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
175 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
176 .toLowerCase().toCharArray();
177 char[] downstream = new String(
178 ds.getSequence(s_end - 1, s_end + dstream_ds)).toLowerCase()
180 char[] coreseq = s.getSequence();
181 char[] nseq = new char[offset + upstream.length + downstream.length
183 char c = core.getGapCharacter();
186 for (; p < offset; p++)
191 System.arraycopy(upstream, 0, nseq, p, upstream.length);
192 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
194 System.arraycopy(downstream, 0, nseq,
195 p + coreseq.length + upstream.length, downstream.length);
196 s.setSequence(new String(nseq));
197 s.setStart(s.getStart() - ustream_ds);
198 s.setEnd(s_end + downstream.length);
200 AlignmentI newAl = new jalview.datamodel.Alignment(
201 sq.toArray(new SequenceI[0]));
202 for (SequenceI s : sq)
204 if (s.getAnnotation() != null)
206 for (AlignmentAnnotation aa : s.getAnnotation())
208 aa.adjustForAlignment(); // JAL-1712 fix
209 newAl.addAnnotation(aa);
213 newAl.setDataset(core.getDataset());
218 * Returns the index (zero-based position) of a sequence in an alignment, or
225 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
229 for (SequenceI alSeq : al.getSequences())
242 * Returns a map of lists of sequences in the alignment, keyed by sequence
243 * name. For use in mapping between different alignment views of the same
246 * @see jalview.datamodel.AlignmentI#getSequencesByName()
248 public static Map<String, List<SequenceI>> getSequencesByName(
251 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
252 for (SequenceI seq : al.getSequences())
254 String name = seq.getName();
257 List<SequenceI> seqs = theMap.get(name);
260 seqs = new ArrayList<SequenceI>();
261 theMap.put(name, seqs);
270 * Build mapping of protein to cDNA alignment. Mappings are made between
271 * sequences where the cDNA translates to the protein sequence. Any new
272 * mappings are added to the protein alignment. Returns true if any mappings
273 * either already exist or were added, else false.
275 * @param proteinAlignment
276 * @param cdnaAlignment
279 public static boolean mapProteinAlignmentToCdna(
280 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
282 if (proteinAlignment == null || cdnaAlignment == null)
287 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
288 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
291 * First pass - map sequences where cross-references exist. This include
292 * 1-to-many mappings to support, for example, variant cDNA.
294 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
295 cdnaAlignment, mappedDna, mappedProtein, true);
298 * Second pass - map sequences where no cross-references exist. This only
299 * does 1-to-1 mappings and assumes corresponding sequences are in the same
300 * order in the alignments.
302 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
303 mappedDna, mappedProtein, false);
304 return mappingPerformed;
308 * Make mappings between compatible sequences (where the cDNA translation
309 * matches the protein).
311 * @param proteinAlignment
312 * @param cdnaAlignment
314 * a set of mapped DNA sequences (to add to)
315 * @param mappedProtein
316 * a set of mapped Protein sequences (to add to)
318 * if true, only map sequences where xrefs exist
321 protected static boolean mapProteinToCdna(
322 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
323 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
326 boolean mappingExistsOrAdded = false;
327 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
328 for (SequenceI aaSeq : thisSeqs)
330 boolean proteinMapped = false;
331 AlignedCodonFrame acf = new AlignedCodonFrame();
333 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
336 * Always try to map if sequences have xref to each other; this supports
337 * variant cDNA or alternative splicing for a protein sequence.
339 * If no xrefs, try to map progressively, assuming that alignments have
340 * mappable sequences in corresponding order. These are not
341 * many-to-many, as that would risk mixing species with similar cDNA
344 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
350 * Don't map non-xrefd sequences more than once each. This heuristic
351 * allows us to pair up similar sequences in ordered alignments.
353 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
354 || mappedDna.contains(cdnaSeq)))
358 if (mappingExists(proteinAlignment.getCodonFrames(),
359 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
361 mappingExistsOrAdded = true;
365 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
368 acf.addMap(cdnaSeq, aaSeq, map);
369 mappingExistsOrAdded = true;
370 proteinMapped = true;
371 mappedDna.add(cdnaSeq);
372 mappedProtein.add(aaSeq);
378 proteinAlignment.addCodonFrame(acf);
381 return mappingExistsOrAdded;
385 * Answers true if the mappings include one between the given (dataset)
388 public static boolean mappingExists(List<AlignedCodonFrame> mappings,
389 SequenceI aaSeq, SequenceI cdnaSeq)
391 if (mappings != null)
393 for (AlignedCodonFrame acf : mappings)
395 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
405 * Builds a mapping (if possible) of a cDNA to a protein sequence.
407 * <li>first checks if the cdna translates exactly to the protein
409 * <li>else checks for translation after removing a STOP codon</li>
410 * <li>else checks for translation after removing a START codon</li>
411 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
413 * Returns null if no mapping is determined.
416 * the aligned protein sequence
418 * the aligned cdna sequence
421 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
425 * Here we handle either dataset sequence set (desktop) or absent (applet).
426 * Use only the char[] form of the sequence to avoid creating possibly large
429 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
430 char[] aaSeqChars = proteinDataset != null
431 ? proteinDataset.getSequence()
432 : proteinSeq.getSequence();
433 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
434 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
435 : cdnaSeq.getSequence();
436 if (aaSeqChars == null || cdnaSeqChars == null)
442 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
444 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
445 int cdnaLength = cdnaSeqChars.length;
446 int cdnaStart = cdnaSeq.getStart();
447 int cdnaEnd = cdnaSeq.getEnd();
448 final int proteinStart = proteinSeq.getStart();
449 final int proteinEnd = proteinSeq.getEnd();
452 * If lengths don't match, try ignoring stop codon (if present)
454 if (cdnaLength != mappedLength && cdnaLength > 2)
456 String lastCodon = String.valueOf(cdnaSeqChars,
457 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
458 for (String stop : ResidueProperties.STOP)
460 if (lastCodon.equals(stop))
462 cdnaEnd -= CODON_LENGTH;
463 cdnaLength -= CODON_LENGTH;
470 * If lengths still don't match, try ignoring start codon.
473 if (cdnaLength != mappedLength && cdnaLength > 2
474 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
475 .equals(ResidueProperties.START))
477 startOffset += CODON_LENGTH;
478 cdnaStart += CODON_LENGTH;
479 cdnaLength -= CODON_LENGTH;
482 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
485 * protein is translation of dna (+/- start/stop codons)
487 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
489 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
494 * translation failed - try mapping CDS annotated regions of dna
496 return mapCdsToProtein(cdnaSeq, proteinSeq);
500 * Test whether the given cdna sequence, starting at the given offset,
501 * translates to the given amino acid sequence, using the standard translation
502 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
504 * @param cdnaSeqChars
509 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
512 if (cdnaSeqChars == null || aaSeqChars == null)
518 int dnaPos = cdnaStart;
519 for (; dnaPos < cdnaSeqChars.length - 2
520 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
522 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
523 final String translated = ResidueProperties.codonTranslate(codon);
526 * allow * in protein to match untranslatable in dna
528 final char aaRes = aaSeqChars[aaPos];
529 if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
533 if (translated == null || !(aaRes == translated.charAt(0)))
536 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
537 // + codon + "(" + translated + ") != " + aaRes));
543 * check we matched all of the protein sequence
545 if (aaPos != aaSeqChars.length)
551 * check we matched all of the dna except
552 * for optional trailing STOP codon
554 if (dnaPos == cdnaSeqChars.length)
558 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
560 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
561 if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
570 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
571 * currently assumes that we are aligning cDNA to match protein.
574 * the sequence to be realigned
576 * the alignment whose sequence alignment is to be 'copied'
578 * character string represent a gap in the realigned sequence
579 * @param preserveUnmappedGaps
580 * @param preserveMappedGaps
581 * @return true if the sequence was realigned, false if it could not be
583 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
584 String gap, boolean preserveMappedGaps,
585 boolean preserveUnmappedGaps)
588 * Get any mappings from the source alignment to the target (dataset)
591 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
592 // all mappings. Would it help to constrain this?
593 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
594 if (mappings == null || mappings.isEmpty())
600 * Locate the aligned source sequence whose dataset sequence is mapped. We
601 * just take the first match here (as we can't align like more than one
604 SequenceI alignFrom = null;
605 AlignedCodonFrame mapping = null;
606 for (AlignedCodonFrame mp : mappings)
608 alignFrom = mp.findAlignedSequence(seq, al);
609 if (alignFrom != null)
616 if (alignFrom == null)
620 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
621 preserveMappedGaps, preserveUnmappedGaps);
626 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
627 * match residues and codons. Flags control whether existing gaps in unmapped
628 * (intron) and mapped (exon) regions are preserved or not. Gaps between
629 * intron and exon are only retained if both flags are set.
636 * @param preserveUnmappedGaps
637 * @param preserveMappedGaps
639 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
640 AlignedCodonFrame mapping, String myGap, char sourceGap,
641 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
643 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
645 // aligned and dataset sequence positions, all base zero
649 int basesWritten = 0;
650 char myGapChar = myGap.charAt(0);
651 int ratio = myGap.length();
653 int fromOffset = alignFrom.getStart() - 1;
654 int toOffset = alignTo.getStart() - 1;
655 int sourceGapMappedLength = 0;
656 boolean inExon = false;
657 final char[] thisSeq = alignTo.getSequence();
658 final char[] thatAligned = alignFrom.getSequence();
659 StringBuilder thisAligned = new StringBuilder(2 * thisSeq.length);
662 * Traverse the 'model' aligned sequence
664 for (char sourceChar : thatAligned)
666 if (sourceChar == sourceGap)
668 sourceGapMappedLength += ratio;
673 * Found a non-gap character. Locate its mapped region if any.
676 // Note mapping positions are base 1, our sequence positions base 0
677 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
678 sourceDsPos + fromOffset);
679 if (mappedPos == null)
682 * unmapped position; treat like a gap
684 sourceGapMappedLength += ratio;
685 // System.err.println("Can't align: no codon mapping to residue "
686 // + sourceDsPos + "(" + sourceChar + ")");
691 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
692 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
693 StringBuilder trailingCopiedGap = new StringBuilder();
696 * Copy dna sequence up to and including this codon. Optionally, include
697 * gaps before the codon starts (in introns) and/or after the codon starts
700 * Note this only works for 'linear' splicing, not reverse or interleaved.
701 * But then 'align dna as protein' doesn't make much sense otherwise.
703 int intronLength = 0;
704 while (basesWritten + toOffset < mappedCodonEnd
705 && thisSeqPos < thisSeq.length)
707 final char c = thisSeq[thisSeqPos++];
711 int sourcePosition = basesWritten + toOffset;
712 if (sourcePosition < mappedCodonStart)
715 * Found an unmapped (intron) base. First add in any preceding gaps
718 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
720 thisAligned.append(trailingCopiedGap.toString());
721 intronLength += trailingCopiedGap.length();
722 trailingCopiedGap = new StringBuilder();
729 final boolean startOfCodon = sourcePosition == mappedCodonStart;
730 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
731 preserveUnmappedGaps, sourceGapMappedLength, inExon,
732 trailingCopiedGap.length(), intronLength, startOfCodon);
733 for (int i = 0; i < gapsToAdd; i++)
735 thisAligned.append(myGapChar);
737 sourceGapMappedLength = 0;
740 thisAligned.append(c);
741 trailingCopiedGap = new StringBuilder();
745 if (inExon && preserveMappedGaps)
747 trailingCopiedGap.append(myGapChar);
749 else if (!inExon && preserveUnmappedGaps)
751 trailingCopiedGap.append(myGapChar);
758 * At end of model aligned sequence. Copy any remaining target sequence, optionally
759 * including (intron) gaps.
761 while (thisSeqPos < thisSeq.length)
763 final char c = thisSeq[thisSeqPos++];
764 if (c != myGapChar || preserveUnmappedGaps)
766 thisAligned.append(c);
768 sourceGapMappedLength--;
772 * finally add gaps to pad for any trailing source gaps or
773 * unmapped characters
775 if (preserveUnmappedGaps)
777 while (sourceGapMappedLength > 0)
779 thisAligned.append(myGapChar);
780 sourceGapMappedLength--;
785 * All done aligning, set the aligned sequence.
787 alignTo.setSequence(new String(thisAligned));
791 * Helper method to work out how many gaps to insert when realigning.
793 * @param preserveMappedGaps
794 * @param preserveUnmappedGaps
795 * @param sourceGapMappedLength
797 * @param trailingCopiedGap
798 * @param intronLength
799 * @param startOfCodon
802 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
803 boolean preserveUnmappedGaps, int sourceGapMappedLength,
804 boolean inExon, int trailingGapLength, int intronLength,
805 final boolean startOfCodon)
811 * Reached start of codon. Ignore trailing gaps in intron unless we are
812 * preserving gaps in both exon and intron. Ignore them anyway if the
813 * protein alignment introduces a gap at least as large as the intronic
816 if (inExon && !preserveMappedGaps)
818 trailingGapLength = 0;
820 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
822 trailingGapLength = 0;
826 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
830 if (intronLength + trailingGapLength <= sourceGapMappedLength)
832 gapsToAdd = sourceGapMappedLength - intronLength;
836 gapsToAdd = Math.min(
837 intronLength + trailingGapLength - sourceGapMappedLength,
845 * second or third base of codon; check for any gaps in dna
847 if (!preserveMappedGaps)
849 trailingGapLength = 0;
851 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
857 * Realigns the given protein to match the alignment of the dna, using codon
858 * mappings to translate aligned codon positions to protein residues.
861 * the alignment whose sequences are realigned by this method
863 * the dna alignment whose alignment we are 'copying'
864 * @return the number of sequences that were realigned
866 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
868 if (protein.isNucleotide() || !dna.isNucleotide())
870 System.err.println("Wrong alignment type in alignProteinAsDna");
873 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
874 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
875 protein, dna, unmappedProtein);
876 return alignProteinAs(protein, alignedCodons, unmappedProtein);
880 * Realigns the given dna to match the alignment of the protein, using codon
881 * mappings to translate aligned peptide positions to codons.
883 * Always produces a padded CDS alignment.
886 * the alignment whose sequences are realigned by this method
888 * the protein alignment whose alignment we are 'copying'
889 * @return the number of sequences that were realigned
891 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
893 if (protein.isNucleotide() || !dna.isNucleotide())
895 System.err.println("Wrong alignment type in alignProteinAsDna");
898 // todo: implement this
899 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
900 int alignedCount = 0;
901 int width = 0; // alignment width for padding CDS
902 for (SequenceI dnaSeq : dna.getSequences())
904 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
905 dna.getGapCharacter()))
909 width = Math.max(dnaSeq.getLength(), width);
913 for (SequenceI dnaSeq : dna.getSequences())
915 oldwidth = dnaSeq.getLength();
916 diff = width - oldwidth;
919 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
926 * Helper method to align (if possible) the dna sequence to match the
927 * alignment of a mapped protein sequence. This is currently limited to
928 * handling coding sequence only.
936 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
937 AlignmentI protein, List<AlignedCodonFrame> mappings,
940 SequenceI cdsDss = cdsSeq.getDatasetSequence();
944 .println("alignCdsSequenceAsProtein needs aligned sequence!");
948 List<AlignedCodonFrame> dnaMappings = MappingUtils
949 .findMappingsForSequence(cdsSeq, mappings);
950 for (AlignedCodonFrame mapping : dnaMappings)
952 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
955 int peptideLength = peptide.getLength();
956 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
959 MapList mapList = map.getMap();
960 if (map.getTo() == peptide.getDatasetSequence())
962 mapList = mapList.getInverse();
964 int cdsLength = cdsDss.getLength();
965 int mappedFromLength = MappingUtils
966 .getLength(mapList.getFromRanges());
967 int mappedToLength = MappingUtils
968 .getLength(mapList.getToRanges());
969 boolean addStopCodon = (cdsLength == mappedFromLength
970 * CODON_LENGTH + CODON_LENGTH)
971 || (peptide.getDatasetSequence()
972 .getLength() == mappedFromLength - 1);
973 if (cdsLength != mappedToLength && !addStopCodon)
975 System.err.println(String.format(
976 "Can't align cds as protein (length mismatch %d/%d): %s",
977 cdsLength, mappedToLength, cdsSeq.getName()));
981 * pre-fill the aligned cds sequence with gaps
983 char[] alignedCds = new char[peptideLength * CODON_LENGTH
984 + (addStopCodon ? CODON_LENGTH : 0)];
985 Arrays.fill(alignedCds, gapChar);
988 * walk over the aligned peptide sequence and insert mapped
989 * codons for residues in the aligned cds sequence
991 char[] alignedPeptide = peptide.getSequence();
992 char[] nucleotides = cdsDss.getSequence();
994 int cdsStart = cdsDss.getStart();
995 int proteinPos = peptide.getStart() - 1;
997 for (char residue : alignedPeptide)
999 if (Comparison.isGap(residue))
1001 cdsCol += CODON_LENGTH;
1006 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1009 // e.g. incomplete start codon, X in peptide
1010 cdsCol += CODON_LENGTH;
1014 for (int j = codon[0]; j <= codon[1]; j++)
1016 char mappedBase = nucleotides[j - cdsStart];
1017 alignedCds[cdsCol++] = mappedBase;
1025 * append stop codon if not mapped from protein,
1026 * closing it up to the end of the mapped sequence
1028 if (copiedBases == nucleotides.length - CODON_LENGTH)
1030 for (int i = alignedCds.length - 1; i >= 0; i--)
1032 if (!Comparison.isGap(alignedCds[i]))
1034 cdsCol = i + 1; // gap just after end of sequence
1038 for (int i = nucleotides.length
1039 - CODON_LENGTH; i < nucleotides.length; i++)
1041 alignedCds[cdsCol++] = nucleotides[i];
1044 cdsSeq.setSequence(new String(alignedCds));
1053 * Builds a map whose key is an aligned codon position (3 alignment column
1054 * numbers base 0), and whose value is a map from protein sequence to each
1055 * protein's peptide residue for that codon. The map generates an ordering of
1056 * the codons, and allows us to read off the peptides at each position in
1057 * order to assemble 'aligned' protein sequences.
1060 * the protein alignment
1062 * the coding dna alignment
1063 * @param unmappedProtein
1064 * any unmapped proteins are added to this list
1067 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1068 AlignmentI protein, AlignmentI dna,
1069 List<SequenceI> unmappedProtein)
1072 * maintain a list of any proteins with no mappings - these will be
1073 * rendered 'as is' in the protein alignment as we can't align them
1075 unmappedProtein.addAll(protein.getSequences());
1077 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1080 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1081 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1082 * comparator keeps the codon positions ordered.
1084 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
1085 new CodonComparator());
1087 for (SequenceI dnaSeq : dna.getSequences())
1089 for (AlignedCodonFrame mapping : mappings)
1091 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1094 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1095 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1097 unmappedProtein.remove(prot);
1103 * Finally add any unmapped peptide start residues (e.g. for incomplete
1104 * codons) as if at the codon position before the second residue
1106 // TODO resolve JAL-2022 so this fudge can be removed
1107 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1108 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1110 return alignedCodons;
1114 * Scans for any protein mapped from position 2 (meaning unmapped start
1115 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1116 * preceding position in the alignment
1118 * @param alignedCodons
1119 * the codon-to-peptide map
1120 * @param mappedSequenceCount
1121 * the number of distinct sequences in the map
1123 protected static void addUnmappedPeptideStarts(
1124 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1125 int mappedSequenceCount)
1127 // TODO delete this ugly hack once JAL-2022 is resolved
1128 // i.e. we can model startPhase > 0 (incomplete start codon)
1130 List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
1131 AlignedCodon lastCodon = null;
1132 Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
1134 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1137 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1140 SequenceI seq = sequenceCodon.getKey();
1141 if (sequencesChecked.contains(seq))
1145 sequencesChecked.add(seq);
1146 AlignedCodon codon = sequenceCodon.getValue();
1147 if (codon.peptideCol > 1)
1150 "Problem mapping protein with >1 unmapped start positions: "
1153 else if (codon.peptideCol == 1)
1156 * first position (peptideCol == 0) was unmapped - add it
1158 if (lastCodon != null)
1160 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1161 lastCodon.pos2, lastCodon.pos3,
1162 String.valueOf(seq.getCharAt(0)), 0);
1163 toAdd.put(seq, firstPeptide);
1168 * unmapped residue at start of alignment (no prior column) -
1169 * 'insert' at nominal codon [0, 0, 0]
1171 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1172 String.valueOf(seq.getCharAt(0)), 0);
1173 toAdd.put(seq, firstPeptide);
1176 if (sequencesChecked.size() == mappedSequenceCount)
1178 // no need to check past first mapped position in all sequences
1182 lastCodon = entry.getKey();
1186 * add any new codons safely after iterating over the map
1188 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1190 addCodonToMap(alignedCodons, startCodon.getValue(),
1191 startCodon.getKey());
1196 * Update the aligned protein sequences to match the codon alignments given in
1200 * @param alignedCodons
1201 * an ordered map of codon positions (columns), with sequence/peptide
1202 * values present in each column
1203 * @param unmappedProtein
1206 protected static int alignProteinAs(AlignmentI protein,
1207 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1208 List<SequenceI> unmappedProtein)
1211 * Prefill aligned sequences with gaps before inserting aligned protein
1214 int alignedWidth = alignedCodons.size();
1215 char[] gaps = new char[alignedWidth];
1216 Arrays.fill(gaps, protein.getGapCharacter());
1217 String allGaps = String.valueOf(gaps);
1218 for (SequenceI seq : protein.getSequences())
1220 if (!unmappedProtein.contains(seq))
1222 seq.setSequence(allGaps);
1227 for (AlignedCodon codon : alignedCodons.keySet())
1229 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1231 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1233 // place translated codon at its column position in sequence
1234 entry.getKey().getSequence()[column] = entry.getValue().product
1243 * Populate the map of aligned codons by traversing the given sequence
1244 * mapping, locating the aligned positions of mapped codons, and adding those
1245 * positions and their translation products to the map.
1248 * the aligned sequence we are mapping from
1250 * the sequence to be aligned to the codons
1252 * the gap character in the dna sequence
1254 * a mapping to a sequence translation
1255 * @param alignedCodons
1256 * the map we are building up
1258 static void addCodonPositions(SequenceI dna, SequenceI protein,
1259 char gapChar, Mapping seqMap,
1260 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1262 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1265 * add codon positions, and their peptide translations, to the alignment
1266 * map, while remembering the first codon mapped
1268 while (codons.hasNext())
1272 AlignedCodon codon = codons.next();
1273 addCodonToMap(alignedCodons, codon, protein);
1274 } catch (IncompleteCodonException e)
1276 // possible incomplete trailing codon - ignore
1277 } catch (NoSuchElementException e)
1279 // possibly peptide lacking STOP
1285 * Helper method to add a codon-to-peptide entry to the aligned codons map
1287 * @param alignedCodons
1291 protected static void addCodonToMap(
1292 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1293 AlignedCodon codon, SequenceI protein)
1295 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1296 if (seqProduct == null)
1298 seqProduct = new HashMap<SequenceI, AlignedCodon>();
1299 alignedCodons.put(codon, seqProduct);
1301 seqProduct.put(protein, codon);
1305 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1306 * between at least one pair of sequences in the two alignments. Currently,
1309 * <li>One alignment must be nucleotide, and the other protein</li>
1310 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1311 * <li>Mappable means the nucleotide translation matches the protein
1313 * <li>The translation may ignore start and stop codons if present in the
1321 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1323 if (al1 == null || al2 == null)
1329 * Require one nucleotide and one protein
1331 if (al1.isNucleotide() == al2.isNucleotide())
1335 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1336 AlignmentI protein = dna == al1 ? al2 : al1;
1337 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1338 for (SequenceI dnaSeq : dna.getSequences())
1340 for (SequenceI proteinSeq : protein.getSequences())
1342 if (isMappable(dnaSeq, proteinSeq, mappings))
1352 * Returns true if the dna sequence is mapped, or could be mapped, to the
1360 protected static boolean isMappable(SequenceI dnaSeq,
1361 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1363 if (dnaSeq == null || proteinSeq == null)
1368 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1369 : dnaSeq.getDatasetSequence();
1370 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1372 : proteinSeq.getDatasetSequence();
1374 for (AlignedCodonFrame mapping : mappings)
1376 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1386 * Just try to make a mapping (it is not yet stored), test whether
1389 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1393 * Finds any reference annotations associated with the sequences in
1394 * sequenceScope, that are not already added to the alignment, and adds them
1395 * to the 'candidates' map. Also populates a lookup table of annotation
1396 * labels, keyed by calcId, for use in constructing tooltips or the like.
1398 * @param sequenceScope
1399 * the sequences to scan for reference annotations
1400 * @param labelForCalcId
1401 * (optional) map to populate with label for calcId
1403 * map to populate with annotations for sequence
1405 * the alignment to check for presence of annotations
1407 public static void findAddableReferenceAnnotations(
1408 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1409 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1412 if (sequenceScope == null)
1418 * For each sequence in scope, make a list of any annotations on the
1419 * underlying dataset sequence which are not already on the alignment.
1421 * Add to a map of { alignmentSequence, <List of annotations to add> }
1423 for (SequenceI seq : sequenceScope)
1425 SequenceI dataset = seq.getDatasetSequence();
1426 if (dataset == null)
1430 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1431 if (datasetAnnotations == null)
1435 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1436 for (AlignmentAnnotation dsann : datasetAnnotations)
1439 * Find matching annotations on the alignment. If none is found, then
1440 * add this annotation to the list of 'addable' annotations for this
1443 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1444 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1445 if (!matchedAlignmentAnnotations.iterator().hasNext())
1448 if (labelForCalcId != null)
1450 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1455 * Save any addable annotations for this sequence
1457 if (!result.isEmpty())
1459 candidates.put(seq, result);
1465 * Adds annotations to the top of the alignment annotations, in the same order
1466 * as their related sequences.
1468 * @param annotations
1469 * the annotations to add
1471 * the alignment to add them to
1472 * @param selectionGroup
1473 * current selection group (or null if none)
1475 public static void addReferenceAnnotations(
1476 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1477 final AlignmentI alignment, final SequenceGroup selectionGroup)
1479 for (SequenceI seq : annotations.keySet())
1481 for (AlignmentAnnotation ann : annotations.get(seq))
1483 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1485 int endRes = ann.annotations.length;
1486 if (selectionGroup != null)
1488 startRes = selectionGroup.getStartRes();
1489 endRes = selectionGroup.getEndRes();
1491 copyAnn.restrict(startRes, endRes);
1494 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1495 * original annotation is already on the sequence.
1497 if (!seq.hasAnnotation(ann))
1499 seq.addAlignmentAnnotation(copyAnn);
1502 copyAnn.adjustForAlignment();
1503 // add to the alignment and set visible
1504 alignment.addAnnotation(copyAnn);
1505 copyAnn.visible = true;
1511 * Set visibility of alignment annotations of specified types (labels), for
1512 * specified sequences. This supports controls like "Show all secondary
1513 * structure", "Hide all Temp factor", etc.
1515 * @al the alignment to scan for annotations
1517 * the types (labels) of annotations to be updated
1518 * @param forSequences
1519 * if not null, only annotations linked to one of these sequences are
1520 * in scope for update; if null, acts on all sequence annotations
1522 * if this flag is true, 'types' is ignored (label not checked)
1524 * if true, set visibility on, else set off
1526 public static void showOrHideSequenceAnnotations(AlignmentI al,
1527 Collection<String> types, List<SequenceI> forSequences,
1528 boolean anyType, boolean doShow)
1530 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1533 for (AlignmentAnnotation aa : anns)
1535 if (anyType || types.contains(aa.label))
1537 if ((aa.sequenceRef != null) && (forSequences == null
1538 || forSequences.contains(aa.sequenceRef)))
1540 aa.visible = doShow;
1548 * Returns true if either sequence has a cross-reference to the other
1554 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1556 // Note: moved here from class CrossRef as the latter class has dependencies
1557 // not availability to the applet's classpath
1558 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1562 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1563 * that sequence name is structured as Source|AccessionId.
1569 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1571 if (seq1 == null || seq2 == null)
1575 String name = seq2.getName();
1576 final DBRefEntry[] xrefs = seq1.getDBRefs();
1579 for (DBRefEntry xref : xrefs)
1581 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1582 // case-insensitive test, consistent with DBRefEntry.equalRef()
1583 if (xrefName.equalsIgnoreCase(name))
1593 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1594 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1595 * added to the original alignment's dataset, which is shared by the new
1596 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1597 * added to the alignment dataset.
1600 * aligned nucleotide (dna or cds) sequences
1602 * the alignment dataset the sequences belong to
1604 * (optional) to restrict results to CDS that map to specified
1606 * @return an alignment whose sequences are the cds-only parts of the dna
1607 * sequences (or null if no mappings are found)
1609 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1610 AlignmentI dataset, SequenceI[] products)
1612 if (dataset == null || dataset.getDataset() != null)
1614 throw new IllegalArgumentException(
1615 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1617 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
1618 List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
1619 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1620 HashSet<SequenceI> productSeqs = null;
1621 if (products != null)
1623 productSeqs = new HashSet<SequenceI>();
1624 for (SequenceI seq : products)
1626 productSeqs.add(seq.getDatasetSequence() == null ? seq
1627 : seq.getDatasetSequence());
1632 * Construct CDS sequences from mappings on the alignment dataset.
1634 * - find the protein product(s) mapped to from each dna sequence
1635 * - if the mapping covers the whole dna sequence (give or take start/stop
1636 * codon), take the dna as the CDS sequence
1637 * - else search dataset mappings for a suitable dna sequence, i.e. one
1638 * whose whole sequence is mapped to the protein
1639 * - if no sequence found, construct one from the dna sequence and mapping
1640 * (and add it to dataset so it is found if this is repeated)
1642 for (SequenceI dnaSeq : dna)
1644 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1645 : dnaSeq.getDatasetSequence();
1647 List<AlignedCodonFrame> seqMappings = MappingUtils
1648 .findMappingsForSequence(dnaSeq, mappings);
1649 for (AlignedCodonFrame mapping : seqMappings)
1651 List<Mapping> mappingsFromSequence = mapping
1652 .getMappingsFromSequence(dnaSeq);
1654 for (Mapping aMapping : mappingsFromSequence)
1656 MapList mapList = aMapping.getMap();
1657 if (mapList.getFromRatio() == 1)
1660 * not a dna-to-protein mapping (likely dna-to-cds)
1666 * skip if mapping is not to one of the target set of proteins
1668 SequenceI proteinProduct = aMapping.getTo();
1669 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1675 * try to locate the CDS from the dataset mappings;
1676 * guard against duplicate results (for the case that protein has
1677 * dbrefs to both dna and cds sequences)
1679 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1680 seqMappings, aMapping);
1683 if (!foundSeqs.contains(cdsSeq))
1685 foundSeqs.add(cdsSeq);
1686 SequenceI derivedSequence = cdsSeq.deriveSequence();
1687 cdsSeqs.add(derivedSequence);
1688 if (!dataset.getSequences().contains(cdsSeq))
1690 dataset.addSequence(cdsSeq);
1697 * didn't find mapped CDS sequence - construct it and add
1698 * its dataset sequence to the dataset
1700 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1701 dataset).deriveSequence();
1702 // cdsSeq has a name constructed as CDS|<dbref>
1703 // <dbref> will be either the accession for the coding sequence,
1704 // marked in the /via/ dbref to the protein product accession
1705 // or it will be the original nucleotide accession.
1706 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1708 cdsSeqs.add(cdsSeq);
1710 if (!dataset.getSequences().contains(cdsSeqDss))
1712 // check if this sequence is a newly created one
1713 // so needs adding to the dataset
1714 dataset.addSequence(cdsSeqDss);
1718 * add a mapping from CDS to the (unchanged) mapped to range
1720 List<int[]> cdsRange = Collections
1721 .singletonList(new int[]
1722 { 1, cdsSeq.getLength() });
1723 MapList cdsToProteinMap = new MapList(cdsRange,
1724 mapList.getToRanges(), mapList.getFromRatio(),
1725 mapList.getToRatio());
1726 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1727 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1731 * guard against duplicating the mapping if repeating this action
1733 if (!mappings.contains(cdsToProteinMapping))
1735 mappings.add(cdsToProteinMapping);
1738 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1739 proteinProduct, aMapping);
1741 * add another mapping from original 'from' range to CDS
1743 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1744 MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1746 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1748 if (!mappings.contains(dnaToCdsMapping))
1750 mappings.add(dnaToCdsMapping);
1754 * add DBRef with mapping from protein to CDS
1755 * (this enables Get Cross-References from protein alignment)
1756 * This is tricky because we can't have two DBRefs with the
1757 * same source and accession, so need a different accession for
1758 * the CDS from the dna sequence
1761 // specific use case:
1762 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1763 // ENSG02, ENSG03, with transcripts and products similarly named.
1764 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1766 // JBPNote: ?? can't actually create an example that demonstrates we
1768 // synthesize an xref.
1770 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1772 // creates a complementary cross-reference to the source sequence's
1773 // primary reference.
1775 DBRefEntry cdsCrossRef = new DBRefEntry(primRef.getSource(),
1776 primRef.getSource() + ":" + primRef.getVersion(),
1777 primRef.getAccessionId());
1779 .setMap(new Mapping(dnaDss, new MapList(dnaToCdsMap)));
1780 cdsSeqDss.addDBRef(cdsCrossRef);
1782 // problem here is that the cross-reference is synthesized -
1783 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1785 // assuming cds version same as dna ?!?
1787 DBRefEntry proteinToCdsRef = new DBRefEntry(primRef.getSource(),
1788 primRef.getVersion(), cdsSeq.getName());
1790 proteinToCdsRef.setMap(
1791 new Mapping(cdsSeqDss, cdsToProteinMap.getInverse()));
1792 proteinProduct.addDBRef(proteinToCdsRef);
1796 * transfer any features on dna that overlap the CDS
1798 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1799 SequenceOntologyI.CDS);
1804 AlignmentI cds = new Alignment(
1805 cdsSeqs.toArray(new SequenceI[cdsSeqs.size()]));
1806 cds.setDataset(dataset);
1812 * A helper method that finds a CDS sequence in the alignment dataset that is
1813 * mapped to the given protein sequence, and either is, or has a mapping from,
1814 * the given dna sequence.
1817 * set of all mappings on the dataset
1819 * a dna (or cds) sequence we are searching from
1820 * @param seqMappings
1821 * the set of mappings involving dnaSeq
1823 * an initial candidate from seqMappings
1826 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1827 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1831 * TODO a better dna-cds-protein mapping data representation to allow easy
1832 * navigation; until then this clunky looping around lists of mappings
1834 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1835 : dnaSeq.getDatasetSequence();
1836 SequenceI proteinProduct = aMapping.getTo();
1839 * is this mapping from the whole dna sequence (i.e. CDS)?
1840 * allowing for possible stop codon on dna but not peptide
1842 int mappedFromLength = MappingUtils
1843 .getLength(aMapping.getMap().getFromRanges());
1844 int dnaLength = seqDss.getLength();
1845 if (mappedFromLength == dnaLength
1846 || mappedFromLength == dnaLength - CODON_LENGTH)
1852 * looks like we found the dna-to-protein mapping; search for the
1853 * corresponding cds-to-protein mapping
1855 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1856 .findMappingsForSequence(proteinProduct, mappings);
1857 for (AlignedCodonFrame acf : mappingsToPeptide)
1859 for (SequenceToSequenceMapping map : acf.getMappings())
1861 Mapping mapping = map.getMapping();
1862 if (mapping != aMapping
1863 && mapping.getMap().getFromRatio() == CODON_LENGTH
1864 && proteinProduct == mapping.getTo()
1865 && seqDss != map.getFromSeq())
1867 mappedFromLength = MappingUtils
1868 .getLength(mapping.getMap().getFromRanges());
1869 if (mappedFromLength == map.getFromSeq().getLength())
1872 * found a 3:1 mapping to the protein product which covers
1873 * the whole dna sequence i.e. is from CDS; finally check it
1874 * is from the dna start sequence
1876 SequenceI cdsSeq = map.getFromSeq();
1877 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1878 .findMappingsForSequence(cdsSeq, seqMappings);
1879 if (!dnaToCdsMaps.isEmpty())
1891 * Helper method that makes a CDS sequence as defined by the mappings from the
1892 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1893 * forward or reverse strand).
1898 * - existing dataset. We check for sequences that look like the CDS
1899 * we are about to construct, if one exists already, then we will
1900 * just return that one.
1901 * @return CDS sequence (as a dataset sequence)
1903 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1906 char[] seqChars = seq.getSequence();
1907 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1908 int cdsWidth = MappingUtils.getLength(fromRanges);
1909 char[] newSeqChars = new char[cdsWidth];
1912 for (int[] range : fromRanges)
1914 if (range[0] <= range[1])
1916 // forward strand mapping - just copy the range
1917 int length = range[1] - range[0] + 1;
1918 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1924 // reverse strand mapping - copy and complement one by one
1925 for (int i = range[0]; i >= range[1]; i--)
1927 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1933 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
1934 * else generate a sequence name
1936 String mapFromId = mapping.getMappedFromId();
1937 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
1938 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
1939 if (dataset != null)
1941 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
1942 if (matches != null)
1944 boolean matched = false;
1945 for (SequenceI mtch : matches)
1947 if (mtch.getStart() != newSeq.getStart())
1951 if (mtch.getEnd() != newSeq.getEnd())
1955 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
1967 "JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
1973 // newSeq.setDescription(mapFromId);
1979 * add any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
1980 * the given mapping.
1985 * @return list of DBRefEntrys added.
1987 public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
1988 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
1991 // gather direct refs from contig congrent with mapping
1992 List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
1993 HashSet<String> directSources = new HashSet<String>();
1994 if (contig.getDBRefs() != null)
1996 for (DBRefEntry dbr : contig.getDBRefs())
1998 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
2000 MapList map = dbr.getMap().getMap();
2001 // check if map is the CDS mapping
2002 if (mapping.getMap().equals(map))
2005 directSources.add(dbr.getSource());
2010 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2011 proteinProduct.getDBRefs(),
2012 directSources.toArray(new String[0]));
2013 List<DBRefEntry> propagated = new ArrayList<DBRefEntry>();
2015 // and generate appropriate mappings
2016 for (DBRefEntry cdsref : direct)
2018 // clone maplist and mapping
2019 MapList cdsposmap = new MapList(
2020 Arrays.asList(new int[][]
2021 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2022 cdsref.getMap().getMap().getToRanges(), 3, 1);
2023 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(),
2024 cdsref.getMap().getMap());
2027 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2028 cdsref.getVersion(), cdsref.getAccessionId(),
2029 new Mapping(cdsmap.getTo(), cdsposmap));
2031 // and see if we can map to the protein product for this mapping.
2032 // onSource is the filtered set of accessions on protein that we are
2033 // tranferring, so we assume accession is the same.
2034 if (cdsmap.getTo() == null && onSource != null)
2036 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2037 cdsref.getAccessionId());
2038 if (sourceRefs != null)
2040 for (DBRefEntry srcref : sourceRefs)
2042 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2044 // we have found a complementary dbref on the protein product, so
2045 // update mapping's getTo
2046 newref.getMap().setTo(proteinProduct);
2051 cdsSeq.addDBRef(newref);
2052 propagated.add(newref);
2058 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2059 * feature start/end ranges, optionally omitting specified feature types.
2060 * Returns the number of features copied.
2065 * if not null, only features of this type are copied (including
2066 * subtypes in the Sequence Ontology)
2068 * the mapping from 'fromSeq' to 'toSeq'
2071 public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2072 MapList mapping, String select, String... omitting)
2074 SequenceI copyTo = toSeq;
2075 while (copyTo.getDatasetSequence() != null)
2077 copyTo = copyTo.getDatasetSequence();
2080 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2082 SequenceFeature[] sfs = fromSeq.getSequenceFeatures();
2085 for (SequenceFeature sf : sfs)
2087 String type = sf.getType();
2088 if (select != null && !so.isA(type, select))
2092 boolean omit = false;
2093 for (String toOmit : omitting)
2095 if (type.equals(toOmit))
2106 * locate the mapped range - null if either start or end is
2107 * not mapped (no partial overlaps are calculated)
2109 int start = sf.getBegin();
2110 int end = sf.getEnd();
2111 int[] mappedTo = mapping.locateInTo(start, end);
2113 * if whole exon range doesn't map, try interpreting it
2114 * as 5' or 3' exon overlapping the CDS range
2116 if (mappedTo == null)
2118 mappedTo = mapping.locateInTo(end, end);
2119 if (mappedTo != null)
2122 * end of exon is in CDS range - 5' overlap
2123 * to a range from the start of the peptide
2128 if (mappedTo == null)
2130 mappedTo = mapping.locateInTo(start, start);
2131 if (mappedTo != null)
2134 * start of exon is in CDS range - 3' overlap
2135 * to a range up to the end of the peptide
2137 mappedTo[1] = toSeq.getLength();
2140 if (mappedTo != null)
2142 SequenceFeature copy = new SequenceFeature(sf);
2143 copy.setBegin(Math.min(mappedTo[0], mappedTo[1]));
2144 copy.setEnd(Math.max(mappedTo[0], mappedTo[1]));
2145 copyTo.addSequenceFeature(copy);
2154 * Returns a mapping from dna to protein by inspecting sequence features of
2155 * type "CDS" on the dna.
2161 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2162 SequenceI proteinSeq)
2164 List<int[]> ranges = findCdsPositions(dnaSeq);
2165 int mappedDnaLength = MappingUtils.getLength(ranges);
2167 int proteinLength = proteinSeq.getLength();
2168 int proteinStart = proteinSeq.getStart();
2169 int proteinEnd = proteinSeq.getEnd();
2172 * incomplete start codon may mean X at start of peptide
2173 * we ignore both for mapping purposes
2175 if (proteinSeq.getCharAt(0) == 'X')
2177 // todo JAL-2022 support startPhase > 0
2181 List<int[]> proteinRange = new ArrayList<int[]>();
2184 * dna length should map to protein (or protein plus stop codon)
2186 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2187 if (codesForResidues == (proteinLength + 1))
2189 // assuming extra codon is for STOP and not in peptide
2192 if (codesForResidues == proteinLength)
2194 proteinRange.add(new int[] { proteinStart, proteinEnd });
2195 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2201 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2202 * start/end positions of sequence features of type "CDS" (or a sub-type of
2203 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2204 * position order, so this method is only valid for linear CDS in the same
2205 * sense as the protein product.
2210 public static List<int[]> findCdsPositions(SequenceI dnaSeq)
2212 List<int[]> result = new ArrayList<int[]>();
2213 SequenceFeature[] sfs = dnaSeq.getSequenceFeatures();
2219 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2222 for (SequenceFeature sf : sfs)
2225 * process a CDS feature (or a sub-type of CDS)
2227 if (so.isA(sf.getType(), SequenceOntologyI.CDS))
2232 phase = Integer.parseInt(sf.getPhase());
2233 } catch (NumberFormatException e)
2238 * phase > 0 on first codon means 5' incomplete - skip to the start
2239 * of the next codon; example ENST00000496384
2241 int begin = sf.getBegin();
2242 int end = sf.getEnd();
2243 if (result.isEmpty())
2248 // shouldn't happen!
2250 "Error: start phase extends beyond start CDS in "
2251 + dnaSeq.getName());
2254 result.add(new int[] { begin, end });
2259 * remove 'startPhase' positions (usually 0) from the first range
2260 * so we begin at the start of a complete codon
2262 if (!result.isEmpty())
2264 // TODO JAL-2022 correctly model start phase > 0
2265 result.get(0)[0] += startPhase;
2269 * Finally sort ranges by start position. This avoids a dependency on
2270 * keeping features in order on the sequence (if they are in order anyway,
2271 * the sort will have almost no work to do). The implicit assumption is CDS
2272 * ranges are assembled in order. Other cases should not use this method,
2273 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2275 Collections.sort(result, new RangeComparator(true));
2280 * Maps exon features from dna to protein, and computes variants in peptide
2281 * product generated by variants in dna, and adds them as sequence_variant
2282 * features on the protein sequence. Returns the number of variant features
2287 * @param dnaToProtein
2289 public static int computeProteinFeatures(SequenceI dnaSeq,
2290 SequenceI peptide, MapList dnaToProtein)
2292 while (dnaSeq.getDatasetSequence() != null)
2294 dnaSeq = dnaSeq.getDatasetSequence();
2296 while (peptide.getDatasetSequence() != null)
2298 peptide = peptide.getDatasetSequence();
2301 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2304 * compute protein variants from dna variants and codon mappings;
2305 * NB - alternatively we could retrieve this using the REST service e.g.
2306 * http://rest.ensembl.org/overlap/translation
2307 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2308 * which would be a bit slower but possibly more reliable
2312 * build a map with codon variations for each potentially varying peptide
2314 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2315 dnaSeq, dnaToProtein);
2318 * scan codon variations, compute peptide variants and add to peptide sequence
2321 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2323 int peptidePos = variant.getKey();
2324 List<DnaVariant>[] codonVariants = variant.getValue();
2325 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2329 * sort to get sequence features in start position order
2330 * - would be better to store in Sequence as a TreeSet or NCList?
2332 if (peptide.getSequenceFeatures() != null)
2334 Arrays.sort(peptide.getSequenceFeatures(),
2335 new Comparator<SequenceFeature>()
2338 public int compare(SequenceFeature o1, SequenceFeature o2)
2340 int c = Integer.compare(o1.getBegin(), o2.getBegin());
2341 return c == 0 ? Integer.compare(o1.getEnd(), o2.getEnd())
2350 * Computes non-synonymous peptide variants from codon variants and adds them
2351 * as sequence_variant features on the protein sequence (one feature per
2352 * allele variant). Selected attributes (variant id, clinical significance)
2353 * are copied over to the new features.
2356 * the protein sequence
2358 * the position to compute peptide variants for
2359 * @param codonVariants
2360 * a list of dna variants per codon position
2361 * @return the number of features added
2363 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2364 List<DnaVariant>[] codonVariants)
2366 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2368 String base1 = codonVariants[0].get(0).base;
2369 String base2 = codonVariants[1].get(0).base;
2370 String base3 = codonVariants[2].get(0).base;
2373 * variants in first codon base
2375 for (DnaVariant var : codonVariants[0])
2377 if (var.variant != null)
2379 String alleles = (String) var.variant.getValue("alleles");
2380 if (alleles != null)
2382 for (String base : alleles.split(","))
2384 String codon = base + base2 + base3;
2385 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2395 * variants in second codon base
2397 for (DnaVariant var : codonVariants[1])
2399 if (var.variant != null)
2401 String alleles = (String) var.variant.getValue("alleles");
2402 if (alleles != null)
2404 for (String base : alleles.split(","))
2406 String codon = base1 + base + base3;
2407 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2417 * variants in third codon base
2419 for (DnaVariant var : codonVariants[2])
2421 if (var.variant != null)
2423 String alleles = (String) var.variant.getValue("alleles");
2424 if (alleles != null)
2426 for (String base : alleles.split(","))
2428 String codon = base1 + base2 + base;
2429 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2442 * Helper method that adds a peptide variant feature, provided the given codon
2443 * translates to a value different to the current residue (is a non-synonymous
2444 * variant). ID and clinical_significance attributes of the dna variant (if
2445 * present) are copied to the new feature.
2452 * @return true if a feature was added, else false
2454 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2455 String residue, DnaVariant var, String codon)
2458 * get peptide translation of codon e.g. GAT -> D
2459 * note that variants which are not single alleles,
2460 * e.g. multibase variants or HGMD_MUTATION etc
2461 * are currently ignored here
2463 String trans = codon.contains("-") ? "-"
2464 : (codon.length() > CODON_LENGTH ? null
2465 : ResidueProperties.codonTranslate(codon));
2466 if (trans != null && !trans.equals(residue))
2468 String residue3Char = StringUtils
2469 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2470 String trans3Char = StringUtils
2471 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2472 String desc = "p." + residue3Char + peptidePos + trans3Char;
2473 // set score to 0f so 'graduated colour' option is offered! JAL-2060
2474 SequenceFeature sf = new SequenceFeature(
2475 SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
2476 peptidePos, 0f, var.getSource());
2477 StringBuilder attributes = new StringBuilder(32);
2478 String id = (String) var.variant.getValue(ID);
2481 if (id.startsWith(SEQUENCE_VARIANT))
2483 id = id.substring(SEQUENCE_VARIANT.length());
2485 sf.setValue(ID, id);
2486 attributes.append(ID).append("=").append(id);
2487 // TODO handle other species variants JAL-2064
2488 StringBuilder link = new StringBuilder(32);
2491 link.append(desc).append(" ").append(id).append(
2492 "|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2493 .append(URLEncoder.encode(id, "UTF-8"));
2494 sf.addLink(link.toString());
2495 } catch (UnsupportedEncodingException e)
2500 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2501 if (clinSig != null)
2503 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2504 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2507 peptide.addSequenceFeature(sf);
2508 if (attributes.length() > 0)
2510 sf.setAttributes(attributes.toString());
2518 * Builds a map whose key is position in the protein sequence, and value is a
2519 * list of the base and all variants for each corresponding codon position
2522 * @param dnaToProtein
2525 @SuppressWarnings("unchecked")
2526 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2527 SequenceI dnaSeq, MapList dnaToProtein)
2530 * map from peptide position to all variants of the codon which codes for it
2531 * LinkedHashMap ensures we keep the peptide features in sequence order
2533 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
2534 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2536 SequenceFeature[] dnaFeatures = dnaSeq.getSequenceFeatures();
2537 if (dnaFeatures == null)
2542 int dnaStart = dnaSeq.getStart();
2543 int[] lastCodon = null;
2544 int lastPeptidePostion = 0;
2547 * build a map of codon variations for peptides
2549 for (SequenceFeature sf : dnaFeatures)
2551 int dnaCol = sf.getBegin();
2552 if (dnaCol != sf.getEnd())
2554 // not handling multi-locus variant features
2557 if (so.isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
2559 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2562 // feature doesn't lie within coding region
2565 int peptidePosition = mapsTo[0];
2566 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2567 if (codonVariants == null)
2569 codonVariants = new ArrayList[CODON_LENGTH];
2570 codonVariants[0] = new ArrayList<DnaVariant>();
2571 codonVariants[1] = new ArrayList<DnaVariant>();
2572 codonVariants[2] = new ArrayList<DnaVariant>();
2573 variants.put(peptidePosition, codonVariants);
2577 * extract dna variants to a string array
2579 String alls = (String) sf.getValue("alleles");
2584 String[] alleles = alls.toUpperCase().split(",");
2586 for (String allele : alleles)
2588 alleles[i++] = allele.trim(); // lose any space characters "A, G"
2592 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2594 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2595 : MappingUtils.flattenRanges(dnaToProtein
2596 .locateInFrom(peptidePosition, peptidePosition));
2597 lastPeptidePostion = peptidePosition;
2601 * save nucleotide (and any variant) for each codon position
2603 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2605 String nucleotide = String
2606 .valueOf(dnaSeq.getCharAt(codon[codonPos] - dnaStart))
2608 List<DnaVariant> codonVariant = codonVariants[codonPos];
2609 if (codon[codonPos] == dnaCol)
2611 if (!codonVariant.isEmpty()
2612 && codonVariant.get(0).variant == null)
2615 * already recorded base value, add this variant
2617 codonVariant.get(0).variant = sf;
2622 * add variant with base value
2624 codonVariant.add(new DnaVariant(nucleotide, sf));
2627 else if (codonVariant.isEmpty())
2630 * record (possibly non-varying) base value
2632 codonVariant.add(new DnaVariant(nucleotide));
2641 * Makes an alignment with a copy of the given sequences, adding in any
2642 * non-redundant sequences which are mapped to by the cross-referenced
2648 * the alignment dataset shared by the new copy
2651 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2652 SequenceI[] xrefs, AlignmentI dataset)
2654 AlignmentI copy = new Alignment(new Alignment(seqs));
2655 copy.setDataset(dataset);
2656 boolean isProtein = !copy.isNucleotide();
2657 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2660 for (SequenceI xref : xrefs)
2662 DBRefEntry[] dbrefs = xref.getDBRefs();
2665 for (DBRefEntry dbref : dbrefs)
2667 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2668 || dbref.getMap().getTo().isProtein() != isProtein)
2672 SequenceI mappedTo = dbref.getMap().getTo();
2673 SequenceI match = matcher.findIdMatch(mappedTo);
2676 matcher.add(mappedTo);
2677 copy.addSequence(mappedTo);
2687 * Try to align sequences in 'unaligned' to match the alignment of their
2688 * mapped regions in 'aligned'. For example, could use this to align CDS
2689 * sequences which are mapped to their parent cDNA sequences.
2691 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2692 * dna-to-protein or protein-to-dna use alternative methods.
2695 * sequences to be aligned
2697 * holds aligned sequences and their mappings
2700 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2703 * easy case - aligning a copy of aligned sequences
2705 if (alignAsSameSequences(unaligned, aligned))
2707 return unaligned.getHeight();
2711 * fancy case - aligning via mappings between sequences
2713 List<SequenceI> unmapped = new ArrayList<SequenceI>();
2714 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2715 unaligned, aligned, unmapped);
2716 int width = columnMap.size();
2717 char gap = unaligned.getGapCharacter();
2718 int realignedCount = 0;
2719 // TODO: verify this loop scales sensibly for very wide/high alignments
2721 for (SequenceI seq : unaligned.getSequences())
2723 if (!unmapped.contains(seq))
2725 char[] newSeq = new char[width];
2726 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2727 // Integer iteration below
2732 * traverse the map to find columns populated
2735 for (Integer column : columnMap.keySet())
2737 Character c = columnMap.get(column).get(seq);
2741 * sequence has a character at this position
2751 * trim trailing gaps
2753 if (lastCol < width)
2755 char[] tmp = new char[lastCol + 1];
2756 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2759 // TODO: optimise SequenceI to avoid char[]->String->char[]
2760 seq.setSequence(String.valueOf(newSeq));
2764 return realignedCount;
2768 * If unaligned and aligned sequences share the same dataset sequences, then
2769 * simply copies the aligned sequences to the unaligned sequences and returns
2770 * true; else returns false
2773 * - sequences to be aligned based on aligned
2775 * - 'guide' alignment containing sequences derived from same dataset
2779 static boolean alignAsSameSequences(AlignmentI unaligned,
2782 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2784 return false; // should only pass alignments with datasets here
2787 // map from dataset sequence to alignment sequence(s)
2788 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<SequenceI, List<SequenceI>>();
2789 for (SequenceI seq : aligned.getSequences())
2791 SequenceI ds = seq.getDatasetSequence();
2792 if (alignedDatasets.get(ds) == null)
2794 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2796 alignedDatasets.get(ds).add(seq);
2800 * first pass - check whether all sequences to be aligned share a dataset
2801 * sequence with an aligned sequence
2803 for (SequenceI seq : unaligned.getSequences())
2805 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2812 * second pass - copy aligned sequences;
2813 * heuristic rule: pair off sequences in order for the case where
2814 * more than one shares the same dataset sequence
2816 for (SequenceI seq : unaligned.getSequences())
2818 List<SequenceI> alignedSequences = alignedDatasets
2819 .get(seq.getDatasetSequence());
2820 // TODO: getSequenceAsString() will be deprecated in the future
2821 // TODO: need to leave to SequenceI implementor to update gaps
2822 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2823 if (alignedSequences.size() > 0)
2825 // pop off aligned sequences (except the last one)
2826 alignedSequences.remove(0);
2834 * Returns a map whose key is alignment column number (base 1), and whose
2835 * values are a map of sequence characters in that column.
2842 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2843 AlignmentI unaligned, AlignmentI aligned,
2844 List<SequenceI> unmapped)
2847 * Map will hold, for each aligned column position, a map of
2848 * {unalignedSequence, characterPerSequence} at that position.
2849 * TreeMap keeps the entries in ascending column order.
2851 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<Integer, Map<SequenceI, Character>>();
2854 * record any sequences that have no mapping so can't be realigned
2856 unmapped.addAll(unaligned.getSequences());
2858 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2860 for (SequenceI seq : unaligned.getSequences())
2862 for (AlignedCodonFrame mapping : mappings)
2864 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2865 if (fromSeq != null)
2867 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2868 if (addMappedPositions(seq, fromSeq, seqMap, map))
2870 unmapped.remove(seq);
2879 * Helper method that adds to a map the mapped column positions of a sequence.
2881 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2882 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2886 * the sequence whose column positions we are recording
2888 * a sequence that is mapped to the first sequence
2890 * the mapping from 'fromSeq' to 'seq'
2892 * a map to add the column positions (in fromSeq) of the mapped
2896 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2897 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2905 * invert mapping if it is from unaligned to aligned sequence
2907 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2909 seqMap = new Mapping(seq.getDatasetSequence(),
2910 seqMap.getMap().getInverse());
2913 char[] fromChars = fromSeq.getSequence();
2914 int toStart = seq.getStart();
2915 char[] toChars = seq.getSequence();
2918 * traverse [start, end, start, end...] ranges in fromSeq
2920 for (int[] fromRange : seqMap.getMap().getFromRanges())
2922 for (int i = 0; i < fromRange.length - 1; i += 2)
2924 boolean forward = fromRange[i + 1] >= fromRange[i];
2927 * find the range mapped to (sequence positions base 1)
2929 int[] range = seqMap.locateMappedRange(fromRange[i],
2933 System.err.println("Error in mapping " + seqMap + " from "
2934 + fromSeq.getName());
2937 int fromCol = fromSeq.findIndex(fromRange[i]);
2938 int mappedCharPos = range[0];
2941 * walk over the 'from' aligned sequence in forward or reverse
2942 * direction; when a non-gap is found, record the column position
2943 * of the next character of the mapped-to sequence; stop when all
2944 * the characters of the range have been counted
2946 while (mappedCharPos <= range[1] && fromCol <= fromChars.length
2949 if (!Comparison.isGap(fromChars[fromCol - 1]))
2952 * mapped from sequence has a character in this column
2953 * record the column position for the mapped to character
2955 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2956 if (seqsMap == null)
2958 seqsMap = new HashMap<SequenceI, Character>();
2959 map.put(fromCol, seqsMap);
2961 seqsMap.put(seq, toChars[mappedCharPos - toStart]);
2964 fromCol += (forward ? 1 : -1);
2971 // strictly temporary hack until proper criteria for aligning protein to cds
2972 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2973 public static boolean looksLikeEnsembl(AlignmentI alignment)
2975 for (SequenceI seq : alignment.getSequences())
2977 String name = seq.getName();
2978 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))