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.io.gff.SequenceOntologyFactory;
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.MapList;
44 import jalview.util.MappingUtils;
45 import jalview.util.RangeComparator;
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.Comparator;
55 import java.util.HashMap;
56 import java.util.HashSet;
57 import java.util.Iterator;
58 import java.util.LinkedHashMap;
59 import java.util.List;
61 import java.util.Map.Entry;
62 import java.util.NoSuchElementException;
64 import java.util.SortedMap;
65 import java.util.TreeMap;
68 * grab bag of useful alignment manipulation operations Expect these to be
69 * refactored elsewhere at some point.
74 public class AlignmentUtils
77 private static final int CODON_LENGTH = 3;
79 private static final String SEQUENCE_VARIANT = "sequence_variant:";
81 private static final String ID = "ID";
84 * A data model to hold the 'normal' base value at a position, and an optional
85 * sequence variant feature
87 static final class DnaVariant
91 SequenceFeature variant;
93 DnaVariant(String nuc)
99 DnaVariant(String nuc, SequenceFeature var)
105 public String getSource()
107 return variant == null ? null : variant.getFeatureGroup();
112 * given an existing alignment, create a new alignment including all, or up to
113 * flankSize additional symbols from each sequence's dataset sequence
119 public static AlignmentI expandContext(AlignmentI core, int flankSize)
121 List<SequenceI> sq = new ArrayList<SequenceI>();
123 for (SequenceI s : core.getSequences())
125 SequenceI newSeq = s.deriveSequence();
126 final int newSeqStart = newSeq.getStart() - 1;
127 if (newSeqStart > maxoffset
128 && newSeq.getDatasetSequence().getStart() < s.getStart())
130 maxoffset = newSeqStart;
136 maxoffset = Math.min(maxoffset, flankSize);
140 * now add offset left and right to create an expanded alignment
142 for (SequenceI s : sq)
145 while (ds.getDatasetSequence() != null)
147 ds = ds.getDatasetSequence();
149 int s_end = s.findPosition(s.getStart() + s.getLength());
150 // find available flanking residues for sequence
151 int ustream_ds = s.getStart() - ds.getStart();
152 int dstream_ds = ds.getEnd() - s_end;
154 // build new flanked sequence
156 // compute gap padding to start of flanking sequence
157 int offset = maxoffset - ustream_ds;
159 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
162 if (flankSize < ustream_ds)
164 // take up to flankSize residues
165 offset = maxoffset - flankSize;
166 ustream_ds = flankSize;
168 if (flankSize <= dstream_ds)
170 dstream_ds = flankSize - 1;
173 // TODO use Character.toLowerCase to avoid creating String objects?
174 char[] upstream = new String(ds.getSequence(s.getStart() - 1
175 - ustream_ds, s.getStart() - 1)).toLowerCase().toCharArray();
176 char[] downstream = new String(ds.getSequence(s_end - 1, s_end
177 + dstream_ds)).toLowerCase().toCharArray();
178 char[] coreseq = s.getSequence();
179 char[] nseq = new char[offset + upstream.length + downstream.length
181 char c = core.getGapCharacter();
184 for (; p < offset; p++)
189 System.arraycopy(upstream, 0, nseq, p, upstream.length);
190 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
192 System.arraycopy(downstream, 0, nseq, p + coreseq.length
193 + upstream.length, downstream.length);
194 s.setSequence(new String(nseq));
195 s.setStart(s.getStart() - ustream_ds);
196 s.setEnd(s_end + downstream.length);
198 AlignmentI newAl = new jalview.datamodel.Alignment(
199 sq.toArray(new SequenceI[0]));
200 for (SequenceI s : sq)
202 if (s.getAnnotation() != null)
204 for (AlignmentAnnotation aa : s.getAnnotation())
206 aa.adjustForAlignment(); // JAL-1712 fix
207 newAl.addAnnotation(aa);
211 newAl.setDataset(core.getDataset());
216 * Returns the index (zero-based position) of a sequence in an alignment, or
223 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
227 for (SequenceI alSeq : al.getSequences())
240 * Returns a map of lists of sequences in the alignment, keyed by sequence
241 * name. For use in mapping between different alignment views of the same
244 * @see jalview.datamodel.AlignmentI#getSequencesByName()
246 public static Map<String, List<SequenceI>> getSequencesByName(
249 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
250 for (SequenceI seq : al.getSequences())
252 String name = seq.getName();
255 List<SequenceI> seqs = theMap.get(name);
258 seqs = new ArrayList<SequenceI>();
259 theMap.put(name, seqs);
268 * Build mapping of protein to cDNA alignment. Mappings are made between
269 * sequences where the cDNA translates to the protein sequence. Any new
270 * mappings are added to the protein alignment. Returns true if any mappings
271 * either already exist or were added, else false.
273 * @param proteinAlignment
274 * @param cdnaAlignment
277 public static boolean mapProteinAlignmentToCdna(
278 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
280 if (proteinAlignment == null || cdnaAlignment == null)
285 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
286 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
289 * First pass - map sequences where cross-references exist. This include
290 * 1-to-many mappings to support, for example, variant cDNA.
292 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
293 cdnaAlignment, mappedDna, mappedProtein, true);
296 * Second pass - map sequences where no cross-references exist. This only
297 * does 1-to-1 mappings and assumes corresponding sequences are in the same
298 * order in the alignments.
300 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
301 mappedDna, mappedProtein, false);
302 return mappingPerformed;
306 * Make mappings between compatible sequences (where the cDNA translation
307 * matches the protein).
309 * @param proteinAlignment
310 * @param cdnaAlignment
312 * a set of mapped DNA sequences (to add to)
313 * @param mappedProtein
314 * a set of mapped Protein sequences (to add to)
316 * if true, only map sequences where xrefs exist
319 protected static boolean mapProteinToCdna(
320 final AlignmentI proteinAlignment,
321 final AlignmentI cdnaAlignment, Set<SequenceI> mappedDna,
322 Set<SequenceI> mappedProtein, boolean xrefsOnly)
324 boolean mappingExistsOrAdded = false;
325 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
326 for (SequenceI aaSeq : thisSeqs)
328 boolean proteinMapped = false;
329 AlignedCodonFrame acf = new AlignedCodonFrame();
331 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
334 * Always try to map if sequences have xref to each other; this supports
335 * variant cDNA or alternative splicing for a protein sequence.
337 * If no xrefs, try to map progressively, assuming that alignments have
338 * mappable sequences in corresponding order. These are not
339 * many-to-many, as that would risk mixing species with similar cDNA
342 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
348 * Don't map non-xrefd sequences more than once each. This heuristic
349 * allows us to pair up similar sequences in ordered alignments.
352 && (mappedProtein.contains(aaSeq) || mappedDna
357 if (mappingExists(proteinAlignment.getCodonFrames(),
358 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
360 mappingExistsOrAdded = true;
364 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
367 acf.addMap(cdnaSeq, aaSeq, map);
368 mappingExistsOrAdded = true;
369 proteinMapped = true;
370 mappedDna.add(cdnaSeq);
371 mappedProtein.add(aaSeq);
377 proteinAlignment.addCodonFrame(acf);
380 return mappingExistsOrAdded;
384 * Answers true if the mappings include one between the given (dataset)
387 public static boolean mappingExists(List<AlignedCodonFrame> mappings,
388 SequenceI aaSeq, SequenceI cdnaSeq)
390 if (mappings != null)
392 for (AlignedCodonFrame acf : mappings)
394 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
404 * Builds a mapping (if possible) of a cDNA to a protein sequence.
406 * <li>first checks if the cdna translates exactly to the protein sequence</li>
407 * <li>else checks for translation after removing a STOP codon</li>
408 * <li>else checks for translation after removing a START codon</li>
409 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
411 * Returns null if no mapping is determined.
414 * the aligned protein sequence
416 * the aligned cdna sequence
419 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
423 * Here we handle either dataset sequence set (desktop) or absent (applet).
424 * Use only the char[] form of the sequence to avoid creating possibly large
427 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
428 char[] aaSeqChars = proteinDataset != null ? proteinDataset
429 .getSequence() : proteinSeq.getSequence();
430 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
431 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
432 : cdnaSeq.getSequence();
433 if (aaSeqChars == null || cdnaSeqChars == null)
439 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
441 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
442 int cdnaLength = cdnaSeqChars.length;
443 int cdnaStart = cdnaSeq.getStart();
444 int cdnaEnd = cdnaSeq.getEnd();
445 final int proteinStart = proteinSeq.getStart();
446 final int proteinEnd = proteinSeq.getEnd();
449 * If lengths don't match, try ignoring stop codon (if present)
451 if (cdnaLength != mappedLength && cdnaLength > 2)
453 String lastCodon = String.valueOf(cdnaSeqChars,
454 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
455 for (String stop : ResidueProperties.STOP)
457 if (lastCodon.equals(stop))
459 cdnaEnd -= CODON_LENGTH;
460 cdnaLength -= CODON_LENGTH;
467 * If lengths still don't match, try ignoring start codon.
470 if (cdnaLength != mappedLength
472 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
473 .equals(ResidueProperties.START))
475 startOffset += CODON_LENGTH;
476 cdnaStart += CODON_LENGTH;
477 cdnaLength -= CODON_LENGTH;
480 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
483 * protein is translation of dna (+/- start/stop codons)
485 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd }, new int[]
486 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
491 * translation failed - try mapping CDS annotated regions of dna
493 return mapCdsToProtein(cdnaSeq, proteinSeq);
497 * Test whether the given cdna sequence, starting at the given offset,
498 * translates to the given amino acid sequence, using the standard translation
499 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
501 * @param cdnaSeqChars
506 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
509 if (cdnaSeqChars == null || aaSeqChars == null)
515 int dnaPos = cdnaStart;
516 for (; dnaPos < cdnaSeqChars.length - 2 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
518 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
519 final String translated = ResidueProperties.codonTranslate(codon);
522 * allow * in protein to match untranslatable in dna
524 final char aaRes = aaSeqChars[aaPos];
525 if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
529 if (translated == null || !(aaRes == translated.charAt(0)))
532 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
533 // + codon + "(" + translated + ") != " + aaRes));
539 * check we matched all of the protein sequence
541 if (aaPos != aaSeqChars.length)
547 * check we matched all of the dna except
548 * for optional trailing STOP codon
550 if (dnaPos == cdnaSeqChars.length)
554 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
556 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
557 if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
566 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
567 * currently assumes that we are aligning cDNA to match protein.
570 * the sequence to be realigned
572 * the alignment whose sequence alignment is to be 'copied'
574 * character string represent a gap in the realigned sequence
575 * @param preserveUnmappedGaps
576 * @param preserveMappedGaps
577 * @return true if the sequence was realigned, false if it could not be
579 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
580 String gap, boolean preserveMappedGaps,
581 boolean preserveUnmappedGaps)
584 * Get any mappings from the source alignment to the target (dataset)
587 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
588 // all mappings. Would it help to constrain this?
589 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
590 if (mappings == null || mappings.isEmpty())
596 * Locate the aligned source sequence whose dataset sequence is mapped. We
597 * just take the first match here (as we can't align like more than one
600 SequenceI alignFrom = null;
601 AlignedCodonFrame mapping = null;
602 for (AlignedCodonFrame mp : mappings)
604 alignFrom = mp.findAlignedSequence(seq, al);
605 if (alignFrom != null)
612 if (alignFrom == null)
616 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
617 preserveMappedGaps, preserveUnmappedGaps);
622 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
623 * match residues and codons. Flags control whether existing gaps in unmapped
624 * (intron) and mapped (exon) regions are preserved or not. Gaps between
625 * intron and exon are only retained if both flags are set.
632 * @param preserveUnmappedGaps
633 * @param preserveMappedGaps
635 public static void alignSequenceAs(SequenceI alignTo,
636 SequenceI alignFrom, AlignedCodonFrame mapping, String myGap,
637 char sourceGap, boolean preserveMappedGaps,
638 boolean preserveUnmappedGaps)
640 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
642 // aligned and dataset sequence positions, all base zero
646 int basesWritten = 0;
647 char myGapChar = myGap.charAt(0);
648 int ratio = myGap.length();
650 int fromOffset = alignFrom.getStart() - 1;
651 int toOffset = alignTo.getStart() - 1;
652 int sourceGapMappedLength = 0;
653 boolean inExon = false;
654 final char[] thisSeq = alignTo.getSequence();
655 final char[] thatAligned = alignFrom.getSequence();
656 StringBuilder thisAligned = new StringBuilder(2 * thisSeq.length);
659 * Traverse the 'model' aligned sequence
661 for (char sourceChar : thatAligned)
663 if (sourceChar == sourceGap)
665 sourceGapMappedLength += ratio;
670 * Found a non-gap character. Locate its mapped region if any.
673 // Note mapping positions are base 1, our sequence positions base 0
674 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
675 sourceDsPos + fromOffset);
676 if (mappedPos == null)
679 * unmapped position; treat like a gap
681 sourceGapMappedLength += ratio;
682 // System.err.println("Can't align: no codon mapping to residue "
683 // + sourceDsPos + "(" + sourceChar + ")");
688 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
689 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
690 StringBuilder trailingCopiedGap = new StringBuilder();
693 * Copy dna sequence up to and including this codon. Optionally, include
694 * gaps before the codon starts (in introns) and/or after the codon starts
697 * Note this only works for 'linear' splicing, not reverse or interleaved.
698 * But then 'align dna as protein' doesn't make much sense otherwise.
700 int intronLength = 0;
701 while (basesWritten + toOffset < mappedCodonEnd
702 && thisSeqPos < thisSeq.length)
704 final char c = thisSeq[thisSeqPos++];
708 int sourcePosition = basesWritten + toOffset;
709 if (sourcePosition < mappedCodonStart)
712 * Found an unmapped (intron) base. First add in any preceding gaps
715 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
717 thisAligned.append(trailingCopiedGap.toString());
718 intronLength += trailingCopiedGap.length();
719 trailingCopiedGap = new StringBuilder();
726 final boolean startOfCodon = sourcePosition == mappedCodonStart;
727 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
728 preserveUnmappedGaps, sourceGapMappedLength, inExon,
729 trailingCopiedGap.length(), intronLength, startOfCodon);
730 for (int i = 0; i < gapsToAdd; i++)
732 thisAligned.append(myGapChar);
734 sourceGapMappedLength = 0;
737 thisAligned.append(c);
738 trailingCopiedGap = new StringBuilder();
742 if (inExon && preserveMappedGaps)
744 trailingCopiedGap.append(myGapChar);
746 else if (!inExon && preserveUnmappedGaps)
748 trailingCopiedGap.append(myGapChar);
755 * At end of model aligned sequence. Copy any remaining target sequence, optionally
756 * including (intron) gaps.
758 while (thisSeqPos < thisSeq.length)
760 final char c = thisSeq[thisSeqPos++];
761 if (c != myGapChar || preserveUnmappedGaps)
763 thisAligned.append(c);
765 sourceGapMappedLength--;
769 * finally add gaps to pad for any trailing source gaps or
770 * unmapped characters
772 if (preserveUnmappedGaps)
774 while (sourceGapMappedLength > 0)
776 thisAligned.append(myGapChar);
777 sourceGapMappedLength--;
782 * All done aligning, set the aligned sequence.
784 alignTo.setSequence(new String(thisAligned));
788 * Helper method to work out how many gaps to insert when realigning.
790 * @param preserveMappedGaps
791 * @param preserveUnmappedGaps
792 * @param sourceGapMappedLength
794 * @param trailingCopiedGap
795 * @param intronLength
796 * @param startOfCodon
799 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
800 boolean preserveUnmappedGaps, int sourceGapMappedLength,
801 boolean inExon, int trailingGapLength, int intronLength,
802 final boolean startOfCodon)
808 * Reached start of codon. Ignore trailing gaps in intron unless we are
809 * preserving gaps in both exon and intron. Ignore them anyway if the
810 * protein alignment introduces a gap at least as large as the intronic
813 if (inExon && !preserveMappedGaps)
815 trailingGapLength = 0;
817 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
819 trailingGapLength = 0;
823 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
827 if (intronLength + trailingGapLength <= sourceGapMappedLength)
829 gapsToAdd = sourceGapMappedLength - intronLength;
833 gapsToAdd = Math.min(intronLength + trailingGapLength
834 - sourceGapMappedLength, trailingGapLength);
841 * second or third base of codon; check for any gaps in dna
843 if (!preserveMappedGaps)
845 trailingGapLength = 0;
847 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
853 * Realigns the given protein to match the alignment of the dna, using codon
854 * mappings to translate aligned codon positions to protein residues.
857 * the alignment whose sequences are realigned by this method
859 * the dna alignment whose alignment we are 'copying'
860 * @return the number of sequences that were realigned
862 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
864 if (protein.isNucleotide() || !dna.isNucleotide())
866 System.err.println("Wrong alignment type in alignProteinAsDna");
869 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
870 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
871 protein, dna, unmappedProtein);
872 return alignProteinAs(protein, alignedCodons, unmappedProtein);
876 * Realigns the given dna to match the alignment of the protein, using codon
877 * mappings to translate aligned peptide positions to codons.
879 * Always produces a padded CDS alignment.
882 * the alignment whose sequences are realigned by this method
884 * the protein alignment whose alignment we are 'copying'
885 * @return the number of sequences that were realigned
887 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
889 if (protein.isNucleotide() || !dna.isNucleotide())
891 System.err.println("Wrong alignment type in alignProteinAsDna");
894 // todo: implement this
895 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
896 int alignedCount = 0;
897 int width = 0; // alignment width for padding CDS
898 for (SequenceI dnaSeq : dna.getSequences())
900 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
901 dna.getGapCharacter()))
905 width = Math.max(dnaSeq.getLength(), width);
909 for (SequenceI dnaSeq : dna.getSequences())
911 oldwidth = dnaSeq.getLength();
912 diff = width - oldwidth;
915 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
922 * Helper method to align (if possible) the dna sequence to match the
923 * alignment of a mapped protein sequence. This is currently limited to
924 * handling coding sequence only.
932 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
933 AlignmentI protein, List<AlignedCodonFrame> mappings, char gapChar)
935 SequenceI cdsDss = cdsSeq.getDatasetSequence();
939 .println("alignCdsSequenceAsProtein needs aligned sequence!");
943 List<AlignedCodonFrame> dnaMappings = MappingUtils
944 .findMappingsForSequence(cdsSeq, mappings);
945 for (AlignedCodonFrame mapping : dnaMappings)
947 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
950 int peptideLength = peptide.getLength();
951 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
954 MapList mapList = map.getMap();
955 if (map.getTo() == peptide.getDatasetSequence())
957 mapList = mapList.getInverse();
959 int cdsLength = cdsDss.getLength();
960 int mappedFromLength = MappingUtils.getLength(mapList
962 int mappedToLength = MappingUtils
963 .getLength(mapList.getToRanges());
964 boolean addStopCodon = (cdsLength == mappedFromLength
965 * CODON_LENGTH + CODON_LENGTH)
966 || (peptide.getDatasetSequence().getLength() == mappedFromLength - 1);
967 if (cdsLength != mappedToLength && !addStopCodon)
971 .format("Can't align cds as protein (length mismatch %d/%d): %s",
972 cdsLength, mappedToLength,
977 * pre-fill the aligned cds sequence with gaps
979 char[] alignedCds = new char[peptideLength * CODON_LENGTH
980 + (addStopCodon ? CODON_LENGTH : 0)];
981 Arrays.fill(alignedCds, gapChar);
984 * walk over the aligned peptide sequence and insert mapped
985 * codons for residues in the aligned cds sequence
987 char[] alignedPeptide = peptide.getSequence();
988 char[] nucleotides = cdsDss.getSequence();
990 int cdsStart = cdsDss.getStart();
991 int proteinPos = peptide.getStart() - 1;
993 for (char residue : alignedPeptide)
995 if (Comparison.isGap(residue))
997 cdsCol += CODON_LENGTH;
1002 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1005 // e.g. incomplete start codon, X in peptide
1006 cdsCol += CODON_LENGTH;
1010 for (int j = codon[0]; j <= codon[1]; j++)
1012 char mappedBase = nucleotides[j - cdsStart];
1013 alignedCds[cdsCol++] = mappedBase;
1021 * append stop codon if not mapped from protein,
1022 * closing it up to the end of the mapped sequence
1024 if (copiedBases == nucleotides.length - CODON_LENGTH)
1026 for (int i = alignedCds.length - 1; i >= 0; i--)
1028 if (!Comparison.isGap(alignedCds[i]))
1030 cdsCol = i + 1; // gap just after end of sequence
1034 for (int i = nucleotides.length - CODON_LENGTH; i < nucleotides.length; i++)
1036 alignedCds[cdsCol++] = nucleotides[i];
1039 cdsSeq.setSequence(new String(alignedCds));
1048 * Builds a map whose key is an aligned codon position (3 alignment column
1049 * numbers base 0), and whose value is a map from protein sequence to each
1050 * protein's peptide residue for that codon. The map generates an ordering of
1051 * the codons, and allows us to read off the peptides at each position in
1052 * order to assemble 'aligned' protein sequences.
1055 * the protein alignment
1057 * the coding dna alignment
1058 * @param unmappedProtein
1059 * any unmapped proteins are added to this list
1062 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1063 AlignmentI protein, AlignmentI dna,
1064 List<SequenceI> unmappedProtein)
1067 * maintain a list of any proteins with no mappings - these will be
1068 * rendered 'as is' in the protein alignment as we can't align them
1070 unmappedProtein.addAll(protein.getSequences());
1072 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1075 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1076 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1077 * comparator keeps the codon positions ordered.
1079 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
1080 new CodonComparator());
1082 for (SequenceI dnaSeq : dna.getSequences())
1084 for (AlignedCodonFrame mapping : mappings)
1086 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1089 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1090 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
1091 seqMap, alignedCodons);
1092 unmappedProtein.remove(prot);
1098 * Finally add any unmapped peptide start residues (e.g. for incomplete
1099 * codons) as if at the codon position before the second residue
1101 // TODO resolve JAL-2022 so this fudge can be removed
1102 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1103 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1105 return alignedCodons;
1109 * Scans for any protein mapped from position 2 (meaning unmapped start
1110 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1111 * preceding position in the alignment
1113 * @param alignedCodons
1114 * the codon-to-peptide map
1115 * @param mappedSequenceCount
1116 * the number of distinct sequences in the map
1118 protected static void addUnmappedPeptideStarts(
1119 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1120 int mappedSequenceCount)
1122 // TODO delete this ugly hack once JAL-2022 is resolved
1123 // i.e. we can model startPhase > 0 (incomplete start codon)
1125 List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
1126 AlignedCodon lastCodon = null;
1127 Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
1129 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1132 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1135 SequenceI seq = sequenceCodon.getKey();
1136 if (sequencesChecked.contains(seq))
1140 sequencesChecked.add(seq);
1141 AlignedCodon codon = sequenceCodon.getValue();
1142 if (codon.peptideCol > 1)
1145 .println("Problem mapping protein with >1 unmapped start positions: "
1148 else if (codon.peptideCol == 1)
1151 * first position (peptideCol == 0) was unmapped - add it
1153 if (lastCodon != null)
1155 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1156 lastCodon.pos2, lastCodon.pos3, String.valueOf(seq
1158 toAdd.put(seq, firstPeptide);
1163 * unmapped residue at start of alignment (no prior column) -
1164 * 'insert' at nominal codon [0, 0, 0]
1166 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1167 String.valueOf(seq.getCharAt(0)), 0);
1168 toAdd.put(seq, firstPeptide);
1171 if (sequencesChecked.size() == mappedSequenceCount)
1173 // no need to check past first mapped position in all sequences
1177 lastCodon = entry.getKey();
1181 * add any new codons safely after iterating over the map
1183 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1185 addCodonToMap(alignedCodons, startCodon.getValue(),
1186 startCodon.getKey());
1191 * Update the aligned protein sequences to match the codon alignments given in
1195 * @param alignedCodons
1196 * an ordered map of codon positions (columns), with sequence/peptide
1197 * values present in each column
1198 * @param unmappedProtein
1201 protected static int alignProteinAs(AlignmentI protein,
1202 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1203 List<SequenceI> unmappedProtein)
1206 * Prefill aligned sequences with gaps before inserting aligned protein
1209 int alignedWidth = alignedCodons.size();
1210 char[] gaps = new char[alignedWidth];
1211 Arrays.fill(gaps, protein.getGapCharacter());
1212 String allGaps = String.valueOf(gaps);
1213 for (SequenceI seq : protein.getSequences())
1215 if (!unmappedProtein.contains(seq))
1217 seq.setSequence(allGaps);
1222 for (AlignedCodon codon : alignedCodons.keySet())
1224 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1226 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1228 // place translated codon at its column position in sequence
1229 entry.getKey().getSequence()[column] = entry.getValue().product
1238 * Populate the map of aligned codons by traversing the given sequence
1239 * mapping, locating the aligned positions of mapped codons, and adding those
1240 * positions and their translation products to the map.
1243 * the aligned sequence we are mapping from
1245 * the sequence to be aligned to the codons
1247 * the gap character in the dna sequence
1249 * a mapping to a sequence translation
1250 * @param alignedCodons
1251 * the map we are building up
1253 static void addCodonPositions(SequenceI dna, SequenceI protein,
1254 char gapChar, Mapping seqMap,
1255 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1257 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1260 * add codon positions, and their peptide translations, to the alignment
1261 * map, while remembering the first codon mapped
1263 while (codons.hasNext())
1267 AlignedCodon codon = codons.next();
1268 addCodonToMap(alignedCodons, codon, protein);
1269 } catch (IncompleteCodonException e)
1271 // possible incomplete trailing codon - ignore
1272 } catch (NoSuchElementException e)
1274 // possibly peptide lacking STOP
1280 * Helper method to add a codon-to-peptide entry to the aligned codons map
1282 * @param alignedCodons
1286 protected static void addCodonToMap(
1287 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1288 AlignedCodon codon, SequenceI protein)
1290 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1291 if (seqProduct == null)
1293 seqProduct = new HashMap<SequenceI, AlignedCodon>();
1294 alignedCodons.put(codon, seqProduct);
1296 seqProduct.put(protein, codon);
1300 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1301 * between at least one pair of sequences in the two alignments. Currently,
1304 * <li>One alignment must be nucleotide, and the other protein</li>
1305 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1306 * <li>Mappable means the nucleotide translation matches the protein sequence</li>
1307 * <li>The translation may ignore start and stop codons if present in the
1315 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1317 if (al1 == null || al2 == null)
1323 * Require one nucleotide and one protein
1325 if (al1.isNucleotide() == al2.isNucleotide())
1329 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1330 AlignmentI protein = dna == al1 ? al2 : al1;
1331 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1332 for (SequenceI dnaSeq : dna.getSequences())
1334 for (SequenceI proteinSeq : protein.getSequences())
1336 if (isMappable(dnaSeq, proteinSeq, mappings))
1346 * Returns true if the dna sequence is mapped, or could be mapped, to the
1354 protected static boolean isMappable(SequenceI dnaSeq,
1355 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1357 if (dnaSeq == null || proteinSeq == null)
1362 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq : dnaSeq
1363 .getDatasetSequence();
1364 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null ? proteinSeq
1365 : proteinSeq.getDatasetSequence();
1367 for (AlignedCodonFrame mapping : mappings)
1369 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1379 * Just try to make a mapping (it is not yet stored), test whether
1382 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1386 * Finds any reference annotations associated with the sequences in
1387 * sequenceScope, that are not already added to the alignment, and adds them
1388 * to the 'candidates' map. Also populates a lookup table of annotation
1389 * labels, keyed by calcId, for use in constructing tooltips or the like.
1391 * @param sequenceScope
1392 * the sequences to scan for reference annotations
1393 * @param labelForCalcId
1394 * (optional) map to populate with label for calcId
1396 * map to populate with annotations for sequence
1398 * the alignment to check for presence of annotations
1400 public static void findAddableReferenceAnnotations(
1401 List<SequenceI> sequenceScope,
1402 Map<String, String> labelForCalcId,
1403 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1406 if (sequenceScope == null)
1412 * For each sequence in scope, make a list of any annotations on the
1413 * underlying dataset sequence which are not already on the alignment.
1415 * Add to a map of { alignmentSequence, <List of annotations to add> }
1417 for (SequenceI seq : sequenceScope)
1419 SequenceI dataset = seq.getDatasetSequence();
1420 if (dataset == null)
1424 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1425 if (datasetAnnotations == null)
1429 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1430 for (AlignmentAnnotation dsann : datasetAnnotations)
1433 * Find matching annotations on the alignment. If none is found, then
1434 * add this annotation to the list of 'addable' annotations for this
1437 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1438 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1439 if (!matchedAlignmentAnnotations.iterator().hasNext())
1442 if (labelForCalcId != null)
1444 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1449 * Save any addable annotations for this sequence
1451 if (!result.isEmpty())
1453 candidates.put(seq, result);
1459 * Adds annotations to the top of the alignment annotations, in the same order
1460 * as their related sequences.
1462 * @param annotations
1463 * the annotations to add
1465 * the alignment to add them to
1466 * @param selectionGroup
1467 * current selection group (or null if none)
1469 public static void addReferenceAnnotations(
1470 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1471 final AlignmentI alignment, final SequenceGroup selectionGroup)
1473 for (SequenceI seq : annotations.keySet())
1475 for (AlignmentAnnotation ann : annotations.get(seq))
1477 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1479 int endRes = ann.annotations.length;
1480 if (selectionGroup != null)
1482 startRes = selectionGroup.getStartRes();
1483 endRes = selectionGroup.getEndRes();
1485 copyAnn.restrict(startRes, endRes);
1488 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1489 * original annotation is already on the sequence.
1491 if (!seq.hasAnnotation(ann))
1493 seq.addAlignmentAnnotation(copyAnn);
1496 copyAnn.adjustForAlignment();
1497 // add to the alignment and set visible
1498 alignment.addAnnotation(copyAnn);
1499 copyAnn.visible = true;
1505 * Set visibility of alignment annotations of specified types (labels), for
1506 * specified sequences. This supports controls like
1507 * "Show all secondary structure", "Hide all Temp factor", etc.
1509 * @al the alignment to scan for annotations
1511 * the types (labels) of annotations to be updated
1512 * @param forSequences
1513 * if not null, only annotations linked to one of these sequences are
1514 * in scope for update; if null, acts on all sequence annotations
1516 * if this flag is true, 'types' is ignored (label not checked)
1518 * if true, set visibility on, else set off
1520 public static void showOrHideSequenceAnnotations(AlignmentI al,
1521 Collection<String> types, List<SequenceI> forSequences,
1522 boolean anyType, boolean doShow)
1524 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1527 for (AlignmentAnnotation aa : anns)
1529 if (anyType || types.contains(aa.label))
1531 if ((aa.sequenceRef != null)
1532 && (forSequences == null || forSequences
1533 .contains(aa.sequenceRef)))
1535 aa.visible = doShow;
1543 * Returns true if either sequence has a cross-reference to the other
1549 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1551 // Note: moved here from class CrossRef as the latter class has dependencies
1552 // not availability to the applet's classpath
1553 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1557 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1558 * that sequence name is structured as Source|AccessionId.
1564 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1566 if (seq1 == null || seq2 == null)
1570 String name = seq2.getName();
1571 final DBRefEntry[] xrefs = seq1.getDBRefs();
1574 for (DBRefEntry xref : xrefs)
1576 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1577 // case-insensitive test, consistent with DBRefEntry.equalRef()
1578 if (xrefName.equalsIgnoreCase(name))
1588 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1589 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1590 * added to the original alignment's dataset, which is shared by the new
1591 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1592 * added to the alignment dataset.
1595 * aligned nucleotide (dna or cds) sequences
1597 * the alignment dataset the sequences belong to
1599 * (optional) to restrict results to CDS that map to specified
1601 * @return an alignment whose sequences are the cds-only parts of the dna
1602 * sequences (or null if no mappings are found)
1604 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1605 AlignmentI dataset, SequenceI[] products)
1607 if (dataset == null || dataset.getDataset() != null)
1609 throw new IllegalArgumentException(
1610 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1612 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
1613 List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
1614 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1615 HashSet<SequenceI> productSeqs = null;
1616 if (products != null)
1618 productSeqs = new HashSet<SequenceI>();
1619 for (SequenceI seq : products)
1621 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1622 .getDatasetSequence());
1627 * Construct CDS sequences from mappings on the alignment dataset.
1629 * - find the protein product(s) mapped to from each dna sequence
1630 * - if the mapping covers the whole dna sequence (give or take start/stop
1631 * codon), take the dna as the CDS sequence
1632 * - else search dataset mappings for a suitable dna sequence, i.e. one
1633 * whose whole sequence is mapped to the protein
1634 * - if no sequence found, construct one from the dna sequence and mapping
1635 * (and add it to dataset so it is found if this is repeated)
1637 for (SequenceI dnaSeq : dna)
1639 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1640 : dnaSeq.getDatasetSequence();
1642 List<AlignedCodonFrame> seqMappings = MappingUtils
1643 .findMappingsForSequence(dnaSeq, mappings);
1644 for (AlignedCodonFrame mapping : seqMappings)
1646 List<Mapping> mappingsFromSequence = mapping
1647 .getMappingsFromSequence(dnaSeq);
1649 for (Mapping aMapping : mappingsFromSequence)
1651 MapList mapList = aMapping.getMap();
1652 if (mapList.getFromRatio() == 1)
1655 * not a dna-to-protein mapping (likely dna-to-cds)
1661 * skip if mapping is not to one of the target set of proteins
1663 SequenceI proteinProduct = aMapping.getTo();
1664 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1670 * try to locate the CDS from the dataset mappings;
1671 * guard against duplicate results (for the case that protein has
1672 * dbrefs to both dna and cds sequences)
1674 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1675 seqMappings, aMapping);
1678 if (!foundSeqs.contains(cdsSeq))
1680 foundSeqs.add(cdsSeq);
1681 SequenceI derivedSequence = cdsSeq.deriveSequence();
1682 cdsSeqs.add(derivedSequence);
1683 if (!dataset.getSequences().contains(cdsSeq))
1685 dataset.addSequence(cdsSeq);
1692 * didn't find mapped CDS sequence - construct it and add
1693 * its dataset sequence to the dataset
1695 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1696 dataset).deriveSequence();
1697 // cdsSeq has a name constructed as CDS|<dbref>
1698 // <dbref> will be either the accession for the coding sequence,
1699 // marked in the /via/ dbref to the protein product accession
1700 // or it will be the original nucleotide accession.
1701 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1703 cdsSeqs.add(cdsSeq);
1705 if (!dataset.getSequences().contains(cdsSeqDss))
1707 // check if this sequence is a newly created one
1708 // so needs adding to the dataset
1709 dataset.addSequence(cdsSeqDss);
1713 * add a mapping from CDS to the (unchanged) mapped to range
1715 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1716 cdsSeq.getLength() });
1717 MapList cdsToProteinMap = new MapList(cdsRange,
1718 mapList.getToRanges(), mapList.getFromRatio(),
1719 mapList.getToRatio());
1720 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1721 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1725 * guard against duplicating the mapping if repeating this action
1727 if (!mappings.contains(cdsToProteinMapping))
1729 mappings.add(cdsToProteinMapping);
1732 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1733 proteinProduct, aMapping);
1735 * add another mapping from original 'from' range to CDS
1737 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1738 MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1740 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1742 if (!mappings.contains(dnaToCdsMapping))
1744 mappings.add(dnaToCdsMapping);
1748 * add DBRef with mapping from protein to CDS
1749 * (this enables Get Cross-References from protein alignment)
1750 * This is tricky because we can't have two DBRefs with the
1751 * same source and accession, so need a different accession for
1752 * the CDS from the dna sequence
1755 // specific use case:
1756 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1757 // ENSG02, ENSG03, with transcripts and products similarly named.
1758 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1760 // JBPNote: ?? can't actually create an example that demonstrates we
1762 // synthesize an xref.
1764 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1766 // creates a complementary cross-reference to the source sequence's
1767 // primary reference.
1769 DBRefEntry cdsCrossRef = new DBRefEntry(primRef.getSource(),
1770 primRef.getSource() + ":" + primRef.getVersion(),
1771 primRef.getAccessionId());
1773 .setMap(new Mapping(dnaDss, new MapList(dnaToCdsMap)));
1774 cdsSeqDss.addDBRef(cdsCrossRef);
1776 // problem here is that the cross-reference is synthesized -
1777 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1779 // assuming cds version same as dna ?!?
1781 DBRefEntry proteinToCdsRef = new DBRefEntry(
1782 primRef.getSource(), primRef.getVersion(),
1785 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1787 proteinProduct.addDBRef(proteinToCdsRef);
1791 * transfer any features on dna that overlap the CDS
1793 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1794 SequenceOntologyI.CDS);
1799 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1801 cds.setDataset(dataset);
1807 * A helper method that finds a CDS sequence in the alignment dataset that is
1808 * mapped to the given protein sequence, and either is, or has a mapping from,
1809 * the given dna sequence.
1812 * set of all mappings on the dataset
1814 * a dna (or cds) sequence we are searching from
1815 * @param seqMappings
1816 * the set of mappings involving dnaSeq
1818 * an initial candidate from seqMappings
1821 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1822 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1826 * TODO a better dna-cds-protein mapping data representation to allow easy
1827 * navigation; until then this clunky looping around lists of mappings
1829 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1830 : dnaSeq.getDatasetSequence();
1831 SequenceI proteinProduct = aMapping.getTo();
1834 * is this mapping from the whole dna sequence (i.e. CDS)?
1835 * allowing for possible stop codon on dna but not peptide
1837 int mappedFromLength = MappingUtils.getLength(aMapping.getMap()
1839 int dnaLength = seqDss.getLength();
1840 if (mappedFromLength == dnaLength
1841 || mappedFromLength == dnaLength - CODON_LENGTH)
1847 * looks like we found the dna-to-protein mapping; search for the
1848 * corresponding cds-to-protein mapping
1850 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1851 .findMappingsForSequence(proteinProduct, mappings);
1852 for (AlignedCodonFrame acf : mappingsToPeptide)
1854 for (SequenceToSequenceMapping map : acf.getMappings())
1856 Mapping mapping = map.getMapping();
1857 if (mapping != aMapping
1858 && mapping.getMap().getFromRatio() == CODON_LENGTH
1859 && proteinProduct == mapping.getTo()
1860 && seqDss != map.getFromSeq())
1862 mappedFromLength = MappingUtils.getLength(mapping.getMap()
1864 if (mappedFromLength == map.getFromSeq().getLength())
1867 * found a 3:1 mapping to the protein product which covers
1868 * the whole dna sequence i.e. is from CDS; finally check it
1869 * is from the dna start sequence
1871 SequenceI cdsSeq = map.getFromSeq();
1872 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1873 .findMappingsForSequence(cdsSeq, seqMappings);
1874 if (!dnaToCdsMaps.isEmpty())
1886 * Helper method that makes a CDS sequence as defined by the mappings from the
1887 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1888 * forward or reverse strand).
1893 * - existing dataset. We check for sequences that look like the CDS
1894 * we are about to construct, if one exists already, then we will
1895 * just return that one.
1896 * @return CDS sequence (as a dataset sequence)
1898 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1901 char[] seqChars = seq.getSequence();
1902 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1903 int cdsWidth = MappingUtils.getLength(fromRanges);
1904 char[] newSeqChars = new char[cdsWidth];
1907 for (int[] range : fromRanges)
1909 if (range[0] <= range[1])
1911 // forward strand mapping - just copy the range
1912 int length = range[1] - range[0] + 1;
1913 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1919 // reverse strand mapping - copy and complement one by one
1920 for (int i = range[0]; i >= range[1]; i--)
1922 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1928 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
1929 * else generate a sequence name
1931 String mapFromId = mapping.getMappedFromId();
1932 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
1933 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
1934 if (dataset != null)
1936 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
1937 if (matches != null)
1939 boolean matched = false;
1940 for (SequenceI mtch : matches)
1942 if (mtch.getStart() != newSeq.getStart())
1946 if (mtch.getEnd() != newSeq.getEnd())
1950 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
1962 .println("JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
1968 // newSeq.setDescription(mapFromId);
1974 * add any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
1975 * the given mapping.
1980 * @return list of DBRefEntrys added.
1982 public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
1983 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
1986 // gather direct refs from contig congrent with mapping
1987 List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
1988 HashSet<String> directSources = new HashSet<String>();
1989 if (contig.getDBRefs() != null)
1991 for (DBRefEntry dbr : contig.getDBRefs())
1993 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
1995 MapList map = dbr.getMap().getMap();
1996 // check if map is the CDS mapping
1997 if (mapping.getMap().equals(map))
2000 directSources.add(dbr.getSource());
2005 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2006 proteinProduct.getDBRefs(),
2007 directSources.toArray(new String[0]));
2008 List<DBRefEntry> propagated = new ArrayList<DBRefEntry>();
2010 // and generate appropriate mappings
2011 for (DBRefEntry cdsref : direct)
2013 // clone maplist and mapping
2014 MapList cdsposmap = new MapList(Arrays.asList(new int[][] { new int[]
2015 { cdsSeq.getStart(), cdsSeq.getEnd() } }), cdsref.getMap().getMap()
2016 .getToRanges(), 3, 1);
2017 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(), cdsref.getMap()
2021 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2022 cdsref.getVersion(), cdsref.getAccessionId(), new Mapping(
2023 cdsmap.getTo(), cdsposmap));
2025 // and see if we can map to the protein product for this mapping.
2026 // onSource is the filtered set of accessions on protein that we are
2027 // tranferring, so we assume accession is the same.
2028 if (cdsmap.getTo() == null && onSource != null)
2030 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2031 cdsref.getAccessionId());
2032 if (sourceRefs != null)
2034 for (DBRefEntry srcref : sourceRefs)
2036 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2038 // we have found a complementary dbref on the protein product, so
2039 // update mapping's getTo
2040 newref.getMap().setTo(proteinProduct);
2045 cdsSeq.addDBRef(newref);
2046 propagated.add(newref);
2052 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2053 * feature start/end ranges, optionally omitting specified feature types.
2054 * Returns the number of features copied.
2059 * if not null, only features of this type are copied (including
2060 * subtypes in the Sequence Ontology)
2062 * the mapping from 'fromSeq' to 'toSeq'
2065 public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2066 MapList mapping, String select, String... omitting)
2068 SequenceI copyTo = toSeq;
2069 while (copyTo.getDatasetSequence() != null)
2071 copyTo = copyTo.getDatasetSequence();
2074 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2076 SequenceFeature[] sfs = fromSeq.getSequenceFeatures();
2079 for (SequenceFeature sf : sfs)
2081 String type = sf.getType();
2082 if (select != null && !so.isA(type, select))
2086 boolean omit = false;
2087 for (String toOmit : omitting)
2089 if (type.equals(toOmit))
2100 * locate the mapped range - null if either start or end is
2101 * not mapped (no partial overlaps are calculated)
2103 int start = sf.getBegin();
2104 int end = sf.getEnd();
2105 int[] mappedTo = mapping.locateInTo(start, end);
2107 * if whole exon range doesn't map, try interpreting it
2108 * as 5' or 3' exon overlapping the CDS range
2110 if (mappedTo == null)
2112 mappedTo = mapping.locateInTo(end, end);
2113 if (mappedTo != null)
2116 * end of exon is in CDS range - 5' overlap
2117 * to a range from the start of the peptide
2122 if (mappedTo == null)
2124 mappedTo = mapping.locateInTo(start, start);
2125 if (mappedTo != null)
2128 * start of exon is in CDS range - 3' overlap
2129 * to a range up to the end of the peptide
2131 mappedTo[1] = toSeq.getLength();
2134 if (mappedTo != null)
2136 SequenceFeature copy = new SequenceFeature(sf);
2137 copy.setBegin(Math.min(mappedTo[0], mappedTo[1]));
2138 copy.setEnd(Math.max(mappedTo[0], mappedTo[1]));
2139 copyTo.addSequenceFeature(copy);
2148 * Returns a mapping from dna to protein by inspecting sequence features of
2149 * type "CDS" on the dna.
2155 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2156 SequenceI proteinSeq)
2158 List<int[]> ranges = findCdsPositions(dnaSeq);
2159 int mappedDnaLength = MappingUtils.getLength(ranges);
2161 int proteinLength = proteinSeq.getLength();
2162 int proteinStart = proteinSeq.getStart();
2163 int proteinEnd = proteinSeq.getEnd();
2166 * incomplete start codon may mean X at start of peptide
2167 * we ignore both for mapping purposes
2169 if (proteinSeq.getCharAt(0) == 'X')
2171 // todo JAL-2022 support startPhase > 0
2175 List<int[]> proteinRange = new ArrayList<int[]>();
2178 * dna length should map to protein (or protein plus stop codon)
2180 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2181 if (codesForResidues == (proteinLength + 1))
2183 // assuming extra codon is for STOP and not in peptide
2186 if (codesForResidues == proteinLength)
2188 proteinRange.add(new int[] { proteinStart, proteinEnd });
2189 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2195 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2196 * start/end positions of sequence features of type "CDS" (or a sub-type of
2197 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2198 * position order, so this method is only valid for linear CDS in the same
2199 * sense as the protein product.
2204 public static List<int[]> findCdsPositions(SequenceI dnaSeq)
2206 List<int[]> result = new ArrayList<int[]>();
2207 SequenceFeature[] sfs = dnaSeq.getSequenceFeatures();
2213 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2216 for (SequenceFeature sf : sfs)
2219 * process a CDS feature (or a sub-type of CDS)
2221 if (so.isA(sf.getType(), SequenceOntologyI.CDS))
2226 phase = Integer.parseInt(sf.getPhase());
2227 } catch (NumberFormatException e)
2232 * phase > 0 on first codon means 5' incomplete - skip to the start
2233 * of the next codon; example ENST00000496384
2235 int begin = sf.getBegin();
2236 int end = sf.getEnd();
2237 if (result.isEmpty())
2242 // shouldn't happen!
2244 .println("Error: start phase extends beyond start CDS in "
2245 + dnaSeq.getName());
2248 result.add(new int[] { begin, end });
2253 * remove 'startPhase' positions (usually 0) from the first range
2254 * so we begin at the start of a complete codon
2256 if (!result.isEmpty())
2258 // TODO JAL-2022 correctly model start phase > 0
2259 result.get(0)[0] += startPhase;
2263 * Finally sort ranges by start position. This avoids a dependency on
2264 * keeping features in order on the sequence (if they are in order anyway,
2265 * the sort will have almost no work to do). The implicit assumption is CDS
2266 * ranges are assembled in order. Other cases should not use this method,
2267 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2269 Collections.sort(result, new RangeComparator(true));
2274 * Maps exon features from dna to protein, and computes variants in peptide
2275 * product generated by variants in dna, and adds them as sequence_variant
2276 * features on the protein sequence. Returns the number of variant features
2281 * @param dnaToProtein
2283 public static int computeProteinFeatures(SequenceI dnaSeq,
2284 SequenceI peptide, MapList dnaToProtein)
2286 while (dnaSeq.getDatasetSequence() != null)
2288 dnaSeq = dnaSeq.getDatasetSequence();
2290 while (peptide.getDatasetSequence() != null)
2292 peptide = peptide.getDatasetSequence();
2295 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2298 * compute protein variants from dna variants and codon mappings;
2299 * NB - alternatively we could retrieve this using the REST service e.g.
2300 * http://rest.ensembl.org/overlap/translation
2301 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2302 * which would be a bit slower but possibly more reliable
2306 * build a map with codon variations for each potentially varying peptide
2308 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2309 dnaSeq, dnaToProtein);
2312 * scan codon variations, compute peptide variants and add to peptide sequence
2315 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2317 int peptidePos = variant.getKey();
2318 List<DnaVariant>[] codonVariants = variant.getValue();
2319 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2323 * sort to get sequence features in start position order
2324 * - would be better to store in Sequence as a TreeSet or NCList?
2326 if (peptide.getSequenceFeatures() != null)
2328 Arrays.sort(peptide.getSequenceFeatures(),
2329 new Comparator<SequenceFeature>()
2332 public int compare(SequenceFeature o1, SequenceFeature o2)
2334 int c = Integer.compare(o1.getBegin(), o2.getBegin());
2335 return c == 0 ? Integer.compare(o1.getEnd(), o2.getEnd())
2344 * Computes non-synonymous peptide variants from codon variants and adds them
2345 * as sequence_variant features on the protein sequence (one feature per
2346 * allele variant). Selected attributes (variant id, clinical significance)
2347 * are copied over to the new features.
2350 * the protein sequence
2352 * the position to compute peptide variants for
2353 * @param codonVariants
2354 * a list of dna variants per codon position
2355 * @return the number of features added
2357 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2358 List<DnaVariant>[] codonVariants)
2360 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2362 String base1 = codonVariants[0].get(0).base;
2363 String base2 = codonVariants[1].get(0).base;
2364 String base3 = codonVariants[2].get(0).base;
2367 * variants in first codon base
2369 for (DnaVariant var : codonVariants[0])
2371 if (var.variant != null)
2373 String alleles = (String) var.variant.getValue("alleles");
2374 if (alleles != null)
2376 for (String base : alleles.split(","))
2378 String codon = base + base2 + base3;
2379 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2389 * variants in second codon base
2391 for (DnaVariant var : codonVariants[1])
2393 if (var.variant != null)
2395 String alleles = (String) var.variant.getValue("alleles");
2396 if (alleles != null)
2398 for (String base : alleles.split(","))
2400 String codon = base1 + base + base3;
2401 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2411 * variants in third codon base
2413 for (DnaVariant var : codonVariants[2])
2415 if (var.variant != null)
2417 String alleles = (String) var.variant.getValue("alleles");
2418 if (alleles != null)
2420 for (String base : alleles.split(","))
2422 String codon = base1 + base2 + base;
2423 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2436 * Helper method that adds a peptide variant feature, provided the given codon
2437 * translates to a value different to the current residue (is a non-synonymous
2438 * variant). ID and clinical_significance attributes of the dna variant (if
2439 * present) are copied to the new feature.
2446 * @return true if a feature was added, else false
2448 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2449 String residue, DnaVariant var, String codon)
2452 * get peptide translation of codon e.g. GAT -> D
2453 * note that variants which are not single alleles,
2454 * e.g. multibase variants or HGMD_MUTATION etc
2455 * are currently ignored here
2457 String trans = codon.contains("-") ? "-"
2458 : (codon.length() > CODON_LENGTH ? null : ResidueProperties
2459 .codonTranslate(codon));
2460 if (trans != null && !trans.equals(residue))
2462 String residue3Char = StringUtils
2463 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2464 String trans3Char = StringUtils
2465 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2466 String desc = "p." + residue3Char + peptidePos + trans3Char;
2467 // set score to 0f so 'graduated colour' option is offered! JAL-2060
2468 SequenceFeature sf = new SequenceFeature(
2469 SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
2470 peptidePos, 0f, var.getSource());
2471 StringBuilder attributes = new StringBuilder(32);
2472 String id = (String) var.variant.getValue(ID);
2475 if (id.startsWith(SEQUENCE_VARIANT))
2477 id = id.substring(SEQUENCE_VARIANT.length());
2479 sf.setValue(ID, id);
2480 attributes.append(ID).append("=").append(id);
2481 // TODO handle other species variants JAL-2064
2482 StringBuilder link = new StringBuilder(32);
2488 .append("|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2489 .append(URLEncoder.encode(id, "UTF-8"));
2490 sf.addLink(link.toString());
2491 } catch (UnsupportedEncodingException e)
2496 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2497 if (clinSig != null)
2499 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2500 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2503 peptide.addSequenceFeature(sf);
2504 if (attributes.length() > 0)
2506 sf.setAttributes(attributes.toString());
2514 * Builds a map whose key is position in the protein sequence, and value is a
2515 * list of the base and all variants for each corresponding codon position
2518 * @param dnaToProtein
2521 @SuppressWarnings("unchecked")
2522 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2523 SequenceI dnaSeq, MapList dnaToProtein)
2526 * map from peptide position to all variants of the codon which codes for it
2527 * LinkedHashMap ensures we keep the peptide features in sequence order
2529 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
2530 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2532 SequenceFeature[] dnaFeatures = dnaSeq.getSequenceFeatures();
2533 if (dnaFeatures == null)
2538 int dnaStart = dnaSeq.getStart();
2539 int[] lastCodon = null;
2540 int lastPeptidePostion = 0;
2543 * build a map of codon variations for peptides
2545 for (SequenceFeature sf : dnaFeatures)
2547 int dnaCol = sf.getBegin();
2548 if (dnaCol != sf.getEnd())
2550 // not handling multi-locus variant features
2553 if (so.isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
2555 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2558 // feature doesn't lie within coding region
2561 int peptidePosition = mapsTo[0];
2562 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2563 if (codonVariants == null)
2565 codonVariants = new ArrayList[CODON_LENGTH];
2566 codonVariants[0] = new ArrayList<DnaVariant>();
2567 codonVariants[1] = new ArrayList<DnaVariant>();
2568 codonVariants[2] = new ArrayList<DnaVariant>();
2569 variants.put(peptidePosition, codonVariants);
2573 * extract dna variants to a string array
2575 String alls = (String) sf.getValue("alleles");
2580 String[] alleles = alls.toUpperCase().split(",");
2582 for (String allele : alleles)
2584 alleles[i++] = allele.trim(); // lose any space characters "A, G"
2588 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2590 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2591 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2592 peptidePosition, peptidePosition));
2593 lastPeptidePostion = peptidePosition;
2597 * save nucleotide (and any variant) for each codon position
2599 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2601 String nucleotide = String.valueOf(
2602 dnaSeq.getCharAt(codon[codonPos] - dnaStart))
2604 List<DnaVariant> codonVariant = codonVariants[codonPos];
2605 if (codon[codonPos] == dnaCol)
2607 if (!codonVariant.isEmpty()
2608 && codonVariant.get(0).variant == null)
2611 * already recorded base value, add this variant
2613 codonVariant.get(0).variant = sf;
2618 * add variant with base value
2620 codonVariant.add(new DnaVariant(nucleotide, sf));
2623 else if (codonVariant.isEmpty())
2626 * record (possibly non-varying) base value
2628 codonVariant.add(new DnaVariant(nucleotide));
2637 * Makes an alignment with a copy of the given sequences, adding in any
2638 * non-redundant sequences which are mapped to by the cross-referenced
2644 * the alignment dataset shared by the new copy
2647 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2648 SequenceI[] xrefs, AlignmentI dataset)
2650 AlignmentI copy = new Alignment(new Alignment(seqs));
2651 copy.setDataset(dataset);
2652 boolean isProtein = !copy.isNucleotide();
2653 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2656 for (SequenceI xref : xrefs)
2658 DBRefEntry[] dbrefs = xref.getDBRefs();
2661 for (DBRefEntry dbref : dbrefs)
2663 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2664 || dbref.getMap().getTo().isProtein() != isProtein)
2668 SequenceI mappedTo = dbref.getMap().getTo();
2669 SequenceI match = matcher.findIdMatch(mappedTo);
2672 matcher.add(mappedTo);
2673 copy.addSequence(mappedTo);
2683 * Try to align sequences in 'unaligned' to match the alignment of their
2684 * mapped regions in 'aligned'. For example, could use this to align CDS
2685 * sequences which are mapped to their parent cDNA sequences.
2687 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2688 * dna-to-protein or protein-to-dna use alternative methods.
2691 * sequences to be aligned
2693 * holds aligned sequences and their mappings
2696 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2699 * easy case - aligning a copy of aligned sequences
2701 if (alignAsSameSequences(unaligned, aligned))
2703 return unaligned.getHeight();
2707 * fancy case - aligning via mappings between sequences
2709 List<SequenceI> unmapped = new ArrayList<SequenceI>();
2710 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2711 unaligned, aligned, unmapped);
2712 int width = columnMap.size();
2713 char gap = unaligned.getGapCharacter();
2714 int realignedCount = 0;
2715 // TODO: verify this loop scales sensibly for very wide/high alignments
2717 for (SequenceI seq : unaligned.getSequences())
2719 if (!unmapped.contains(seq))
2721 char[] newSeq = new char[width];
2722 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2723 // Integer iteration below
2728 * traverse the map to find columns populated
2731 for (Integer column : columnMap.keySet())
2733 Character c = columnMap.get(column).get(seq);
2737 * sequence has a character at this position
2747 * trim trailing gaps
2749 if (lastCol < width)
2751 char[] tmp = new char[lastCol + 1];
2752 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2755 // TODO: optimise SequenceI to avoid char[]->String->char[]
2756 seq.setSequence(String.valueOf(newSeq));
2760 return realignedCount;
2764 * If unaligned and aligned sequences share the same dataset sequences, then
2765 * simply copies the aligned sequences to the unaligned sequences and returns
2766 * true; else returns false
2769 * - sequences to be aligned based on aligned
2771 * - 'guide' alignment containing sequences derived from same dataset
2775 static boolean alignAsSameSequences(AlignmentI unaligned,
2778 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2780 return false; // should only pass alignments with datasets here
2783 // map from dataset sequence to alignment sequence(s)
2784 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<SequenceI, List<SequenceI>>();
2785 for (SequenceI seq : aligned.getSequences())
2787 SequenceI ds = seq.getDatasetSequence();
2788 if (alignedDatasets.get(ds) == null)
2790 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2792 alignedDatasets.get(ds).add(seq);
2796 * first pass - check whether all sequences to be aligned share a dataset
2797 * sequence with an aligned sequence
2799 for (SequenceI seq : unaligned.getSequences())
2801 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2808 * second pass - copy aligned sequences;
2809 * heuristic rule: pair off sequences in order for the case where
2810 * more than one shares the same dataset sequence
2812 for (SequenceI seq : unaligned.getSequences())
2814 List<SequenceI> alignedSequences = alignedDatasets.get(seq
2815 .getDatasetSequence());
2816 // TODO: getSequenceAsString() will be deprecated in the future
2817 // TODO: need to leave to SequenceI implementor to update gaps
2818 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2819 if (alignedSequences.size() > 0)
2821 // pop off aligned sequences (except the last one)
2822 alignedSequences.remove(0);
2830 * Returns a map whose key is alignment column number (base 1), and whose
2831 * values are a map of sequence characters in that column.
2838 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2839 AlignmentI unaligned, AlignmentI aligned, List<SequenceI> unmapped)
2842 * Map will hold, for each aligned column position, a map of
2843 * {unalignedSequence, characterPerSequence} at that position.
2844 * TreeMap keeps the entries in ascending column order.
2846 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<Integer, Map<SequenceI, Character>>();
2849 * record any sequences that have no mapping so can't be realigned
2851 unmapped.addAll(unaligned.getSequences());
2853 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2855 for (SequenceI seq : unaligned.getSequences())
2857 for (AlignedCodonFrame mapping : mappings)
2859 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2860 if (fromSeq != null)
2862 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2863 if (addMappedPositions(seq, fromSeq, seqMap, map))
2865 unmapped.remove(seq);
2874 * Helper method that adds to a map the mapped column positions of a sequence. <br>
2875 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2876 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2880 * the sequence whose column positions we are recording
2882 * a sequence that is mapped to the first sequence
2884 * the mapping from 'fromSeq' to 'seq'
2886 * a map to add the column positions (in fromSeq) of the mapped
2890 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2891 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2899 * invert mapping if it is from unaligned to aligned sequence
2901 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2903 seqMap = new Mapping(seq.getDatasetSequence(), seqMap.getMap()
2907 char[] fromChars = fromSeq.getSequence();
2908 int toStart = seq.getStart();
2909 char[] toChars = seq.getSequence();
2912 * traverse [start, end, start, end...] ranges in fromSeq
2914 for (int[] fromRange : seqMap.getMap().getFromRanges())
2916 for (int i = 0; i < fromRange.length - 1; i += 2)
2918 boolean forward = fromRange[i + 1] >= fromRange[i];
2921 * find the range mapped to (sequence positions base 1)
2923 int[] range = seqMap.locateMappedRange(fromRange[i],
2927 System.err.println("Error in mapping " + seqMap + " from "
2928 + fromSeq.getName());
2931 int fromCol = fromSeq.findIndex(fromRange[i]);
2932 int mappedCharPos = range[0];
2935 * walk over the 'from' aligned sequence in forward or reverse
2936 * direction; when a non-gap is found, record the column position
2937 * of the next character of the mapped-to sequence; stop when all
2938 * the characters of the range have been counted
2940 while (mappedCharPos <= range[1] && fromCol <= fromChars.length
2943 if (!Comparison.isGap(fromChars[fromCol - 1]))
2946 * mapped from sequence has a character in this column
2947 * record the column position for the mapped to character
2949 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2950 if (seqsMap == null)
2952 seqsMap = new HashMap<SequenceI, Character>();
2953 map.put(fromCol, seqsMap);
2955 seqsMap.put(seq, toChars[mappedCharPos - toStart]);
2958 fromCol += (forward ? 1 : -1);
2965 // strictly temporary hack until proper criteria for aligning protein to cds
2966 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2967 public static boolean looksLikeEnsembl(AlignmentI alignment)
2969 for (SequenceI seq : alignment.getSequences())
2971 String name = seq.getName();
2972 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))