation; until then this clunky looping around lists of mappings
1908 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1909 : dnaSeq.getDatasetSequence();
1910 SequenceI proteinProduct = aMapping.getTo();
1913 * is this mapping from the whole dna sequence (i.e. CDS)?
1914 * allowing for possible stop codon on dna but not peptide
1916 int mappedFromLength = MappingUtils
1917 .getLength(aMapping.getMap().getFromRanges());
1918 int dnaLength = seqDss.getLength();
1919 if (mappedFromLength == dnaLength
1920 || mappedFromLength == dnaLength - CODON_LENGTH)
1923 * if sequence has CDS features, this is a transcript with no UTR
1924 * - do not take this as the CDS sequence! (JAL-2789)
1926 if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
1934 * looks like we found the dna-to-protein mapping; search for the
1935 * corresponding cds-to-protein mapping
1937 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1938 .findMappingsForSequence(proteinProduct, mappings);
1939 for (AlignedCodonFrame acf : mappingsToPeptide)
1941 for (SequenceToSequenceMapping map : acf.getMappings())
1943 Mapping mapping = map.getMapping();
1944 if (mapping != aMapping
1945 && mapping.getMap().getFromRatio() == CODON_LENGTH
1946 && proteinProduct == mapping.getTo()
1947 && seqDss != map.getFromSeq())
1949 mappedFromLength = MappingUtils
1950 .getLength(mapping.getMap().getFromRanges());
1951 if (mappedFromLength == map.getFromSeq().getLength())
1954 * found a 3:1 mapping to the protein product which covers
1955 * the whole dna sequence i.e. is from CDS; finally check the CDS
1956 * is mapped from the given dna start sequence
1958 SequenceI cdsSeq = map.getFromSeq();
1959 // todo this test is weak if seqMappings contains multiple mappings;
1960 // we get away with it if transcript:cds relationship is 1:1
1961 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1962 .findMappingsForSequence(cdsSeq, seqMappings);
1963 if (!dnaToCdsMaps.isEmpty())
1975 * Helper method that makes a CDS sequence as defined by the mappings from the
1976 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1977 * forward or reverse strand).
1982 * - existing dataset. We check for sequences that look like the CDS
1983 * we are about to construct, if one exists already, then we will
1984 * just return that one.
1985 * @return CDS sequence (as a dataset sequence)
1987 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1990 char[] seqChars = seq.getSequence();
1991 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1992 int cdsWidth = MappingUtils.getLength(fromRanges);
1993 char[] newSeqChars = new char[cdsWidth];
1996 for (int[] range : fromRanges)
1998 if (range[0] <= range[1])
2000 // forward strand mapping - just copy the range
2001 int length = range[1] - range[0] + 1;
2002 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
2008 // reverse strand mapping - copy and complement one by one
2009 for (int i = range[0]; i >= range[1]; i--)
2011 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
2017 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
2018 * else generate a sequence name
2020 String mapFromId = mapping.getMappedFromId();
2021 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
2022 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2023 if (dataset != null)
2025 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2026 if (matches != null)
2028 boolean matched = false;
2029 for (SequenceI mtch : matches)
2031 if (mtch.getStart() != newSeq.getStart())
2035 if (mtch.getEnd() != newSeq.getEnd())
2039 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2051 "JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
2057 // newSeq.setDescription(mapFromId);
2063 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2064 * the given mapping.
2068 * @param proteinProduct
2070 * @return list of DBRefEntrys added
2072 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2073 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2076 // gather direct refs from contig congruent with mapping
2077 List<DBRefEntry> direct = new ArrayList<>();
2078 HashSet<String> directSources = new HashSet<>();
2080 if (contig.getDBRefs() != null)
2082 for (DBRefEntry dbr : contig.getDBRefs())
2084 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
2086 MapList map = dbr.getMap().getMap();
2087 // check if map is the CDS mapping
2088 if (mapping.getMap().equals(map))
2091 directSources.add(dbr.getSource());
2096 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2097 proteinProduct.getDBRefs(),
2098 directSources.toArray(new String[0]));
2099 List<DBRefEntry> propagated = new ArrayList<>();
2101 // and generate appropriate mappings
2102 for (DBRefEntry cdsref : direct)
2104 // clone maplist and mapping
2105 MapList cdsposmap = new MapList(
2106 Arrays.asList(new int[][]
2107 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2108 cdsref.getMap().getMap().getToRanges(), 3, 1);
2109 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(),
2110 cdsref.getMap().getMap());
2113 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2114 cdsref.getVersion(), cdsref.getAccessionId(),
2115 new Mapping(cdsmap.getTo(), cdsposmap));
2117 // and see if we can map to the protein product for this mapping.
2118 // onSource is the filtered set of accessions on protein that we are
2119 // tranferring, so we assume accession is the same.
2120 if (cdsmap.getTo() == null && onSource != null)
2122 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2123 cdsref.getAccessionId());
2124 if (sourceRefs != null)
2126 for (DBRefEntry srcref : sourceRefs)
2128 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2130 // we have found a complementary dbref on the protein product, so
2131 // update mapping's getTo
2132 newref.getMap().setTo(proteinProduct);
2137 cdsSeq.addDBRef(newref);
2138 propagated.add(newref);
2144 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2145 * feature start/end ranges, optionally omitting specified feature types.
2146 * Returns the number of features copied.
2151 * the mapping from 'fromSeq' to 'toSeq'
2153 * if not null, only features of this type are copied (including
2154 * subtypes in the Sequence Ontology)
2157 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2158 MapList mapping, String select, String... omitting)
2160 SequenceI copyTo = toSeq;
2161 while (copyTo.getDatasetSequence() != null)
2163 copyTo = copyTo.getDatasetSequence();
2167 * get features, optionally restricted by an ontology term
2169 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2170 .getPositionalFeatures() : fromSeq.getFeatures()
2171 .getFeaturesByOntology(select);
2174 for (SequenceFeature sf : sfs)
2176 String type = sf.getType();
2177 boolean omit = false;
2178 for (String toOmit : omitting)
2180 if (type.equals(toOmit))
2191 * locate the mapped range - null if either start or end is
2192 * not mapped (no partial overlaps are calculated)
2194 int start = sf.getBegin();
2195 int end = sf.getEnd();
2196 int[] mappedTo = mapping.locateInTo(start, end);
2198 * if whole exon range doesn't map, try interpreting it
2199 * as 5' or 3' exon overlapping the CDS range
2201 if (mappedTo == null)
2203 mappedTo = mapping.locateInTo(end, end);
2204 if (mappedTo != null)
2207 * end of exon is in CDS range - 5' overlap
2208 * to a range from the start of the peptide
2213 if (mappedTo == null)
2215 mappedTo = mapping.locateInTo(start, start);
2216 if (mappedTo != null)
2219 * start of exon is in CDS range - 3' overlap
2220 * to a range up to the end of the peptide
2222 mappedTo[1] = toSeq.getLength();
2225 if (mappedTo != null)
2227 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2228 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2229 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2230 sf.getFeatureGroup(), sf.getScore());
2231 copyTo.addSequenceFeature(copy);
2239 * Returns a mapping from dna to protein by inspecting sequence features of
2240 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2241 * length is 3 times the peptide length (optionally after dropping a trailing
2242 * stop codon). This method does not check whether the CDS nucleotide sequence
2243 * translates to the peptide sequence.
2249 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2250 SequenceI proteinSeq)
2252 List<int[]> ranges = findCdsPositions(dnaSeq);
2253 int mappedDnaLength = MappingUtils.getLength(ranges);
2256 * if not a whole number of codons, truncate mapping
2258 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2259 if (codonRemainder > 0)
2261 mappedDnaLength -= codonRemainder;
2262 MappingUtils.removeEndPositions(codonRemainder, ranges);
2265 int proteinLength = proteinSeq.getLength();
2266 int proteinStart = proteinSeq.getStart();
2267 int proteinEnd = proteinSeq.getEnd();
2270 * incomplete start codon may mean X at start of peptide
2271 * we ignore both for mapping purposes
2273 if (proteinSeq.getCharAt(0) == 'X')
2275 // todo JAL-2022 support startPhase > 0
2279 List<int[]> proteinRange = new ArrayList<>();
2282 * dna length should map to protein (or protein plus stop codon)
2284 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2285 if (codesForResidues == (proteinLength + 1))
2287 // assuming extra codon is for STOP and not in peptide
2288 // todo: check trailing codon is indeed a STOP codon
2290 mappedDnaLength -= CODON_LENGTH;
2291 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2294 if (codesForResidues == proteinLength)
2296 proteinRange.add(new int[] { proteinStart, proteinEnd });
2297 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2303 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2304 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2305 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2306 * position order, so this method is only valid for linear CDS in the same
2307 * sense as the protein product.
2312 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2314 List<int[]> result = new ArrayList<>();
2316 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2317 SequenceOntologyI.CDS);
2322 SequenceFeatures.sortFeatures(sfs, true);
2324 for (SequenceFeature sf : sfs)
2329 phase = Integer.parseInt(sf.getPhase());
2330 } catch (NumberFormatException e)
2335 * phase > 0 on first codon means 5' incomplete - skip to the start
2336 * of the next codon; example ENST00000496384
2338 int begin = sf.getBegin();
2339 int end = sf.getEnd();
2340 if (result.isEmpty() && phase > 0)
2345 // shouldn't happen!
2347 .println("Error: start phase extends beyond start CDS in "
2348 + dnaSeq.getName());
2351 result.add(new int[] { begin, end });
2355 * Finally sort ranges by start position. This avoids a dependency on
2356 * keeping features in order on the sequence (if they are in order anyway,
2357 * the sort will have almost no work to do). The implicit assumption is CDS
2358 * ranges are assembled in order. Other cases should not use this method,
2359 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2361 Collections.sort(result, IntRangeComparator.ASCENDING);
2366 * Maps exon features from dna to protein, and computes variants in peptide
2367 * product generated by variants in dna, and adds them as sequence_variant
2368 * features on the protein sequence. Returns the number of variant features
2373 * @param dnaToProtein
2375 public static int computeProteinFeatures(SequenceI dnaSeq,
2376 SequenceI peptide, MapList dnaToProtein)
2378 while (dnaSeq.getDatasetSequence() != null)
2380 dnaSeq = dnaSeq.getDatasetSequence();
2382 while (peptide.getDatasetSequence() != null)
2384 peptide = peptide.getDatasetSequence();
2387 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2390 * compute protein variants from dna variants and codon mappings;
2391 * NB - alternatively we could retrieve this using the REST service e.g.
2392 * http://rest.ensembl.org/overlap/translation
2393 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2394 * which would be a bit slower but possibly more reliable
2398 * build a map with codon variations for each potentially varying peptide
2400 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2401 dnaSeq, dnaToProtein);
2404 * scan codon variations, compute peptide variants and add to peptide sequence
2407 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2409 int peptidePos = variant.getKey();
2410 List<DnaVariant>[] codonVariants = variant.getValue();
2411 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2418 * Computes non-synonymous peptide variants from codon variants and adds them
2419 * as sequence_variant features on the protein sequence (one feature per
2420 * allele variant). Selected attributes (variant id, clinical significance)
2421 * are copied over to the new features.
2424 * the protein sequence
2426 * the position to compute peptide variants for
2427 * @param codonVariants
2428 * a list of dna variants per codon position
2429 * @return the number of features added
2431 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2432 List<DnaVariant>[] codonVariants)
2434 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2436 String base1 = codonVariants[0].get(0).base;
2437 String base2 = codonVariants[1].get(0).base;
2438 String base3 = codonVariants[2].get(0).base;
2441 * variants in first codon base
2443 for (DnaVariant var : codonVariants[0])
2445 if (var.variant != null)
2447 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2448 if (alleles != null)
2450 for (String base : alleles.split(","))
2452 if (!base1.equalsIgnoreCase(base))
2454 String codon = base.toUpperCase() + base2.toLowerCase()
2455 + base3.toLowerCase();
2456 String canonical = base1.toUpperCase() + base2.toLowerCase()
2457 + base3.toLowerCase();
2458 if (addPeptideVariant(peptide, peptidePos, residue, var,
2470 * variants in second codon base
2472 for (DnaVariant var : codonVariants[1])
2474 if (var.variant != null)
2476 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2477 if (alleles != null)
2479 for (String base : alleles.split(","))
2481 if (!base2.equalsIgnoreCase(base))
2483 String codon = base1.toLowerCase() + base.toUpperCase()
2484 + base3.toLowerCase();
2485 String canonical = base1.toLowerCase() + base2.toUpperCase()
2486 + base3.toLowerCase();
2487 if (addPeptideVariant(peptide, peptidePos, residue, var,
2499 * variants in third codon base
2501 for (DnaVariant var : codonVariants[2])
2503 if (var.variant != null)
2505 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2506 if (alleles != null)
2508 for (String base : alleles.split(","))
2510 if (!base3.equalsIgnoreCase(base))
2512 String codon = base1.toLowerCase() + base2.toLowerCase()
2513 + base.toUpperCase();
2514 String canonical = base1.toLowerCase() + base2.toLowerCase()
2515 + base3.toUpperCase();
2516 if (addPeptideVariant(peptide, peptidePos, residue, var,
2531 * Helper method that adds a peptide variant feature. ID and
2532 * clinical_significance attributes of the dna variant (if present) are copied
2533 * to the new feature.
2540 * the variant codon e.g. aCg
2542 * the 'normal' codon e.g. aTg
2543 * @return true if a feature was added, else false
2545 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2546 String residue, DnaVariant var, String codon, String canonical)
2549 * get peptide translation of codon e.g. GAT -> D
2550 * note that variants which are not single alleles,
2551 * e.g. multibase variants or HGMD_MUTATION etc
2552 * are currently ignored here
2554 String trans = codon.contains("-") ? null
2555 : (codon.length() > CODON_LENGTH ? null
2556 : ResidueProperties.codonTranslate(codon));
2561 String desc = canonical + "/" + codon;
2562 String featureType = "";
2563 if (trans.equals(residue))
2565 featureType = SequenceOntologyI.SYNONYMOUS_VARIANT;
2567 else if (ResidueProperties.STOP.equals(trans))
2569 featureType = SequenceOntologyI.STOP_GAINED;
2573 String residue3Char = StringUtils
2574 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2575 String trans3Char = StringUtils
2576 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2577 desc = "p." + residue3Char + peptidePos + trans3Char;
2578 featureType = SequenceOntologyI.NONSYNONYMOUS_VARIANT;
2580 SequenceFeature sf = new SequenceFeature(featureType, desc, peptidePos,
2581 peptidePos, var.getSource());
2583 StringBuilder attributes = new StringBuilder(32);
2584 String id = (String) var.variant.getValue(VARIANT_ID);
2587 if (id.startsWith(SEQUENCE_VARIANT))
2589 id = id.substring(SEQUENCE_VARIANT.length());
2591 sf.setValue(VARIANT_ID, id);
2592 attributes.append(VARIANT_ID).append("=").append(id);
2593 // TODO handle other species variants JAL-2064
2594 StringBuilder link = new StringBuilder(32);
2597 link.append(desc).append(" ").append(id).append(
2598 "|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2599 .append(URLEncoder.encode(id, "UTF-8"));
2600 sf.addLink(link.toString());
2601 } catch (UnsupportedEncodingException e)
2606 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2607 if (clinSig != null)
2609 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2610 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2613 peptide.addSequenceFeature(sf);
2614 if (attributes.length() > 0)
2616 sf.setAttributes(attributes.toString());
2622 * Builds a map whose key is position in the protein sequence, and value is a
2623 * list of the base and all variants for each corresponding codon position.
2625 * This depends on dna variants being held as a comma-separated list as
2626 * property "alleles" on variant features.
2629 * @param dnaToProtein
2632 @SuppressWarnings("unchecked")
2633 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2634 SequenceI dnaSeq, MapList dnaToProtein)
2637 * map from peptide position to all variants of the codon which codes for it
2638 * LinkedHashMap ensures we keep the peptide features in sequence order
2640 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<>();
2642 List<SequenceFeature> dnaFeatures = dnaSeq.getFeatures()
2643 .getFeaturesByOntology(SequenceOntologyI.SEQUENCE_VARIANT);
2644 if (dnaFeatures.isEmpty())
2649 int dnaStart = dnaSeq.getStart();
2650 int[] lastCodon = null;
2651 int lastPeptidePostion = 0;
2654 * build a map of codon variations for peptides
2656 for (SequenceFeature sf : dnaFeatures)
2658 int dnaCol = sf.getBegin();
2659 if (dnaCol != sf.getEnd())
2661 // not handling multi-locus variant features
2666 * ignore variant if not a SNP
2668 String alls = (String) sf.getValue(Gff3Helper.ALLELES);
2671 continue; // non-SNP VCF variant perhaps - can't process this
2674 String[] alleles = alls.toUpperCase().split(",");
2675 boolean isSnp = true;
2676 for (String allele : alleles)
2678 if (allele.trim().length() > 1)
2688 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2691 // feature doesn't lie within coding region
2694 int peptidePosition = mapsTo[0];
2695 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2696 if (codonVariants == null)
2698 codonVariants = new ArrayList[CODON_LENGTH];
2699 codonVariants[0] = new ArrayList<>();
2700 codonVariants[1] = new ArrayList<>();
2701 codonVariants[2] = new ArrayList<>();
2702 variants.put(peptidePosition, codonVariants);
2706 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2708 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2709 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2710 peptidePosition, peptidePosition));
2711 lastPeptidePostion = peptidePosition;
2715 * save nucleotide (and any variant) for each codon position
2717 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2719 String nucleotide = String.valueOf(
2720 dnaSeq.getCharAt(codon[codonPos] - dnaStart)).toUpperCase();
2721 List<DnaVariant> codonVariant = codonVariants[codonPos];
2722 if (codon[codonPos] == dnaCol)
2724 if (!codonVariant.isEmpty()
2725 && codonVariant.get(0).variant == null)
2728 * already recorded base value, add this variant
2730 codonVariant.get(0).variant = sf;
2735 * add variant with base value
2737 codonVariant.add(new DnaVariant(nucleotide, sf));
2740 else if (codonVariant.isEmpty())
2743 * record (possibly non-varying) base value
2745 codonVariant.add(new DnaVariant(nucleotide));
2753 * Makes an alignment with a copy of the given sequences, adding in any
2754 * non-redundant sequences which are mapped to by the cross-referenced
2760 * the alignment dataset shared by the new copy
2763 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2764 SequenceI[] xrefs, AlignmentI dataset)
2766 AlignmentI copy = new Alignment(new Alignment(seqs));
2767 copy.setDataset(dataset);
2768 boolean isProtein = !copy.isNucleotide();
2769 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2772 for (SequenceI xref : xrefs)
2774 DBRefEntry[] dbrefs = xref.getDBRefs();
2777 for (DBRefEntry dbref : dbrefs)
2779 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2780 || dbref.getMap().getTo().isProtein() != isProtein)
2784 SequenceI mappedTo = dbref.getMap().getTo();
2785 SequenceI match = matcher.findIdMatch(mappedTo);
2788 matcher.add(mappedTo);
2789 copy.addSequence(mappedTo);
2799 * Try to align sequences in 'unaligned' to match the alignment of their
2800 * mapped regions in 'aligned'. For example, could use this to align CDS
2801 * sequences which are mapped to their parent cDNA sequences.
2803 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2804 * dna-to-protein or protein-to-dna use alternative methods.
2807 * sequences to be aligned
2809 * holds aligned sequences and their mappings
2812 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2815 * easy case - aligning a copy of aligned sequences
2817 if (alignAsSameSequences(unaligned, aligned))
2819 return unaligned.getHeight();
2823 * fancy case - aligning via mappings between sequences
2825 List<SequenceI> unmapped = new ArrayList<>();
2826 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2827 unaligned, aligned, unmapped);
2828 int width = columnMap.size();
2829 char gap = unaligned.getGapCharacter();
2830 int realignedCount = 0;
2831 // TODO: verify this loop scales sensibly for very wide/high alignments
2833 for (SequenceI seq : unaligned.getSequences())
2835 if (!unmapped.contains(seq))
2837 char[] newSeq = new char[width];
2838 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2839 // Integer iteration below
2844 * traverse the map to find columns populated
2847 for (Integer column : columnMap.keySet())
2849 Character c = columnMap.get(column).get(seq);
2853 * sequence has a character at this position
2863 * trim trailing gaps
2865 if (lastCol < width)
2867 char[] tmp = new char[lastCol + 1];
2868 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2871 // TODO: optimise SequenceI to avoid char[]->String->char[]
2872 seq.setSequence(String.valueOf(newSeq));
2876 return realignedCount;
2880 * If unaligned and aligned sequences share the same dataset sequences, then
2881 * simply copies the aligned sequences to the unaligned sequences and returns
2882 * true; else returns false
2885 * - sequences to be aligned based on aligned
2887 * - 'guide' alignment containing sequences derived from same dataset
2891 static boolean alignAsSameSequences(AlignmentI unaligned,
2894 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2896 return false; // should only pass alignments with datasets here
2899 // map from dataset sequence to alignment sequence(s)
2900 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2901 for (SequenceI seq : aligned.getSequences())
2903 SequenceI ds = seq.getDatasetSequence();
2904 if (alignedDatasets.get(ds) == null)
2906 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2908 alignedDatasets.get(ds).add(seq);
2912 * first pass - check whether all sequences to be aligned share a dataset
2913 * sequence with an aligned sequence
2915 for (SequenceI seq : unaligned.getSequences())
2917 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2924 * second pass - copy aligned sequences;
2925 * heuristic rule: pair off sequences in order for the case where
2926 * more than one shares the same dataset sequence
2928 for (SequenceI seq : unaligned.getSequences())
2930 List<SequenceI> alignedSequences = alignedDatasets
2931 .get(seq.getDatasetSequence());
2932 // TODO: getSequenceAsString() will be deprecated in the future
2933 // TODO: need to leave to SequenceI implementor to update gaps
2934 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2935 if (alignedSequences.size() > 0)
2937 // pop off aligned sequences (except the last one)
2938 alignedSequences.remove(0);
2946 * Returns a map whose key is alignment column number (base 1), and whose
2947 * values are a map of sequence characters in that column.
2954 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2955 AlignmentI unaligned, AlignmentI aligned,
2956 List<SequenceI> unmapped)
2959 * Map will hold, for each aligned column position, a map of
2960 * {unalignedSequence, characterPerSequence} at that position.
2961 * TreeMap keeps the entries in ascending column order.
2963 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2966 * record any sequences that have no mapping so can't be realigned
2968 unmapped.addAll(unaligned.getSequences());
2970 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2972 for (SequenceI seq : unaligned.getSequences())
2974 for (AlignedCodonFrame mapping : mappings)
2976 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2977 if (fromSeq != null)
2979 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2980 if (addMappedPositions(seq, fromSeq, seqMap, map))
2982 unmapped.remove(seq);
2991 * Helper method that adds to a map the mapped column positions of a sequence.
2993 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2994 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2998 * the sequence whose column positions we are recording
3000 * a sequence that is mapped to the first sequence
3002 * the mapping from 'fromSeq' to 'seq'
3004 * a map to add the column positions (in fromSeq) of the mapped
3008 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
3009 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
3017 * invert mapping if it is from unaligned to aligned sequence
3019 if (seqMap.getTo() == fromSeq.getDatasetSequence())
3021 seqMap = new Mapping(seq.getDatasetSequence(),
3022 seqMap.getMap().getInverse());
3025 int toStart = seq.getStart();
3028 * traverse [start, end, start, end...] ranges in fromSeq
3030 for (int[] fromRange : seqMap.getMap().getFromRanges())
3032 for (int i = 0; i < fromRange.length - 1; i += 2)
3034 boolean forward = fromRange[i + 1] >= fromRange[i];
3037 * find the range mapped to (sequence positions base 1)
3039 int[] range = seqMap.locateMappedRange(fromRange[i],
3043 System.err.println("Error in mapping " + seqMap + " from "
3044 + fromSeq.getName());
3047 int fromCol = fromSeq.findIndex(fromRange[i]);
3048 int mappedCharPos = range[0];
3051 * walk over the 'from' aligned sequence in forward or reverse
3052 * direction; when a non-gap is found, record the column position
3053 * of the next character of the mapped-to sequence; stop when all
3054 * the characters of the range have been counted
3056 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
3059 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
3062 * mapped from sequence has a character in this column
3063 * record the column position for the mapped to character
3065 Map<SequenceI, Character> seqsMap = map.get(fromCol);
3066 if (seqsMap == null)
3068 seqsMap = new HashMap<>();
3069 map.put(fromCol, seqsMap);
3071 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
3074 fromCol += (forward ? 1 : -1);
3081 // strictly temporary hack until proper criteria for aligning protein to cds
3082 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
3083 public static boolean looksLikeEnsembl(AlignmentI alignment)
3085 for (SequenceI seq : alignment.getSequences())
3087 String name = seq.getName();
3088 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))