2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.analysis;
23 import static jalview.io.gff.GffConstants.CLINICAL_SIGNIFICANCE;
25 import jalview.datamodel.AlignedCodon;
26 import jalview.datamodel.AlignedCodonFrame;
27 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
28 import jalview.datamodel.Alignment;
29 import jalview.datamodel.AlignmentAnnotation;
30 import jalview.datamodel.AlignmentI;
31 import jalview.datamodel.DBRefEntry;
32 import jalview.datamodel.GeneLociI;
33 import jalview.datamodel.IncompleteCodonException;
34 import jalview.datamodel.Mapping;
35 import jalview.datamodel.Sequence;
36 import jalview.datamodel.SequenceFeature;
37 import jalview.datamodel.SequenceGroup;
38 import jalview.datamodel.SequenceI;
39 import jalview.datamodel.features.SequenceFeatures;
40 import jalview.io.gff.Gff3Helper;
41 import jalview.io.gff.SequenceOntologyI;
42 import jalview.schemes.ResidueProperties;
43 import jalview.util.Comparison;
44 import jalview.util.DBRefUtils;
45 import jalview.util.IntRangeComparator;
46 import jalview.util.MapList;
47 import jalview.util.MappingUtils;
48 import jalview.util.StringUtils;
50 import java.io.UnsupportedEncodingException;
51 import java.net.URLEncoder;
52 import java.util.ArrayList;
53 import java.util.Arrays;
54 import java.util.Collection;
55 import java.util.Collections;
56 import java.util.HashMap;
57 import java.util.HashSet;
58 import java.util.Iterator;
59 import java.util.LinkedHashMap;
60 import java.util.List;
62 import java.util.Map.Entry;
63 import java.util.NoSuchElementException;
65 import java.util.SortedMap;
66 import java.util.TreeMap;
69 * grab bag of useful alignment manipulation operations Expect these to be
70 * refactored elsewhere at some point.
75 public class AlignmentUtils
78 private static final int CODON_LENGTH = 3;
80 private static final String SEQUENCE_VARIANT = "sequence_variant:";
82 private static final String ID = "ID";
85 * A data model to hold the 'normal' base value at a position, and an optional
86 * sequence variant feature
88 static final class DnaVariant
92 SequenceFeature variant;
94 DnaVariant(String nuc)
100 DnaVariant(String nuc, SequenceFeature var)
106 public String getSource()
108 return variant == null ? null : variant.getFeatureGroup();
112 * toString for aid in the debugger only
115 public String toString()
117 return base + ":" + (variant == null ? "" : variant.getDescription());
122 * given an existing alignment, create a new alignment including all, or up to
123 * flankSize additional symbols from each sequence's dataset sequence
129 public static AlignmentI expandContext(AlignmentI core, int flankSize)
131 List<SequenceI> sq = new ArrayList<SequenceI>();
133 for (SequenceI s : core.getSequences())
135 SequenceI newSeq = s.deriveSequence();
136 final int newSeqStart = newSeq.getStart() - 1;
137 if (newSeqStart > maxoffset
138 && newSeq.getDatasetSequence().getStart() < s.getStart())
140 maxoffset = newSeqStart;
146 maxoffset = Math.min(maxoffset, flankSize);
150 * now add offset left and right to create an expanded alignment
152 for (SequenceI s : sq)
155 while (ds.getDatasetSequence() != null)
157 ds = ds.getDatasetSequence();
159 int s_end = s.findPosition(s.getStart() + s.getLength());
160 // find available flanking residues for sequence
161 int ustream_ds = s.getStart() - ds.getStart();
162 int dstream_ds = ds.getEnd() - s_end;
164 // build new flanked sequence
166 // compute gap padding to start of flanking sequence
167 int offset = maxoffset - ustream_ds;
169 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
172 if (flankSize < ustream_ds)
174 // take up to flankSize residues
175 offset = maxoffset - flankSize;
176 ustream_ds = flankSize;
178 if (flankSize <= dstream_ds)
180 dstream_ds = flankSize - 1;
183 // TODO use Character.toLowerCase to avoid creating String objects?
184 char[] upstream = new String(ds
185 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
186 .toLowerCase().toCharArray();
187 char[] downstream = new String(
188 ds.getSequence(s_end - 1, s_end + dstream_ds)).toLowerCase()
190 char[] coreseq = s.getSequence();
191 char[] nseq = new char[offset + upstream.length + downstream.length
193 char c = core.getGapCharacter();
196 for (; p < offset; p++)
201 System.arraycopy(upstream, 0, nseq, p, upstream.length);
202 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
204 System.arraycopy(downstream, 0, nseq,
205 p + coreseq.length + upstream.length, downstream.length);
206 s.setSequence(new String(nseq));
207 s.setStart(s.getStart() - ustream_ds);
208 s.setEnd(s_end + downstream.length);
210 AlignmentI newAl = new jalview.datamodel.Alignment(
211 sq.toArray(new SequenceI[0]));
212 for (SequenceI s : sq)
214 if (s.getAnnotation() != null)
216 for (AlignmentAnnotation aa : s.getAnnotation())
218 aa.adjustForAlignment(); // JAL-1712 fix
219 newAl.addAnnotation(aa);
223 newAl.setDataset(core.getDataset());
228 * Returns the index (zero-based position) of a sequence in an alignment, or
235 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
239 for (SequenceI alSeq : al.getSequences())
252 * Returns a map of lists of sequences in the alignment, keyed by sequence
253 * name. For use in mapping between different alignment views of the same
256 * @see jalview.datamodel.AlignmentI#getSequencesByName()
258 public static Map<String, List<SequenceI>> getSequencesByName(
261 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
262 for (SequenceI seq : al.getSequences())
264 String name = seq.getName();
267 List<SequenceI> seqs = theMap.get(name);
270 seqs = new ArrayList<SequenceI>();
271 theMap.put(name, seqs);
280 * Build mapping of protein to cDNA alignment. Mappings are made between
281 * sequences where the cDNA translates to the protein sequence. Any new
282 * mappings are added to the protein alignment. Returns true if any mappings
283 * either already exist or were added, else false.
285 * @param proteinAlignment
286 * @param cdnaAlignment
289 public static boolean mapProteinAlignmentToCdna(
290 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
292 if (proteinAlignment == null || cdnaAlignment == null)
297 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
298 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
301 * First pass - map sequences where cross-references exist. This include
302 * 1-to-many mappings to support, for example, variant cDNA.
304 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
305 cdnaAlignment, mappedDna, mappedProtein, true);
308 * Second pass - map sequences where no cross-references exist. This only
309 * does 1-to-1 mappings and assumes corresponding sequences are in the same
310 * order in the alignments.
312 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
313 mappedDna, mappedProtein, false);
314 return mappingPerformed;
318 * Make mappings between compatible sequences (where the cDNA translation
319 * matches the protein).
321 * @param proteinAlignment
322 * @param cdnaAlignment
324 * a set of mapped DNA sequences (to add to)
325 * @param mappedProtein
326 * a set of mapped Protein sequences (to add to)
328 * if true, only map sequences where xrefs exist
331 protected static boolean mapProteinToCdna(
332 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
333 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
336 boolean mappingExistsOrAdded = false;
337 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
338 for (SequenceI aaSeq : thisSeqs)
340 boolean proteinMapped = false;
341 AlignedCodonFrame acf = new AlignedCodonFrame();
343 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
346 * Always try to map if sequences have xref to each other; this supports
347 * variant cDNA or alternative splicing for a protein sequence.
349 * If no xrefs, try to map progressively, assuming that alignments have
350 * mappable sequences in corresponding order. These are not
351 * many-to-many, as that would risk mixing species with similar cDNA
354 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
360 * Don't map non-xrefd sequences more than once each. This heuristic
361 * allows us to pair up similar sequences in ordered alignments.
363 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
364 || mappedDna.contains(cdnaSeq)))
368 if (mappingExists(proteinAlignment.getCodonFrames(),
369 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
371 mappingExistsOrAdded = true;
375 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
378 acf.addMap(cdnaSeq, aaSeq, map);
379 mappingExistsOrAdded = true;
380 proteinMapped = true;
381 mappedDna.add(cdnaSeq);
382 mappedProtein.add(aaSeq);
388 proteinAlignment.addCodonFrame(acf);
391 return mappingExistsOrAdded;
395 * Answers true if the mappings include one between the given (dataset)
398 protected static boolean mappingExists(List<AlignedCodonFrame> mappings,
399 SequenceI aaSeq, SequenceI cdnaSeq)
401 if (mappings != null)
403 for (AlignedCodonFrame acf : mappings)
405 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
415 * Builds a mapping (if possible) of a cDNA to a protein sequence.
417 * <li>first checks if the cdna translates exactly to the protein
419 * <li>else checks for translation after removing a STOP codon</li>
420 * <li>else checks for translation after removing a START codon</li>
421 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
423 * Returns null if no mapping is determined.
426 * the aligned protein sequence
428 * the aligned cdna sequence
431 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
435 * Here we handle either dataset sequence set (desktop) or absent (applet).
436 * Use only the char[] form of the sequence to avoid creating possibly large
439 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
440 char[] aaSeqChars = proteinDataset != null
441 ? proteinDataset.getSequence()
442 : proteinSeq.getSequence();
443 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
444 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
445 : cdnaSeq.getSequence();
446 if (aaSeqChars == null || cdnaSeqChars == null)
452 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
454 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
455 int cdnaLength = cdnaSeqChars.length;
456 int cdnaStart = cdnaSeq.getStart();
457 int cdnaEnd = cdnaSeq.getEnd();
458 final int proteinStart = proteinSeq.getStart();
459 final int proteinEnd = proteinSeq.getEnd();
462 * If lengths don't match, try ignoring stop codon (if present)
464 if (cdnaLength != mappedLength && cdnaLength > 2)
466 String lastCodon = String.valueOf(cdnaSeqChars,
467 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
468 for (String stop : ResidueProperties.STOP)
470 if (lastCodon.equals(stop))
472 cdnaEnd -= CODON_LENGTH;
473 cdnaLength -= CODON_LENGTH;
480 * If lengths still don't match, try ignoring start codon.
483 if (cdnaLength != mappedLength && cdnaLength > 2
484 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
485 .equals(ResidueProperties.START))
487 startOffset += CODON_LENGTH;
488 cdnaStart += CODON_LENGTH;
489 cdnaLength -= CODON_LENGTH;
492 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
495 * protein is translation of dna (+/- start/stop codons)
497 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
499 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
504 * translation failed - try mapping CDS annotated regions of dna
506 return mapCdsToProtein(cdnaSeq, proteinSeq);
510 * Test whether the given cdna sequence, starting at the given offset,
511 * translates to the given amino acid sequence, using the standard translation
512 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
514 * @param cdnaSeqChars
519 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
522 if (cdnaSeqChars == null || aaSeqChars == null)
528 int dnaPos = cdnaStart;
529 for (; dnaPos < cdnaSeqChars.length - 2
530 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
532 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
533 final String translated = ResidueProperties.codonTranslate(codon);
536 * allow * in protein to match untranslatable in dna
538 final char aaRes = aaSeqChars[aaPos];
539 if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
543 if (translated == null || !(aaRes == translated.charAt(0)))
546 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
547 // + codon + "(" + translated + ") != " + aaRes));
553 * check we matched all of the protein sequence
555 if (aaPos != aaSeqChars.length)
561 * check we matched all of the dna except
562 * for optional trailing STOP codon
564 if (dnaPos == cdnaSeqChars.length)
568 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
570 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
571 if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
580 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
581 * currently assumes that we are aligning cDNA to match protein.
584 * the sequence to be realigned
586 * the alignment whose sequence alignment is to be 'copied'
588 * character string represent a gap in the realigned sequence
589 * @param preserveUnmappedGaps
590 * @param preserveMappedGaps
591 * @return true if the sequence was realigned, false if it could not be
593 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
594 String gap, boolean preserveMappedGaps,
595 boolean preserveUnmappedGaps)
598 * Get any mappings from the source alignment to the target (dataset)
601 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
602 // all mappings. Would it help to constrain this?
603 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
604 if (mappings == null || mappings.isEmpty())
610 * Locate the aligned source sequence whose dataset sequence is mapped. We
611 * just take the first match here (as we can't align like more than one
614 SequenceI alignFrom = null;
615 AlignedCodonFrame mapping = null;
616 for (AlignedCodonFrame mp : mappings)
618 alignFrom = mp.findAlignedSequence(seq, al);
619 if (alignFrom != null)
626 if (alignFrom == null)
630 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
631 preserveMappedGaps, preserveUnmappedGaps);
636 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
637 * match residues and codons. Flags control whether existing gaps in unmapped
638 * (intron) and mapped (exon) regions are preserved or not. Gaps between
639 * intron and exon are only retained if both flags are set.
646 * @param preserveUnmappedGaps
647 * @param preserveMappedGaps
649 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
650 AlignedCodonFrame mapping, String myGap, char sourceGap,
651 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
653 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
655 // aligned and dataset sequence positions, all base zero
659 int basesWritten = 0;
660 char myGapChar = myGap.charAt(0);
661 int ratio = myGap.length();
663 int fromOffset = alignFrom.getStart() - 1;
664 int toOffset = alignTo.getStart() - 1;
665 int sourceGapMappedLength = 0;
666 boolean inExon = false;
667 final int toLength = alignTo.getLength();
668 final int fromLength = alignFrom.getLength();
669 StringBuilder thisAligned = new StringBuilder(2 * toLength);
672 * Traverse the 'model' aligned sequence
674 for (int i = 0; i < fromLength; i++)
676 char sourceChar = alignFrom.getCharAt(i);
677 if (sourceChar == sourceGap)
679 sourceGapMappedLength += ratio;
684 * Found a non-gap character. Locate its mapped region if any.
687 // Note mapping positions are base 1, our sequence positions base 0
688 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
689 sourceDsPos + fromOffset);
690 if (mappedPos == null)
693 * unmapped position; treat like a gap
695 sourceGapMappedLength += ratio;
696 // System.err.println("Can't align: no codon mapping to residue "
697 // + sourceDsPos + "(" + sourceChar + ")");
702 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
703 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
704 StringBuilder trailingCopiedGap = new StringBuilder();
707 * Copy dna sequence up to and including this codon. Optionally, include
708 * gaps before the codon starts (in introns) and/or after the codon starts
711 * Note this only works for 'linear' splicing, not reverse or interleaved.
712 * But then 'align dna as protein' doesn't make much sense otherwise.
714 int intronLength = 0;
715 while (basesWritten + toOffset < mappedCodonEnd
716 && thisSeqPos < toLength)
718 final char c = alignTo.getCharAt(thisSeqPos++);
722 int sourcePosition = basesWritten + toOffset;
723 if (sourcePosition < mappedCodonStart)
726 * Found an unmapped (intron) base. First add in any preceding gaps
729 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
731 thisAligned.append(trailingCopiedGap.toString());
732 intronLength += trailingCopiedGap.length();
733 trailingCopiedGap = new StringBuilder();
740 final boolean startOfCodon = sourcePosition == mappedCodonStart;
741 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
742 preserveUnmappedGaps, sourceGapMappedLength, inExon,
743 trailingCopiedGap.length(), intronLength, startOfCodon);
744 for (int k = 0; k < gapsToAdd; k++)
746 thisAligned.append(myGapChar);
748 sourceGapMappedLength = 0;
751 thisAligned.append(c);
752 trailingCopiedGap = new StringBuilder();
756 if (inExon && preserveMappedGaps)
758 trailingCopiedGap.append(myGapChar);
760 else if (!inExon && preserveUnmappedGaps)
762 trailingCopiedGap.append(myGapChar);
769 * At end of model aligned sequence. Copy any remaining target sequence, optionally
770 * including (intron) gaps.
772 while (thisSeqPos < toLength)
774 final char c = alignTo.getCharAt(thisSeqPos++);
775 if (c != myGapChar || preserveUnmappedGaps)
777 thisAligned.append(c);
779 sourceGapMappedLength--;
783 * finally add gaps to pad for any trailing source gaps or
784 * unmapped characters
786 if (preserveUnmappedGaps)
788 while (sourceGapMappedLength > 0)
790 thisAligned.append(myGapChar);
791 sourceGapMappedLength--;
796 * All done aligning, set the aligned sequence.
798 alignTo.setSequence(new String(thisAligned));
802 * Helper method to work out how many gaps to insert when realigning.
804 * @param preserveMappedGaps
805 * @param preserveUnmappedGaps
806 * @param sourceGapMappedLength
808 * @param trailingCopiedGap
809 * @param intronLength
810 * @param startOfCodon
813 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
814 boolean preserveUnmappedGaps, int sourceGapMappedLength,
815 boolean inExon, int trailingGapLength, int intronLength,
816 final boolean startOfCodon)
822 * Reached start of codon. Ignore trailing gaps in intron unless we are
823 * preserving gaps in both exon and intron. Ignore them anyway if the
824 * protein alignment introduces a gap at least as large as the intronic
827 if (inExon && !preserveMappedGaps)
829 trailingGapLength = 0;
831 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
833 trailingGapLength = 0;
837 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
841 if (intronLength + trailingGapLength <= sourceGapMappedLength)
843 gapsToAdd = sourceGapMappedLength - intronLength;
847 gapsToAdd = Math.min(
848 intronLength + trailingGapLength - sourceGapMappedLength,
856 * second or third base of codon; check for any gaps in dna
858 if (!preserveMappedGaps)
860 trailingGapLength = 0;
862 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
868 * Realigns the given protein to match the alignment of the dna, using codon
869 * mappings to translate aligned codon positions to protein residues.
872 * the alignment whose sequences are realigned by this method
874 * the dna alignment whose alignment we are 'copying'
875 * @return the number of sequences that were realigned
877 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
879 if (protein.isNucleotide() || !dna.isNucleotide())
881 System.err.println("Wrong alignment type in alignProteinAsDna");
884 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
885 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
886 protein, dna, unmappedProtein);
887 return alignProteinAs(protein, alignedCodons, unmappedProtein);
891 * Realigns the given dna to match the alignment of the protein, using codon
892 * mappings to translate aligned peptide positions to codons.
894 * Always produces a padded CDS alignment.
897 * the alignment whose sequences are realigned by this method
899 * the protein alignment whose alignment we are 'copying'
900 * @return the number of sequences that were realigned
902 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
904 if (protein.isNucleotide() || !dna.isNucleotide())
906 System.err.println("Wrong alignment type in alignProteinAsDna");
909 // todo: implement this
910 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
911 int alignedCount = 0;
912 int width = 0; // alignment width for padding CDS
913 for (SequenceI dnaSeq : dna.getSequences())
915 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
916 dna.getGapCharacter()))
920 width = Math.max(dnaSeq.getLength(), width);
924 for (SequenceI dnaSeq : dna.getSequences())
926 oldwidth = dnaSeq.getLength();
927 diff = width - oldwidth;
930 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
937 * Helper method to align (if possible) the dna sequence to match the
938 * alignment of a mapped protein sequence. This is currently limited to
939 * handling coding sequence only.
947 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
948 AlignmentI protein, List<AlignedCodonFrame> mappings,
951 SequenceI cdsDss = cdsSeq.getDatasetSequence();
955 .println("alignCdsSequenceAsProtein needs aligned sequence!");
959 List<AlignedCodonFrame> dnaMappings = MappingUtils
960 .findMappingsForSequence(cdsSeq, mappings);
961 for (AlignedCodonFrame mapping : dnaMappings)
963 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
966 final int peptideLength = peptide.getLength();
967 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
970 MapList mapList = map.getMap();
971 if (map.getTo() == peptide.getDatasetSequence())
973 mapList = mapList.getInverse();
975 final int cdsLength = cdsDss.getLength();
976 int mappedFromLength = MappingUtils.getLength(mapList
978 int mappedToLength = MappingUtils
979 .getLength(mapList.getToRanges());
980 boolean addStopCodon = (cdsLength == mappedFromLength
981 * CODON_LENGTH + CODON_LENGTH)
982 || (peptide.getDatasetSequence()
983 .getLength() == mappedFromLength - 1);
984 if (cdsLength != mappedToLength && !addStopCodon)
986 System.err.println(String.format(
987 "Can't align cds as protein (length mismatch %d/%d): %s",
988 cdsLength, mappedToLength, cdsSeq.getName()));
992 * pre-fill the aligned cds sequence with gaps
994 char[] alignedCds = new char[peptideLength * CODON_LENGTH
995 + (addStopCodon ? CODON_LENGTH : 0)];
996 Arrays.fill(alignedCds, gapChar);
999 * walk over the aligned peptide sequence and insert mapped
1000 * codons for residues in the aligned cds sequence
1002 int copiedBases = 0;
1003 int cdsStart = cdsDss.getStart();
1004 int proteinPos = peptide.getStart() - 1;
1007 for (int col = 0; col < peptideLength; col++)
1009 char residue = peptide.getCharAt(col);
1011 if (Comparison.isGap(residue))
1013 cdsCol += CODON_LENGTH;
1018 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1021 // e.g. incomplete start codon, X in peptide
1022 cdsCol += CODON_LENGTH;
1026 for (int j = codon[0]; j <= codon[1]; j++)
1028 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1029 alignedCds[cdsCol++] = mappedBase;
1037 * append stop codon if not mapped from protein,
1038 * closing it up to the end of the mapped sequence
1040 if (copiedBases == cdsLength - CODON_LENGTH)
1042 for (int i = alignedCds.length - 1; i >= 0; i--)
1044 if (!Comparison.isGap(alignedCds[i]))
1046 cdsCol = i + 1; // gap just after end of sequence
1050 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1052 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1055 cdsSeq.setSequence(new String(alignedCds));
1064 * Builds a map whose key is an aligned codon position (3 alignment column
1065 * numbers base 0), and whose value is a map from protein sequence to each
1066 * protein's peptide residue for that codon. The map generates an ordering of
1067 * the codons, and allows us to read off the peptides at each position in
1068 * order to assemble 'aligned' protein sequences.
1071 * the protein alignment
1073 * the coding dna alignment
1074 * @param unmappedProtein
1075 * any unmapped proteins are added to this list
1078 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1079 AlignmentI protein, AlignmentI dna,
1080 List<SequenceI> unmappedProtein)
1083 * maintain a list of any proteins with no mappings - these will be
1084 * rendered 'as is' in the protein alignment as we can't align them
1086 unmappedProtein.addAll(protein.getSequences());
1088 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1091 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1092 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1093 * comparator keeps the codon positions ordered.
1095 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
1096 new CodonComparator());
1098 for (SequenceI dnaSeq : dna.getSequences())
1100 for (AlignedCodonFrame mapping : mappings)
1102 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1105 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1106 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1108 unmappedProtein.remove(prot);
1114 * Finally add any unmapped peptide start residues (e.g. for incomplete
1115 * codons) as if at the codon position before the second residue
1117 // TODO resolve JAL-2022 so this fudge can be removed
1118 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1119 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1121 return alignedCodons;
1125 * Scans for any protein mapped from position 2 (meaning unmapped start
1126 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1127 * preceding position in the alignment
1129 * @param alignedCodons
1130 * the codon-to-peptide map
1131 * @param mappedSequenceCount
1132 * the number of distinct sequences in the map
1134 protected static void addUnmappedPeptideStarts(
1135 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1136 int mappedSequenceCount)
1138 // TODO delete this ugly hack once JAL-2022 is resolved
1139 // i.e. we can model startPhase > 0 (incomplete start codon)
1141 List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
1142 AlignedCodon lastCodon = null;
1143 Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
1145 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1148 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1151 SequenceI seq = sequenceCodon.getKey();
1152 if (sequencesChecked.contains(seq))
1156 sequencesChecked.add(seq);
1157 AlignedCodon codon = sequenceCodon.getValue();
1158 if (codon.peptideCol > 1)
1161 "Problem mapping protein with >1 unmapped start positions: "
1164 else if (codon.peptideCol == 1)
1167 * first position (peptideCol == 0) was unmapped - add it
1169 if (lastCodon != null)
1171 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1172 lastCodon.pos2, lastCodon.pos3,
1173 String.valueOf(seq.getCharAt(0)), 0);
1174 toAdd.put(seq, firstPeptide);
1179 * unmapped residue at start of alignment (no prior column) -
1180 * 'insert' at nominal codon [0, 0, 0]
1182 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1183 String.valueOf(seq.getCharAt(0)), 0);
1184 toAdd.put(seq, firstPeptide);
1187 if (sequencesChecked.size() == mappedSequenceCount)
1189 // no need to check past first mapped position in all sequences
1193 lastCodon = entry.getKey();
1197 * add any new codons safely after iterating over the map
1199 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1201 addCodonToMap(alignedCodons, startCodon.getValue(),
1202 startCodon.getKey());
1207 * Update the aligned protein sequences to match the codon alignments given in
1211 * @param alignedCodons
1212 * an ordered map of codon positions (columns), with sequence/peptide
1213 * values present in each column
1214 * @param unmappedProtein
1217 protected static int alignProteinAs(AlignmentI protein,
1218 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1219 List<SequenceI> unmappedProtein)
1222 * prefill peptide sequences with gaps
1224 int alignedWidth = alignedCodons.size();
1225 char[] gaps = new char[alignedWidth];
1226 Arrays.fill(gaps, protein.getGapCharacter());
1227 Map<SequenceI, char[]> peptides = new HashMap<>();
1228 for (SequenceI seq : protein.getSequences())
1230 if (!unmappedProtein.contains(seq))
1232 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1237 * Traverse the codons left to right (as defined by CodonComparator)
1238 * and insert peptides in each column where the sequence is mapped.
1239 * This gives a peptide 'alignment' where residues are aligned if their
1240 * corresponding codons occupy the same columns in the cdna alignment.
1243 for (AlignedCodon codon : alignedCodons.keySet())
1245 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1247 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1249 char residue = entry.getValue().product.charAt(0);
1250 peptides.get(entry.getKey())[column] = residue;
1256 * and finally set the constructed sequences
1258 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1260 entry.getKey().setSequence(new String(entry.getValue()));
1267 * Populate the map of aligned codons by traversing the given sequence
1268 * mapping, locating the aligned positions of mapped codons, and adding those
1269 * positions and their translation products to the map.
1272 * the aligned sequence we are mapping from
1274 * the sequence to be aligned to the codons
1276 * the gap character in the dna sequence
1278 * a mapping to a sequence translation
1279 * @param alignedCodons
1280 * the map we are building up
1282 static void addCodonPositions(SequenceI dna, SequenceI protein,
1283 char gapChar, Mapping seqMap,
1284 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1286 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1289 * add codon positions, and their peptide translations, to the alignment
1290 * map, while remembering the first codon mapped
1292 while (codons.hasNext())
1296 AlignedCodon codon = codons.next();
1297 addCodonToMap(alignedCodons, codon, protein);
1298 } catch (IncompleteCodonException e)
1300 // possible incomplete trailing codon - ignore
1301 } catch (NoSuchElementException e)
1303 // possibly peptide lacking STOP
1309 * Helper method to add a codon-to-peptide entry to the aligned codons map
1311 * @param alignedCodons
1315 protected static void addCodonToMap(
1316 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1317 AlignedCodon codon, SequenceI protein)
1319 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1320 if (seqProduct == null)
1322 seqProduct = new HashMap<SequenceI, AlignedCodon>();
1323 alignedCodons.put(codon, seqProduct);
1325 seqProduct.put(protein, codon);
1329 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1330 * between at least one pair of sequences in the two alignments. Currently,
1333 * <li>One alignment must be nucleotide, and the other protein</li>
1334 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1335 * <li>Mappable means the nucleotide translation matches the protein
1337 * <li>The translation may ignore start and stop codons if present in the
1345 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1347 if (al1 == null || al2 == null)
1353 * Require one nucleotide and one protein
1355 if (al1.isNucleotide() == al2.isNucleotide())
1359 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1360 AlignmentI protein = dna == al1 ? al2 : al1;
1361 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1362 for (SequenceI dnaSeq : dna.getSequences())
1364 for (SequenceI proteinSeq : protein.getSequences())
1366 if (isMappable(dnaSeq, proteinSeq, mappings))
1376 * Returns true if the dna sequence is mapped, or could be mapped, to the
1384 protected static boolean isMappable(SequenceI dnaSeq,
1385 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1387 if (dnaSeq == null || proteinSeq == null)
1392 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1393 : dnaSeq.getDatasetSequence();
1394 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1396 : proteinSeq.getDatasetSequence();
1398 for (AlignedCodonFrame mapping : mappings)
1400 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1410 * Just try to make a mapping (it is not yet stored), test whether
1413 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1417 * Finds any reference annotations associated with the sequences in
1418 * sequenceScope, that are not already added to the alignment, and adds them
1419 * to the 'candidates' map. Also populates a lookup table of annotation
1420 * labels, keyed by calcId, for use in constructing tooltips or the like.
1422 * @param sequenceScope
1423 * the sequences to scan for reference annotations
1424 * @param labelForCalcId
1425 * (optional) map to populate with label for calcId
1427 * map to populate with annotations for sequence
1429 * the alignment to check for presence of annotations
1431 public static void findAddableReferenceAnnotations(
1432 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1433 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1436 if (sequenceScope == null)
1442 * For each sequence in scope, make a list of any annotations on the
1443 * underlying dataset sequence which are not already on the alignment.
1445 * Add to a map of { alignmentSequence, <List of annotations to add> }
1447 for (SequenceI seq : sequenceScope)
1449 SequenceI dataset = seq.getDatasetSequence();
1450 if (dataset == null)
1454 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1455 if (datasetAnnotations == null)
1459 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1460 for (AlignmentAnnotation dsann : datasetAnnotations)
1463 * Find matching annotations on the alignment. If none is found, then
1464 * add this annotation to the list of 'addable' annotations for this
1467 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1468 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1469 if (!matchedAlignmentAnnotations.iterator().hasNext())
1472 if (labelForCalcId != null)
1474 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1479 * Save any addable annotations for this sequence
1481 if (!result.isEmpty())
1483 candidates.put(seq, result);
1489 * Adds annotations to the top of the alignment annotations, in the same order
1490 * as their related sequences.
1492 * @param annotations
1493 * the annotations to add
1495 * the alignment to add them to
1496 * @param selectionGroup
1497 * current selection group (or null if none)
1499 public static void addReferenceAnnotations(
1500 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1501 final AlignmentI alignment, final SequenceGroup selectionGroup)
1503 for (SequenceI seq : annotations.keySet())
1505 for (AlignmentAnnotation ann : annotations.get(seq))
1507 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1509 int endRes = ann.annotations.length;
1510 if (selectionGroup != null)
1512 startRes = selectionGroup.getStartRes();
1513 endRes = selectionGroup.getEndRes();
1515 copyAnn.restrict(startRes, endRes);
1518 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1519 * original annotation is already on the sequence.
1521 if (!seq.hasAnnotation(ann))
1523 seq.addAlignmentAnnotation(copyAnn);
1526 copyAnn.adjustForAlignment();
1527 // add to the alignment and set visible
1528 alignment.addAnnotation(copyAnn);
1529 copyAnn.visible = true;
1535 * Set visibility of alignment annotations of specified types (labels), for
1536 * specified sequences. This supports controls like "Show all secondary
1537 * structure", "Hide all Temp factor", etc.
1539 * @al the alignment to scan for annotations
1541 * the types (labels) of annotations to be updated
1542 * @param forSequences
1543 * if not null, only annotations linked to one of these sequences are
1544 * in scope for update; if null, acts on all sequence annotations
1546 * if this flag is true, 'types' is ignored (label not checked)
1548 * if true, set visibility on, else set off
1550 public static void showOrHideSequenceAnnotations(AlignmentI al,
1551 Collection<String> types, List<SequenceI> forSequences,
1552 boolean anyType, boolean doShow)
1554 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1557 for (AlignmentAnnotation aa : anns)
1559 if (anyType || types.contains(aa.label))
1561 if ((aa.sequenceRef != null) && (forSequences == null
1562 || forSequences.contains(aa.sequenceRef)))
1564 aa.visible = doShow;
1572 * Returns true if either sequence has a cross-reference to the other
1578 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1580 // Note: moved here from class CrossRef as the latter class has dependencies
1581 // not availability to the applet's classpath
1582 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1586 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1587 * that sequence name is structured as Source|AccessionId.
1593 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1595 if (seq1 == null || seq2 == null)
1599 String name = seq2.getName();
1600 final DBRefEntry[] xrefs = seq1.getDBRefs();
1603 for (DBRefEntry xref : xrefs)
1605 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1606 // case-insensitive test, consistent with DBRefEntry.equalRef()
1607 if (xrefName.equalsIgnoreCase(name))
1617 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1618 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1619 * added to the original alignment's dataset, which is shared by the new
1620 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1621 * added to the alignment dataset.
1624 * aligned nucleotide (dna or cds) sequences
1626 * the alignment dataset the sequences belong to
1628 * (optional) to restrict results to CDS that map to specified
1630 * @return an alignment whose sequences are the cds-only parts of the dna
1631 * sequences (or null if no mappings are found)
1633 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1634 AlignmentI dataset, SequenceI[] products)
1636 if (dataset == null || dataset.getDataset() != null)
1638 throw new IllegalArgumentException(
1639 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1641 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
1642 List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
1643 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1644 HashSet<SequenceI> productSeqs = null;
1645 if (products != null)
1647 productSeqs = new HashSet<SequenceI>();
1648 for (SequenceI seq : products)
1650 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1651 .getDatasetSequence());
1656 * Construct CDS sequences from mappings on the alignment dataset.
1658 * - find the protein product(s) mapped to from each dna sequence
1659 * - if the mapping covers the whole dna sequence (give or take start/stop
1660 * codon), take the dna as the CDS sequence
1661 * - else search dataset mappings for a suitable dna sequence, i.e. one
1662 * whose whole sequence is mapped to the protein
1663 * - if no sequence found, construct one from the dna sequence and mapping
1664 * (and add it to dataset so it is found if this is repeated)
1666 for (SequenceI dnaSeq : dna)
1668 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1669 : dnaSeq.getDatasetSequence();
1671 List<AlignedCodonFrame> seqMappings = MappingUtils
1672 .findMappingsForSequence(dnaSeq, mappings);
1673 for (AlignedCodonFrame mapping : seqMappings)
1675 List<Mapping> mappingsFromSequence = mapping
1676 .getMappingsFromSequence(dnaSeq);
1678 for (Mapping aMapping : mappingsFromSequence)
1680 MapList mapList = aMapping.getMap();
1681 if (mapList.getFromRatio() == 1)
1684 * not a dna-to-protein mapping (likely dna-to-cds)
1690 * skip if mapping is not to one of the target set of proteins
1692 SequenceI proteinProduct = aMapping.getTo();
1693 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1699 * try to locate the CDS from the dataset mappings;
1700 * guard against duplicate results (for the case that protein has
1701 * dbrefs to both dna and cds sequences)
1703 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1704 seqMappings, aMapping);
1707 if (!foundSeqs.contains(cdsSeq))
1709 foundSeqs.add(cdsSeq);
1710 SequenceI derivedSequence = cdsSeq.deriveSequence();
1711 cdsSeqs.add(derivedSequence);
1712 if (!dataset.getSequences().contains(cdsSeq))
1714 dataset.addSequence(cdsSeq);
1721 * didn't find mapped CDS sequence - construct it and add
1722 * its dataset sequence to the dataset
1724 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1725 dataset).deriveSequence();
1726 // cdsSeq has a name constructed as CDS|<dbref>
1727 // <dbref> will be either the accession for the coding sequence,
1728 // marked in the /via/ dbref to the protein product accession
1729 // or it will be the original nucleotide accession.
1730 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1732 cdsSeqs.add(cdsSeq);
1734 if (!dataset.getSequences().contains(cdsSeqDss))
1736 // check if this sequence is a newly created one
1737 // so needs adding to the dataset
1738 dataset.addSequence(cdsSeqDss);
1742 * add a mapping from CDS to the (unchanged) mapped to range
1744 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1745 cdsSeq.getLength() });
1746 MapList cdsToProteinMap = new MapList(cdsRange,
1747 mapList.getToRanges(), mapList.getFromRatio(),
1748 mapList.getToRatio());
1749 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1750 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1754 * guard against duplicating the mapping if repeating this action
1756 if (!mappings.contains(cdsToProteinMapping))
1758 mappings.add(cdsToProteinMapping);
1761 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1762 proteinProduct, aMapping);
1764 * add another mapping from original 'from' range to CDS
1766 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1767 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1769 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1771 if (!mappings.contains(dnaToCdsMapping))
1773 mappings.add(dnaToCdsMapping);
1777 * transfer dna chromosomal loci (if known) to the CDS
1778 * sequence (via the mapping)
1780 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1781 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1784 * add DBRef with mapping from protein to CDS
1785 * (this enables Get Cross-References from protein alignment)
1786 * This is tricky because we can't have two DBRefs with the
1787 * same source and accession, so need a different accession for
1788 * the CDS from the dna sequence
1791 // specific use case:
1792 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1793 // ENSG02, ENSG03, with transcripts and products similarly named.
1794 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1796 // JBPNote: ?? can't actually create an example that demonstrates we
1798 // synthesize an xref.
1800 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1803 * create a cross-reference from CDS to the source sequence's
1804 * primary reference and vice versa
1806 String source = primRef.getSource();
1807 String version = primRef.getVersion();
1808 DBRefEntry cdsCrossRef = new DBRefEntry(source, source + ":"
1809 + version, primRef.getAccessionId());
1810 cdsCrossRef.setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
1811 cdsSeqDss.addDBRef(cdsCrossRef);
1813 dnaSeq.addDBRef(new DBRefEntry(source, version, cdsSeq
1814 .getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
1816 // problem here is that the cross-reference is synthesized -
1817 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1819 // assuming cds version same as dna ?!?
1821 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
1824 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1826 proteinProduct.addDBRef(proteinToCdsRef);
1830 * transfer any features on dna that overlap the CDS
1832 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1833 SequenceOntologyI.CDS);
1838 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1840 cds.setDataset(dataset);
1846 * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
1847 * toSeq, mediated by the given mapping between the sequences
1850 * @param targetToFrom
1854 protected static void transferGeneLoci(SequenceI fromSeq,
1855 MapList targetToFrom, SequenceI targetSeq)
1857 if (targetSeq.getGeneLoci() != null)
1859 // already have - don't override
1862 GeneLociI fromLoci = fromSeq.getGeneLoci();
1863 if (fromLoci == null)
1868 MapList newMap = targetToFrom.traverse(fromLoci.getMap());
1872 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
1873 fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
1878 * A helper method that finds a CDS sequence in the alignment dataset that is
1879 * mapped to the given protein sequence, and either is, or has a mapping from,
1880 * the given dna sequence.
1883 * set of all mappings on the dataset
1885 * a dna (or cds) sequence we are searching from
1886 * @param seqMappings
1887 * the set of mappings involving dnaSeq
1889 * an initial candidate from seqMappings
1892 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1893 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1897 * TODO a better dna-cds-protein mapping data representation to allow easy
1898 * navigation; until then this clunky looping around lists of mappings
1900 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1901 : dnaSeq.getDatasetSequence();
1902 SequenceI proteinProduct = aMapping.getTo();
1905 * is this mapping from the whole dna sequence (i.e. CDS)?
1906 * allowing for possible stop codon on dna but not peptide
1908 int mappedFromLength = MappingUtils
1909 .getLength(aMapping.getMap().getFromRanges());
1910 int dnaLength = seqDss.getLength();
1911 if (mappedFromLength == dnaLength
1912 || mappedFromLength == dnaLength - CODON_LENGTH)
1918 * looks like we found the dna-to-protein mapping; search for the
1919 * corresponding cds-to-protein mapping
1921 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1922 .findMappingsForSequence(proteinProduct, mappings);
1923 for (AlignedCodonFrame acf : mappingsToPeptide)
1925 for (SequenceToSequenceMapping map : acf.getMappings())
1927 Mapping mapping = map.getMapping();
1928 if (mapping != aMapping
1929 && mapping.getMap().getFromRatio() == CODON_LENGTH
1930 && proteinProduct == mapping.getTo()
1931 && seqDss != map.getFromSeq())
1933 mappedFromLength = MappingUtils
1934 .getLength(mapping.getMap().getFromRanges());
1935 if (mappedFromLength == map.getFromSeq().getLength())
1938 * found a 3:1 mapping to the protein product which covers
1939 * the whole dna sequence i.e. is from CDS; finally check it
1940 * is from the dna start sequence
1942 SequenceI cdsSeq = map.getFromSeq();
1943 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1944 .findMappingsForSequence(cdsSeq, seqMappings);
1945 if (!dnaToCdsMaps.isEmpty())
1957 * Helper method that makes a CDS sequence as defined by the mappings from the
1958 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1959 * forward or reverse strand).
1964 * - existing dataset. We check for sequences that look like the CDS
1965 * we are about to construct, if one exists already, then we will
1966 * just return that one.
1967 * @return CDS sequence (as a dataset sequence)
1969 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1972 char[] seqChars = seq.getSequence();
1973 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1974 int cdsWidth = MappingUtils.getLength(fromRanges);
1975 char[] newSeqChars = new char[cdsWidth];
1978 for (int[] range : fromRanges)
1980 if (range[0] <= range[1])
1982 // forward strand mapping - just copy the range
1983 int length = range[1] - range[0] + 1;
1984 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1990 // reverse strand mapping - copy and complement one by one
1991 for (int i = range[0]; i >= range[1]; i--)
1993 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1999 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
2000 * else generate a sequence name
2002 String mapFromId = mapping.getMappedFromId();
2003 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
2004 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2005 if (dataset != null)
2007 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2008 if (matches != null)
2010 boolean matched = false;
2011 for (SequenceI mtch : matches)
2013 if (mtch.getStart() != newSeq.getStart())
2017 if (mtch.getEnd() != newSeq.getEnd())
2021 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2033 "JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
2039 // newSeq.setDescription(mapFromId);
2045 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2046 * the given mapping.
2050 * @param proteinProduct
2052 * @return list of DBRefEntrys added
2054 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2055 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2057 // gather direct refs from contig congruent with mapping
2058 List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
2059 HashSet<String> directSources = new HashSet<String>();
2060 if (contig.getDBRefs() != null)
2062 for (DBRefEntry dbr : contig.getDBRefs())
2064 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
2066 MapList map = dbr.getMap().getMap();
2067 // check if map is the CDS mapping
2068 if (mapping.getMap().equals(map))
2071 directSources.add(dbr.getSource());
2076 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2077 proteinProduct.getDBRefs(),
2078 directSources.toArray(new String[0]));
2079 List<DBRefEntry> propagated = new ArrayList<DBRefEntry>();
2081 // and generate appropriate mappings
2082 for (DBRefEntry cdsref : direct)
2084 // clone maplist and mapping
2085 MapList cdsposmap = new MapList(
2086 Arrays.asList(new int[][]
2087 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2088 cdsref.getMap().getMap().getToRanges(), 3, 1);
2089 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(),
2090 cdsref.getMap().getMap());
2093 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2094 cdsref.getVersion(), cdsref.getAccessionId(),
2095 new Mapping(cdsmap.getTo(), cdsposmap));
2097 // and see if we can map to the protein product for this mapping.
2098 // onSource is the filtered set of accessions on protein that we are
2099 // tranferring, so we assume accession is the same.
2100 if (cdsmap.getTo() == null && onSource != null)
2102 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2103 cdsref.getAccessionId());
2104 if (sourceRefs != null)
2106 for (DBRefEntry srcref : sourceRefs)
2108 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2110 // we have found a complementary dbref on the protein product, so
2111 // update mapping's getTo
2112 newref.getMap().setTo(proteinProduct);
2117 cdsSeq.addDBRef(newref);
2118 propagated.add(newref);
2124 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2125 * feature start/end ranges, optionally omitting specified feature types.
2126 * Returns the number of features copied.
2131 * the mapping from 'fromSeq' to 'toSeq'
2133 * if not null, only features of this type are copied (including
2134 * subtypes in the Sequence Ontology)
2137 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2138 MapList mapping, String select, String... omitting)
2140 SequenceI copyTo = toSeq;
2141 while (copyTo.getDatasetSequence() != null)
2143 copyTo = copyTo.getDatasetSequence();
2147 * get features, optionally restricted by an ontology term
2149 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2150 .getPositionalFeatures() : fromSeq.getFeatures()
2151 .getFeaturesByOntology(select);
2154 for (SequenceFeature sf : sfs)
2156 String type = sf.getType();
2157 boolean omit = false;
2158 for (String toOmit : omitting)
2160 if (type.equals(toOmit))
2171 * locate the mapped range - null if either start or end is
2172 * not mapped (no partial overlaps are calculated)
2174 int start = sf.getBegin();
2175 int end = sf.getEnd();
2176 int[] mappedTo = mapping.locateInTo(start, end);
2178 * if whole exon range doesn't map, try interpreting it
2179 * as 5' or 3' exon overlapping the CDS range
2181 if (mappedTo == null)
2183 mappedTo = mapping.locateInTo(end, end);
2184 if (mappedTo != null)
2187 * end of exon is in CDS range - 5' overlap
2188 * to a range from the start of the peptide
2193 if (mappedTo == null)
2195 mappedTo = mapping.locateInTo(start, start);
2196 if (mappedTo != null)
2199 * start of exon is in CDS range - 3' overlap
2200 * to a range up to the end of the peptide
2202 mappedTo[1] = toSeq.getLength();
2205 if (mappedTo != null)
2207 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2208 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2209 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2210 sf.getFeatureGroup(), sf.getScore());
2211 copyTo.addSequenceFeature(copy);
2219 * Returns a mapping from dna to protein by inspecting sequence features of
2220 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2221 * length is 3 times the peptide length (optionally after dropping a trailing
2222 * stop codon). This method does not check whether the CDS nucleotide sequence
2223 * translates to the peptide sequence.
2229 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2230 SequenceI proteinSeq)
2232 List<int[]> ranges = findCdsPositions(dnaSeq);
2233 int mappedDnaLength = MappingUtils.getLength(ranges);
2236 * if not a whole number of codons, something is wrong,
2239 if (mappedDnaLength % CODON_LENGTH > 0)
2244 int proteinLength = proteinSeq.getLength();
2245 int proteinStart = proteinSeq.getStart();
2246 int proteinEnd = proteinSeq.getEnd();
2249 * incomplete start codon may mean X at start of peptide
2250 * we ignore both for mapping purposes
2252 if (proteinSeq.getCharAt(0) == 'X')
2254 // todo JAL-2022 support startPhase > 0
2258 List<int[]> proteinRange = new ArrayList<int[]>();
2261 * dna length should map to protein (or protein plus stop codon)
2263 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2264 if (codesForResidues == (proteinLength + 1))
2266 // assuming extra codon is for STOP and not in peptide
2267 // todo: check trailing codon is indeed a STOP codon
2269 mappedDnaLength -= CODON_LENGTH;
2270 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2273 if (codesForResidues == proteinLength)
2275 proteinRange.add(new int[] { proteinStart, proteinEnd });
2276 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2282 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2283 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2284 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2285 * position order, so this method is only valid for linear CDS in the same
2286 * sense as the protein product.
2291 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2293 List<int[]> result = new ArrayList<int[]>();
2295 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2296 SequenceOntologyI.CDS);
2301 SequenceFeatures.sortFeatures(sfs, true);
2303 for (SequenceFeature sf : sfs)
2308 phase = Integer.parseInt(sf.getPhase());
2309 } catch (NumberFormatException e)
2314 * phase > 0 on first codon means 5' incomplete - skip to the start
2315 * of the next codon; example ENST00000496384
2317 int begin = sf.getBegin();
2318 int end = sf.getEnd();
2319 if (result.isEmpty() && phase > 0)
2324 // shouldn't happen!
2326 .println("Error: start phase extends beyond start CDS in "
2327 + dnaSeq.getName());
2330 result.add(new int[] { begin, end });
2334 * Finally sort ranges by start position. This avoids a dependency on
2335 * keeping features in order on the sequence (if they are in order anyway,
2336 * the sort will have almost no work to do). The implicit assumption is CDS
2337 * ranges are assembled in order. Other cases should not use this method,
2338 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2340 Collections.sort(result, IntRangeComparator.ASCENDING);
2345 * Maps exon features from dna to protein, and computes variants in peptide
2346 * product generated by variants in dna, and adds them as sequence_variant
2347 * features on the protein sequence. Returns the number of variant features
2352 * @param dnaToProtein
2354 public static int computeProteinFeatures(SequenceI dnaSeq,
2355 SequenceI peptide, MapList dnaToProtein)
2357 while (dnaSeq.getDatasetSequence() != null)
2359 dnaSeq = dnaSeq.getDatasetSequence();
2361 while (peptide.getDatasetSequence() != null)
2363 peptide = peptide.getDatasetSequence();
2366 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2369 * compute protein variants from dna variants and codon mappings;
2370 * NB - alternatively we could retrieve this using the REST service e.g.
2371 * http://rest.ensembl.org/overlap/translation
2372 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2373 * which would be a bit slower but possibly more reliable
2377 * build a map with codon variations for each potentially varying peptide
2379 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2380 dnaSeq, dnaToProtein);
2383 * scan codon variations, compute peptide variants and add to peptide sequence
2386 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2388 int peptidePos = variant.getKey();
2389 List<DnaVariant>[] codonVariants = variant.getValue();
2390 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2397 * Computes non-synonymous peptide variants from codon variants and adds them
2398 * as sequence_variant features on the protein sequence (one feature per
2399 * allele variant). Selected attributes (variant id, clinical significance)
2400 * are copied over to the new features.
2403 * the protein sequence
2405 * the position to compute peptide variants for
2406 * @param codonVariants
2407 * a list of dna variants per codon position
2408 * @return the number of features added
2410 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2411 List<DnaVariant>[] codonVariants)
2413 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2415 String base1 = codonVariants[0].get(0).base;
2416 String base2 = codonVariants[1].get(0).base;
2417 String base3 = codonVariants[2].get(0).base;
2420 * variants in first codon base
2422 for (DnaVariant var : codonVariants[0])
2424 if (var.variant != null)
2426 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2427 if (alleles != null)
2429 for (String base : alleles.split(","))
2431 String codon = base + base2 + base3;
2432 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2442 * variants in second codon base
2444 for (DnaVariant var : codonVariants[1])
2446 if (var.variant != null)
2448 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2449 if (alleles != null)
2451 for (String base : alleles.split(","))
2453 String codon = base1 + base + base3;
2454 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2464 * variants in third codon base
2466 for (DnaVariant var : codonVariants[2])
2468 if (var.variant != null)
2470 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2471 if (alleles != null)
2473 for (String base : alleles.split(","))
2475 String codon = base1 + base2 + base;
2476 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2489 * Helper method that adds a peptide variant feature, provided the given codon
2490 * translates to a value different to the current residue (is a non-synonymous
2491 * variant). ID and clinical_significance attributes of the dna variant (if
2492 * present) are copied to the new feature.
2499 * @return true if a feature was added, else false
2501 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2502 String residue, DnaVariant var, String codon)
2505 * get peptide translation of codon e.g. GAT -> D
2506 * note that variants which are not single alleles,
2507 * e.g. multibase variants or HGMD_MUTATION etc
2508 * are currently ignored here
2510 String trans = codon.contains("-") ? "-"
2511 : (codon.length() > CODON_LENGTH ? null
2512 : ResidueProperties.codonTranslate(codon));
2513 if (trans != null && !trans.equals(residue))
2515 String residue3Char = StringUtils
2516 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2517 String trans3Char = StringUtils
2518 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2519 String desc = "p." + residue3Char + peptidePos + trans3Char;
2520 SequenceFeature sf = new SequenceFeature(
2521 SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
2522 peptidePos, var.getSource());
2523 StringBuilder attributes = new StringBuilder(32);
2524 String id = (String) var.variant.getValue(ID);
2527 if (id.startsWith(SEQUENCE_VARIANT))
2529 id = id.substring(SEQUENCE_VARIANT.length());
2531 sf.setValue(ID, id);
2532 attributes.append(ID).append("=").append(id);
2533 // TODO handle other species variants JAL-2064
2534 StringBuilder link = new StringBuilder(32);
2537 link.append(desc).append(" ").append(id).append(
2538 "|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2539 .append(URLEncoder.encode(id, "UTF-8"));
2540 sf.addLink(link.toString());
2541 } catch (UnsupportedEncodingException e)
2546 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2547 if (clinSig != null)
2549 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2550 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2553 peptide.addSequenceFeature(sf);
2554 if (attributes.length() > 0)
2556 sf.setAttributes(attributes.toString());
2564 * Builds a map whose key is position in the protein sequence, and value is a
2565 * list of the base and all variants for each corresponding codon position.
2567 * This depends on dna variants being held as a comma-separated list as
2568 * property "alleles" on variant features.
2571 * @param dnaToProtein
2574 @SuppressWarnings("unchecked")
2575 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2576 SequenceI dnaSeq, MapList dnaToProtein)
2579 * map from peptide position to all variants of the codon which codes for it
2580 * LinkedHashMap ensures we keep the peptide features in sequence order
2582 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
2584 List<SequenceFeature> dnaFeatures = dnaSeq.getFeatures()
2585 .getFeaturesByOntology(SequenceOntologyI.SEQUENCE_VARIANT);
2586 if (dnaFeatures.isEmpty())
2591 int dnaStart = dnaSeq.getStart();
2592 int[] lastCodon = null;
2593 int lastPeptidePostion = 0;
2596 * build a map of codon variations for peptides
2598 for (SequenceFeature sf : dnaFeatures)
2600 int dnaCol = sf.getBegin();
2601 if (dnaCol != sf.getEnd())
2603 // not handling multi-locus variant features
2608 * ignore variant if not a SNP
2610 String alls = (String) sf.getValue(Gff3Helper.ALLELES);
2613 continue; // non-SNP VCF variant perhaps - can't process this
2616 String[] alleles = alls.toUpperCase().split(",");
2617 boolean isSnp = true;
2618 for (String allele : alleles)
2620 if (allele.trim().length() > 1)
2630 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2633 // feature doesn't lie within coding region
2636 int peptidePosition = mapsTo[0];
2637 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2638 if (codonVariants == null)
2640 codonVariants = new ArrayList[CODON_LENGTH];
2641 codonVariants[0] = new ArrayList<DnaVariant>();
2642 codonVariants[1] = new ArrayList<DnaVariant>();
2643 codonVariants[2] = new ArrayList<DnaVariant>();
2644 variants.put(peptidePosition, codonVariants);
2648 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2650 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2651 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2652 peptidePosition, peptidePosition));
2653 lastPeptidePostion = peptidePosition;
2657 * save nucleotide (and any variant) for each codon position
2659 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2661 String nucleotide = String.valueOf(
2662 dnaSeq.getCharAt(codon[codonPos] - dnaStart)).toUpperCase();
2663 List<DnaVariant> codonVariant = codonVariants[codonPos];
2664 if (codon[codonPos] == dnaCol)
2666 if (!codonVariant.isEmpty()
2667 && codonVariant.get(0).variant == null)
2670 * already recorded base value, add this variant
2672 codonVariant.get(0).variant = sf;
2677 * add variant with base value
2679 codonVariant.add(new DnaVariant(nucleotide, sf));
2682 else if (codonVariant.isEmpty())
2685 * record (possibly non-varying) base value
2687 codonVariant.add(new DnaVariant(nucleotide));
2695 * Makes an alignment with a copy of the given sequences, adding in any
2696 * non-redundant sequences which are mapped to by the cross-referenced
2702 * the alignment dataset shared by the new copy
2705 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2706 SequenceI[] xrefs, AlignmentI dataset)
2708 AlignmentI copy = new Alignment(new Alignment(seqs));
2709 copy.setDataset(dataset);
2710 boolean isProtein = !copy.isNucleotide();
2711 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2714 for (SequenceI xref : xrefs)
2716 DBRefEntry[] dbrefs = xref.getDBRefs();
2719 for (DBRefEntry dbref : dbrefs)
2721 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2722 || dbref.getMap().getTo().isProtein() != isProtein)
2726 SequenceI mappedTo = dbref.getMap().getTo();
2727 SequenceI match = matcher.findIdMatch(mappedTo);
2730 matcher.add(mappedTo);
2731 copy.addSequence(mappedTo);
2741 * Try to align sequences in 'unaligned' to match the alignment of their
2742 * mapped regions in 'aligned'. For example, could use this to align CDS
2743 * sequences which are mapped to their parent cDNA sequences.
2745 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2746 * dna-to-protein or protein-to-dna use alternative methods.
2749 * sequences to be aligned
2751 * holds aligned sequences and their mappings
2754 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2757 * easy case - aligning a copy of aligned sequences
2759 if (alignAsSameSequences(unaligned, aligned))
2761 return unaligned.getHeight();
2765 * fancy case - aligning via mappings between sequences
2767 List<SequenceI> unmapped = new ArrayList<SequenceI>();
2768 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2769 unaligned, aligned, unmapped);
2770 int width = columnMap.size();
2771 char gap = unaligned.getGapCharacter();
2772 int realignedCount = 0;
2773 // TODO: verify this loop scales sensibly for very wide/high alignments
2775 for (SequenceI seq : unaligned.getSequences())
2777 if (!unmapped.contains(seq))
2779 char[] newSeq = new char[width];
2780 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2781 // Integer iteration below
2786 * traverse the map to find columns populated
2789 for (Integer column : columnMap.keySet())
2791 Character c = columnMap.get(column).get(seq);
2795 * sequence has a character at this position
2805 * trim trailing gaps
2807 if (lastCol < width)
2809 char[] tmp = new char[lastCol + 1];
2810 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2813 // TODO: optimise SequenceI to avoid char[]->String->char[]
2814 seq.setSequence(String.valueOf(newSeq));
2818 return realignedCount;
2822 * If unaligned and aligned sequences share the same dataset sequences, then
2823 * simply copies the aligned sequences to the unaligned sequences and returns
2824 * true; else returns false
2827 * - sequences to be aligned based on aligned
2829 * - 'guide' alignment containing sequences derived from same dataset
2833 static boolean alignAsSameSequences(AlignmentI unaligned,
2836 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2838 return false; // should only pass alignments with datasets here
2841 // map from dataset sequence to alignment sequence(s)
2842 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<SequenceI, List<SequenceI>>();
2843 for (SequenceI seq : aligned.getSequences())
2845 SequenceI ds = seq.getDatasetSequence();
2846 if (alignedDatasets.get(ds) == null)
2848 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2850 alignedDatasets.get(ds).add(seq);
2854 * first pass - check whether all sequences to be aligned share a dataset
2855 * sequence with an aligned sequence
2857 for (SequenceI seq : unaligned.getSequences())
2859 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2866 * second pass - copy aligned sequences;
2867 * heuristic rule: pair off sequences in order for the case where
2868 * more than one shares the same dataset sequence
2870 for (SequenceI seq : unaligned.getSequences())
2872 List<SequenceI> alignedSequences = alignedDatasets
2873 .get(seq.getDatasetSequence());
2874 // TODO: getSequenceAsString() will be deprecated in the future
2875 // TODO: need to leave to SequenceI implementor to update gaps
2876 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2877 if (alignedSequences.size() > 0)
2879 // pop off aligned sequences (except the last one)
2880 alignedSequences.remove(0);
2888 * Returns a map whose key is alignment column number (base 1), and whose
2889 * values are a map of sequence characters in that column.
2896 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2897 AlignmentI unaligned, AlignmentI aligned,
2898 List<SequenceI> unmapped)
2901 * Map will hold, for each aligned column position, a map of
2902 * {unalignedSequence, characterPerSequence} at that position.
2903 * TreeMap keeps the entries in ascending column order.
2905 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<Integer, Map<SequenceI, Character>>();
2908 * record any sequences that have no mapping so can't be realigned
2910 unmapped.addAll(unaligned.getSequences());
2912 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2914 for (SequenceI seq : unaligned.getSequences())
2916 for (AlignedCodonFrame mapping : mappings)
2918 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2919 if (fromSeq != null)
2921 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2922 if (addMappedPositions(seq, fromSeq, seqMap, map))
2924 unmapped.remove(seq);
2933 * Helper method that adds to a map the mapped column positions of a sequence.
2935 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2936 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2940 * the sequence whose column positions we are recording
2942 * a sequence that is mapped to the first sequence
2944 * the mapping from 'fromSeq' to 'seq'
2946 * a map to add the column positions (in fromSeq) of the mapped
2950 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2951 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2959 * invert mapping if it is from unaligned to aligned sequence
2961 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2963 seqMap = new Mapping(seq.getDatasetSequence(),
2964 seqMap.getMap().getInverse());
2967 int toStart = seq.getStart();
2970 * traverse [start, end, start, end...] ranges in fromSeq
2972 for (int[] fromRange : seqMap.getMap().getFromRanges())
2974 for (int i = 0; i < fromRange.length - 1; i += 2)
2976 boolean forward = fromRange[i + 1] >= fromRange[i];
2979 * find the range mapped to (sequence positions base 1)
2981 int[] range = seqMap.locateMappedRange(fromRange[i],
2985 System.err.println("Error in mapping " + seqMap + " from "
2986 + fromSeq.getName());
2989 int fromCol = fromSeq.findIndex(fromRange[i]);
2990 int mappedCharPos = range[0];
2993 * walk over the 'from' aligned sequence in forward or reverse
2994 * direction; when a non-gap is found, record the column position
2995 * of the next character of the mapped-to sequence; stop when all
2996 * the characters of the range have been counted
2998 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
3001 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
3004 * mapped from sequence has a character in this column
3005 * record the column position for the mapped to character
3007 Map<SequenceI, Character> seqsMap = map.get(fromCol);
3008 if (seqsMap == null)
3010 seqsMap = new HashMap<SequenceI, Character>();
3011 map.put(fromCol, seqsMap);
3013 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
3016 fromCol += (forward ? 1 : -1);
3023 // strictly temporary hack until proper criteria for aligning protein to cds
3024 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
3025 public static boolean looksLikeEnsembl(AlignmentI alignment)
3027 for (SequenceI seq : alignment.getSequences())
3029 String name = seq.getName();
3030 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))