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.commands.RemoveGapColCommand;
26 import jalview.datamodel.AlignedCodon;
27 import jalview.datamodel.AlignedCodonFrame;
28 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
29 import jalview.datamodel.Alignment;
30 import jalview.datamodel.AlignmentAnnotation;
31 import jalview.datamodel.AlignmentI;
32 import jalview.datamodel.DBRefEntry;
33 import jalview.datamodel.GeneLociI;
34 import jalview.datamodel.IncompleteCodonException;
35 import jalview.datamodel.Mapping;
36 import jalview.datamodel.Sequence;
37 import jalview.datamodel.SequenceFeature;
38 import jalview.datamodel.SequenceGroup;
39 import jalview.datamodel.SequenceI;
40 import jalview.datamodel.features.SequenceFeatures;
41 import jalview.io.gff.Gff3Helper;
42 import jalview.io.gff.SequenceOntologyI;
43 import jalview.schemes.ResidueProperties;
44 import jalview.util.Comparison;
45 import jalview.util.DBRefUtils;
46 import jalview.util.IntRangeComparator;
47 import jalview.util.MapList;
48 import jalview.util.MappingUtils;
49 import jalview.util.StringUtils;
51 import java.io.UnsupportedEncodingException;
52 import java.net.URLEncoder;
53 import java.util.ArrayList;
54 import java.util.Arrays;
55 import java.util.Collection;
56 import java.util.Collections;
57 import java.util.HashMap;
58 import java.util.HashSet;
59 import java.util.Iterator;
60 import java.util.LinkedHashMap;
61 import java.util.List;
63 import java.util.Map.Entry;
64 import java.util.NoSuchElementException;
66 import java.util.SortedMap;
67 import java.util.TreeMap;
70 * grab bag of useful alignment manipulation operations Expect these to be
71 * refactored elsewhere at some point.
76 public class AlignmentUtils
78 private static final int CODON_LENGTH = 3;
80 private static final String SEQUENCE_VARIANT = "sequence_variant:";
83 * the 'id' attribute is provided for variant features fetched from
84 * Ensembl using its REST service with JSON format
86 public static final String VARIANT_ID = "id";
89 * A data model to hold the 'normal' base value at a position, and an optional
90 * sequence variant feature
92 static final class DnaVariant
96 SequenceFeature variant;
98 DnaVariant(String nuc)
104 DnaVariant(String nuc, SequenceFeature var)
110 public String getSource()
112 return variant == null ? null : variant.getFeatureGroup();
116 * toString for aid in the debugger only
119 public String toString()
121 return base + ":" + (variant == null ? "" : variant.getDescription());
126 * given an existing alignment, create a new alignment including all, or up to
127 * flankSize additional symbols from each sequence's dataset sequence
133 public static AlignmentI expandContext(AlignmentI core, int flankSize)
135 List<SequenceI> sq = new ArrayList<>();
137 for (SequenceI s : core.getSequences())
139 SequenceI newSeq = s.deriveSequence();
140 final int newSeqStart = newSeq.getStart() - 1;
141 if (newSeqStart > maxoffset
142 && newSeq.getDatasetSequence().getStart() < s.getStart())
144 maxoffset = newSeqStart;
150 maxoffset = Math.min(maxoffset, flankSize);
154 * now add offset left and right to create an expanded alignment
156 for (SequenceI s : sq)
159 while (ds.getDatasetSequence() != null)
161 ds = ds.getDatasetSequence();
163 int s_end = s.findPosition(s.getStart() + s.getLength());
164 // find available flanking residues for sequence
165 int ustream_ds = s.getStart() - ds.getStart();
166 int dstream_ds = ds.getEnd() - s_end;
168 // build new flanked sequence
170 // compute gap padding to start of flanking sequence
171 int offset = maxoffset - ustream_ds;
173 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
176 if (flankSize < ustream_ds)
178 // take up to flankSize residues
179 offset = maxoffset - flankSize;
180 ustream_ds = flankSize;
182 if (flankSize <= dstream_ds)
184 dstream_ds = flankSize - 1;
187 // TODO use Character.toLowerCase to avoid creating String objects?
188 char[] upstream = new String(ds
189 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
190 .toLowerCase().toCharArray();
191 char[] downstream = new String(
192 ds.getSequence(s_end - 1, s_end + dstream_ds)).toLowerCase()
194 char[] coreseq = s.getSequence();
195 char[] nseq = new char[offset + upstream.length + downstream.length
197 char c = core.getGapCharacter();
200 for (; p < offset; p++)
205 System.arraycopy(upstream, 0, nseq, p, upstream.length);
206 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
208 System.arraycopy(downstream, 0, nseq,
209 p + coreseq.length + upstream.length, downstream.length);
210 s.setSequence(new String(nseq));
211 s.setStart(s.getStart() - ustream_ds);
212 s.setEnd(s_end + downstream.length);
214 AlignmentI newAl = new jalview.datamodel.Alignment(
215 sq.toArray(new SequenceI[0]));
216 for (SequenceI s : sq)
218 if (s.getAnnotation() != null)
220 for (AlignmentAnnotation aa : s.getAnnotation())
222 aa.adjustForAlignment(); // JAL-1712 fix
223 newAl.addAnnotation(aa);
227 newAl.setDataset(core.getDataset());
232 * Returns the index (zero-based position) of a sequence in an alignment, or
239 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
243 for (SequenceI alSeq : al.getSequences())
256 * Returns a map of lists of sequences in the alignment, keyed by sequence
257 * name. For use in mapping between different alignment views of the same
260 * @see jalview.datamodel.AlignmentI#getSequencesByName()
262 public static Map<String, List<SequenceI>> getSequencesByName(
265 Map<String, List<SequenceI>> theMap = new LinkedHashMap<>();
266 for (SequenceI seq : al.getSequences())
268 String name = seq.getName();
271 List<SequenceI> seqs = theMap.get(name);
274 seqs = new ArrayList<>();
275 theMap.put(name, seqs);
284 * Build mapping of protein to cDNA alignment. Mappings are made between
285 * sequences where the cDNA translates to the protein sequence. Any new
286 * mappings are added to the protein alignment. Returns true if any mappings
287 * either already exist or were added, else false.
289 * @param proteinAlignment
290 * @param cdnaAlignment
293 public static boolean mapProteinAlignmentToCdna(
294 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
296 if (proteinAlignment == null || cdnaAlignment == null)
301 Set<SequenceI> mappedDna = new HashSet<>();
302 Set<SequenceI> mappedProtein = new HashSet<>();
305 * First pass - map sequences where cross-references exist. This include
306 * 1-to-many mappings to support, for example, variant cDNA.
308 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
309 cdnaAlignment, mappedDna, mappedProtein, true);
312 * Second pass - map sequences where no cross-references exist. This only
313 * does 1-to-1 mappings and assumes corresponding sequences are in the same
314 * order in the alignments.
316 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
317 mappedDna, mappedProtein, false);
318 return mappingPerformed;
322 * Make mappings between compatible sequences (where the cDNA translation
323 * matches the protein).
325 * @param proteinAlignment
326 * @param cdnaAlignment
328 * a set of mapped DNA sequences (to add to)
329 * @param mappedProtein
330 * a set of mapped Protein sequences (to add to)
332 * if true, only map sequences where xrefs exist
335 protected static boolean mapProteinToCdna(
336 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
337 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
340 boolean mappingExistsOrAdded = false;
341 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
342 for (SequenceI aaSeq : thisSeqs)
344 boolean proteinMapped = false;
345 AlignedCodonFrame acf = new AlignedCodonFrame();
347 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
350 * Always try to map if sequences have xref to each other; this supports
351 * variant cDNA or alternative splicing for a protein sequence.
353 * If no xrefs, try to map progressively, assuming that alignments have
354 * mappable sequences in corresponding order. These are not
355 * many-to-many, as that would risk mixing species with similar cDNA
358 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
364 * Don't map non-xrefd sequences more than once each. This heuristic
365 * allows us to pair up similar sequences in ordered alignments.
367 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
368 || mappedDna.contains(cdnaSeq)))
372 if (mappingExists(proteinAlignment.getCodonFrames(),
373 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
375 mappingExistsOrAdded = true;
379 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
382 acf.addMap(cdnaSeq, aaSeq, map);
383 mappingExistsOrAdded = true;
384 proteinMapped = true;
385 mappedDna.add(cdnaSeq);
386 mappedProtein.add(aaSeq);
392 proteinAlignment.addCodonFrame(acf);
395 return mappingExistsOrAdded;
399 * Answers true if the mappings include one between the given (dataset)
402 protected static boolean mappingExists(List<AlignedCodonFrame> mappings,
403 SequenceI aaSeq, SequenceI cdnaSeq)
405 if (mappings != null)
407 for (AlignedCodonFrame acf : mappings)
409 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
419 * Builds a mapping (if possible) of a cDNA to a protein sequence.
421 * <li>first checks if the cdna translates exactly to the protein
423 * <li>else checks for translation after removing a STOP codon</li>
424 * <li>else checks for translation after removing a START codon</li>
425 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
427 * Returns null if no mapping is determined.
430 * the aligned protein sequence
432 * the aligned cdna sequence
435 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
439 * Here we handle either dataset sequence set (desktop) or absent (applet).
440 * Use only the char[] form of the sequence to avoid creating possibly large
443 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
444 char[] aaSeqChars = proteinDataset != null
445 ? proteinDataset.getSequence()
446 : proteinSeq.getSequence();
447 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
448 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
449 : cdnaSeq.getSequence();
450 if (aaSeqChars == null || cdnaSeqChars == null)
456 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
458 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
459 int cdnaLength = cdnaSeqChars.length;
460 int cdnaStart = cdnaSeq.getStart();
461 int cdnaEnd = cdnaSeq.getEnd();
462 final int proteinStart = proteinSeq.getStart();
463 final int proteinEnd = proteinSeq.getEnd();
466 * If lengths don't match, try ignoring stop codon (if present)
468 if (cdnaLength != mappedLength && cdnaLength > 2)
470 String lastCodon = String.valueOf(cdnaSeqChars,
471 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
472 for (String stop : ResidueProperties.STOP_CODONS)
474 if (lastCodon.equals(stop))
476 cdnaEnd -= CODON_LENGTH;
477 cdnaLength -= CODON_LENGTH;
484 * If lengths still don't match, try ignoring start codon.
487 if (cdnaLength != mappedLength && cdnaLength > 2
488 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
489 .equals(ResidueProperties.START))
491 startOffset += CODON_LENGTH;
492 cdnaStart += CODON_LENGTH;
493 cdnaLength -= CODON_LENGTH;
496 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
499 * protein is translation of dna (+/- start/stop codons)
501 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
503 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
508 * translation failed - try mapping CDS annotated regions of dna
510 return mapCdsToProtein(cdnaSeq, proteinSeq);
514 * Test whether the given cdna sequence, starting at the given offset,
515 * translates to the given amino acid sequence, using the standard translation
516 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
518 * @param cdnaSeqChars
523 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
526 if (cdnaSeqChars == null || aaSeqChars == null)
532 int dnaPos = cdnaStart;
533 for (; dnaPos < cdnaSeqChars.length - 2
534 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
536 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
537 final String translated = ResidueProperties.codonTranslate(codon);
540 * allow * in protein to match untranslatable in dna
542 final char aaRes = aaSeqChars[aaPos];
543 if ((translated == null || ResidueProperties.STOP.equals(translated))
548 if (translated == null || !(aaRes == translated.charAt(0)))
551 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
552 // + codon + "(" + translated + ") != " + aaRes));
558 * check we matched all of the protein sequence
560 if (aaPos != aaSeqChars.length)
566 * check we matched all of the dna except
567 * for optional trailing STOP codon
569 if (dnaPos == cdnaSeqChars.length)
573 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
575 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
576 if (ResidueProperties.STOP
577 .equals(ResidueProperties.codonTranslate(codon)))
586 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
587 * currently assumes that we are aligning cDNA to match protein.
590 * the sequence to be realigned
592 * the alignment whose sequence alignment is to be 'copied'
594 * character string represent a gap in the realigned sequence
595 * @param preserveUnmappedGaps
596 * @param preserveMappedGaps
597 * @return true if the sequence was realigned, false if it could not be
599 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
600 String gap, boolean preserveMappedGaps,
601 boolean preserveUnmappedGaps)
604 * Get any mappings from the source alignment to the target (dataset)
607 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
608 // all mappings. Would it help to constrain this?
609 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
610 if (mappings == null || mappings.isEmpty())
616 * Locate the aligned source sequence whose dataset sequence is mapped. We
617 * just take the first match here (as we can't align like more than one
620 SequenceI alignFrom = null;
621 AlignedCodonFrame mapping = null;
622 for (AlignedCodonFrame mp : mappings)
624 alignFrom = mp.findAlignedSequence(seq, al);
625 if (alignFrom != null)
632 if (alignFrom == null)
636 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
637 preserveMappedGaps, preserveUnmappedGaps);
642 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
643 * match residues and codons. Flags control whether existing gaps in unmapped
644 * (intron) and mapped (exon) regions are preserved or not. Gaps between
645 * intron and exon are only retained if both flags are set.
652 * @param preserveUnmappedGaps
653 * @param preserveMappedGaps
655 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
656 AlignedCodonFrame mapping, String myGap, char sourceGap,
657 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
659 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
661 // aligned and dataset sequence positions, all base zero
665 int basesWritten = 0;
666 char myGapChar = myGap.charAt(0);
667 int ratio = myGap.length();
669 int fromOffset = alignFrom.getStart() - 1;
670 int toOffset = alignTo.getStart() - 1;
671 int sourceGapMappedLength = 0;
672 boolean inExon = false;
673 final int toLength = alignTo.getLength();
674 final int fromLength = alignFrom.getLength();
675 StringBuilder thisAligned = new StringBuilder(2 * toLength);
678 * Traverse the 'model' aligned sequence
680 for (int i = 0; i < fromLength; i++)
682 char sourceChar = alignFrom.getCharAt(i);
683 if (sourceChar == sourceGap)
685 sourceGapMappedLength += ratio;
690 * Found a non-gap character. Locate its mapped region if any.
693 // Note mapping positions are base 1, our sequence positions base 0
694 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
695 sourceDsPos + fromOffset);
696 if (mappedPos == null)
699 * unmapped position; treat like a gap
701 sourceGapMappedLength += ratio;
702 // System.err.println("Can't align: no codon mapping to residue "
703 // + sourceDsPos + "(" + sourceChar + ")");
708 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
709 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
710 StringBuilder trailingCopiedGap = new StringBuilder();
713 * Copy dna sequence up to and including this codon. Optionally, include
714 * gaps before the codon starts (in introns) and/or after the codon starts
717 * Note this only works for 'linear' splicing, not reverse or interleaved.
718 * But then 'align dna as protein' doesn't make much sense otherwise.
720 int intronLength = 0;
721 while (basesWritten + toOffset < mappedCodonEnd
722 && thisSeqPos < toLength)
724 final char c = alignTo.getCharAt(thisSeqPos++);
728 int sourcePosition = basesWritten + toOffset;
729 if (sourcePosition < mappedCodonStart)
732 * Found an unmapped (intron) base. First add in any preceding gaps
735 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
737 thisAligned.append(trailingCopiedGap.toString());
738 intronLength += trailingCopiedGap.length();
739 trailingCopiedGap = new StringBuilder();
746 final boolean startOfCodon = sourcePosition == mappedCodonStart;
747 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
748 preserveUnmappedGaps, sourceGapMappedLength, inExon,
749 trailingCopiedGap.length(), intronLength, startOfCodon);
750 for (int k = 0; k < gapsToAdd; k++)
752 thisAligned.append(myGapChar);
754 sourceGapMappedLength = 0;
757 thisAligned.append(c);
758 trailingCopiedGap = new StringBuilder();
762 if (inExon && preserveMappedGaps)
764 trailingCopiedGap.append(myGapChar);
766 else if (!inExon && preserveUnmappedGaps)
768 trailingCopiedGap.append(myGapChar);
775 * At end of model aligned sequence. Copy any remaining target sequence, optionally
776 * including (intron) gaps.
778 while (thisSeqPos < toLength)
780 final char c = alignTo.getCharAt(thisSeqPos++);
781 if (c != myGapChar || preserveUnmappedGaps)
783 thisAligned.append(c);
785 sourceGapMappedLength--;
789 * finally add gaps to pad for any trailing source gaps or
790 * unmapped characters
792 if (preserveUnmappedGaps)
794 while (sourceGapMappedLength > 0)
796 thisAligned.append(myGapChar);
797 sourceGapMappedLength--;
802 * All done aligning, set the aligned sequence.
804 alignTo.setSequence(new String(thisAligned));
808 * Helper method to work out how many gaps to insert when realigning.
810 * @param preserveMappedGaps
811 * @param preserveUnmappedGaps
812 * @param sourceGapMappedLength
814 * @param trailingCopiedGap
815 * @param intronLength
816 * @param startOfCodon
819 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
820 boolean preserveUnmappedGaps, int sourceGapMappedLength,
821 boolean inExon, int trailingGapLength, int intronLength,
822 final boolean startOfCodon)
828 * Reached start of codon. Ignore trailing gaps in intron unless we are
829 * preserving gaps in both exon and intron. Ignore them anyway if the
830 * protein alignment introduces a gap at least as large as the intronic
833 if (inExon && !preserveMappedGaps)
835 trailingGapLength = 0;
837 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
839 trailingGapLength = 0;
843 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
847 if (intronLength + trailingGapLength <= sourceGapMappedLength)
849 gapsToAdd = sourceGapMappedLength - intronLength;
853 gapsToAdd = Math.min(
854 intronLength + trailingGapLength - sourceGapMappedLength,
862 * second or third base of codon; check for any gaps in dna
864 if (!preserveMappedGaps)
866 trailingGapLength = 0;
868 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
874 * Realigns the given protein to match the alignment of the dna, using codon
875 * mappings to translate aligned codon positions to protein residues.
878 * the alignment whose sequences are realigned by this method
880 * the dna alignment whose alignment we are 'copying'
881 * @return the number of sequences that were realigned
883 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
885 if (protein.isNucleotide() || !dna.isNucleotide())
887 System.err.println("Wrong alignment type in alignProteinAsDna");
890 List<SequenceI> unmappedProtein = new ArrayList<>();
891 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
892 protein, dna, unmappedProtein);
893 return alignProteinAs(protein, alignedCodons, unmappedProtein);
897 * Realigns the given dna to match the alignment of the protein, using codon
898 * mappings to translate aligned peptide positions to codons.
900 * Always produces a padded CDS alignment.
903 * the alignment whose sequences are realigned by this method
905 * the protein alignment whose alignment we are 'copying'
906 * @return the number of sequences that were realigned
908 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
910 if (protein.isNucleotide() || !dna.isNucleotide())
912 System.err.println("Wrong alignment type in alignProteinAsDna");
915 // todo: implement this
916 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
917 int alignedCount = 0;
918 int width = 0; // alignment width for padding CDS
919 for (SequenceI dnaSeq : dna.getSequences())
921 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
922 dna.getGapCharacter()))
926 width = Math.max(dnaSeq.getLength(), width);
930 for (SequenceI dnaSeq : dna.getSequences())
932 oldwidth = dnaSeq.getLength();
933 diff = width - oldwidth;
936 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
943 * Helper method to align (if possible) the dna sequence to match the
944 * alignment of a mapped protein sequence. This is currently limited to
945 * handling coding sequence only.
953 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
954 AlignmentI protein, List<AlignedCodonFrame> mappings,
957 SequenceI cdsDss = cdsSeq.getDatasetSequence();
961 .println("alignCdsSequenceAsProtein needs aligned sequence!");
965 List<AlignedCodonFrame> dnaMappings = MappingUtils
966 .findMappingsForSequence(cdsSeq, mappings);
967 for (AlignedCodonFrame mapping : dnaMappings)
969 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
972 final int peptideLength = peptide.getLength();
973 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
976 MapList mapList = map.getMap();
977 if (map.getTo() == peptide.getDatasetSequence())
979 mapList = mapList.getInverse();
981 final int cdsLength = cdsDss.getLength();
982 int mappedFromLength = MappingUtils.getLength(mapList
984 int mappedToLength = MappingUtils
985 .getLength(mapList.getToRanges());
986 boolean addStopCodon = (cdsLength == mappedFromLength
987 * CODON_LENGTH + CODON_LENGTH)
988 || (peptide.getDatasetSequence()
989 .getLength() == mappedFromLength - 1);
990 if (cdsLength != mappedToLength && !addStopCodon)
992 System.err.println(String.format(
993 "Can't align cds as protein (length mismatch %d/%d): %s",
994 cdsLength, mappedToLength, cdsSeq.getName()));
998 * pre-fill the aligned cds sequence with gaps
1000 char[] alignedCds = new char[peptideLength * CODON_LENGTH
1001 + (addStopCodon ? CODON_LENGTH : 0)];
1002 Arrays.fill(alignedCds, gapChar);
1005 * walk over the aligned peptide sequence and insert mapped
1006 * codons for residues in the aligned cds sequence
1008 int copiedBases = 0;
1009 int cdsStart = cdsDss.getStart();
1010 int proteinPos = peptide.getStart() - 1;
1013 for (int col = 0; col < peptideLength; col++)
1015 char residue = peptide.getCharAt(col);
1017 if (Comparison.isGap(residue))
1019 cdsCol += CODON_LENGTH;
1024 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1027 // e.g. incomplete start codon, X in peptide
1028 cdsCol += CODON_LENGTH;
1032 for (int j = codon[0]; j <= codon[1]; j++)
1034 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1035 alignedCds[cdsCol++] = mappedBase;
1043 * append stop codon if not mapped from protein,
1044 * closing it up to the end of the mapped sequence
1046 if (copiedBases == cdsLength - CODON_LENGTH)
1048 for (int i = alignedCds.length - 1; i >= 0; i--)
1050 if (!Comparison.isGap(alignedCds[i]))
1052 cdsCol = i + 1; // gap just after end of sequence
1056 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1058 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1061 cdsSeq.setSequence(new String(alignedCds));
1070 * Builds a map whose key is an aligned codon position (3 alignment column
1071 * numbers base 0), and whose value is a map from protein sequence to each
1072 * protein's peptide residue for that codon. The map generates an ordering of
1073 * the codons, and allows us to read off the peptides at each position in
1074 * order to assemble 'aligned' protein sequences.
1077 * the protein alignment
1079 * the coding dna alignment
1080 * @param unmappedProtein
1081 * any unmapped proteins are added to this list
1084 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1085 AlignmentI protein, AlignmentI dna,
1086 List<SequenceI> unmappedProtein)
1089 * maintain a list of any proteins with no mappings - these will be
1090 * rendered 'as is' in the protein alignment as we can't align them
1092 unmappedProtein.addAll(protein.getSequences());
1094 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1097 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1098 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1099 * comparator keeps the codon positions ordered.
1101 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<>(
1102 new CodonComparator());
1104 for (SequenceI dnaSeq : dna.getSequences())
1106 for (AlignedCodonFrame mapping : mappings)
1108 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1111 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1112 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1114 unmappedProtein.remove(prot);
1120 * Finally add any unmapped peptide start residues (e.g. for incomplete
1121 * codons) as if at the codon position before the second residue
1123 // TODO resolve JAL-2022 so this fudge can be removed
1124 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1125 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1127 return alignedCodons;
1131 * Scans for any protein mapped from position 2 (meaning unmapped start
1132 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1133 * preceding position in the alignment
1135 * @param alignedCodons
1136 * the codon-to-peptide map
1137 * @param mappedSequenceCount
1138 * the number of distinct sequences in the map
1140 protected static void addUnmappedPeptideStarts(
1141 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1142 int mappedSequenceCount)
1144 // TODO delete this ugly hack once JAL-2022 is resolved
1145 // i.e. we can model startPhase > 0 (incomplete start codon)
1147 List<SequenceI> sequencesChecked = new ArrayList<>();
1148 AlignedCodon lastCodon = null;
1149 Map<SequenceI, AlignedCodon> toAdd = new HashMap<>();
1151 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1154 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1157 SequenceI seq = sequenceCodon.getKey();
1158 if (sequencesChecked.contains(seq))
1162 sequencesChecked.add(seq);
1163 AlignedCodon codon = sequenceCodon.getValue();
1164 if (codon.peptideCol > 1)
1167 "Problem mapping protein with >1 unmapped start positions: "
1170 else if (codon.peptideCol == 1)
1173 * first position (peptideCol == 0) was unmapped - add it
1175 if (lastCodon != null)
1177 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1178 lastCodon.pos2, lastCodon.pos3,
1179 String.valueOf(seq.getCharAt(0)), 0);
1180 toAdd.put(seq, firstPeptide);
1185 * unmapped residue at start of alignment (no prior column) -
1186 * 'insert' at nominal codon [0, 0, 0]
1188 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1189 String.valueOf(seq.getCharAt(0)), 0);
1190 toAdd.put(seq, firstPeptide);
1193 if (sequencesChecked.size() == mappedSequenceCount)
1195 // no need to check past first mapped position in all sequences
1199 lastCodon = entry.getKey();
1203 * add any new codons safely after iterating over the map
1205 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1207 addCodonToMap(alignedCodons, startCodon.getValue(),
1208 startCodon.getKey());
1213 * Update the aligned protein sequences to match the codon alignments given in
1217 * @param alignedCodons
1218 * an ordered map of codon positions (columns), with sequence/peptide
1219 * values present in each column
1220 * @param unmappedProtein
1223 protected static int alignProteinAs(AlignmentI protein,
1224 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1225 List<SequenceI> unmappedProtein)
1228 * prefill peptide sequences with gaps
1230 int alignedWidth = alignedCodons.size();
1231 char[] gaps = new char[alignedWidth];
1232 Arrays.fill(gaps, protein.getGapCharacter());
1233 Map<SequenceI, char[]> peptides = new HashMap<>();
1234 for (SequenceI seq : protein.getSequences())
1236 if (!unmappedProtein.contains(seq))
1238 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1243 * Traverse the codons left to right (as defined by CodonComparator)
1244 * and insert peptides in each column where the sequence is mapped.
1245 * This gives a peptide 'alignment' where residues are aligned if their
1246 * corresponding codons occupy the same columns in the cdna alignment.
1249 for (AlignedCodon codon : alignedCodons.keySet())
1251 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1253 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1255 char residue = entry.getValue().product.charAt(0);
1256 peptides.get(entry.getKey())[column] = residue;
1262 * and finally set the constructed sequences
1264 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1266 entry.getKey().setSequence(new String(entry.getValue()));
1273 * Populate the map of aligned codons by traversing the given sequence
1274 * mapping, locating the aligned positions of mapped codons, and adding those
1275 * positions and their translation products to the map.
1278 * the aligned sequence we are mapping from
1280 * the sequence to be aligned to the codons
1282 * the gap character in the dna sequence
1284 * a mapping to a sequence translation
1285 * @param alignedCodons
1286 * the map we are building up
1288 static void addCodonPositions(SequenceI dna, SequenceI protein,
1289 char gapChar, Mapping seqMap,
1290 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1292 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1295 * add codon positions, and their peptide translations, to the alignment
1296 * map, while remembering the first codon mapped
1298 while (codons.hasNext())
1302 AlignedCodon codon = codons.next();
1303 addCodonToMap(alignedCodons, codon, protein);
1304 } catch (IncompleteCodonException e)
1306 // possible incomplete trailing codon - ignore
1307 } catch (NoSuchElementException e)
1309 // possibly peptide lacking STOP
1315 * Helper method to add a codon-to-peptide entry to the aligned codons map
1317 * @param alignedCodons
1321 protected static void addCodonToMap(
1322 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1323 AlignedCodon codon, SequenceI protein)
1325 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1326 if (seqProduct == null)
1328 seqProduct = new HashMap<>();
1329 alignedCodons.put(codon, seqProduct);
1331 seqProduct.put(protein, codon);
1335 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1336 * between at least one pair of sequences in the two alignments. Currently,
1339 * <li>One alignment must be nucleotide, and the other protein</li>
1340 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1341 * <li>Mappable means the nucleotide translation matches the protein
1343 * <li>The translation may ignore start and stop codons if present in the
1351 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1353 if (al1 == null || al2 == null)
1359 * Require one nucleotide and one protein
1361 if (al1.isNucleotide() == al2.isNucleotide())
1365 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1366 AlignmentI protein = dna == al1 ? al2 : al1;
1367 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1368 for (SequenceI dnaSeq : dna.getSequences())
1370 for (SequenceI proteinSeq : protein.getSequences())
1372 if (isMappable(dnaSeq, proteinSeq, mappings))
1382 * Returns true if the dna sequence is mapped, or could be mapped, to the
1390 protected static boolean isMappable(SequenceI dnaSeq,
1391 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1393 if (dnaSeq == null || proteinSeq == null)
1398 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1399 : dnaSeq.getDatasetSequence();
1400 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1402 : proteinSeq.getDatasetSequence();
1404 for (AlignedCodonFrame mapping : mappings)
1406 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1416 * Just try to make a mapping (it is not yet stored), test whether
1419 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1423 * Finds any reference annotations associated with the sequences in
1424 * sequenceScope, that are not already added to the alignment, and adds them
1425 * to the 'candidates' map. Also populates a lookup table of annotation
1426 * labels, keyed by calcId, for use in constructing tooltips or the like.
1428 * @param sequenceScope
1429 * the sequences to scan for reference annotations
1430 * @param labelForCalcId
1431 * (optional) map to populate with label for calcId
1433 * map to populate with annotations for sequence
1435 * the alignment to check for presence of annotations
1437 public static void findAddableReferenceAnnotations(
1438 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1439 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1442 if (sequenceScope == null)
1448 * For each sequence in scope, make a list of any annotations on the
1449 * underlying dataset sequence which are not already on the alignment.
1451 * Add to a map of { alignmentSequence, <List of annotations to add> }
1453 for (SequenceI seq : sequenceScope)
1455 SequenceI dataset = seq.getDatasetSequence();
1456 if (dataset == null)
1460 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1461 if (datasetAnnotations == null)
1465 final List<AlignmentAnnotation> result = new ArrayList<>();
1466 for (AlignmentAnnotation dsann : datasetAnnotations)
1469 * Find matching annotations on the alignment. If none is found, then
1470 * add this annotation to the list of 'addable' annotations for this
1473 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1474 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1475 if (!matchedAlignmentAnnotations.iterator().hasNext())
1478 if (labelForCalcId != null)
1480 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1485 * Save any addable annotations for this sequence
1487 if (!result.isEmpty())
1489 candidates.put(seq, result);
1495 * Adds annotations to the top of the alignment annotations, in the same order
1496 * as their related sequences.
1498 * @param annotations
1499 * the annotations to add
1501 * the alignment to add them to
1502 * @param selectionGroup
1503 * current selection group (or null if none)
1505 public static void addReferenceAnnotations(
1506 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1507 final AlignmentI alignment, final SequenceGroup selectionGroup)
1509 for (SequenceI seq : annotations.keySet())
1511 for (AlignmentAnnotation ann : annotations.get(seq))
1513 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1515 int endRes = ann.annotations.length;
1516 if (selectionGroup != null)
1518 startRes = selectionGroup.getStartRes();
1519 endRes = selectionGroup.getEndRes();
1521 copyAnn.restrict(startRes, endRes);
1524 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1525 * original annotation is already on the sequence.
1527 if (!seq.hasAnnotation(ann))
1529 seq.addAlignmentAnnotation(copyAnn);
1532 copyAnn.adjustForAlignment();
1533 // add to the alignment and set visible
1534 alignment.addAnnotation(copyAnn);
1535 copyAnn.visible = true;
1541 * Set visibility of alignment annotations of specified types (labels), for
1542 * specified sequences. This supports controls like "Show all secondary
1543 * structure", "Hide all Temp factor", etc.
1545 * @al the alignment to scan for annotations
1547 * the types (labels) of annotations to be updated
1548 * @param forSequences
1549 * if not null, only annotations linked to one of these sequences are
1550 * in scope for update; if null, acts on all sequence annotations
1552 * if this flag is true, 'types' is ignored (label not checked)
1554 * if true, set visibility on, else set off
1556 public static void showOrHideSequenceAnnotations(AlignmentI al,
1557 Collection<String> types, List<SequenceI> forSequences,
1558 boolean anyType, boolean doShow)
1560 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1563 for (AlignmentAnnotation aa : anns)
1565 if (anyType || types.contains(aa.label))
1567 if ((aa.sequenceRef != null) && (forSequences == null
1568 || forSequences.contains(aa.sequenceRef)))
1570 aa.visible = doShow;
1578 * Returns true if either sequence has a cross-reference to the other
1584 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1586 // Note: moved here from class CrossRef as the latter class has dependencies
1587 // not availability to the applet's classpath
1588 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1592 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1593 * that sequence name is structured as Source|AccessionId.
1599 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1601 if (seq1 == null || seq2 == null)
1605 String name = seq2.getName();
1606 final DBRefEntry[] xrefs = seq1.getDBRefs();
1609 for (DBRefEntry xref : xrefs)
1611 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1612 // case-insensitive test, consistent with DBRefEntry.equalRef()
1613 if (xrefName.equalsIgnoreCase(name))
1623 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1624 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1625 * added to the original alignment's dataset, which is shared by the new
1626 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1627 * added to the alignment dataset.
1630 * aligned nucleotide (dna or cds) sequences
1632 * the alignment dataset the sequences belong to
1634 * (optional) to restrict results to CDS that map to specified
1636 * @return an alignment whose sequences are the cds-only parts of the dna
1637 * sequences (or null if no mappings are found)
1639 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1640 AlignmentI dataset, SequenceI[] products)
1642 if (dataset == null || dataset.getDataset() != null)
1644 throw new IllegalArgumentException(
1645 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1647 List<SequenceI> foundSeqs = new ArrayList<>();
1648 List<SequenceI> cdsSeqs = new ArrayList<>();
1649 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1650 HashSet<SequenceI> productSeqs = null;
1651 if (products != null)
1653 productSeqs = new HashSet<>();
1654 for (SequenceI seq : products)
1656 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1657 .getDatasetSequence());
1662 * Construct CDS sequences from mappings on the alignment dataset.
1664 * - find the protein product(s) mapped to from each dna sequence
1665 * - if the mapping covers the whole dna sequence (give or take start/stop
1666 * codon), take the dna as the CDS sequence
1667 * - else search dataset mappings for a suitable dna sequence, i.e. one
1668 * whose whole sequence is mapped to the protein
1669 * - if no sequence found, construct one from the dna sequence and mapping
1670 * (and add it to dataset so it is found if this is repeated)
1672 for (SequenceI dnaSeq : dna)
1674 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1675 : dnaSeq.getDatasetSequence();
1677 List<AlignedCodonFrame> seqMappings = MappingUtils
1678 .findMappingsForSequence(dnaSeq, mappings);
1679 for (AlignedCodonFrame mapping : seqMappings)
1681 List<Mapping> mappingsFromSequence = mapping
1682 .getMappingsFromSequence(dnaSeq);
1684 for (Mapping aMapping : mappingsFromSequence)
1686 MapList mapList = aMapping.getMap();
1687 if (mapList.getFromRatio() == 1)
1690 * not a dna-to-protein mapping (likely dna-to-cds)
1696 * skip if mapping is not to one of the target set of proteins
1698 SequenceI proteinProduct = aMapping.getTo();
1699 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1705 * try to locate the CDS from the dataset mappings;
1706 * guard against duplicate results (for the case that protein has
1707 * dbrefs to both dna and cds sequences)
1709 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1710 seqMappings, aMapping);
1713 if (!foundSeqs.contains(cdsSeq))
1715 foundSeqs.add(cdsSeq);
1716 SequenceI derivedSequence = cdsSeq.deriveSequence();
1717 cdsSeqs.add(derivedSequence);
1718 if (!dataset.getSequences().contains(cdsSeq))
1720 dataset.addSequence(cdsSeq);
1727 * didn't find mapped CDS sequence - construct it and add
1728 * its dataset sequence to the dataset
1730 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1731 dataset).deriveSequence();
1732 // cdsSeq has a name constructed as CDS|<dbref>
1733 // <dbref> will be either the accession for the coding sequence,
1734 // marked in the /via/ dbref to the protein product accession
1735 // or it will be the original nucleotide accession.
1736 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1738 cdsSeqs.add(cdsSeq);
1740 if (!dataset.getSequences().contains(cdsSeqDss))
1742 // check if this sequence is a newly created one
1743 // so needs adding to the dataset
1744 dataset.addSequence(cdsSeqDss);
1748 * add a mapping from CDS to the (unchanged) mapped to range
1750 List<int[]> cdsRange = Collections
1751 .singletonList(new int[]
1752 { cdsSeq.getStart(),
1753 cdsSeq.getLength() + cdsSeq.getStart() - 1 });
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.getMapping());
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,
1991 * construct CDS sequence name as "CDS|" with 'from id' held in the mapping
1992 * if set (e.g. EMBL protein_id), else sequence name appended
1994 String mapFromId = mapping.getMappedFromId();
1995 final String seqId = "CDS|"
1996 + (mapFromId != null ? mapFromId : seq.getName());
1998 SequenceI newSeq = null;
2000 final MapList maplist = mapping.getMap();
2001 if (maplist.isContiguous() && maplist.isFromForwardStrand())
2004 * just a subsequence, keep same dataset sequence
2006 int start = maplist.getFromLowest();
2007 int end = maplist.getFromHighest();
2008 newSeq = seq.getSubSequence(start - 1, end);
2009 newSeq.setName(seqId);
2014 * construct by splicing mapped from ranges
2016 char[] seqChars = seq.getSequence();
2017 List<int[]> fromRanges = maplist.getFromRanges();
2018 int cdsWidth = MappingUtils.getLength(fromRanges);
2019 char[] newSeqChars = new char[cdsWidth];
2022 for (int[] range : fromRanges)
2024 if (range[0] <= range[1])
2026 // forward strand mapping - just copy the range
2027 int length = range[1] - range[0] + 1;
2028 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
2034 // reverse strand mapping - copy and complement one by one
2035 for (int i = range[0]; i >= range[1]; i--)
2037 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
2042 newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2045 if (dataset != null)
2047 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2048 if (matches != null)
2050 boolean matched = false;
2051 for (SequenceI mtch : matches)
2053 if (mtch.getStart() != newSeq.getStart())
2057 if (mtch.getEnd() != newSeq.getEnd())
2061 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2073 "JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
2079 // newSeq.setDescription(mapFromId);
2085 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2086 * the given mapping.
2090 * @param proteinProduct
2092 * @return list of DBRefEntrys added
2094 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2095 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2098 // gather direct refs from contig congruent with mapping
2099 List<DBRefEntry> direct = new ArrayList<>();
2100 HashSet<String> directSources = new HashSet<>();
2102 if (contig.getDBRefs() != null)
2104 for (DBRefEntry dbr : contig.getDBRefs())
2106 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
2108 MapList map = dbr.getMap().getMap();
2109 // check if map is the CDS mapping
2110 if (mapping.getMap().equals(map))
2113 directSources.add(dbr.getSource());
2118 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2119 proteinProduct.getDBRefs(),
2120 directSources.toArray(new String[0]));
2121 List<DBRefEntry> propagated = new ArrayList<>();
2123 // and generate appropriate mappings
2124 for (DBRefEntry cdsref : direct)
2126 // clone maplist and mapping
2127 MapList cdsposmap = new MapList(
2128 Arrays.asList(new int[][]
2129 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2130 cdsref.getMap().getMap().getToRanges(), 3, 1);
2131 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(),
2132 cdsref.getMap().getMap());
2135 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2136 cdsref.getVersion(), cdsref.getAccessionId(),
2137 new Mapping(cdsmap.getTo(), cdsposmap));
2139 // and see if we can map to the protein product for this mapping.
2140 // onSource is the filtered set of accessions on protein that we are
2141 // tranferring, so we assume accession is the same.
2142 if (cdsmap.getTo() == null && onSource != null)
2144 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2145 cdsref.getAccessionId());
2146 if (sourceRefs != null)
2148 for (DBRefEntry srcref : sourceRefs)
2150 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2152 // we have found a complementary dbref on the protein product, so
2153 // update mapping's getTo
2154 newref.getMap().setTo(proteinProduct);
2159 cdsSeq.addDBRef(newref);
2160 propagated.add(newref);
2166 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2167 * feature start/end ranges, optionally omitting specified feature types.
2168 * Returns the number of features copied.
2173 * the mapping from 'fromSeq' to 'toSeq'
2175 * if not null, only features of this type are copied (including
2176 * subtypes in the Sequence Ontology)
2179 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2180 MapList mapping, String select, String... omitting)
2182 SequenceI copyTo = toSeq;
2183 while (copyTo.getDatasetSequence() != null)
2185 copyTo = copyTo.getDatasetSequence();
2187 if (fromSeq == copyTo || fromSeq.getDatasetSequence() == copyTo)
2189 return 0; // shared dataset sequence
2193 * get features, optionally restricted by an ontology term
2195 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2196 .getPositionalFeatures() : fromSeq.getFeatures()
2197 .getFeaturesByOntology(select);
2200 for (SequenceFeature sf : sfs)
2202 String type = sf.getType();
2203 boolean omit = false;
2204 for (String toOmit : omitting)
2206 if (type.equals(toOmit))
2217 * locate the mapped range - null if either start or end is
2218 * not mapped (no partial overlaps are calculated)
2220 int start = sf.getBegin();
2221 int end = sf.getEnd();
2222 int[] mappedTo = mapping.locateInTo(start, end);
2224 * if whole exon range doesn't map, try interpreting it
2225 * as 5' or 3' exon overlapping the CDS range
2227 if (mappedTo == null)
2229 mappedTo = mapping.locateInTo(end, end);
2230 if (mappedTo != null)
2233 * end of exon is in CDS range - 5' overlap
2234 * to a range from the start of the peptide
2239 if (mappedTo == null)
2241 mappedTo = mapping.locateInTo(start, start);
2242 if (mappedTo != null)
2245 * start of exon is in CDS range - 3' overlap
2246 * to a range up to the end of the peptide
2248 mappedTo[1] = toSeq.getLength();
2251 if (mappedTo != null)
2253 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2254 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2255 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2256 sf.getFeatureGroup(), sf.getScore());
2257 copyTo.addSequenceFeature(copy);
2265 * Returns a mapping from dna to protein by inspecting sequence features of
2266 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2267 * length is 3 times the peptide length (optionally after dropping a trailing
2268 * stop codon). This method does not check whether the CDS nucleotide sequence
2269 * translates to the peptide sequence.
2275 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2276 SequenceI proteinSeq)
2278 List<int[]> ranges = findCdsPositions(dnaSeq);
2279 int mappedDnaLength = MappingUtils.getLength(ranges);
2282 * if not a whole number of codons, truncate mapping
2284 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2285 if (codonRemainder > 0)
2287 mappedDnaLength -= codonRemainder;
2288 MappingUtils.removeEndPositions(codonRemainder, ranges);
2291 int proteinLength = proteinSeq.getLength();
2292 int proteinStart = proteinSeq.getStart();
2293 int proteinEnd = proteinSeq.getEnd();
2296 * incomplete start codon may mean X at start of peptide
2297 * we ignore both for mapping purposes
2299 if (proteinSeq.getCharAt(0) == 'X')
2301 // todo JAL-2022 support startPhase > 0
2305 List<int[]> proteinRange = new ArrayList<>();
2308 * dna length should map to protein (or protein plus stop codon)
2310 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2311 if (codesForResidues == (proteinLength + 1))
2313 // assuming extra codon is for STOP and not in peptide
2314 // todo: check trailing codon is indeed a STOP codon
2316 mappedDnaLength -= CODON_LENGTH;
2317 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2320 if (codesForResidues == proteinLength)
2322 proteinRange.add(new int[] { proteinStart, proteinEnd });
2323 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2329 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2330 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2331 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2332 * position order, so this method is only valid for linear CDS in the same
2333 * sense as the protein product.
2338 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2340 List<int[]> result = new ArrayList<>();
2342 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2343 SequenceOntologyI.CDS);
2348 SequenceFeatures.sortFeatures(sfs, true);
2350 for (SequenceFeature sf : sfs)
2355 phase = Integer.parseInt(sf.getPhase());
2356 } catch (NumberFormatException e)
2361 * phase > 0 on first codon means 5' incomplete - skip to the start
2362 * of the next codon; example ENST00000496384
2364 int begin = sf.getBegin();
2365 int end = sf.getEnd();
2366 if (result.isEmpty() && phase > 0)
2371 // shouldn't happen!
2373 .println("Error: start phase extends beyond start CDS in "
2374 + dnaSeq.getName());
2377 result.add(new int[] { begin, end });
2381 * Finally sort ranges by start position. This avoids a dependency on
2382 * keeping features in order on the sequence (if they are in order anyway,
2383 * the sort will have almost no work to do). The implicit assumption is CDS
2384 * ranges are assembled in order. Other cases should not use this method,
2385 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2387 Collections.sort(result, IntRangeComparator.ASCENDING);
2392 * Helper method that adds a peptide variant feature. ID and
2393 * clinical_significance attributes of the dna variant (if present) are copied
2394 * to the new feature.
2401 * the variant codon e.g. aCg
2403 * the 'normal' codon e.g. aTg
2404 * @return true if a feature was added, else false
2406 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2407 String residue, DnaVariant var, String codon, String canonical)
2410 * get peptide translation of codon e.g. GAT -> D
2411 * note that variants which are not single alleles,
2412 * e.g. multibase variants or HGMD_MUTATION etc
2413 * are currently ignored here
2415 String trans = codon.contains("-") ? null
2416 : (codon.length() > CODON_LENGTH ? null
2417 : ResidueProperties.codonTranslate(codon));
2422 String desc = canonical + "/" + codon;
2423 String featureType = "";
2424 if (trans.equals(residue))
2426 featureType = SequenceOntologyI.SYNONYMOUS_VARIANT;
2428 else if (ResidueProperties.STOP.equals(trans))
2430 featureType = SequenceOntologyI.STOP_GAINED;
2434 String residue3Char = StringUtils
2435 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2436 String trans3Char = StringUtils
2437 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2438 desc = "p." + residue3Char + peptidePos + trans3Char;
2439 featureType = SequenceOntologyI.NONSYNONYMOUS_VARIANT;
2441 SequenceFeature sf = new SequenceFeature(featureType, desc, peptidePos,
2442 peptidePos, var.getSource());
2444 StringBuilder attributes = new StringBuilder(32);
2445 String id = (String) var.variant.getValue(VARIANT_ID);
2448 if (id.startsWith(SEQUENCE_VARIANT))
2450 id = id.substring(SEQUENCE_VARIANT.length());
2452 sf.setValue(VARIANT_ID, id);
2453 attributes.append(VARIANT_ID).append("=").append(id);
2454 // TODO handle other species variants JAL-2064
2455 StringBuilder link = new StringBuilder(32);
2458 link.append(desc).append(" ").append(id).append(
2459 "|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2460 .append(URLEncoder.encode(id, "UTF-8"));
2461 sf.addLink(link.toString());
2462 } catch (UnsupportedEncodingException e)
2467 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2468 if (clinSig != null)
2470 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2471 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2474 peptide.addSequenceFeature(sf);
2475 if (attributes.length() > 0)
2477 sf.setAttributes(attributes.toString());
2483 * Makes an alignment with a copy of the given sequences, adding in any
2484 * non-redundant sequences which are mapped to by the cross-referenced
2490 * the alignment dataset shared by the new copy
2493 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2494 SequenceI[] xrefs, AlignmentI dataset)
2496 AlignmentI copy = new Alignment(new Alignment(seqs));
2497 copy.setDataset(dataset);
2498 boolean isProtein = !copy.isNucleotide();
2499 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2502 for (SequenceI xref : xrefs)
2504 DBRefEntry[] dbrefs = xref.getDBRefs();
2507 for (DBRefEntry dbref : dbrefs)
2509 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2510 || dbref.getMap().getTo().isProtein() != isProtein)
2514 SequenceI mappedTo = dbref.getMap().getTo();
2515 SequenceI match = matcher.findIdMatch(mappedTo);
2518 matcher.add(mappedTo);
2519 copy.addSequence(mappedTo);
2529 * Try to align sequences in 'unaligned' to match the alignment of their
2530 * mapped regions in 'aligned'. For example, could use this to align CDS
2531 * sequences which are mapped to their parent cDNA sequences.
2533 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2534 * dna-to-protein or protein-to-dna use alternative methods.
2537 * sequences to be aligned
2539 * holds aligned sequences and their mappings
2542 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2545 * easy case - aligning a copy of aligned sequences
2547 if (alignAsSameSequences(unaligned, aligned))
2549 return unaligned.getHeight();
2553 * fancy case - aligning via mappings between sequences
2555 List<SequenceI> unmapped = new ArrayList<>();
2556 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2557 unaligned, aligned, unmapped);
2558 int width = columnMap.size();
2559 char gap = unaligned.getGapCharacter();
2560 int realignedCount = 0;
2561 // TODO: verify this loop scales sensibly for very wide/high alignments
2563 for (SequenceI seq : unaligned.getSequences())
2565 if (!unmapped.contains(seq))
2567 char[] newSeq = new char[width];
2568 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2569 // Integer iteration below
2574 * traverse the map to find columns populated
2577 for (Integer column : columnMap.keySet())
2579 Character c = columnMap.get(column).get(seq);
2583 * sequence has a character at this position
2593 * trim trailing gaps
2595 if (lastCol < width)
2597 char[] tmp = new char[lastCol + 1];
2598 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2601 // TODO: optimise SequenceI to avoid char[]->String->char[]
2602 seq.setSequence(String.valueOf(newSeq));
2606 return realignedCount;
2610 * If unaligned and aligned sequences share the same dataset sequences, then
2611 * simply copies the aligned sequences to the unaligned sequences and returns
2612 * true; else returns false
2615 * - sequences to be aligned based on aligned
2617 * - 'guide' alignment containing sequences derived from same
2618 * dataset as unaligned
2621 static boolean alignAsSameSequences(AlignmentI unaligned,
2624 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2626 return false; // should only pass alignments with datasets here
2629 // map from dataset sequence to alignment sequence(s)
2630 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2631 for (SequenceI seq : aligned.getSequences())
2633 SequenceI ds = seq.getDatasetSequence();
2634 if (alignedDatasets.get(ds) == null)
2636 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2638 alignedDatasets.get(ds).add(seq);
2642 * first pass - check whether all sequences to be aligned share a
2643 * dataset sequence with an aligned sequence; also note the leftmost
2644 * ungapped column from which to copy
2646 int leftmost = Integer.MAX_VALUE;
2647 for (SequenceI seq : unaligned.getSequences())
2649 final SequenceI ds = seq.getDatasetSequence();
2650 if (!alignedDatasets.containsKey(ds))
2654 SequenceI alignedSeq = alignedDatasets.get(ds)
2656 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2657 leftmost = Math.min(leftmost, startCol);
2661 * second pass - copy aligned sequences;
2662 * heuristic rule: pair off sequences in order for the case where
2663 * more than one shares the same dataset sequence
2665 final char gapCharacter = aligned.getGapCharacter();
2666 for (SequenceI seq : unaligned.getSequences())
2668 List<SequenceI> alignedSequences = alignedDatasets
2669 .get(seq.getDatasetSequence());
2670 SequenceI alignedSeq = alignedSequences.get(0);
2673 * gap fill for leading (5') UTR if any
2675 // TODO this copies intron columns - wrong!
2676 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2677 int endCol = alignedSeq.findIndex(seq.getEnd());
2678 char[] seqchars = new char[endCol - leftmost + 1];
2679 Arrays.fill(seqchars, gapCharacter);
2680 char[] toCopy = alignedSeq.getSequence(startCol - 1, endCol);
2681 System.arraycopy(toCopy, 0, seqchars, startCol - leftmost,
2683 seq.setSequence(String.valueOf(seqchars));
2684 if (alignedSequences.size() > 0)
2686 // pop off aligned sequences (except the last one)
2687 alignedSequences.remove(0);
2692 * finally remove gapped columns (e.g. introns)
2694 new RemoveGapColCommand("", unaligned.getSequencesArray(), 0,
2695 unaligned.getWidth() - 1, unaligned);
2701 * Returns a map whose key is alignment column number (base 1), and whose
2702 * values are a map of sequence characters in that column.
2709 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2710 AlignmentI unaligned, AlignmentI aligned,
2711 List<SequenceI> unmapped)
2714 * Map will hold, for each aligned column position, a map of
2715 * {unalignedSequence, characterPerSequence} at that position.
2716 * TreeMap keeps the entries in ascending column order.
2718 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2721 * record any sequences that have no mapping so can't be realigned
2723 unmapped.addAll(unaligned.getSequences());
2725 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2727 for (SequenceI seq : unaligned.getSequences())
2729 for (AlignedCodonFrame mapping : mappings)
2731 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2732 if (fromSeq != null)
2734 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2735 if (addMappedPositions(seq, fromSeq, seqMap, map))
2737 unmapped.remove(seq);
2746 * Helper method that adds to a map the mapped column positions of a sequence.
2748 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2749 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2753 * the sequence whose column positions we are recording
2755 * a sequence that is mapped to the first sequence
2757 * the mapping from 'fromSeq' to 'seq'
2759 * a map to add the column positions (in fromSeq) of the mapped
2763 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2764 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2772 * invert mapping if it is from unaligned to aligned sequence
2774 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2776 seqMap = new Mapping(seq.getDatasetSequence(),
2777 seqMap.getMap().getInverse());
2780 int toStart = seq.getStart();
2783 * traverse [start, end, start, end...] ranges in fromSeq
2785 for (int[] fromRange : seqMap.getMap().getFromRanges())
2787 for (int i = 0; i < fromRange.length - 1; i += 2)
2789 boolean forward = fromRange[i + 1] >= fromRange[i];
2792 * find the range mapped to (sequence positions base 1)
2794 int[] range = seqMap.locateMappedRange(fromRange[i],
2798 System.err.println("Error in mapping " + seqMap + " from "
2799 + fromSeq.getName());
2802 int fromCol = fromSeq.findIndex(fromRange[i]);
2803 int mappedCharPos = range[0];
2806 * walk over the 'from' aligned sequence in forward or reverse
2807 * direction; when a non-gap is found, record the column position
2808 * of the next character of the mapped-to sequence; stop when all
2809 * the characters of the range have been counted
2811 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
2814 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
2817 * mapped from sequence has a character in this column
2818 * record the column position for the mapped to character
2820 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2821 if (seqsMap == null)
2823 seqsMap = new HashMap<>();
2824 map.put(fromCol, seqsMap);
2826 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
2829 fromCol += (forward ? 1 : -1);
2836 // strictly temporary hack until proper criteria for aligning protein to cds
2837 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2838 public static boolean looksLikeEnsembl(AlignmentI alignment)
2840 for (SequenceI seq : alignment.getSequences())
2842 String name = seq.getName();
2843 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))