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 char[] thisSeq = alignTo.getSequence();
654 final char[] thatAligned = alignFrom.getSequence();
655 StringBuilder thisAligned = new StringBuilder(2 * thisSeq.length);
658 * Traverse the 'model' aligned sequence
660 for (char sourceChar : thatAligned)
662 if (sourceChar == sourceGap)
664 sourceGapMappedLength += ratio;
669 * Found a non-gap character. Locate its mapped region if any.
672 // Note mapping positions are base 1, our sequence positions base 0
673 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
674 sourceDsPos + fromOffset);
675 if (mappedPos == null)
678 * unmapped position; treat like a gap
680 sourceGapMappedLength += ratio;
681 // System.err.println("Can't align: no codon mapping to residue "
682 // + sourceDsPos + "(" + sourceChar + ")");
687 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
688 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
689 StringBuilder trailingCopiedGap = new StringBuilder();
692 * Copy dna sequence up to and including this codon. Optionally, include
693 * gaps before the codon starts (in introns) and/or after the codon starts
696 * Note this only works for 'linear' splicing, not reverse or interleaved.
697 * But then 'align dna as protein' doesn't make much sense otherwise.
699 int intronLength = 0;
700 while (basesWritten + toOffset < mappedCodonEnd
701 && thisSeqPos < thisSeq.length)
703 final char c = thisSeq[thisSeqPos++];
707 int sourcePosition = basesWritten + toOffset;
708 if (sourcePosition < mappedCodonStart)
711 * Found an unmapped (intron) base. First add in any preceding gaps
714 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
716 thisAligned.append(trailingCopiedGap.toString());
717 intronLength += trailingCopiedGap.length();
718 trailingCopiedGap = new StringBuilder();
725 final boolean startOfCodon = sourcePosition == mappedCodonStart;
726 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
727 preserveUnmappedGaps, sourceGapMappedLength, inExon,
728 trailingCopiedGap.length(), intronLength, startOfCodon);
729 for (int i = 0; i < gapsToAdd; i++)
731 thisAligned.append(myGapChar);
733 sourceGapMappedLength = 0;
736 thisAligned.append(c);
737 trailingCopiedGap = new StringBuilder();
741 if (inExon && preserveMappedGaps)
743 trailingCopiedGap.append(myGapChar);
745 else if (!inExon && preserveUnmappedGaps)
747 trailingCopiedGap.append(myGapChar);
754 * At end of model aligned sequence. Copy any remaining target sequence, optionally
755 * including (intron) gaps.
757 while (thisSeqPos < thisSeq.length)
759 final char c = thisSeq[thisSeqPos++];
760 if (c != myGapChar || preserveUnmappedGaps)
762 thisAligned.append(c);
764 sourceGapMappedLength--;
768 * finally add gaps to pad for any trailing source gaps or
769 * unmapped characters
771 if (preserveUnmappedGaps)
773 while (sourceGapMappedLength > 0)
775 thisAligned.append(myGapChar);
776 sourceGapMappedLength--;
781 * All done aligning, set the aligned sequence.
783 alignTo.setSequence(new String(thisAligned));
787 * Helper method to work out how many gaps to insert when realigning.
789 * @param preserveMappedGaps
790 * @param preserveUnmappedGaps
791 * @param sourceGapMappedLength
793 * @param trailingCopiedGap
794 * @param intronLength
795 * @param startOfCodon
798 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
799 boolean preserveUnmappedGaps, int sourceGapMappedLength,
800 boolean inExon, int trailingGapLength, int intronLength,
801 final boolean startOfCodon)
807 * Reached start of codon. Ignore trailing gaps in intron unless we are
808 * preserving gaps in both exon and intron. Ignore them anyway if the
809 * protein alignment introduces a gap at least as large as the intronic
812 if (inExon && !preserveMappedGaps)
814 trailingGapLength = 0;
816 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
818 trailingGapLength = 0;
822 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
826 if (intronLength + trailingGapLength <= sourceGapMappedLength)
828 gapsToAdd = sourceGapMappedLength - intronLength;
832 gapsToAdd = Math.min(intronLength + trailingGapLength
833 - sourceGapMappedLength, trailingGapLength);
840 * second or third base of codon; check for any gaps in dna
842 if (!preserveMappedGaps)
844 trailingGapLength = 0;
846 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
852 * Realigns the given protein to match the alignment of the dna, using codon
853 * mappings to translate aligned codon positions to protein residues.
856 * the alignment whose sequences are realigned by this method
858 * the dna alignment whose alignment we are 'copying'
859 * @return the number of sequences that were realigned
861 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
863 if (protein.isNucleotide() || !dna.isNucleotide())
865 System.err.println("Wrong alignment type in alignProteinAsDna");
868 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
869 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
870 protein, dna, unmappedProtein);
871 return alignProteinAs(protein, alignedCodons, unmappedProtein);
875 * Realigns the given dna to match the alignment of the protein, using codon
876 * mappings to translate aligned peptide positions to codons.
878 * Always produces a padded CDS alignment.
881 * the alignment whose sequences are realigned by this method
883 * the protein alignment whose alignment we are 'copying'
884 * @return the number of sequences that were realigned
886 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
888 if (protein.isNucleotide() || !dna.isNucleotide())
890 System.err.println("Wrong alignment type in alignProteinAsDna");
893 // todo: implement this
894 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
895 int alignedCount = 0;
896 int width = 0; // alignment width for padding CDS
897 for (SequenceI dnaSeq : dna.getSequences())
899 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
900 dna.getGapCharacter()))
904 width = Math.max(dnaSeq.getLength(), width);
908 for (SequenceI dnaSeq : dna.getSequences())
910 oldwidth = dnaSeq.getLength();
911 diff = width - oldwidth;
914 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
921 * Helper method to align (if possible) the dna sequence to match the
922 * alignment of a mapped protein sequence. This is currently limited to
923 * handling coding sequence only.
931 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
932 AlignmentI protein, List<AlignedCodonFrame> mappings, char gapChar)
934 SequenceI cdsDss = cdsSeq.getDatasetSequence();
938 .println("alignCdsSequenceAsProtein needs aligned sequence!");
942 List<AlignedCodonFrame> dnaMappings = MappingUtils
943 .findMappingsForSequence(cdsSeq, mappings);
944 for (AlignedCodonFrame mapping : dnaMappings)
946 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
949 int peptideLength = peptide.getLength();
950 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
953 MapList mapList = map.getMap();
954 if (map.getTo() == peptide.getDatasetSequence())
956 mapList = mapList.getInverse();
958 int cdsLength = cdsDss.getLength();
959 int mappedFromLength = MappingUtils.getLength(mapList
961 int mappedToLength = MappingUtils
962 .getLength(mapList.getToRanges());
963 boolean addStopCodon = (cdsLength == mappedFromLength
964 * CODON_LENGTH + CODON_LENGTH)
965 || (peptide.getDatasetSequence().getLength() == mappedFromLength - 1);
966 if (cdsLength != mappedToLength && !addStopCodon)
970 .format("Can't align cds as protein (length mismatch %d/%d): %s",
971 cdsLength, mappedToLength,
976 * pre-fill the aligned cds sequence with gaps
978 char[] alignedCds = new char[peptideLength * CODON_LENGTH
979 + (addStopCodon ? CODON_LENGTH : 0)];
980 Arrays.fill(alignedCds, gapChar);
983 * walk over the aligned peptide sequence and insert mapped
984 * codons for residues in the aligned cds sequence
986 char[] alignedPeptide = peptide.getSequence();
987 char[] nucleotides = cdsDss.getSequence();
989 int cdsStart = cdsDss.getStart();
990 int proteinPos = peptide.getStart() - 1;
992 for (char residue : alignedPeptide)
994 if (Comparison.isGap(residue))
996 cdsCol += CODON_LENGTH;
1001 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1004 // e.g. incomplete start codon, X in peptide
1005 cdsCol += CODON_LENGTH;
1009 for (int j = codon[0]; j <= codon[1]; j++)
1011 char mappedBase = nucleotides[j - cdsStart];
1012 alignedCds[cdsCol++] = mappedBase;
1020 * append stop codon if not mapped from protein,
1021 * closing it up to the end of the mapped sequence
1023 if (copiedBases == nucleotides.length - CODON_LENGTH)
1025 for (int i = alignedCds.length - 1; i >= 0; i--)
1027 if (!Comparison.isGap(alignedCds[i]))
1029 cdsCol = i + 1; // gap just after end of sequence
1033 for (int i = nucleotides.length - CODON_LENGTH; i < nucleotides.length; i++)
1035 alignedCds[cdsCol++] = nucleotides[i];
1038 cdsSeq.setSequence(new String(alignedCds));
1047 * Builds a map whose key is an aligned codon position (3 alignment column
1048 * numbers base 0), and whose value is a map from protein sequence to each
1049 * protein's peptide residue for that codon. The map generates an ordering of
1050 * the codons, and allows us to read off the peptides at each position in
1051 * order to assemble 'aligned' protein sequences.
1054 * the protein alignment
1056 * the coding dna alignment
1057 * @param unmappedProtein
1058 * any unmapped proteins are added to this list
1061 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1062 AlignmentI protein, AlignmentI dna,
1063 List<SequenceI> unmappedProtein)
1066 * maintain a list of any proteins with no mappings - these will be
1067 * rendered 'as is' in the protein alignment as we can't align them
1069 unmappedProtein.addAll(protein.getSequences());
1071 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1074 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1075 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1076 * comparator keeps the codon positions ordered.
1078 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
1079 new CodonComparator());
1081 for (SequenceI dnaSeq : dna.getSequences())
1083 for (AlignedCodonFrame mapping : mappings)
1085 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1088 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1089 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
1090 seqMap, alignedCodons);
1091 unmappedProtein.remove(prot);
1097 * Finally add any unmapped peptide start residues (e.g. for incomplete
1098 * codons) as if at the codon position before the second residue
1100 // TODO resolve JAL-2022 so this fudge can be removed
1101 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1102 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1104 return alignedCodons;
1108 * Scans for any protein mapped from position 2 (meaning unmapped start
1109 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1110 * preceding position in the alignment
1112 * @param alignedCodons
1113 * the codon-to-peptide map
1114 * @param mappedSequenceCount
1115 * the number of distinct sequences in the map
1117 protected static void addUnmappedPeptideStarts(
1118 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1119 int mappedSequenceCount)
1121 // TODO delete this ugly hack once JAL-2022 is resolved
1122 // i.e. we can model startPhase > 0 (incomplete start codon)
1124 List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
1125 AlignedCodon lastCodon = null;
1126 Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
1128 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1131 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1134 SequenceI seq = sequenceCodon.getKey();
1135 if (sequencesChecked.contains(seq))
1139 sequencesChecked.add(seq);
1140 AlignedCodon codon = sequenceCodon.getValue();
1141 if (codon.peptideCol > 1)
1144 .println("Problem mapping protein with >1 unmapped start positions: "
1147 else if (codon.peptideCol == 1)
1150 * first position (peptideCol == 0) was unmapped - add it
1152 if (lastCodon != null)
1154 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1155 lastCodon.pos2, lastCodon.pos3, String.valueOf(seq
1157 toAdd.put(seq, firstPeptide);
1162 * unmapped residue at start of alignment (no prior column) -
1163 * 'insert' at nominal codon [0, 0, 0]
1165 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1166 String.valueOf(seq.getCharAt(0)), 0);
1167 toAdd.put(seq, firstPeptide);
1170 if (sequencesChecked.size() == mappedSequenceCount)
1172 // no need to check past first mapped position in all sequences
1176 lastCodon = entry.getKey();
1180 * add any new codons safely after iterating over the map
1182 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1184 addCodonToMap(alignedCodons, startCodon.getValue(),
1185 startCodon.getKey());
1190 * Update the aligned protein sequences to match the codon alignments given in
1194 * @param alignedCodons
1195 * an ordered map of codon positions (columns), with sequence/peptide
1196 * values present in each column
1197 * @param unmappedProtein
1200 protected static int alignProteinAs(AlignmentI protein,
1201 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1202 List<SequenceI> unmappedProtein)
1205 * Prefill aligned sequences with gaps before inserting aligned protein
1208 int alignedWidth = alignedCodons.size();
1209 char[] gaps = new char[alignedWidth];
1210 Arrays.fill(gaps, protein.getGapCharacter());
1211 String allGaps = String.valueOf(gaps);
1212 for (SequenceI seq : protein.getSequences())
1214 if (!unmappedProtein.contains(seq))
1216 seq.setSequence(allGaps);
1221 for (AlignedCodon codon : alignedCodons.keySet())
1223 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1225 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1227 // place translated codon at its column position in sequence
1228 entry.getKey().getSequence()[column] = entry.getValue().product
1237 * Populate the map of aligned codons by traversing the given sequence
1238 * mapping, locating the aligned positions of mapped codons, and adding those
1239 * positions and their translation products to the map.
1242 * the aligned sequence we are mapping from
1244 * the sequence to be aligned to the codons
1246 * the gap character in the dna sequence
1248 * a mapping to a sequence translation
1249 * @param alignedCodons
1250 * the map we are building up
1252 static void addCodonPositions(SequenceI dna, SequenceI protein,
1253 char gapChar, Mapping seqMap,
1254 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1256 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1259 * add codon positions, and their peptide translations, to the alignment
1260 * map, while remembering the first codon mapped
1262 while (codons.hasNext())
1266 AlignedCodon codon = codons.next();
1267 addCodonToMap(alignedCodons, codon, protein);
1268 } catch (IncompleteCodonException e)
1270 // possible incomplete trailing codon - ignore
1271 } catch (NoSuchElementException e)
1273 // possibly peptide lacking STOP
1279 * Helper method to add a codon-to-peptide entry to the aligned codons map
1281 * @param alignedCodons
1285 protected static void addCodonToMap(
1286 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1287 AlignedCodon codon, SequenceI protein)
1289 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1290 if (seqProduct == null)
1292 seqProduct = new HashMap<SequenceI, AlignedCodon>();
1293 alignedCodons.put(codon, seqProduct);
1295 seqProduct.put(protein, codon);
1299 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1300 * between at least one pair of sequences in the two alignments. Currently,
1303 * <li>One alignment must be nucleotide, and the other protein</li>
1304 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1305 * <li>Mappable means the nucleotide translation matches the protein sequence</li>
1306 * <li>The translation may ignore start and stop codons if present in the
1314 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1316 if (al1 == null || al2 == null)
1322 * Require one nucleotide and one protein
1324 if (al1.isNucleotide() == al2.isNucleotide())
1328 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1329 AlignmentI protein = dna == al1 ? al2 : al1;
1330 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1331 for (SequenceI dnaSeq : dna.getSequences())
1333 for (SequenceI proteinSeq : protein.getSequences())
1335 if (isMappable(dnaSeq, proteinSeq, mappings))
1345 * Returns true if the dna sequence is mapped, or could be mapped, to the
1353 protected static boolean isMappable(SequenceI dnaSeq,
1354 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1356 if (dnaSeq == null || proteinSeq == null)
1361 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq : dnaSeq
1362 .getDatasetSequence();
1363 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null ? proteinSeq
1364 : proteinSeq.getDatasetSequence();
1366 for (AlignedCodonFrame mapping : mappings)
1368 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1378 * Just try to make a mapping (it is not yet stored), test whether
1381 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1385 * Finds any reference annotations associated with the sequences in
1386 * sequenceScope, that are not already added to the alignment, and adds them
1387 * to the 'candidates' map. Also populates a lookup table of annotation
1388 * labels, keyed by calcId, for use in constructing tooltips or the like.
1390 * @param sequenceScope
1391 * the sequences to scan for reference annotations
1392 * @param labelForCalcId
1393 * (optional) map to populate with label for calcId
1395 * map to populate with annotations for sequence
1397 * the alignment to check for presence of annotations
1399 public static void findAddableReferenceAnnotations(
1400 List<SequenceI> sequenceScope,
1401 Map<String, String> labelForCalcId,
1402 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1405 if (sequenceScope == null)
1411 * For each sequence in scope, make a list of any annotations on the
1412 * underlying dataset sequence which are not already on the alignment.
1414 * Add to a map of { alignmentSequence, <List of annotations to add> }
1416 for (SequenceI seq : sequenceScope)
1418 SequenceI dataset = seq.getDatasetSequence();
1419 if (dataset == null)
1423 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1424 if (datasetAnnotations == null)
1428 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1429 for (AlignmentAnnotation dsann : datasetAnnotations)
1432 * Find matching annotations on the alignment. If none is found, then
1433 * add this annotation to the list of 'addable' annotations for this
1436 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1437 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1438 if (!matchedAlignmentAnnotations.iterator().hasNext())
1441 if (labelForCalcId != null)
1443 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1448 * Save any addable annotations for this sequence
1450 if (!result.isEmpty())
1452 candidates.put(seq, result);
1458 * Adds annotations to the top of the alignment annotations, in the same order
1459 * as their related sequences.
1461 * @param annotations
1462 * the annotations to add
1464 * the alignment to add them to
1465 * @param selectionGroup
1466 * current selection group (or null if none)
1468 public static void addReferenceAnnotations(
1469 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1470 final AlignmentI alignment, final SequenceGroup selectionGroup)
1472 for (SequenceI seq : annotations.keySet())
1474 for (AlignmentAnnotation ann : annotations.get(seq))
1476 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1478 int endRes = ann.annotations.length;
1479 if (selectionGroup != null)
1481 startRes = selectionGroup.getStartRes();
1482 endRes = selectionGroup.getEndRes();
1484 copyAnn.restrict(startRes, endRes);
1487 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1488 * original annotation is already on the sequence.
1490 if (!seq.hasAnnotation(ann))
1492 seq.addAlignmentAnnotation(copyAnn);
1495 copyAnn.adjustForAlignment();
1496 // add to the alignment and set visible
1497 alignment.addAnnotation(copyAnn);
1498 copyAnn.visible = true;
1504 * Set visibility of alignment annotations of specified types (labels), for
1505 * specified sequences. This supports controls like
1506 * "Show all secondary structure", "Hide all Temp factor", etc.
1508 * @al the alignment to scan for annotations
1510 * the types (labels) of annotations to be updated
1511 * @param forSequences
1512 * if not null, only annotations linked to one of these sequences are
1513 * in scope for update; if null, acts on all sequence annotations
1515 * if this flag is true, 'types' is ignored (label not checked)
1517 * if true, set visibility on, else set off
1519 public static void showOrHideSequenceAnnotations(AlignmentI al,
1520 Collection<String> types, List<SequenceI> forSequences,
1521 boolean anyType, boolean doShow)
1523 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1526 for (AlignmentAnnotation aa : anns)
1528 if (anyType || types.contains(aa.label))
1530 if ((aa.sequenceRef != null)
1531 && (forSequences == null || forSequences
1532 .contains(aa.sequenceRef)))
1534 aa.visible = doShow;
1542 * Returns true if either sequence has a cross-reference to the other
1548 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1550 // Note: moved here from class CrossRef as the latter class has dependencies
1551 // not availability to the applet's classpath
1552 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1556 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1557 * that sequence name is structured as Source|AccessionId.
1563 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1565 if (seq1 == null || seq2 == null)
1569 String name = seq2.getName();
1570 final DBRefEntry[] xrefs = seq1.getDBRefs();
1573 for (DBRefEntry xref : xrefs)
1575 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1576 // case-insensitive test, consistent with DBRefEntry.equalRef()
1577 if (xrefName.equalsIgnoreCase(name))
1587 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1588 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1589 * added to the original alignment's dataset, which is shared by the new
1590 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1591 * added to the alignment dataset.
1594 * aligned nucleotide (dna or cds) sequences
1596 * the alignment dataset the sequences belong to
1598 * (optional) to restrict results to CDS that map to specified
1600 * @return an alignment whose sequences are the cds-only parts of the dna
1601 * sequences (or null if no mappings are found)
1603 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1604 AlignmentI dataset, SequenceI[] products)
1606 if (dataset == null || dataset.getDataset() != null)
1608 throw new IllegalArgumentException(
1609 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1611 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
1612 List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
1613 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1614 HashSet<SequenceI> productSeqs = null;
1615 if (products != null)
1617 productSeqs = new HashSet<SequenceI>();
1618 for (SequenceI seq : products)
1620 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1621 .getDatasetSequence());
1626 * Construct CDS sequences from mappings on the alignment dataset.
1628 * - find the protein product(s) mapped to from each dna sequence
1629 * - if the mapping covers the whole dna sequence (give or take start/stop
1630 * codon), take the dna as the CDS sequence
1631 * - else search dataset mappings for a suitable dna sequence, i.e. one
1632 * whose whole sequence is mapped to the protein
1633 * - if no sequence found, construct one from the dna sequence and mapping
1634 * (and add it to dataset so it is found if this is repeated)
1636 for (SequenceI dnaSeq : dna)
1638 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1639 : dnaSeq.getDatasetSequence();
1641 List<AlignedCodonFrame> seqMappings = MappingUtils
1642 .findMappingsForSequence(dnaSeq, mappings);
1643 for (AlignedCodonFrame mapping : seqMappings)
1645 List<Mapping> mappingsFromSequence = mapping
1646 .getMappingsFromSequence(dnaSeq);
1648 for (Mapping aMapping : mappingsFromSequence)
1650 MapList mapList = aMapping.getMap();
1651 if (mapList.getFromRatio() == 1)
1654 * not a dna-to-protein mapping (likely dna-to-cds)
1660 * skip if mapping is not to one of the target set of proteins
1662 SequenceI proteinProduct = aMapping.getTo();
1663 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1669 * try to locate the CDS from the dataset mappings;
1670 * guard against duplicate results (for the case that protein has
1671 * dbrefs to both dna and cds sequences)
1673 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1674 seqMappings, aMapping);
1677 if (!foundSeqs.contains(cdsSeq))
1679 foundSeqs.add(cdsSeq);
1680 SequenceI derivedSequence = cdsSeq.deriveSequence();
1681 cdsSeqs.add(derivedSequence);
1682 if (!dataset.getSequences().contains(cdsSeq))
1684 dataset.addSequence(cdsSeq);
1691 * didn't find mapped CDS sequence - construct it and add
1692 * its dataset sequence to the dataset
1694 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1695 dataset).deriveSequence();
1696 // cdsSeq has a name constructed as CDS|<dbref>
1697 // <dbref> will be either the accession for the coding sequence,
1698 // marked in the /via/ dbref to the protein product accession
1699 // or it will be the original nucleotide accession.
1700 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1702 cdsSeqs.add(cdsSeq);
1704 if (!dataset.getSequences().contains(cdsSeqDss))
1706 // check if this sequence is a newly created one
1707 // so needs adding to the dataset
1708 dataset.addSequence(cdsSeqDss);
1712 * add a mapping from CDS to the (unchanged) mapped to range
1714 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1715 cdsSeq.getLength() });
1716 MapList cdsToProteinMap = new MapList(cdsRange,
1717 mapList.getToRanges(), mapList.getFromRatio(),
1718 mapList.getToRatio());
1719 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1720 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1724 * guard against duplicating the mapping if repeating this action
1726 if (!mappings.contains(cdsToProteinMapping))
1728 mappings.add(cdsToProteinMapping);
1731 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1732 proteinProduct, aMapping);
1734 * add another mapping from original 'from' range to CDS
1736 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1737 MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1739 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1741 if (!mappings.contains(dnaToCdsMapping))
1743 mappings.add(dnaToCdsMapping);
1747 * add DBRef with mapping from protein to CDS
1748 * (this enables Get Cross-References from protein alignment)
1749 * This is tricky because we can't have two DBRefs with the
1750 * same source and accession, so need a different accession for
1751 * the CDS from the dna sequence
1754 // specific use case:
1755 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1756 // ENSG02, ENSG03, with transcripts and products similarly named.
1757 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1759 // JBPNote: ?? can't actually create an example that demonstrates we
1761 // synthesize an xref.
1763 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1765 // creates a complementary cross-reference to the source sequence's
1766 // primary reference.
1768 DBRefEntry cdsCrossRef = new DBRefEntry(primRef.getSource(),
1769 primRef.getSource() + ":" + primRef.getVersion(),
1770 primRef.getAccessionId());
1772 .setMap(new Mapping(dnaDss, new MapList(dnaToCdsMap)));
1773 cdsSeqDss.addDBRef(cdsCrossRef);
1775 // problem here is that the cross-reference is synthesized -
1776 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1778 // assuming cds version same as dna ?!?
1780 DBRefEntry proteinToCdsRef = new DBRefEntry(
1781 primRef.getSource(), primRef.getVersion(),
1784 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1786 proteinProduct.addDBRef(proteinToCdsRef);
1790 * transfer any features on dna that overlap the CDS
1792 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1793 SequenceOntologyI.CDS);
1798 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1800 cds.setDataset(dataset);
1806 * A helper method that finds a CDS sequence in the alignment dataset that is
1807 * mapped to the given protein sequence, and either is, or has a mapping from,
1808 * the given dna sequence.
1811 * set of all mappings on the dataset
1813 * a dna (or cds) sequence we are searching from
1814 * @param seqMappings
1815 * the set of mappings involving dnaSeq
1817 * an initial candidate from seqMappings
1820 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1821 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1825 * TODO a better dna-cds-protein mapping data representation to allow easy
1826 * navigation; until then this clunky looping around lists of mappings
1828 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1829 : dnaSeq.getDatasetSequence();
1830 SequenceI proteinProduct = aMapping.getTo();
1833 * is this mapping from the whole dna sequence (i.e. CDS)?
1834 * allowing for possible stop codon on dna but not peptide
1836 int mappedFromLength = MappingUtils.getLength(aMapping.getMap()
1838 int dnaLength = seqDss.getLength();
1839 if (mappedFromLength == dnaLength
1840 || mappedFromLength == dnaLength - CODON_LENGTH)
1846 * looks like we found the dna-to-protein mapping; search for the
1847 * corresponding cds-to-protein mapping
1849 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1850 .findMappingsForSequence(proteinProduct, mappings);
1851 for (AlignedCodonFrame acf : mappingsToPeptide)
1853 for (SequenceToSequenceMapping map : acf.getMappings())
1855 Mapping mapping = map.getMapping();
1856 if (mapping != aMapping
1857 && mapping.getMap().getFromRatio() == CODON_LENGTH
1858 && proteinProduct == mapping.getTo()
1859 && seqDss != map.getFromSeq())
1861 mappedFromLength = MappingUtils.getLength(mapping.getMap()
1863 if (mappedFromLength == map.getFromSeq().getLength())
1866 * found a 3:1 mapping to the protein product which covers
1867 * the whole dna sequence i.e. is from CDS; finally check it
1868 * is from the dna start sequence
1870 SequenceI cdsSeq = map.getFromSeq();
1871 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1872 .findMappingsForSequence(cdsSeq, seqMappings);
1873 if (!dnaToCdsMaps.isEmpty())
1885 * Helper method that makes a CDS sequence as defined by the mappings from the
1886 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1887 * forward or reverse strand).
1892 * - existing dataset. We check for sequences that look like the CDS
1893 * we are about to construct, if one exists already, then we will
1894 * just return that one.
1895 * @return CDS sequence (as a dataset sequence)
1897 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1900 char[] seqChars = seq.getSequence();
1901 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1902 int cdsWidth = MappingUtils.getLength(fromRanges);
1903 char[] newSeqChars = new char[cdsWidth];
1906 for (int[] range : fromRanges)
1908 if (range[0] <= range[1])
1910 // forward strand mapping - just copy the range
1911 int length = range[1] - range[0] + 1;
1912 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1918 // reverse strand mapping - copy and complement one by one
1919 for (int i = range[0]; i >= range[1]; i--)
1921 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1927 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
1928 * else generate a sequence name
1930 String mapFromId = mapping.getMappedFromId();
1931 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
1932 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
1933 if (dataset != null)
1935 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
1936 if (matches != null)
1938 boolean matched = false;
1939 for (SequenceI mtch : matches)
1941 if (mtch.getStart() != newSeq.getStart())
1945 if (mtch.getEnd() != newSeq.getEnd())
1949 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
1961 .println("JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
1967 // newSeq.setDescription(mapFromId);
1973 * add any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
1974 * the given mapping.
1979 * @return list of DBRefEntrys added.
1981 public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
1982 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
1985 // gather direct refs from contig congrent with mapping
1986 List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
1987 HashSet<String> directSources = new HashSet<String>();
1988 if (contig.getDBRefs() != null)
1990 for (DBRefEntry dbr : contig.getDBRefs())
1992 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
1994 MapList map = dbr.getMap().getMap();
1995 // check if map is the CDS mapping
1996 if (mapping.getMap().equals(map))
1999 directSources.add(dbr.getSource());
2004 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2005 proteinProduct.getDBRefs(),
2006 directSources.toArray(new String[0]));
2007 List<DBRefEntry> propagated = new ArrayList<DBRefEntry>();
2009 // and generate appropriate mappings
2010 for (DBRefEntry cdsref : direct)
2012 // clone maplist and mapping
2013 MapList cdsposmap = new MapList(Arrays.asList(new int[][] { new int[]
2014 { cdsSeq.getStart(), cdsSeq.getEnd() } }), cdsref.getMap().getMap()
2015 .getToRanges(), 3, 1);
2016 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(), cdsref.getMap()
2020 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2021 cdsref.getVersion(), cdsref.getAccessionId(), new Mapping(
2022 cdsmap.getTo(), cdsposmap));
2024 // and see if we can map to the protein product for this mapping.
2025 // onSource is the filtered set of accessions on protein that we are
2026 // tranferring, so we assume accession is the same.
2027 if (cdsmap.getTo() == null && onSource != null)
2029 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2030 cdsref.getAccessionId());
2031 if (sourceRefs != null)
2033 for (DBRefEntry srcref : sourceRefs)
2035 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2037 // we have found a complementary dbref on the protein product, so
2038 // update mapping's getTo
2039 newref.getMap().setTo(proteinProduct);
2044 cdsSeq.addDBRef(newref);
2045 propagated.add(newref);
2051 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2052 * feature start/end ranges, optionally omitting specified feature types.
2053 * Returns the number of features copied.
2058 * the mapping from 'fromSeq' to 'toSeq'
2060 * if not null, only features of this type are copied (including
2061 * subtypes in the Sequence Ontology)
2064 public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2065 MapList mapping, String select, String... omitting)
2067 SequenceI copyTo = toSeq;
2068 while (copyTo.getDatasetSequence() != null)
2070 copyTo = copyTo.getDatasetSequence();
2074 * get features, optionally restricted by an ontology term
2076 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2077 .getPositionalFeatures() : fromSeq.getFeatures()
2078 .getFeaturesByOntology(select);
2081 for (SequenceFeature sf : sfs)
2083 String type = sf.getType();
2084 boolean omit = false;
2085 for (String toOmit : omitting)
2087 if (type.equals(toOmit))
2098 * locate the mapped range - null if either start or end is
2099 * not mapped (no partial overlaps are calculated)
2101 int start = sf.getBegin();
2102 int end = sf.getEnd();
2103 int[] mappedTo = mapping.locateInTo(start, end);
2105 * if whole exon range doesn't map, try interpreting it
2106 * as 5' or 3' exon overlapping the CDS range
2108 if (mappedTo == null)
2110 mappedTo = mapping.locateInTo(end, end);
2111 if (mappedTo != null)
2114 * end of exon is in CDS range - 5' overlap
2115 * to a range from the start of the peptide
2120 if (mappedTo == null)
2122 mappedTo = mapping.locateInTo(start, start);
2123 if (mappedTo != null)
2126 * start of exon is in CDS range - 3' overlap
2127 * to a range up to the end of the peptide
2129 mappedTo[1] = toSeq.getLength();
2132 if (mappedTo != null)
2134 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2135 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2136 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2137 sf.getFeatureGroup());
2138 copyTo.addSequenceFeature(copy);
2146 * Returns a mapping from dna to protein by inspecting sequence features of
2147 * type "CDS" on the dna.
2153 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2154 SequenceI proteinSeq)
2156 List<int[]> ranges = findCdsPositions(dnaSeq);
2157 int mappedDnaLength = MappingUtils.getLength(ranges);
2159 int proteinLength = proteinSeq.getLength();
2160 int proteinStart = proteinSeq.getStart();
2161 int proteinEnd = proteinSeq.getEnd();
2164 * incomplete start codon may mean X at start of peptide
2165 * we ignore both for mapping purposes
2167 if (proteinSeq.getCharAt(0) == 'X')
2169 // todo JAL-2022 support startPhase > 0
2173 List<int[]> proteinRange = new ArrayList<int[]>();
2176 * dna length should map to protein (or protein plus stop codon)
2178 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2179 if (codesForResidues == (proteinLength + 1))
2181 // assuming extra codon is for STOP and not in peptide
2184 if (codesForResidues == proteinLength)
2186 proteinRange.add(new int[] { proteinStart, proteinEnd });
2187 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2193 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2194 * start/end positions of sequence features of type "CDS" (or a sub-type of
2195 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2196 * position order, so this method is only valid for linear CDS in the same
2197 * sense as the protein product.
2202 public static List<int[]> findCdsPositions(SequenceI dnaSeq)
2204 List<int[]> result = new ArrayList<int[]>();
2206 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2207 SequenceOntologyI.CDS);
2212 SequenceFeatures.sortFeatures(sfs, true);
2215 for (SequenceFeature sf : sfs)
2220 phase = Integer.parseInt(sf.getPhase());
2221 } catch (NumberFormatException e)
2226 * phase > 0 on first codon means 5' incomplete - skip to the start
2227 * of the next codon; example ENST00000496384
2229 int begin = sf.getBegin();
2230 int end = sf.getEnd();
2231 if (result.isEmpty())
2236 // shouldn't happen!
2238 .println("Error: start phase extends beyond start CDS in "
2239 + dnaSeq.getName());
2242 result.add(new int[] { begin, end });
2246 * remove 'startPhase' positions (usually 0) from the first range
2247 * so we begin at the start of a complete codon
2249 if (!result.isEmpty())
2251 // TODO JAL-2022 correctly model start phase > 0
2252 result.get(0)[0] += startPhase;
2256 * Finally sort ranges by start position. This avoids a dependency on
2257 * keeping features in order on the sequence (if they are in order anyway,
2258 * the sort will have almost no work to do). The implicit assumption is CDS
2259 * ranges are assembled in order. Other cases should not use this method,
2260 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2262 Collections.sort(result, IntRangeComparator.ASCENDING);
2267 * Maps exon features from dna to protein, and computes variants in peptide
2268 * product generated by variants in dna, and adds them as sequence_variant
2269 * features on the protein sequence. Returns the number of variant features
2274 * @param dnaToProtein
2276 public static int computeProteinFeatures(SequenceI dnaSeq,
2277 SequenceI peptide, MapList dnaToProtein)
2279 while (dnaSeq.getDatasetSequence() != null)
2281 dnaSeq = dnaSeq.getDatasetSequence();
2283 while (peptide.getDatasetSequence() != null)
2285 peptide = peptide.getDatasetSequence();
2288 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2291 * compute protein variants from dna variants and codon mappings;
2292 * NB - alternatively we could retrieve this using the REST service e.g.
2293 * http://rest.ensembl.org/overlap/translation
2294 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2295 * which would be a bit slower but possibly more reliable
2299 * build a map with codon variations for each potentially varying peptide
2301 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2302 dnaSeq, dnaToProtein);
2305 * scan codon variations, compute peptide variants and add to peptide sequence
2308 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2310 int peptidePos = variant.getKey();
2311 List<DnaVariant>[] codonVariants = variant.getValue();
2312 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2319 * Computes non-synonymous peptide variants from codon variants and adds them
2320 * as sequence_variant features on the protein sequence (one feature per
2321 * allele variant). Selected attributes (variant id, clinical significance)
2322 * are copied over to the new features.
2325 * the protein sequence
2327 * the position to compute peptide variants for
2328 * @param codonVariants
2329 * a list of dna variants per codon position
2330 * @return the number of features added
2332 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2333 List<DnaVariant>[] codonVariants)
2335 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2337 String base1 = codonVariants[0].get(0).base;
2338 String base2 = codonVariants[1].get(0).base;
2339 String base3 = codonVariants[2].get(0).base;
2342 * variants in first codon base
2344 for (DnaVariant var : codonVariants[0])
2346 if (var.variant != null)
2348 String alleles = (String) var.variant.getValue("alleles");
2349 if (alleles != null)
2351 for (String base : alleles.split(","))
2353 String codon = base + base2 + base3;
2354 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2364 * variants in second codon base
2366 for (DnaVariant var : codonVariants[1])
2368 if (var.variant != null)
2370 String alleles = (String) var.variant.getValue("alleles");
2371 if (alleles != null)
2373 for (String base : alleles.split(","))
2375 String codon = base1 + base + base3;
2376 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2386 * variants in third codon base
2388 for (DnaVariant var : codonVariants[2])
2390 if (var.variant != null)
2392 String alleles = (String) var.variant.getValue("alleles");
2393 if (alleles != null)
2395 for (String base : alleles.split(","))
2397 String codon = base1 + base2 + base;
2398 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2411 * Helper method that adds a peptide variant feature, provided the given codon
2412 * translates to a value different to the current residue (is a non-synonymous
2413 * variant). ID and clinical_significance attributes of the dna variant (if
2414 * present) are copied to the new feature.
2421 * @return true if a feature was added, else false
2423 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2424 String residue, DnaVariant var, String codon)
2427 * get peptide translation of codon e.g. GAT -> D
2428 * note that variants which are not single alleles,
2429 * e.g. multibase variants or HGMD_MUTATION etc
2430 * are currently ignored here
2432 String trans = codon.contains("-") ? "-"
2433 : (codon.length() > CODON_LENGTH ? null : ResidueProperties
2434 .codonTranslate(codon));
2435 if (trans != null && !trans.equals(residue))
2437 String residue3Char = StringUtils
2438 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2439 String trans3Char = StringUtils
2440 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2441 String desc = "p." + residue3Char + peptidePos + trans3Char;
2442 // set score to 0f so 'graduated colour' option is offered! JAL-2060
2443 SequenceFeature sf = new SequenceFeature(
2444 SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
2445 peptidePos, 0f, var.getSource());
2446 StringBuilder attributes = new StringBuilder(32);
2447 String id = (String) var.variant.getValue(ID);
2450 if (id.startsWith(SEQUENCE_VARIANT))
2452 id = id.substring(SEQUENCE_VARIANT.length());
2454 sf.setValue(ID, id);
2455 attributes.append(ID).append("=").append(id);
2456 // TODO handle other species variants JAL-2064
2457 StringBuilder link = new StringBuilder(32);
2463 .append("|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2464 .append(URLEncoder.encode(id, "UTF-8"));
2465 sf.addLink(link.toString());
2466 } catch (UnsupportedEncodingException e)
2471 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2472 if (clinSig != null)
2474 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2475 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2478 peptide.addSequenceFeature(sf);
2479 if (attributes.length() > 0)
2481 sf.setAttributes(attributes.toString());
2489 * Builds a map whose key is position in the protein sequence, and value is a
2490 * list of the base and all variants for each corresponding codon position
2493 * @param dnaToProtein
2496 @SuppressWarnings("unchecked")
2497 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2498 SequenceI dnaSeq, MapList dnaToProtein)
2501 * map from peptide position to all variants of the codon which codes for it
2502 * LinkedHashMap ensures we keep the peptide features in sequence order
2504 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
2506 List<SequenceFeature> dnaFeatures = dnaSeq.getFeatures()
2507 .getFeaturesByOntology(SequenceOntologyI.SEQUENCE_VARIANT);
2508 if (dnaFeatures.isEmpty())
2513 int dnaStart = dnaSeq.getStart();
2514 int[] lastCodon = null;
2515 int lastPeptidePostion = 0;
2518 * build a map of codon variations for peptides
2520 for (SequenceFeature sf : dnaFeatures)
2522 int dnaCol = sf.getBegin();
2523 if (dnaCol != sf.getEnd())
2525 // not handling multi-locus variant features
2528 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2531 // feature doesn't lie within coding region
2534 int peptidePosition = mapsTo[0];
2535 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2536 if (codonVariants == null)
2538 codonVariants = new ArrayList[CODON_LENGTH];
2539 codonVariants[0] = new ArrayList<DnaVariant>();
2540 codonVariants[1] = new ArrayList<DnaVariant>();
2541 codonVariants[2] = new ArrayList<DnaVariant>();
2542 variants.put(peptidePosition, codonVariants);
2546 * extract dna variants to a string array
2548 String alls = (String) sf.getValue("alleles");
2553 String[] alleles = alls.toUpperCase().split(",");
2555 for (String allele : alleles)
2557 alleles[i++] = allele.trim(); // lose any space characters "A, G"
2561 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2563 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2564 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2565 peptidePosition, peptidePosition));
2566 lastPeptidePostion = peptidePosition;
2570 * save nucleotide (and any variant) for each codon position
2572 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2574 String nucleotide = String.valueOf(
2575 dnaSeq.getCharAt(codon[codonPos] - dnaStart)).toUpperCase();
2576 List<DnaVariant> codonVariant = codonVariants[codonPos];
2577 if (codon[codonPos] == dnaCol)
2579 if (!codonVariant.isEmpty()
2580 && codonVariant.get(0).variant == null)
2583 * already recorded base value, add this variant
2585 codonVariant.get(0).variant = sf;
2590 * add variant with base value
2592 codonVariant.add(new DnaVariant(nucleotide, sf));
2595 else if (codonVariant.isEmpty())
2598 * record (possibly non-varying) base value
2600 codonVariant.add(new DnaVariant(nucleotide));
2608 * Makes an alignment with a copy of the given sequences, adding in any
2609 * non-redundant sequences which are mapped to by the cross-referenced
2615 * the alignment dataset shared by the new copy
2618 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2619 SequenceI[] xrefs, AlignmentI dataset)
2621 AlignmentI copy = new Alignment(new Alignment(seqs));
2622 copy.setDataset(dataset);
2623 boolean isProtein = !copy.isNucleotide();
2624 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2627 for (SequenceI xref : xrefs)
2629 DBRefEntry[] dbrefs = xref.getDBRefs();
2632 for (DBRefEntry dbref : dbrefs)
2634 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2635 || dbref.getMap().getTo().isProtein() != isProtein)
2639 SequenceI mappedTo = dbref.getMap().getTo();
2640 SequenceI match = matcher.findIdMatch(mappedTo);
2643 matcher.add(mappedTo);
2644 copy.addSequence(mappedTo);
2654 * Try to align sequences in 'unaligned' to match the alignment of their
2655 * mapped regions in 'aligned'. For example, could use this to align CDS
2656 * sequences which are mapped to their parent cDNA sequences.
2658 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2659 * dna-to-protein or protein-to-dna use alternative methods.
2662 * sequences to be aligned
2664 * holds aligned sequences and their mappings
2667 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2670 * easy case - aligning a copy of aligned sequences
2672 if (alignAsSameSequences(unaligned, aligned))
2674 return unaligned.getHeight();
2678 * fancy case - aligning via mappings between sequences
2680 List<SequenceI> unmapped = new ArrayList<SequenceI>();
2681 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2682 unaligned, aligned, unmapped);
2683 int width = columnMap.size();
2684 char gap = unaligned.getGapCharacter();
2685 int realignedCount = 0;
2686 // TODO: verify this loop scales sensibly for very wide/high alignments
2688 for (SequenceI seq : unaligned.getSequences())
2690 if (!unmapped.contains(seq))
2692 char[] newSeq = new char[width];
2693 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2694 // Integer iteration below
2699 * traverse the map to find columns populated
2702 for (Integer column : columnMap.keySet())
2704 Character c = columnMap.get(column).get(seq);
2708 * sequence has a character at this position
2718 * trim trailing gaps
2720 if (lastCol < width)
2722 char[] tmp = new char[lastCol + 1];
2723 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2726 // TODO: optimise SequenceI to avoid char[]->String->char[]
2727 seq.setSequence(String.valueOf(newSeq));
2731 return realignedCount;
2735 * If unaligned and aligned sequences share the same dataset sequences, then
2736 * simply copies the aligned sequences to the unaligned sequences and returns
2737 * true; else returns false
2740 * - sequences to be aligned based on aligned
2742 * - 'guide' alignment containing sequences derived from same dataset
2746 static boolean alignAsSameSequences(AlignmentI unaligned,
2749 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2751 return false; // should only pass alignments with datasets here
2754 // map from dataset sequence to alignment sequence(s)
2755 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<SequenceI, List<SequenceI>>();
2756 for (SequenceI seq : aligned.getSequences())
2758 SequenceI ds = seq.getDatasetSequence();
2759 if (alignedDatasets.get(ds) == null)
2761 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2763 alignedDatasets.get(ds).add(seq);
2767 * first pass - check whether all sequences to be aligned share a dataset
2768 * sequence with an aligned sequence
2770 for (SequenceI seq : unaligned.getSequences())
2772 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2779 * second pass - copy aligned sequences;
2780 * heuristic rule: pair off sequences in order for the case where
2781 * more than one shares the same dataset sequence
2783 for (SequenceI seq : unaligned.getSequences())
2785 List<SequenceI> alignedSequences = alignedDatasets.get(seq
2786 .getDatasetSequence());
2787 // TODO: getSequenceAsString() will be deprecated in the future
2788 // TODO: need to leave to SequenceI implementor to update gaps
2789 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2790 if (alignedSequences.size() > 0)
2792 // pop off aligned sequences (except the last one)
2793 alignedSequences.remove(0);
2801 * Returns a map whose key is alignment column number (base 1), and whose
2802 * values are a map of sequence characters in that column.
2809 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2810 AlignmentI unaligned, AlignmentI aligned, List<SequenceI> unmapped)
2813 * Map will hold, for each aligned column position, a map of
2814 * {unalignedSequence, characterPerSequence} at that position.
2815 * TreeMap keeps the entries in ascending column order.
2817 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<Integer, Map<SequenceI, Character>>();
2820 * record any sequences that have no mapping so can't be realigned
2822 unmapped.addAll(unaligned.getSequences());
2824 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2826 for (SequenceI seq : unaligned.getSequences())
2828 for (AlignedCodonFrame mapping : mappings)
2830 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2831 if (fromSeq != null)
2833 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2834 if (addMappedPositions(seq, fromSeq, seqMap, map))
2836 unmapped.remove(seq);
2845 * Helper method that adds to a map the mapped column positions of a sequence. <br>
2846 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2847 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2851 * the sequence whose column positions we are recording
2853 * a sequence that is mapped to the first sequence
2855 * the mapping from 'fromSeq' to 'seq'
2857 * a map to add the column positions (in fromSeq) of the mapped
2861 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2862 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2870 * invert mapping if it is from unaligned to aligned sequence
2872 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2874 seqMap = new Mapping(seq.getDatasetSequence(), seqMap.getMap()
2878 char[] fromChars = fromSeq.getSequence();
2879 int toStart = seq.getStart();
2880 char[] toChars = seq.getSequence();
2883 * traverse [start, end, start, end...] ranges in fromSeq
2885 for (int[] fromRange : seqMap.getMap().getFromRanges())
2887 for (int i = 0; i < fromRange.length - 1; i += 2)
2889 boolean forward = fromRange[i + 1] >= fromRange[i];
2892 * find the range mapped to (sequence positions base 1)
2894 int[] range = seqMap.locateMappedRange(fromRange[i],
2898 System.err.println("Error in mapping " + seqMap + " from "
2899 + fromSeq.getName());
2902 int fromCol = fromSeq.findIndex(fromRange[i]);
2903 int mappedCharPos = range[0];
2906 * walk over the 'from' aligned sequence in forward or reverse
2907 * direction; when a non-gap is found, record the column position
2908 * of the next character of the mapped-to sequence; stop when all
2909 * the characters of the range have been counted
2911 while (mappedCharPos <= range[1] && fromCol <= fromChars.length
2914 if (!Comparison.isGap(fromChars[fromCol - 1]))
2917 * mapped from sequence has a character in this column
2918 * record the column position for the mapped to character
2920 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2921 if (seqsMap == null)
2923 seqsMap = new HashMap<SequenceI, Character>();
2924 map.put(fromCol, seqsMap);
2926 seqsMap.put(seq, toChars[mappedCharPos - toStart]);
2929 fromCol += (forward ? 1 : -1);
2936 // strictly temporary hack until proper criteria for aligning protein to cds
2937 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2938 public static boolean looksLikeEnsembl(AlignmentI alignment)
2940 for (SequenceI seq : alignment.getSequences())
2942 String name = seq.getName();
2943 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))