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.IncompleteCodonException;
33 import jalview.datamodel.Mapping;
34 import jalview.datamodel.Sequence;
35 import jalview.datamodel.SequenceFeature;
36 import jalview.datamodel.SequenceGroup;
37 import jalview.datamodel.SequenceI;
38 import jalview.datamodel.features.SequenceFeatures;
39 import jalview.io.gff.SequenceOntologyI;
40 import jalview.schemes.ResidueProperties;
41 import jalview.util.Comparison;
42 import jalview.util.DBRefUtils;
43 import jalview.util.IntRangeComparator;
44 import jalview.util.MapList;
45 import jalview.util.MappingUtils;
46 import jalview.util.StringUtils;
48 import java.io.UnsupportedEncodingException;
49 import java.net.URLEncoder;
50 import java.util.ArrayList;
51 import java.util.Arrays;
52 import java.util.Collection;
53 import java.util.Collections;
54 import java.util.HashMap;
55 import java.util.HashSet;
56 import java.util.Iterator;
57 import java.util.LinkedHashMap;
58 import java.util.List;
60 import java.util.Map.Entry;
61 import java.util.NoSuchElementException;
63 import java.util.SortedMap;
64 import java.util.TreeMap;
67 * grab bag of useful alignment manipulation operations Expect these to be
68 * refactored elsewhere at some point.
73 public class AlignmentUtils
76 private static final int CODON_LENGTH = 3;
78 private static final String SEQUENCE_VARIANT = "sequence_variant:";
80 private static final String ID = "ID";
83 * A data model to hold the 'normal' base value at a position, and an optional
84 * sequence variant feature
86 static final class DnaVariant
90 SequenceFeature variant;
92 DnaVariant(String nuc)
98 DnaVariant(String nuc, SequenceFeature var)
104 public String getSource()
106 return variant == null ? null : variant.getFeatureGroup();
111 * given an existing alignment, create a new alignment including all, or up to
112 * flankSize additional symbols from each sequence's dataset sequence
118 public static AlignmentI expandContext(AlignmentI core, int flankSize)
120 List<SequenceI> sq = new ArrayList<SequenceI>();
122 for (SequenceI s : core.getSequences())
124 SequenceI newSeq = s.deriveSequence();
125 final int newSeqStart = newSeq.getStart() - 1;
126 if (newSeqStart > maxoffset
127 && newSeq.getDatasetSequence().getStart() < s.getStart())
129 maxoffset = newSeqStart;
135 maxoffset = Math.min(maxoffset, flankSize);
139 * now add offset left and right to create an expanded alignment
141 for (SequenceI s : sq)
144 while (ds.getDatasetSequence() != null)
146 ds = ds.getDatasetSequence();
148 int s_end = s.findPosition(s.getStart() + s.getLength());
149 // find available flanking residues for sequence
150 int ustream_ds = s.getStart() - ds.getStart();
151 int dstream_ds = ds.getEnd() - s_end;
153 // build new flanked sequence
155 // compute gap padding to start of flanking sequence
156 int offset = maxoffset - ustream_ds;
158 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
161 if (flankSize < ustream_ds)
163 // take up to flankSize residues
164 offset = maxoffset - flankSize;
165 ustream_ds = flankSize;
167 if (flankSize <= dstream_ds)
169 dstream_ds = flankSize - 1;
172 // TODO use Character.toLowerCase to avoid creating String objects?
173 char[] upstream = new String(ds.getSequence(s.getStart() - 1
174 - ustream_ds, s.getStart() - 1)).toLowerCase().toCharArray();
175 char[] downstream = new String(ds.getSequence(s_end - 1, s_end
176 + dstream_ds)).toLowerCase().toCharArray();
177 char[] coreseq = s.getSequence();
178 char[] nseq = new char[offset + upstream.length + downstream.length
180 char c = core.getGapCharacter();
183 for (; p < offset; p++)
188 System.arraycopy(upstream, 0, nseq, p, upstream.length);
189 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
191 System.arraycopy(downstream, 0, nseq, p + coreseq.length
192 + upstream.length, downstream.length);
193 s.setSequence(new String(nseq));
194 s.setStart(s.getStart() - ustream_ds);
195 s.setEnd(s_end + downstream.length);
197 AlignmentI newAl = new jalview.datamodel.Alignment(
198 sq.toArray(new SequenceI[0]));
199 for (SequenceI s : sq)
201 if (s.getAnnotation() != null)
203 for (AlignmentAnnotation aa : s.getAnnotation())
205 aa.adjustForAlignment(); // JAL-1712 fix
206 newAl.addAnnotation(aa);
210 newAl.setDataset(core.getDataset());
215 * Returns the index (zero-based position) of a sequence in an alignment, or
222 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
226 for (SequenceI alSeq : al.getSequences())
239 * Returns a map of lists of sequences in the alignment, keyed by sequence
240 * name. For use in mapping between different alignment views of the same
243 * @see jalview.datamodel.AlignmentI#getSequencesByName()
245 public static Map<String, List<SequenceI>> getSequencesByName(
248 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
249 for (SequenceI seq : al.getSequences())
251 String name = seq.getName();
254 List<SequenceI> seqs = theMap.get(name);
257 seqs = new ArrayList<SequenceI>();
258 theMap.put(name, seqs);
267 * Build mapping of protein to cDNA alignment. Mappings are made between
268 * sequences where the cDNA translates to the protein sequence. Any new
269 * mappings are added to the protein alignment. Returns true if any mappings
270 * either already exist or were added, else false.
272 * @param proteinAlignment
273 * @param cdnaAlignment
276 public static boolean mapProteinAlignmentToCdna(
277 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
279 if (proteinAlignment == null || cdnaAlignment == null)
284 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
285 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
288 * First pass - map sequences where cross-references exist. This include
289 * 1-to-many mappings to support, for example, variant cDNA.
291 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
292 cdnaAlignment, mappedDna, mappedProtein, true);
295 * Second pass - map sequences where no cross-references exist. This only
296 * does 1-to-1 mappings and assumes corresponding sequences are in the same
297 * order in the alignments.
299 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
300 mappedDna, mappedProtein, false);
301 return mappingPerformed;
305 * Make mappings between compatible sequences (where the cDNA translation
306 * matches the protein).
308 * @param proteinAlignment
309 * @param cdnaAlignment
311 * a set of mapped DNA sequences (to add to)
312 * @param mappedProtein
313 * a set of mapped Protein sequences (to add to)
315 * if true, only map sequences where xrefs exist
318 protected static boolean mapProteinToCdna(
319 final AlignmentI proteinAlignment,
320 final AlignmentI cdnaAlignment, Set<SequenceI> mappedDna,
321 Set<SequenceI> mappedProtein, boolean xrefsOnly)
323 boolean mappingExistsOrAdded = false;
324 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
325 for (SequenceI aaSeq : thisSeqs)
327 boolean proteinMapped = false;
328 AlignedCodonFrame acf = new AlignedCodonFrame();
330 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
333 * Always try to map if sequences have xref to each other; this supports
334 * variant cDNA or alternative splicing for a protein sequence.
336 * If no xrefs, try to map progressively, assuming that alignments have
337 * mappable sequences in corresponding order. These are not
338 * many-to-many, as that would risk mixing species with similar cDNA
341 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
347 * Don't map non-xrefd sequences more than once each. This heuristic
348 * allows us to pair up similar sequences in ordered alignments.
351 && (mappedProtein.contains(aaSeq) || mappedDna
356 if (mappingExists(proteinAlignment.getCodonFrames(),
357 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
359 mappingExistsOrAdded = true;
363 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
366 acf.addMap(cdnaSeq, aaSeq, map);
367 mappingExistsOrAdded = true;
368 proteinMapped = true;
369 mappedDna.add(cdnaSeq);
370 mappedProtein.add(aaSeq);
376 proteinAlignment.addCodonFrame(acf);
379 return mappingExistsOrAdded;
383 * Answers true if the mappings include one between the given (dataset)
386 public static boolean mappingExists(List<AlignedCodonFrame> mappings,
387 SequenceI aaSeq, SequenceI cdnaSeq)
389 if (mappings != null)
391 for (AlignedCodonFrame acf : mappings)
393 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
403 * Builds a mapping (if possible) of a cDNA to a protein sequence.
405 * <li>first checks if the cdna translates exactly to the protein sequence</li>
406 * <li>else checks for translation after removing a STOP codon</li>
407 * <li>else checks for translation after removing a START codon</li>
408 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
410 * Returns null if no mapping is determined.
413 * the aligned protein sequence
415 * the aligned cdna sequence
418 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
422 * Here we handle either dataset sequence set (desktop) or absent (applet).
423 * Use only the char[] form of the sequence to avoid creating possibly large
426 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
427 char[] aaSeqChars = proteinDataset != null ? proteinDataset
428 .getSequence() : proteinSeq.getSequence();
429 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
430 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
431 : cdnaSeq.getSequence();
432 if (aaSeqChars == null || cdnaSeqChars == null)
438 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
440 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
441 int cdnaLength = cdnaSeqChars.length;
442 int cdnaStart = cdnaSeq.getStart();
443 int cdnaEnd = cdnaSeq.getEnd();
444 final int proteinStart = proteinSeq.getStart();
445 final int proteinEnd = proteinSeq.getEnd();
448 * If lengths don't match, try ignoring stop codon (if present)
450 if (cdnaLength != mappedLength && cdnaLength > 2)
452 String lastCodon = String.valueOf(cdnaSeqChars,
453 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
454 for (String stop : ResidueProperties.STOP)
456 if (lastCodon.equals(stop))
458 cdnaEnd -= CODON_LENGTH;
459 cdnaLength -= CODON_LENGTH;
466 * If lengths still don't match, try ignoring start codon.
469 if (cdnaLength != mappedLength
471 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
472 .equals(ResidueProperties.START))
474 startOffset += CODON_LENGTH;
475 cdnaStart += CODON_LENGTH;
476 cdnaLength -= CODON_LENGTH;
479 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
482 * protein is translation of dna (+/- start/stop codons)
484 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd }, new int[]
485 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
490 * translation failed - try mapping CDS annotated regions of dna
492 return mapCdsToProtein(cdnaSeq, proteinSeq);
496 * Test whether the given cdna sequence, starting at the given offset,
497 * translates to the given amino acid sequence, using the standard translation
498 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
500 * @param cdnaSeqChars
505 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
508 if (cdnaSeqChars == null || aaSeqChars == null)
514 int dnaPos = cdnaStart;
515 for (; dnaPos < cdnaSeqChars.length - 2 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
517 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
518 final String translated = ResidueProperties.codonTranslate(codon);
521 * allow * in protein to match untranslatable in dna
523 final char aaRes = aaSeqChars[aaPos];
524 if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
528 if (translated == null || !(aaRes == translated.charAt(0)))
531 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
532 // + codon + "(" + translated + ") != " + aaRes));
538 * check we matched all of the protein sequence
540 if (aaPos != aaSeqChars.length)
546 * check we matched all of the dna except
547 * for optional trailing STOP codon
549 if (dnaPos == cdnaSeqChars.length)
553 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
555 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
556 if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
565 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
566 * currently assumes that we are aligning cDNA to match protein.
569 * the sequence to be realigned
571 * the alignment whose sequence alignment is to be 'copied'
573 * character string represent a gap in the realigned sequence
574 * @param preserveUnmappedGaps
575 * @param preserveMappedGaps
576 * @return true if the sequence was realigned, false if it could not be
578 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
579 String gap, boolean preserveMappedGaps,
580 boolean preserveUnmappedGaps)
583 * Get any mappings from the source alignment to the target (dataset)
586 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
587 // all mappings. Would it help to constrain this?
588 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
589 if (mappings == null || mappings.isEmpty())
595 * Locate the aligned source sequence whose dataset sequence is mapped. We
596 * just take the first match here (as we can't align like more than one
599 SequenceI alignFrom = null;
600 AlignedCodonFrame mapping = null;
601 for (AlignedCodonFrame mp : mappings)
603 alignFrom = mp.findAlignedSequence(seq, al);
604 if (alignFrom != null)
611 if (alignFrom == null)
615 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
616 preserveMappedGaps, preserveUnmappedGaps);
621 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
622 * match residues and codons. Flags control whether existing gaps in unmapped
623 * (intron) and mapped (exon) regions are preserved or not. Gaps between
624 * intron and exon are only retained if both flags are set.
631 * @param preserveUnmappedGaps
632 * @param preserveMappedGaps
634 public static void alignSequenceAs(SequenceI alignTo,
635 SequenceI alignFrom, AlignedCodonFrame mapping, String myGap,
636 char sourceGap, boolean preserveMappedGaps,
637 boolean preserveUnmappedGaps)
639 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
641 // aligned and dataset sequence positions, all base zero
645 int basesWritten = 0;
646 char myGapChar = myGap.charAt(0);
647 int ratio = myGap.length();
649 int fromOffset = alignFrom.getStart() - 1;
650 int toOffset = alignTo.getStart() - 1;
651 int sourceGapMappedLength = 0;
652 boolean inExon = false;
653 final int toLength = alignTo.getLength();
654 final int fromLength = alignFrom.getLength();
655 StringBuilder thisAligned = new StringBuilder(2 * toLength);
658 * Traverse the 'model' aligned sequence
660 for (int i = 0; i < fromLength; i++)
662 char sourceChar = alignFrom.getCharAt(i);
663 if (sourceChar == sourceGap)
665 sourceGapMappedLength += ratio;
670 * Found a non-gap character. Locate its mapped region if any.
673 // Note mapping positions are base 1, our sequence positions base 0
674 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
675 sourceDsPos + fromOffset);
676 if (mappedPos == null)
679 * unmapped position; treat like a gap
681 sourceGapMappedLength += ratio;
682 // System.err.println("Can't align: no codon mapping to residue "
683 // + sourceDsPos + "(" + sourceChar + ")");
688 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
689 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
690 StringBuilder trailingCopiedGap = new StringBuilder();
693 * Copy dna sequence up to and including this codon. Optionally, include
694 * gaps before the codon starts (in introns) and/or after the codon starts
697 * Note this only works for 'linear' splicing, not reverse or interleaved.
698 * But then 'align dna as protein' doesn't make much sense otherwise.
700 int intronLength = 0;
701 while (basesWritten + toOffset < mappedCodonEnd
702 && thisSeqPos < toLength)
704 final char c = alignTo.getCharAt(thisSeqPos++);
708 int sourcePosition = basesWritten + toOffset;
709 if (sourcePosition < mappedCodonStart)
712 * Found an unmapped (intron) base. First add in any preceding gaps
715 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
717 thisAligned.append(trailingCopiedGap.toString());
718 intronLength += trailingCopiedGap.length();
719 trailingCopiedGap = new StringBuilder();
726 final boolean startOfCodon = sourcePosition == mappedCodonStart;
727 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
728 preserveUnmappedGaps, sourceGapMappedLength, inExon,
729 trailingCopiedGap.length(), intronLength, startOfCodon);
730 for (int k = 0; k < gapsToAdd; k++)
732 thisAligned.append(myGapChar);
734 sourceGapMappedLength = 0;
737 thisAligned.append(c);
738 trailingCopiedGap = new StringBuilder();
742 if (inExon && preserveMappedGaps)
744 trailingCopiedGap.append(myGapChar);
746 else if (!inExon && preserveUnmappedGaps)
748 trailingCopiedGap.append(myGapChar);
755 * At end of model aligned sequence. Copy any remaining target sequence, optionally
756 * including (intron) gaps.
758 while (thisSeqPos < toLength)
760 final char c = alignTo.getCharAt(thisSeqPos++);
761 if (c != myGapChar || preserveUnmappedGaps)
763 thisAligned.append(c);
765 sourceGapMappedLength--;
769 * finally add gaps to pad for any trailing source gaps or
770 * unmapped characters
772 if (preserveUnmappedGaps)
774 while (sourceGapMappedLength > 0)
776 thisAligned.append(myGapChar);
777 sourceGapMappedLength--;
782 * All done aligning, set the aligned sequence.
784 alignTo.setSequence(new String(thisAligned));
788 * Helper method to work out how many gaps to insert when realigning.
790 * @param preserveMappedGaps
791 * @param preserveUnmappedGaps
792 * @param sourceGapMappedLength
794 * @param trailingCopiedGap
795 * @param intronLength
796 * @param startOfCodon
799 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
800 boolean preserveUnmappedGaps, int sourceGapMappedLength,
801 boolean inExon, int trailingGapLength, int intronLength,
802 final boolean startOfCodon)
808 * Reached start of codon. Ignore trailing gaps in intron unless we are
809 * preserving gaps in both exon and intron. Ignore them anyway if the
810 * protein alignment introduces a gap at least as large as the intronic
813 if (inExon && !preserveMappedGaps)
815 trailingGapLength = 0;
817 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
819 trailingGapLength = 0;
823 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
827 if (intronLength + trailingGapLength <= sourceGapMappedLength)
829 gapsToAdd = sourceGapMappedLength - intronLength;
833 gapsToAdd = Math.min(intronLength + trailingGapLength
834 - sourceGapMappedLength, trailingGapLength);
841 * second or third base of codon; check for any gaps in dna
843 if (!preserveMappedGaps)
845 trailingGapLength = 0;
847 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
853 * Realigns the given protein to match the alignment of the dna, using codon
854 * mappings to translate aligned codon positions to protein residues.
857 * the alignment whose sequences are realigned by this method
859 * the dna alignment whose alignment we are 'copying'
860 * @return the number of sequences that were realigned
862 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
864 if (protein.isNucleotide() || !dna.isNucleotide())
866 System.err.println("Wrong alignment type in alignProteinAsDna");
869 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
870 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
871 protein, dna, unmappedProtein);
872 return alignProteinAs(protein, alignedCodons, unmappedProtein);
876 * Realigns the given dna to match the alignment of the protein, using codon
877 * mappings to translate aligned peptide positions to codons.
879 * Always produces a padded CDS alignment.
882 * the alignment whose sequences are realigned by this method
884 * the protein alignment whose alignment we are 'copying'
885 * @return the number of sequences that were realigned
887 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
889 if (protein.isNucleotide() || !dna.isNucleotide())
891 System.err.println("Wrong alignment type in alignProteinAsDna");
894 // todo: implement this
895 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
896 int alignedCount = 0;
897 int width = 0; // alignment width for padding CDS
898 for (SequenceI dnaSeq : dna.getSequences())
900 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
901 dna.getGapCharacter()))
905 width = Math.max(dnaSeq.getLength(), width);
909 for (SequenceI dnaSeq : dna.getSequences())
911 oldwidth = dnaSeq.getLength();
912 diff = width - oldwidth;
915 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
922 * Helper method to align (if possible) the dna sequence to match the
923 * alignment of a mapped protein sequence. This is currently limited to
924 * handling coding sequence only.
932 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
933 AlignmentI protein, List<AlignedCodonFrame> mappings, char gapChar)
935 SequenceI cdsDss = cdsSeq.getDatasetSequence();
939 .println("alignCdsSequenceAsProtein needs aligned sequence!");
943 List<AlignedCodonFrame> dnaMappings = MappingUtils
944 .findMappingsForSequence(cdsSeq, mappings);
945 for (AlignedCodonFrame mapping : dnaMappings)
947 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
950 final int peptideLength = peptide.getLength();
951 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
954 MapList mapList = map.getMap();
955 if (map.getTo() == peptide.getDatasetSequence())
957 mapList = mapList.getInverse();
959 final int cdsLength = cdsDss.getLength();
960 int mappedFromLength = MappingUtils.getLength(mapList
962 int mappedToLength = MappingUtils
963 .getLength(mapList.getToRanges());
964 boolean addStopCodon = (cdsLength == mappedFromLength
965 * CODON_LENGTH + CODON_LENGTH)
966 || (peptide.getDatasetSequence().getLength() == mappedFromLength - 1);
967 if (cdsLength != mappedToLength && !addStopCodon)
971 .format("Can't align cds as protein (length mismatch %d/%d): %s",
972 cdsLength, mappedToLength,
977 * pre-fill the aligned cds sequence with gaps
979 char[] alignedCds = new char[peptideLength * CODON_LENGTH
980 + (addStopCodon ? CODON_LENGTH : 0)];
981 Arrays.fill(alignedCds, gapChar);
984 * walk over the aligned peptide sequence and insert mapped
985 * codons for residues in the aligned cds sequence
988 int cdsStart = cdsDss.getStart();
989 int proteinPos = peptide.getStart() - 1;
992 for (int col = 0; col < peptideLength; col++)
994 char residue = peptide.getCharAt(col);
996 if (Comparison.isGap(residue))
998 cdsCol += CODON_LENGTH;
1003 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1006 // e.g. incomplete start codon, X in peptide
1007 cdsCol += CODON_LENGTH;
1011 for (int j = codon[0]; j <= codon[1]; j++)
1013 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1014 alignedCds[cdsCol++] = mappedBase;
1022 * append stop codon if not mapped from protein,
1023 * closing it up to the end of the mapped sequence
1025 if (copiedBases == cdsLength - CODON_LENGTH)
1027 for (int i = alignedCds.length - 1; i >= 0; i--)
1029 if (!Comparison.isGap(alignedCds[i]))
1031 cdsCol = i + 1; // gap just after end of sequence
1035 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1037 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1040 cdsSeq.setSequence(new String(alignedCds));
1049 * Builds a map whose key is an aligned codon position (3 alignment column
1050 * numbers base 0), and whose value is a map from protein sequence to each
1051 * protein's peptide residue for that codon. The map generates an ordering of
1052 * the codons, and allows us to read off the peptides at each position in
1053 * order to assemble 'aligned' protein sequences.
1056 * the protein alignment
1058 * the coding dna alignment
1059 * @param unmappedProtein
1060 * any unmapped proteins are added to this list
1063 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1064 AlignmentI protein, AlignmentI dna,
1065 List<SequenceI> unmappedProtein)
1068 * maintain a list of any proteins with no mappings - these will be
1069 * rendered 'as is' in the protein alignment as we can't align them
1071 unmappedProtein.addAll(protein.getSequences());
1073 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1076 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1077 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1078 * comparator keeps the codon positions ordered.
1080 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
1081 new CodonComparator());
1083 for (SequenceI dnaSeq : dna.getSequences())
1085 for (AlignedCodonFrame mapping : mappings)
1087 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1090 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1091 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
1092 seqMap, alignedCodons);
1093 unmappedProtein.remove(prot);
1099 * Finally add any unmapped peptide start residues (e.g. for incomplete
1100 * codons) as if at the codon position before the second residue
1102 // TODO resolve JAL-2022 so this fudge can be removed
1103 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1104 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1106 return alignedCodons;
1110 * Scans for any protein mapped from position 2 (meaning unmapped start
1111 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1112 * preceding position in the alignment
1114 * @param alignedCodons
1115 * the codon-to-peptide map
1116 * @param mappedSequenceCount
1117 * the number of distinct sequences in the map
1119 protected static void addUnmappedPeptideStarts(
1120 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1121 int mappedSequenceCount)
1123 // TODO delete this ugly hack once JAL-2022 is resolved
1124 // i.e. we can model startPhase > 0 (incomplete start codon)
1126 List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
1127 AlignedCodon lastCodon = null;
1128 Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
1130 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1133 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1136 SequenceI seq = sequenceCodon.getKey();
1137 if (sequencesChecked.contains(seq))
1141 sequencesChecked.add(seq);
1142 AlignedCodon codon = sequenceCodon.getValue();
1143 if (codon.peptideCol > 1)
1146 .println("Problem mapping protein with >1 unmapped start positions: "
1149 else if (codon.peptideCol == 1)
1152 * first position (peptideCol == 0) was unmapped - add it
1154 if (lastCodon != null)
1156 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1157 lastCodon.pos2, lastCodon.pos3, String.valueOf(seq
1159 toAdd.put(seq, firstPeptide);
1164 * unmapped residue at start of alignment (no prior column) -
1165 * 'insert' at nominal codon [0, 0, 0]
1167 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1168 String.valueOf(seq.getCharAt(0)), 0);
1169 toAdd.put(seq, firstPeptide);
1172 if (sequencesChecked.size() == mappedSequenceCount)
1174 // no need to check past first mapped position in all sequences
1178 lastCodon = entry.getKey();
1182 * add any new codons safely after iterating over the map
1184 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1186 addCodonToMap(alignedCodons, startCodon.getValue(),
1187 startCodon.getKey());
1192 * Update the aligned protein sequences to match the codon alignments given in
1196 * @param alignedCodons
1197 * an ordered map of codon positions (columns), with sequence/peptide
1198 * values present in each column
1199 * @param unmappedProtein
1202 protected static int alignProteinAs(AlignmentI protein,
1203 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1204 List<SequenceI> unmappedProtein)
1207 * prefill peptide sequences with gaps
1209 int alignedWidth = alignedCodons.size();
1210 char[] gaps = new char[alignedWidth];
1211 Arrays.fill(gaps, protein.getGapCharacter());
1212 Map<SequenceI, char[]> peptides = new HashMap<>();
1213 for (SequenceI seq : protein.getSequences())
1215 if (!unmappedProtein.contains(seq))
1217 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1222 * Traverse the codons left to right (as defined by CodonComparator)
1223 * and insert peptides in each column where the sequence is mapped.
1224 * This gives a peptide 'alignment' where residues are aligned if their
1225 * corresponding codons occupy the same columns in the cdna alignment.
1228 for (AlignedCodon codon : alignedCodons.keySet())
1230 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1232 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1234 char residue = entry.getValue().product.charAt(0);
1235 peptides.get(entry.getKey())[column] = residue;
1241 * and finally set the constructed sequences
1243 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1245 entry.getKey().setSequence(new String(entry.getValue()));
1252 * Populate the map of aligned codons by traversing the given sequence
1253 * mapping, locating the aligned positions of mapped codons, and adding those
1254 * positions and their translation products to the map.
1257 * the aligned sequence we are mapping from
1259 * the sequence to be aligned to the codons
1261 * the gap character in the dna sequence
1263 * a mapping to a sequence translation
1264 * @param alignedCodons
1265 * the map we are building up
1267 static void addCodonPositions(SequenceI dna, SequenceI protein,
1268 char gapChar, Mapping seqMap,
1269 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1271 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1274 * add codon positions, and their peptide translations, to the alignment
1275 * map, while remembering the first codon mapped
1277 while (codons.hasNext())
1281 AlignedCodon codon = codons.next();
1282 addCodonToMap(alignedCodons, codon, protein);
1283 } catch (IncompleteCodonException e)
1285 // possible incomplete trailing codon - ignore
1286 } catch (NoSuchElementException e)
1288 // possibly peptide lacking STOP
1294 * Helper method to add a codon-to-peptide entry to the aligned codons map
1296 * @param alignedCodons
1300 protected static void addCodonToMap(
1301 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1302 AlignedCodon codon, SequenceI protein)
1304 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1305 if (seqProduct == null)
1307 seqProduct = new HashMap<SequenceI, AlignedCodon>();
1308 alignedCodons.put(codon, seqProduct);
1310 seqProduct.put(protein, codon);
1314 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1315 * between at least one pair of sequences in the two alignments. Currently,
1318 * <li>One alignment must be nucleotide, and the other protein</li>
1319 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1320 * <li>Mappable means the nucleotide translation matches the protein sequence</li>
1321 * <li>The translation may ignore start and stop codons if present in the
1329 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1331 if (al1 == null || al2 == null)
1337 * Require one nucleotide and one protein
1339 if (al1.isNucleotide() == al2.isNucleotide())
1343 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1344 AlignmentI protein = dna == al1 ? al2 : al1;
1345 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1346 for (SequenceI dnaSeq : dna.getSequences())
1348 for (SequenceI proteinSeq : protein.getSequences())
1350 if (isMappable(dnaSeq, proteinSeq, mappings))
1360 * Returns true if the dna sequence is mapped, or could be mapped, to the
1368 protected static boolean isMappable(SequenceI dnaSeq,
1369 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1371 if (dnaSeq == null || proteinSeq == null)
1376 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq : dnaSeq
1377 .getDatasetSequence();
1378 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null ? proteinSeq
1379 : proteinSeq.getDatasetSequence();
1381 for (AlignedCodonFrame mapping : mappings)
1383 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1393 * Just try to make a mapping (it is not yet stored), test whether
1396 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1400 * Finds any reference annotations associated with the sequences in
1401 * sequenceScope, that are not already added to the alignment, and adds them
1402 * to the 'candidates' map. Also populates a lookup table of annotation
1403 * labels, keyed by calcId, for use in constructing tooltips or the like.
1405 * @param sequenceScope
1406 * the sequences to scan for reference annotations
1407 * @param labelForCalcId
1408 * (optional) map to populate with label for calcId
1410 * map to populate with annotations for sequence
1412 * the alignment to check for presence of annotations
1414 public static void findAddableReferenceAnnotations(
1415 List<SequenceI> sequenceScope,
1416 Map<String, String> labelForCalcId,
1417 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1420 if (sequenceScope == null)
1426 * For each sequence in scope, make a list of any annotations on the
1427 * underlying dataset sequence which are not already on the alignment.
1429 * Add to a map of { alignmentSequence, <List of annotations to add> }
1431 for (SequenceI seq : sequenceScope)
1433 SequenceI dataset = seq.getDatasetSequence();
1434 if (dataset == null)
1438 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1439 if (datasetAnnotations == null)
1443 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1444 for (AlignmentAnnotation dsann : datasetAnnotations)
1447 * Find matching annotations on the alignment. If none is found, then
1448 * add this annotation to the list of 'addable' annotations for this
1451 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1452 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1453 if (!matchedAlignmentAnnotations.iterator().hasNext())
1456 if (labelForCalcId != null)
1458 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1463 * Save any addable annotations for this sequence
1465 if (!result.isEmpty())
1467 candidates.put(seq, result);
1473 * Adds annotations to the top of the alignment annotations, in the same order
1474 * as their related sequences.
1476 * @param annotations
1477 * the annotations to add
1479 * the alignment to add them to
1480 * @param selectionGroup
1481 * current selection group (or null if none)
1483 public static void addReferenceAnnotations(
1484 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1485 final AlignmentI alignment, final SequenceGroup selectionGroup)
1487 for (SequenceI seq : annotations.keySet())
1489 for (AlignmentAnnotation ann : annotations.get(seq))
1491 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1493 int endRes = ann.annotations.length;
1494 if (selectionGroup != null)
1496 startRes = selectionGroup.getStartRes();
1497 endRes = selectionGroup.getEndRes();
1499 copyAnn.restrict(startRes, endRes);
1502 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1503 * original annotation is already on the sequence.
1505 if (!seq.hasAnnotation(ann))
1507 seq.addAlignmentAnnotation(copyAnn);
1510 copyAnn.adjustForAlignment();
1511 // add to the alignment and set visible
1512 alignment.addAnnotation(copyAnn);
1513 copyAnn.visible = true;
1519 * Set visibility of alignment annotations of specified types (labels), for
1520 * specified sequences. This supports controls like
1521 * "Show all secondary structure", "Hide all Temp factor", etc.
1523 * @al the alignment to scan for annotations
1525 * the types (labels) of annotations to be updated
1526 * @param forSequences
1527 * if not null, only annotations linked to one of these sequences are
1528 * in scope for update; if null, acts on all sequence annotations
1530 * if this flag is true, 'types' is ignored (label not checked)
1532 * if true, set visibility on, else set off
1534 public static void showOrHideSequenceAnnotations(AlignmentI al,
1535 Collection<String> types, List<SequenceI> forSequences,
1536 boolean anyType, boolean doShow)
1538 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1541 for (AlignmentAnnotation aa : anns)
1543 if (anyType || types.contains(aa.label))
1545 if ((aa.sequenceRef != null)
1546 && (forSequences == null || forSequences
1547 .contains(aa.sequenceRef)))
1549 aa.visible = doShow;
1557 * Returns true if either sequence has a cross-reference to the other
1563 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1565 // Note: moved here from class CrossRef as the latter class has dependencies
1566 // not availability to the applet's classpath
1567 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1571 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1572 * that sequence name is structured as Source|AccessionId.
1578 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1580 if (seq1 == null || seq2 == null)
1584 String name = seq2.getName();
1585 final DBRefEntry[] xrefs = seq1.getDBRefs();
1588 for (DBRefEntry xref : xrefs)
1590 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1591 // case-insensitive test, consistent with DBRefEntry.equalRef()
1592 if (xrefName.equalsIgnoreCase(name))
1602 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1603 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1604 * added to the original alignment's dataset, which is shared by the new
1605 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1606 * added to the alignment dataset.
1609 * aligned nucleotide (dna or cds) sequences
1611 * the alignment dataset the sequences belong to
1613 * (optional) to restrict results to CDS that map to specified
1615 * @return an alignment whose sequences are the cds-only parts of the dna
1616 * sequences (or null if no mappings are found)
1618 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1619 AlignmentI dataset, SequenceI[] products)
1621 if (dataset == null || dataset.getDataset() != null)
1623 throw new IllegalArgumentException(
1624 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1626 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
1627 List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
1628 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1629 HashSet<SequenceI> productSeqs = null;
1630 if (products != null)
1632 productSeqs = new HashSet<SequenceI>();
1633 for (SequenceI seq : products)
1635 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1636 .getDatasetSequence());
1641 * Construct CDS sequences from mappings on the alignment dataset.
1643 * - find the protein product(s) mapped to from each dna sequence
1644 * - if the mapping covers the whole dna sequence (give or take start/stop
1645 * codon), take the dna as the CDS sequence
1646 * - else search dataset mappings for a suitable dna sequence, i.e. one
1647 * whose whole sequence is mapped to the protein
1648 * - if no sequence found, construct one from the dna sequence and mapping
1649 * (and add it to dataset so it is found if this is repeated)
1651 for (SequenceI dnaSeq : dna)
1653 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1654 : dnaSeq.getDatasetSequence();
1656 List<AlignedCodonFrame> seqMappings = MappingUtils
1657 .findMappingsForSequence(dnaSeq, mappings);
1658 for (AlignedCodonFrame mapping : seqMappings)
1660 List<Mapping> mappingsFromSequence = mapping
1661 .getMappingsFromSequence(dnaSeq);
1663 for (Mapping aMapping : mappingsFromSequence)
1665 MapList mapList = aMapping.getMap();
1666 if (mapList.getFromRatio() == 1)
1669 * not a dna-to-protein mapping (likely dna-to-cds)
1675 * skip if mapping is not to one of the target set of proteins
1677 SequenceI proteinProduct = aMapping.getTo();
1678 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1684 * try to locate the CDS from the dataset mappings;
1685 * guard against duplicate results (for the case that protein has
1686 * dbrefs to both dna and cds sequences)
1688 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1689 seqMappings, aMapping);
1692 if (!foundSeqs.contains(cdsSeq))
1694 foundSeqs.add(cdsSeq);
1695 SequenceI derivedSequence = cdsSeq.deriveSequence();
1696 cdsSeqs.add(derivedSequence);
1697 if (!dataset.getSequences().contains(cdsSeq))
1699 dataset.addSequence(cdsSeq);
1706 * didn't find mapped CDS sequence - construct it and add
1707 * its dataset sequence to the dataset
1709 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1710 dataset).deriveSequence();
1711 // cdsSeq has a name constructed as CDS|<dbref>
1712 // <dbref> will be either the accession for the coding sequence,
1713 // marked in the /via/ dbref to the protein product accession
1714 // or it will be the original nucleotide accession.
1715 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1717 cdsSeqs.add(cdsSeq);
1719 if (!dataset.getSequences().contains(cdsSeqDss))
1721 // check if this sequence is a newly created one
1722 // so needs adding to the dataset
1723 dataset.addSequence(cdsSeqDss);
1727 * add a mapping from CDS to the (unchanged) mapped to range
1729 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1730 cdsSeq.getLength() });
1731 MapList cdsToProteinMap = new MapList(cdsRange,
1732 mapList.getToRanges(), mapList.getFromRatio(),
1733 mapList.getToRatio());
1734 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1735 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1739 * guard against duplicating the mapping if repeating this action
1741 if (!mappings.contains(cdsToProteinMapping))
1743 mappings.add(cdsToProteinMapping);
1746 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1747 proteinProduct, aMapping);
1749 * add another mapping from original 'from' range to CDS
1751 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1752 MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1754 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1756 if (!mappings.contains(dnaToCdsMapping))
1758 mappings.add(dnaToCdsMapping);
1762 * add DBRef with mapping from protein to CDS
1763 * (this enables Get Cross-References from protein alignment)
1764 * This is tricky because we can't have two DBRefs with the
1765 * same source and accession, so need a different accession for
1766 * the CDS from the dna sequence
1769 // specific use case:
1770 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1771 // ENSG02, ENSG03, with transcripts and products similarly named.
1772 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1774 // JBPNote: ?? can't actually create an example that demonstrates we
1776 // synthesize an xref.
1778 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1780 // creates a complementary cross-reference to the source sequence's
1781 // primary reference.
1783 DBRefEntry cdsCrossRef = new DBRefEntry(primRef.getSource(),
1784 primRef.getSource() + ":" + primRef.getVersion(),
1785 primRef.getAccessionId());
1787 .setMap(new Mapping(dnaDss, new MapList(dnaToCdsMap)));
1788 cdsSeqDss.addDBRef(cdsCrossRef);
1790 // problem here is that the cross-reference is synthesized -
1791 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1793 // assuming cds version same as dna ?!?
1795 DBRefEntry proteinToCdsRef = new DBRefEntry(
1796 primRef.getSource(), primRef.getVersion(),
1799 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1801 proteinProduct.addDBRef(proteinToCdsRef);
1805 * transfer any features on dna that overlap the CDS
1807 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1808 SequenceOntologyI.CDS);
1813 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1815 cds.setDataset(dataset);
1821 * A helper method that finds a CDS sequence in the alignment dataset that is
1822 * mapped to the given protein sequence, and either is, or has a mapping from,
1823 * the given dna sequence.
1826 * set of all mappings on the dataset
1828 * a dna (or cds) sequence we are searching from
1829 * @param seqMappings
1830 * the set of mappings involving dnaSeq
1832 * an initial candidate from seqMappings
1835 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1836 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1840 * TODO a better dna-cds-protein mapping data representation to allow easy
1841 * navigation; until then this clunky looping around lists of mappings
1843 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1844 : dnaSeq.getDatasetSequence();
1845 SequenceI proteinProduct = aMapping.getTo();
1848 * is this mapping from the whole dna sequence (i.e. CDS)?
1849 * allowing for possible stop codon on dna but not peptide
1851 int mappedFromLength = MappingUtils.getLength(aMapping.getMap()
1853 int dnaLength = seqDss.getLength();
1854 if (mappedFromLength == dnaLength
1855 || mappedFromLength == dnaLength - CODON_LENGTH)
1861 * looks like we found the dna-to-protein mapping; search for the
1862 * corresponding cds-to-protein mapping
1864 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1865 .findMappingsForSequence(proteinProduct, mappings);
1866 for (AlignedCodonFrame acf : mappingsToPeptide)
1868 for (SequenceToSequenceMapping map : acf.getMappings())
1870 Mapping mapping = map.getMapping();
1871 if (mapping != aMapping
1872 && mapping.getMap().getFromRatio() == CODON_LENGTH
1873 && proteinProduct == mapping.getTo()
1874 && seqDss != map.getFromSeq())
1876 mappedFromLength = MappingUtils.getLength(mapping.getMap()
1878 if (mappedFromLength == map.getFromSeq().getLength())
1881 * found a 3:1 mapping to the protein product which covers
1882 * the whole dna sequence i.e. is from CDS; finally check it
1883 * is from the dna start sequence
1885 SequenceI cdsSeq = map.getFromSeq();
1886 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1887 .findMappingsForSequence(cdsSeq, seqMappings);
1888 if (!dnaToCdsMaps.isEmpty())
1900 * Helper method that makes a CDS sequence as defined by the mappings from the
1901 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1902 * forward or reverse strand).
1907 * - existing dataset. We check for sequences that look like the CDS
1908 * we are about to construct, if one exists already, then we will
1909 * just return that one.
1910 * @return CDS sequence (as a dataset sequence)
1912 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1915 char[] seqChars = seq.getSequence();
1916 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1917 int cdsWidth = MappingUtils.getLength(fromRanges);
1918 char[] newSeqChars = new char[cdsWidth];
1921 for (int[] range : fromRanges)
1923 if (range[0] <= range[1])
1925 // forward strand mapping - just copy the range
1926 int length = range[1] - range[0] + 1;
1927 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1933 // reverse strand mapping - copy and complement one by one
1934 for (int i = range[0]; i >= range[1]; i--)
1936 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1942 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
1943 * else generate a sequence name
1945 String mapFromId = mapping.getMappedFromId();
1946 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
1947 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
1948 if (dataset != null)
1950 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
1951 if (matches != null)
1953 boolean matched = false;
1954 for (SequenceI mtch : matches)
1956 if (mtch.getStart() != newSeq.getStart())
1960 if (mtch.getEnd() != newSeq.getEnd())
1964 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
1976 .println("JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
1982 // newSeq.setDescription(mapFromId);
1988 * add any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
1989 * the given mapping.
1994 * @return list of DBRefEntrys added.
1996 public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
1997 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2000 // gather direct refs from contig congrent with mapping
2001 List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
2002 HashSet<String> directSources = new HashSet<String>();
2003 if (contig.getDBRefs() != null)
2005 for (DBRefEntry dbr : contig.getDBRefs())
2007 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
2009 MapList map = dbr.getMap().getMap();
2010 // check if map is the CDS mapping
2011 if (mapping.getMap().equals(map))
2014 directSources.add(dbr.getSource());
2019 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2020 proteinProduct.getDBRefs(),
2021 directSources.toArray(new String[0]));
2022 List<DBRefEntry> propagated = new ArrayList<DBRefEntry>();
2024 // and generate appropriate mappings
2025 for (DBRefEntry cdsref : direct)
2027 // clone maplist and mapping
2028 MapList cdsposmap = new MapList(Arrays.asList(new int[][] { new int[]
2029 { cdsSeq.getStart(), cdsSeq.getEnd() } }), cdsref.getMap().getMap()
2030 .getToRanges(), 3, 1);
2031 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(), cdsref.getMap()
2035 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2036 cdsref.getVersion(), cdsref.getAccessionId(), new Mapping(
2037 cdsmap.getTo(), cdsposmap));
2039 // and see if we can map to the protein product for this mapping.
2040 // onSource is the filtered set of accessions on protein that we are
2041 // tranferring, so we assume accession is the same.
2042 if (cdsmap.getTo() == null && onSource != null)
2044 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2045 cdsref.getAccessionId());
2046 if (sourceRefs != null)
2048 for (DBRefEntry srcref : sourceRefs)
2050 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2052 // we have found a complementary dbref on the protein product, so
2053 // update mapping's getTo
2054 newref.getMap().setTo(proteinProduct);
2059 cdsSeq.addDBRef(newref);
2060 propagated.add(newref);
2066 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2067 * feature start/end ranges, optionally omitting specified feature types.
2068 * Returns the number of features copied.
2073 * the mapping from 'fromSeq' to 'toSeq'
2075 * if not null, only features of this type are copied (including
2076 * subtypes in the Sequence Ontology)
2079 public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2080 MapList mapping, String select, String... omitting)
2082 SequenceI copyTo = toSeq;
2083 while (copyTo.getDatasetSequence() != null)
2085 copyTo = copyTo.getDatasetSequence();
2089 * get features, optionally restricted by an ontology term
2091 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2092 .getPositionalFeatures() : fromSeq.getFeatures()
2093 .getFeaturesByOntology(select);
2096 for (SequenceFeature sf : sfs)
2098 String type = sf.getType();
2099 boolean omit = false;
2100 for (String toOmit : omitting)
2102 if (type.equals(toOmit))
2113 * locate the mapped range - null if either start or end is
2114 * not mapped (no partial overlaps are calculated)
2116 int start = sf.getBegin();
2117 int end = sf.getEnd();
2118 int[] mappedTo = mapping.locateInTo(start, end);
2120 * if whole exon range doesn't map, try interpreting it
2121 * as 5' or 3' exon overlapping the CDS range
2123 if (mappedTo == null)
2125 mappedTo = mapping.locateInTo(end, end);
2126 if (mappedTo != null)
2129 * end of exon is in CDS range - 5' overlap
2130 * to a range from the start of the peptide
2135 if (mappedTo == null)
2137 mappedTo = mapping.locateInTo(start, start);
2138 if (mappedTo != null)
2141 * start of exon is in CDS range - 3' overlap
2142 * to a range up to the end of the peptide
2144 mappedTo[1] = toSeq.getLength();
2147 if (mappedTo != null)
2149 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2150 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2151 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2152 sf.getFeatureGroup(), sf.getScore());
2153 copyTo.addSequenceFeature(copy);
2161 * Returns a mapping from dna to protein by inspecting sequence features of
2162 * type "CDS" on the dna.
2168 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2169 SequenceI proteinSeq)
2171 List<int[]> ranges = findCdsPositions(dnaSeq);
2172 int mappedDnaLength = MappingUtils.getLength(ranges);
2174 int proteinLength = proteinSeq.getLength();
2175 int proteinStart = proteinSeq.getStart();
2176 int proteinEnd = proteinSeq.getEnd();
2179 * incomplete start codon may mean X at start of peptide
2180 * we ignore both for mapping purposes
2182 if (proteinSeq.getCharAt(0) == 'X')
2184 // todo JAL-2022 support startPhase > 0
2188 List<int[]> proteinRange = new ArrayList<int[]>();
2191 * dna length should map to protein (or protein plus stop codon)
2193 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2194 if (codesForResidues == (proteinLength + 1))
2196 // assuming extra codon is for STOP and not in peptide
2199 if (codesForResidues == proteinLength)
2201 proteinRange.add(new int[] { proteinStart, proteinEnd });
2202 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2208 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2209 * start/end positions of sequence features of type "CDS" (or a sub-type of
2210 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2211 * position order, so this method is only valid for linear CDS in the same
2212 * sense as the protein product.
2217 public static List<int[]> findCdsPositions(SequenceI dnaSeq)
2219 List<int[]> result = new ArrayList<int[]>();
2221 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2222 SequenceOntologyI.CDS);
2227 SequenceFeatures.sortFeatures(sfs, true);
2230 for (SequenceFeature sf : sfs)
2235 phase = Integer.parseInt(sf.getPhase());
2236 } catch (NumberFormatException e)
2241 * phase > 0 on first codon means 5' incomplete - skip to the start
2242 * of the next codon; example ENST00000496384
2244 int begin = sf.getBegin();
2245 int end = sf.getEnd();
2246 if (result.isEmpty())
2251 // shouldn't happen!
2253 .println("Error: start phase extends beyond start CDS in "
2254 + dnaSeq.getName());
2257 result.add(new int[] { begin, end });
2261 * remove 'startPhase' positions (usually 0) from the first range
2262 * so we begin at the start of a complete codon
2264 if (!result.isEmpty())
2266 // TODO JAL-2022 correctly model start phase > 0
2267 result.get(0)[0] += startPhase;
2271 * Finally sort ranges by start position. This avoids a dependency on
2272 * keeping features in order on the sequence (if they are in order anyway,
2273 * the sort will have almost no work to do). The implicit assumption is CDS
2274 * ranges are assembled in order. Other cases should not use this method,
2275 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2277 Collections.sort(result, IntRangeComparator.ASCENDING);
2282 * Maps exon features from dna to protein, and computes variants in peptide
2283 * product generated by variants in dna, and adds them as sequence_variant
2284 * features on the protein sequence. Returns the number of variant features
2289 * @param dnaToProtein
2291 public static int computeProteinFeatures(SequenceI dnaSeq,
2292 SequenceI peptide, MapList dnaToProtein)
2294 while (dnaSeq.getDatasetSequence() != null)
2296 dnaSeq = dnaSeq.getDatasetSequence();
2298 while (peptide.getDatasetSequence() != null)
2300 peptide = peptide.getDatasetSequence();
2303 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2306 * compute protein variants from dna variants and codon mappings;
2307 * NB - alternatively we could retrieve this using the REST service e.g.
2308 * http://rest.ensembl.org/overlap/translation
2309 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2310 * which would be a bit slower but possibly more reliable
2314 * build a map with codon variations for each potentially varying peptide
2316 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2317 dnaSeq, dnaToProtein);
2320 * scan codon variations, compute peptide variants and add to peptide sequence
2323 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2325 int peptidePos = variant.getKey();
2326 List<DnaVariant>[] codonVariants = variant.getValue();
2327 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2334 * Computes non-synonymous peptide variants from codon variants and adds them
2335 * as sequence_variant features on the protein sequence (one feature per
2336 * allele variant). Selected attributes (variant id, clinical significance)
2337 * are copied over to the new features.
2340 * the protein sequence
2342 * the position to compute peptide variants for
2343 * @param codonVariants
2344 * a list of dna variants per codon position
2345 * @return the number of features added
2347 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2348 List<DnaVariant>[] codonVariants)
2350 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2352 String base1 = codonVariants[0].get(0).base;
2353 String base2 = codonVariants[1].get(0).base;
2354 String base3 = codonVariants[2].get(0).base;
2357 * variants in first codon base
2359 for (DnaVariant var : codonVariants[0])
2361 if (var.variant != null)
2363 String alleles = (String) var.variant.getValue("alleles");
2364 if (alleles != null)
2366 for (String base : alleles.split(","))
2368 String codon = base + base2 + base3;
2369 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2379 * variants in second codon base
2381 for (DnaVariant var : codonVariants[1])
2383 if (var.variant != null)
2385 String alleles = (String) var.variant.getValue("alleles");
2386 if (alleles != null)
2388 for (String base : alleles.split(","))
2390 String codon = base1 + base + base3;
2391 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2401 * variants in third codon base
2403 for (DnaVariant var : codonVariants[2])
2405 if (var.variant != null)
2407 String alleles = (String) var.variant.getValue("alleles");
2408 if (alleles != null)
2410 for (String base : alleles.split(","))
2412 String codon = base1 + base2 + base;
2413 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2426 * Helper method that adds a peptide variant feature, provided the given codon
2427 * translates to a value different to the current residue (is a non-synonymous
2428 * variant). ID and clinical_significance attributes of the dna variant (if
2429 * present) are copied to the new feature.
2436 * @return true if a feature was added, else false
2438 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2439 String residue, DnaVariant var, String codon)
2442 * get peptide translation of codon e.g. GAT -> D
2443 * note that variants which are not single alleles,
2444 * e.g. multibase variants or HGMD_MUTATION etc
2445 * are currently ignored here
2447 String trans = codon.contains("-") ? "-"
2448 : (codon.length() > CODON_LENGTH ? null : ResidueProperties
2449 .codonTranslate(codon));
2450 if (trans != null && !trans.equals(residue))
2452 String residue3Char = StringUtils
2453 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2454 String trans3Char = StringUtils
2455 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2456 String desc = "p." + residue3Char + peptidePos + trans3Char;
2457 SequenceFeature sf = new SequenceFeature(
2458 SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
2459 peptidePos, var.getSource());
2460 StringBuilder attributes = new StringBuilder(32);
2461 String id = (String) var.variant.getValue(ID);
2464 if (id.startsWith(SEQUENCE_VARIANT))
2466 id = id.substring(SEQUENCE_VARIANT.length());
2468 sf.setValue(ID, id);
2469 attributes.append(ID).append("=").append(id);
2470 // TODO handle other species variants JAL-2064
2471 StringBuilder link = new StringBuilder(32);
2477 .append("|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2478 .append(URLEncoder.encode(id, "UTF-8"));
2479 sf.addLink(link.toString());
2480 } catch (UnsupportedEncodingException e)
2485 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2486 if (clinSig != null)
2488 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2489 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2492 peptide.addSequenceFeature(sf);
2493 if (attributes.length() > 0)
2495 sf.setAttributes(attributes.toString());
2503 * Builds a map whose key is position in the protein sequence, and value is a
2504 * list of the base and all variants for each corresponding codon position
2507 * @param dnaToProtein
2510 @SuppressWarnings("unchecked")
2511 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2512 SequenceI dnaSeq, MapList dnaToProtein)
2515 * map from peptide position to all variants of the codon which codes for it
2516 * LinkedHashMap ensures we keep the peptide features in sequence order
2518 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
2520 List<SequenceFeature> dnaFeatures = dnaSeq.getFeatures()
2521 .getFeaturesByOntology(SequenceOntologyI.SEQUENCE_VARIANT);
2522 if (dnaFeatures.isEmpty())
2527 int dnaStart = dnaSeq.getStart();
2528 int[] lastCodon = null;
2529 int lastPeptidePostion = 0;
2532 * build a map of codon variations for peptides
2534 for (SequenceFeature sf : dnaFeatures)
2536 int dnaCol = sf.getBegin();
2537 if (dnaCol != sf.getEnd())
2539 // not handling multi-locus variant features
2542 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2545 // feature doesn't lie within coding region
2548 int peptidePosition = mapsTo[0];
2549 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2550 if (codonVariants == null)
2552 codonVariants = new ArrayList[CODON_LENGTH];
2553 codonVariants[0] = new ArrayList<DnaVariant>();
2554 codonVariants[1] = new ArrayList<DnaVariant>();
2555 codonVariants[2] = new ArrayList<DnaVariant>();
2556 variants.put(peptidePosition, codonVariants);
2560 * extract dna variants to a string array
2562 String alls = (String) sf.getValue("alleles");
2567 String[] alleles = alls.toUpperCase().split(",");
2569 for (String allele : alleles)
2571 alleles[i++] = allele.trim(); // lose any space characters "A, G"
2575 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2577 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2578 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2579 peptidePosition, peptidePosition));
2580 lastPeptidePostion = peptidePosition;
2584 * save nucleotide (and any variant) for each codon position
2586 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2588 String nucleotide = String.valueOf(
2589 dnaSeq.getCharAt(codon[codonPos] - dnaStart)).toUpperCase();
2590 List<DnaVariant> codonVariant = codonVariants[codonPos];
2591 if (codon[codonPos] == dnaCol)
2593 if (!codonVariant.isEmpty()
2594 && codonVariant.get(0).variant == null)
2597 * already recorded base value, add this variant
2599 codonVariant.get(0).variant = sf;
2604 * add variant with base value
2606 codonVariant.add(new DnaVariant(nucleotide, sf));
2609 else if (codonVariant.isEmpty())
2612 * record (possibly non-varying) base value
2614 codonVariant.add(new DnaVariant(nucleotide));
2622 * Makes an alignment with a copy of the given sequences, adding in any
2623 * non-redundant sequences which are mapped to by the cross-referenced
2629 * the alignment dataset shared by the new copy
2632 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2633 SequenceI[] xrefs, AlignmentI dataset)
2635 AlignmentI copy = new Alignment(new Alignment(seqs));
2636 copy.setDataset(dataset);
2637 boolean isProtein = !copy.isNucleotide();
2638 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2641 for (SequenceI xref : xrefs)
2643 DBRefEntry[] dbrefs = xref.getDBRefs();
2646 for (DBRefEntry dbref : dbrefs)
2648 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2649 || dbref.getMap().getTo().isProtein() != isProtein)
2653 SequenceI mappedTo = dbref.getMap().getTo();
2654 SequenceI match = matcher.findIdMatch(mappedTo);
2657 matcher.add(mappedTo);
2658 copy.addSequence(mappedTo);
2668 * Try to align sequences in 'unaligned' to match the alignment of their
2669 * mapped regions in 'aligned'. For example, could use this to align CDS
2670 * sequences which are mapped to their parent cDNA sequences.
2672 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2673 * dna-to-protein or protein-to-dna use alternative methods.
2676 * sequences to be aligned
2678 * holds aligned sequences and their mappings
2681 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2684 * easy case - aligning a copy of aligned sequences
2686 if (alignAsSameSequences(unaligned, aligned))
2688 return unaligned.getHeight();
2692 * fancy case - aligning via mappings between sequences
2694 List<SequenceI> unmapped = new ArrayList<SequenceI>();
2695 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2696 unaligned, aligned, unmapped);
2697 int width = columnMap.size();
2698 char gap = unaligned.getGapCharacter();
2699 int realignedCount = 0;
2700 // TODO: verify this loop scales sensibly for very wide/high alignments
2702 for (SequenceI seq : unaligned.getSequences())
2704 if (!unmapped.contains(seq))
2706 char[] newSeq = new char[width];
2707 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2708 // Integer iteration below
2713 * traverse the map to find columns populated
2716 for (Integer column : columnMap.keySet())
2718 Character c = columnMap.get(column).get(seq);
2722 * sequence has a character at this position
2732 * trim trailing gaps
2734 if (lastCol < width)
2736 char[] tmp = new char[lastCol + 1];
2737 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2740 // TODO: optimise SequenceI to avoid char[]->String->char[]
2741 seq.setSequence(String.valueOf(newSeq));
2745 return realignedCount;
2749 * If unaligned and aligned sequences share the same dataset sequences, then
2750 * simply copies the aligned sequences to the unaligned sequences and returns
2751 * true; else returns false
2754 * - sequences to be aligned based on aligned
2756 * - 'guide' alignment containing sequences derived from same dataset
2760 static boolean alignAsSameSequences(AlignmentI unaligned,
2763 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2765 return false; // should only pass alignments with datasets here
2768 // map from dataset sequence to alignment sequence(s)
2769 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<SequenceI, List<SequenceI>>();
2770 for (SequenceI seq : aligned.getSequences())
2772 SequenceI ds = seq.getDatasetSequence();
2773 if (alignedDatasets.get(ds) == null)
2775 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2777 alignedDatasets.get(ds).add(seq);
2781 * first pass - check whether all sequences to be aligned share a dataset
2782 * sequence with an aligned sequence
2784 for (SequenceI seq : unaligned.getSequences())
2786 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2793 * second pass - copy aligned sequences;
2794 * heuristic rule: pair off sequences in order for the case where
2795 * more than one shares the same dataset sequence
2797 for (SequenceI seq : unaligned.getSequences())
2799 List<SequenceI> alignedSequences = alignedDatasets.get(seq
2800 .getDatasetSequence());
2801 // TODO: getSequenceAsString() will be deprecated in the future
2802 // TODO: need to leave to SequenceI implementor to update gaps
2803 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2804 if (alignedSequences.size() > 0)
2806 // pop off aligned sequences (except the last one)
2807 alignedSequences.remove(0);
2815 * Returns a map whose key is alignment column number (base 1), and whose
2816 * values are a map of sequence characters in that column.
2823 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2824 AlignmentI unaligned, AlignmentI aligned, List<SequenceI> unmapped)
2827 * Map will hold, for each aligned column position, a map of
2828 * {unalignedSequence, characterPerSequence} at that position.
2829 * TreeMap keeps the entries in ascending column order.
2831 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<Integer, Map<SequenceI, Character>>();
2834 * record any sequences that have no mapping so can't be realigned
2836 unmapped.addAll(unaligned.getSequences());
2838 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2840 for (SequenceI seq : unaligned.getSequences())
2842 for (AlignedCodonFrame mapping : mappings)
2844 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2845 if (fromSeq != null)
2847 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2848 if (addMappedPositions(seq, fromSeq, seqMap, map))
2850 unmapped.remove(seq);
2859 * Helper method that adds to a map the mapped column positions of a sequence. <br>
2860 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2861 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2865 * the sequence whose column positions we are recording
2867 * a sequence that is mapped to the first sequence
2869 * the mapping from 'fromSeq' to 'seq'
2871 * a map to add the column positions (in fromSeq) of the mapped
2875 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2876 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2884 * invert mapping if it is from unaligned to aligned sequence
2886 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2888 seqMap = new Mapping(seq.getDatasetSequence(), seqMap.getMap()
2892 int toStart = seq.getStart();
2895 * traverse [start, end, start, end...] ranges in fromSeq
2897 for (int[] fromRange : seqMap.getMap().getFromRanges())
2899 for (int i = 0; i < fromRange.length - 1; i += 2)
2901 boolean forward = fromRange[i + 1] >= fromRange[i];
2904 * find the range mapped to (sequence positions base 1)
2906 int[] range = seqMap.locateMappedRange(fromRange[i],
2910 System.err.println("Error in mapping " + seqMap + " from "
2911 + fromSeq.getName());
2914 int fromCol = fromSeq.findIndex(fromRange[i]);
2915 int mappedCharPos = range[0];
2918 * walk over the 'from' aligned sequence in forward or reverse
2919 * direction; when a non-gap is found, record the column position
2920 * of the next character of the mapped-to sequence; stop when all
2921 * the characters of the range have been counted
2923 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
2926 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
2929 * mapped from sequence has a character in this column
2930 * record the column position for the mapped to character
2932 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2933 if (seqsMap == null)
2935 seqsMap = new HashMap<SequenceI, Character>();
2936 map.put(fromCol, seqsMap);
2938 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
2941 fromCol += (forward ? 1 : -1);
2948 // strictly temporary hack until proper criteria for aligning protein to cds
2949 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2950 public static boolean looksLikeEnsembl(AlignmentI alignment)
2952 for (SequenceI seq : alignment.getSequences())
2954 String name = seq.getName();
2955 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))