2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.analysis;
23 import static jalview.io.gff.GffConstants.CLINICAL_SIGNIFICANCE;
25 import jalview.datamodel.AlignedCodon;
26 import jalview.datamodel.AlignedCodonFrame;
27 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
28 import jalview.datamodel.Alignment;
29 import jalview.datamodel.AlignmentAnnotation;
30 import jalview.datamodel.AlignmentI;
31 import jalview.datamodel.DBRefEntry;
32 import jalview.datamodel.IncompleteCodonException;
33 import jalview.datamodel.Mapping;
34 import jalview.datamodel.Sequence;
35 import jalview.datamodel.SequenceFeature;
36 import jalview.datamodel.SequenceGroup;
37 import jalview.datamodel.SequenceI;
38 import jalview.datamodel.features.SequenceFeatures;
39 import jalview.io.gff.Gff3Helper;
40 import jalview.io.gff.SequenceOntologyI;
41 import jalview.schemes.ResidueProperties;
42 import jalview.util.Comparison;
43 import jalview.util.DBRefUtils;
44 import jalview.util.IntRangeComparator;
45 import jalview.util.MapList;
46 import jalview.util.MappingUtils;
47 import jalview.util.StringUtils;
49 import java.io.UnsupportedEncodingException;
50 import java.net.URLEncoder;
51 import java.util.ArrayList;
52 import java.util.Arrays;
53 import java.util.Collection;
54 import java.util.Collections;
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();
111 * toString for aid in the debugger only
114 public String toString()
116 return base + ":" + (variant == null ? "" : variant.getDescription());
121 * given an existing alignment, create a new alignment including all, or up to
122 * flankSize additional symbols from each sequence's dataset sequence
128 public static AlignmentI expandContext(AlignmentI core, int flankSize)
130 List<SequenceI> sq = new ArrayList<SequenceI>();
132 for (SequenceI s : core.getSequences())
134 SequenceI newSeq = s.deriveSequence();
135 final int newSeqStart = newSeq.getStart() - 1;
136 if (newSeqStart > maxoffset
137 && newSeq.getDatasetSequence().getStart() < s.getStart())
139 maxoffset = newSeqStart;
145 maxoffset = Math.min(maxoffset, flankSize);
149 * now add offset left and right to create an expanded alignment
151 for (SequenceI s : sq)
154 while (ds.getDatasetSequence() != null)
156 ds = ds.getDatasetSequence();
158 int s_end = s.findPosition(s.getStart() + s.getLength());
159 // find available flanking residues for sequence
160 int ustream_ds = s.getStart() - ds.getStart();
161 int dstream_ds = ds.getEnd() - s_end;
163 // build new flanked sequence
165 // compute gap padding to start of flanking sequence
166 int offset = maxoffset - ustream_ds;
168 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
171 if (flankSize < ustream_ds)
173 // take up to flankSize residues
174 offset = maxoffset - flankSize;
175 ustream_ds = flankSize;
177 if (flankSize <= dstream_ds)
179 dstream_ds = flankSize - 1;
182 // TODO use Character.toLowerCase to avoid creating String objects?
183 char[] upstream = new String(ds
184 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
185 .toLowerCase().toCharArray();
186 char[] downstream = new String(
187 ds.getSequence(s_end - 1, s_end + dstream_ds)).toLowerCase()
189 char[] coreseq = s.getSequence();
190 char[] nseq = new char[offset + upstream.length + downstream.length
192 char c = core.getGapCharacter();
195 for (; p < offset; p++)
200 System.arraycopy(upstream, 0, nseq, p, upstream.length);
201 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
203 System.arraycopy(downstream, 0, nseq,
204 p + coreseq.length + upstream.length, downstream.length);
205 s.setSequence(new String(nseq));
206 s.setStart(s.getStart() - ustream_ds);
207 s.setEnd(s_end + downstream.length);
209 AlignmentI newAl = new jalview.datamodel.Alignment(
210 sq.toArray(new SequenceI[0]));
211 for (SequenceI s : sq)
213 if (s.getAnnotation() != null)
215 for (AlignmentAnnotation aa : s.getAnnotation())
217 aa.adjustForAlignment(); // JAL-1712 fix
218 newAl.addAnnotation(aa);
222 newAl.setDataset(core.getDataset());
227 * Returns the index (zero-based position) of a sequence in an alignment, or
234 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
238 for (SequenceI alSeq : al.getSequences())
251 * Returns a map of lists of sequences in the alignment, keyed by sequence
252 * name. For use in mapping between different alignment views of the same
255 * @see jalview.datamodel.AlignmentI#getSequencesByName()
257 public static Map<String, List<SequenceI>> getSequencesByName(
260 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
261 for (SequenceI seq : al.getSequences())
263 String name = seq.getName();
266 List<SequenceI> seqs = theMap.get(name);
269 seqs = new ArrayList<SequenceI>();
270 theMap.put(name, seqs);
279 * Build mapping of protein to cDNA alignment. Mappings are made between
280 * sequences where the cDNA translates to the protein sequence. Any new
281 * mappings are added to the protein alignment. Returns true if any mappings
282 * either already exist or were added, else false.
284 * @param proteinAlignment
285 * @param cdnaAlignment
288 public static boolean mapProteinAlignmentToCdna(
289 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
291 if (proteinAlignment == null || cdnaAlignment == null)
296 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
297 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
300 * First pass - map sequences where cross-references exist. This include
301 * 1-to-many mappings to support, for example, variant cDNA.
303 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
304 cdnaAlignment, mappedDna, mappedProtein, true);
307 * Second pass - map sequences where no cross-references exist. This only
308 * does 1-to-1 mappings and assumes corresponding sequences are in the same
309 * order in the alignments.
311 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
312 mappedDna, mappedProtein, false);
313 return mappingPerformed;
317 * Make mappings between compatible sequences (where the cDNA translation
318 * matches the protein).
320 * @param proteinAlignment
321 * @param cdnaAlignment
323 * a set of mapped DNA sequences (to add to)
324 * @param mappedProtein
325 * a set of mapped Protein sequences (to add to)
327 * if true, only map sequences where xrefs exist
330 protected static boolean mapProteinToCdna(
331 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
332 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
335 boolean mappingExistsOrAdded = false;
336 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
337 for (SequenceI aaSeq : thisSeqs)
339 boolean proteinMapped = false;
340 AlignedCodonFrame acf = new AlignedCodonFrame();
342 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
345 * Always try to map if sequences have xref to each other; this supports
346 * variant cDNA or alternative splicing for a protein sequence.
348 * If no xrefs, try to map progressively, assuming that alignments have
349 * mappable sequences in corresponding order. These are not
350 * many-to-many, as that would risk mixing species with similar cDNA
353 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
359 * Don't map non-xrefd sequences more than once each. This heuristic
360 * allows us to pair up similar sequences in ordered alignments.
362 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
363 || mappedDna.contains(cdnaSeq)))
367 if (mappingExists(proteinAlignment.getCodonFrames(),
368 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
370 mappingExistsOrAdded = true;
374 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
377 acf.addMap(cdnaSeq, aaSeq, map);
378 mappingExistsOrAdded = true;
379 proteinMapped = true;
380 mappedDna.add(cdnaSeq);
381 mappedProtein.add(aaSeq);
387 proteinAlignment.addCodonFrame(acf);
390 return mappingExistsOrAdded;
394 * Answers true if the mappings include one between the given (dataset)
397 public static boolean mappingExists(List<AlignedCodonFrame> mappings,
398 SequenceI aaSeq, SequenceI cdnaSeq)
400 if (mappings != null)
402 for (AlignedCodonFrame acf : mappings)
404 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
414 * Builds a mapping (if possible) of a cDNA to a protein sequence.
416 * <li>first checks if the cdna translates exactly to the protein
418 * <li>else checks for translation after removing a STOP codon</li>
419 * <li>else checks for translation after removing a START codon</li>
420 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
422 * Returns null if no mapping is determined.
425 * the aligned protein sequence
427 * the aligned cdna sequence
430 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
434 * Here we handle either dataset sequence set (desktop) or absent (applet).
435 * Use only the char[] form of the sequence to avoid creating possibly large
438 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
439 char[] aaSeqChars = proteinDataset != null
440 ? proteinDataset.getSequence()
441 : proteinSeq.getSequence();
442 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
443 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
444 : cdnaSeq.getSequence();
445 if (aaSeqChars == null || cdnaSeqChars == null)
451 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
453 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
454 int cdnaLength = cdnaSeqChars.length;
455 int cdnaStart = cdnaSeq.getStart();
456 int cdnaEnd = cdnaSeq.getEnd();
457 final int proteinStart = proteinSeq.getStart();
458 final int proteinEnd = proteinSeq.getEnd();
461 * If lengths don't match, try ignoring stop codon (if present)
463 if (cdnaLength != mappedLength && cdnaLength > 2)
465 String lastCodon = String.valueOf(cdnaSeqChars,
466 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
467 for (String stop : ResidueProperties.STOP)
469 if (lastCodon.equals(stop))
471 cdnaEnd -= CODON_LENGTH;
472 cdnaLength -= CODON_LENGTH;
479 * If lengths still don't match, try ignoring start codon.
482 if (cdnaLength != mappedLength && cdnaLength > 2
483 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
484 .equals(ResidueProperties.START))
486 startOffset += CODON_LENGTH;
487 cdnaStart += CODON_LENGTH;
488 cdnaLength -= CODON_LENGTH;
491 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
494 * protein is translation of dna (+/- start/stop codons)
496 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
498 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
503 * translation failed - try mapping CDS annotated regions of dna
505 return mapCdsToProtein(cdnaSeq, proteinSeq);
509 * Test whether the given cdna sequence, starting at the given offset,
510 * translates to the given amino acid sequence, using the standard translation
511 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
513 * @param cdnaSeqChars
518 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
521 if (cdnaSeqChars == null || aaSeqChars == null)
527 int dnaPos = cdnaStart;
528 for (; dnaPos < cdnaSeqChars.length - 2
529 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
531 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
532 final String translated = ResidueProperties.codonTranslate(codon);
535 * allow * in protein to match untranslatable in dna
537 final char aaRes = aaSeqChars[aaPos];
538 if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
542 if (translated == null || !(aaRes == translated.charAt(0)))
545 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
546 // + codon + "(" + translated + ") != " + aaRes));
552 * check we matched all of the protein sequence
554 if (aaPos != aaSeqChars.length)
560 * check we matched all of the dna except
561 * for optional trailing STOP codon
563 if (dnaPos == cdnaSeqChars.length)
567 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
569 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
570 if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
579 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
580 * currently assumes that we are aligning cDNA to match protein.
583 * the sequence to be realigned
585 * the alignment whose sequence alignment is to be 'copied'
587 * character string represent a gap in the realigned sequence
588 * @param preserveUnmappedGaps
589 * @param preserveMappedGaps
590 * @return true if the sequence was realigned, false if it could not be
592 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
593 String gap, boolean preserveMappedGaps,
594 boolean preserveUnmappedGaps)
597 * Get any mappings from the source alignment to the target (dataset)
600 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
601 // all mappings. Would it help to constrain this?
602 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
603 if (mappings == null || mappings.isEmpty())
609 * Locate the aligned source sequence whose dataset sequence is mapped. We
610 * just take the first match here (as we can't align like more than one
613 SequenceI alignFrom = null;
614 AlignedCodonFrame mapping = null;
615 for (AlignedCodonFrame mp : mappings)
617 alignFrom = mp.findAlignedSequence(seq, al);
618 if (alignFrom != null)
625 if (alignFrom == null)
629 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
630 preserveMappedGaps, preserveUnmappedGaps);
635 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
636 * match residues and codons. Flags control whether existing gaps in unmapped
637 * (intron) and mapped (exon) regions are preserved or not. Gaps between
638 * intron and exon are only retained if both flags are set.
645 * @param preserveUnmappedGaps
646 * @param preserveMappedGaps
648 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
649 AlignedCodonFrame mapping, String myGap, char sourceGap,
650 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
652 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
654 // aligned and dataset sequence positions, all base zero
658 int basesWritten = 0;
659 char myGapChar = myGap.charAt(0);
660 int ratio = myGap.length();
662 int fromOffset = alignFrom.getStart() - 1;
663 int toOffset = alignTo.getStart() - 1;
664 int sourceGapMappedLength = 0;
665 boolean inExon = false;
666 final int toLength = alignTo.getLength();
667 final int fromLength = alignFrom.getLength();
668 StringBuilder thisAligned = new StringBuilder(2 * toLength);
671 * Traverse the 'model' aligned sequence
673 for (int i = 0; i < fromLength; i++)
675 char sourceChar = alignFrom.getCharAt(i);
676 if (sourceChar == sourceGap)
678 sourceGapMappedLength += ratio;
683 * Found a non-gap character. Locate its mapped region if any.
686 // Note mapping positions are base 1, our sequence positions base 0
687 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
688 sourceDsPos + fromOffset);
689 if (mappedPos == null)
692 * unmapped position; treat like a gap
694 sourceGapMappedLength += ratio;
695 // System.err.println("Can't align: no codon mapping to residue "
696 // + sourceDsPos + "(" + sourceChar + ")");
701 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
702 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
703 StringBuilder trailingCopiedGap = new StringBuilder();
706 * Copy dna sequence up to and including this codon. Optionally, include
707 * gaps before the codon starts (in introns) and/or after the codon starts
710 * Note this only works for 'linear' splicing, not reverse or interleaved.
711 * But then 'align dna as protein' doesn't make much sense otherwise.
713 int intronLength = 0;
714 while (basesWritten + toOffset < mappedCodonEnd
715 && thisSeqPos < toLength)
717 final char c = alignTo.getCharAt(thisSeqPos++);
721 int sourcePosition = basesWritten + toOffset;
722 if (sourcePosition < mappedCodonStart)
725 * Found an unmapped (intron) base. First add in any preceding gaps
728 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
730 thisAligned.append(trailingCopiedGap.toString());
731 intronLength += trailingCopiedGap.length();
732 trailingCopiedGap = new StringBuilder();
739 final boolean startOfCodon = sourcePosition == mappedCodonStart;
740 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
741 preserveUnmappedGaps, sourceGapMappedLength, inExon,
742 trailingCopiedGap.length(), intronLength, startOfCodon);
743 for (int k = 0; k < gapsToAdd; k++)
745 thisAligned.append(myGapChar);
747 sourceGapMappedLength = 0;
750 thisAligned.append(c);
751 trailingCopiedGap = new StringBuilder();
755 if (inExon && preserveMappedGaps)
757 trailingCopiedGap.append(myGapChar);
759 else if (!inExon && preserveUnmappedGaps)
761 trailingCopiedGap.append(myGapChar);
768 * At end of model aligned sequence. Copy any remaining target sequence, optionally
769 * including (intron) gaps.
771 while (thisSeqPos < toLength)
773 final char c = alignTo.getCharAt(thisSeqPos++);
774 if (c != myGapChar || preserveUnmappedGaps)
776 thisAligned.append(c);
778 sourceGapMappedLength--;
782 * finally add gaps to pad for any trailing source gaps or
783 * unmapped characters
785 if (preserveUnmappedGaps)
787 while (sourceGapMappedLength > 0)
789 thisAligned.append(myGapChar);
790 sourceGapMappedLength--;
795 * All done aligning, set the aligned sequence.
797 alignTo.setSequence(new String(thisAligned));
801 * Helper method to work out how many gaps to insert when realigning.
803 * @param preserveMappedGaps
804 * @param preserveUnmappedGaps
805 * @param sourceGapMappedLength
807 * @param trailingCopiedGap
808 * @param intronLength
809 * @param startOfCodon
812 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
813 boolean preserveUnmappedGaps, int sourceGapMappedLength,
814 boolean inExon, int trailingGapLength, int intronLength,
815 final boolean startOfCodon)
821 * Reached start of codon. Ignore trailing gaps in intron unless we are
822 * preserving gaps in both exon and intron. Ignore them anyway if the
823 * protein alignment introduces a gap at least as large as the intronic
826 if (inExon && !preserveMappedGaps)
828 trailingGapLength = 0;
830 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
832 trailingGapLength = 0;
836 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
840 if (intronLength + trailingGapLength <= sourceGapMappedLength)
842 gapsToAdd = sourceGapMappedLength - intronLength;
846 gapsToAdd = Math.min(
847 intronLength + trailingGapLength - sourceGapMappedLength,
855 * second or third base of codon; check for any gaps in dna
857 if (!preserveMappedGaps)
859 trailingGapLength = 0;
861 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
867 * Realigns the given protein to match the alignment of the dna, using codon
868 * mappings to translate aligned codon positions to protein residues.
871 * the alignment whose sequences are realigned by this method
873 * the dna alignment whose alignment we are 'copying'
874 * @return the number of sequences that were realigned
876 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
878 if (protein.isNucleotide() || !dna.isNucleotide())
880 System.err.println("Wrong alignment type in alignProteinAsDna");
883 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
884 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
885 protein, dna, unmappedProtein);
886 return alignProteinAs(protein, alignedCodons, unmappedProtein);
890 * Realigns the given dna to match the alignment of the protein, using codon
891 * mappings to translate aligned peptide positions to codons.
893 * Always produces a padded CDS alignment.
896 * the alignment whose sequences are realigned by this method
898 * the protein alignment whose alignment we are 'copying'
899 * @return the number of sequences that were realigned
901 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
903 if (protein.isNucleotide() || !dna.isNucleotide())
905 System.err.println("Wrong alignment type in alignProteinAsDna");
908 // todo: implement this
909 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
910 int alignedCount = 0;
911 int width = 0; // alignment width for padding CDS
912 for (SequenceI dnaSeq : dna.getSequences())
914 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
915 dna.getGapCharacter()))
919 width = Math.max(dnaSeq.getLength(), width);
923 for (SequenceI dnaSeq : dna.getSequences())
925 oldwidth = dnaSeq.getLength();
926 diff = width - oldwidth;
929 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
936 * Helper method to align (if possible) the dna sequence to match the
937 * alignment of a mapped protein sequence. This is currently limited to
938 * handling coding sequence only.
946 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
947 AlignmentI protein, List<AlignedCodonFrame> mappings,
950 SequenceI cdsDss = cdsSeq.getDatasetSequence();
954 .println("alignCdsSequenceAsProtein needs aligned sequence!");
958 List<AlignedCodonFrame> dnaMappings = MappingUtils
959 .findMappingsForSequence(cdsSeq, mappings);
960 for (AlignedCodonFrame mapping : dnaMappings)
962 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
965 final int peptideLength = peptide.getLength();
966 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
969 MapList mapList = map.getMap();
970 if (map.getTo() == peptide.getDatasetSequence())
972 mapList = mapList.getInverse();
974 final int cdsLength = cdsDss.getLength();
975 int mappedFromLength = MappingUtils.getLength(mapList
977 int mappedToLength = MappingUtils
978 .getLength(mapList.getToRanges());
979 boolean addStopCodon = (cdsLength == mappedFromLength
980 * CODON_LENGTH + CODON_LENGTH)
981 || (peptide.getDatasetSequence()
982 .getLength() == mappedFromLength - 1);
983 if (cdsLength != mappedToLength && !addStopCodon)
985 System.err.println(String.format(
986 "Can't align cds as protein (length mismatch %d/%d): %s",
987 cdsLength, mappedToLength, cdsSeq.getName()));
991 * pre-fill the aligned cds sequence with gaps
993 char[] alignedCds = new char[peptideLength * CODON_LENGTH
994 + (addStopCodon ? CODON_LENGTH : 0)];
995 Arrays.fill(alignedCds, gapChar);
998 * walk over the aligned peptide sequence and insert mapped
999 * codons for residues in the aligned cds sequence
1001 int copiedBases = 0;
1002 int cdsStart = cdsDss.getStart();
1003 int proteinPos = peptide.getStart() - 1;
1006 for (int col = 0; col < peptideLength; col++)
1008 char residue = peptide.getCharAt(col);
1010 if (Comparison.isGap(residue))
1012 cdsCol += CODON_LENGTH;
1017 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1020 // e.g. incomplete start codon, X in peptide
1021 cdsCol += CODON_LENGTH;
1025 for (int j = codon[0]; j <= codon[1]; j++)
1027 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1028 alignedCds[cdsCol++] = mappedBase;
1036 * append stop codon if not mapped from protein,
1037 * closing it up to the end of the mapped sequence
1039 if (copiedBases == cdsLength - CODON_LENGTH)
1041 for (int i = alignedCds.length - 1; i >= 0; i--)
1043 if (!Comparison.isGap(alignedCds[i]))
1045 cdsCol = i + 1; // gap just after end of sequence
1049 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1051 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1054 cdsSeq.setSequence(new String(alignedCds));
1063 * Builds a map whose key is an aligned codon position (3 alignment column
1064 * numbers base 0), and whose value is a map from protein sequence to each
1065 * protein's peptide residue for that codon. The map generates an ordering of
1066 * the codons, and allows us to read off the peptides at each position in
1067 * order to assemble 'aligned' protein sequences.
1070 * the protein alignment
1072 * the coding dna alignment
1073 * @param unmappedProtein
1074 * any unmapped proteins are added to this list
1077 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1078 AlignmentI protein, AlignmentI dna,
1079 List<SequenceI> unmappedProtein)
1082 * maintain a list of any proteins with no mappings - these will be
1083 * rendered 'as is' in the protein alignment as we can't align them
1085 unmappedProtein.addAll(protein.getSequences());
1087 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1090 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1091 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1092 * comparator keeps the codon positions ordered.
1094 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
1095 new CodonComparator());
1097 for (SequenceI dnaSeq : dna.getSequences())
1099 for (AlignedCodonFrame mapping : mappings)
1101 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1104 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1105 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1107 unmappedProtein.remove(prot);
1113 * Finally add any unmapped peptide start residues (e.g. for incomplete
1114 * codons) as if at the codon position before the second residue
1116 // TODO resolve JAL-2022 so this fudge can be removed
1117 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1118 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1120 return alignedCodons;
1124 * Scans for any protein mapped from position 2 (meaning unmapped start
1125 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1126 * preceding position in the alignment
1128 * @param alignedCodons
1129 * the codon-to-peptide map
1130 * @param mappedSequenceCount
1131 * the number of distinct sequences in the map
1133 protected static void addUnmappedPeptideStarts(
1134 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1135 int mappedSequenceCount)
1137 // TODO delete this ugly hack once JAL-2022 is resolved
1138 // i.e. we can model startPhase > 0 (incomplete start codon)
1140 List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
1141 AlignedCodon lastCodon = null;
1142 Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
1144 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1147 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1150 SequenceI seq = sequenceCodon.getKey();
1151 if (sequencesChecked.contains(seq))
1155 sequencesChecked.add(seq);
1156 AlignedCodon codon = sequenceCodon.getValue();
1157 if (codon.peptideCol > 1)
1160 "Problem mapping protein with >1 unmapped start positions: "
1163 else if (codon.peptideCol == 1)
1166 * first position (peptideCol == 0) was unmapped - add it
1168 if (lastCodon != null)
1170 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1171 lastCodon.pos2, lastCodon.pos3,
1172 String.valueOf(seq.getCharAt(0)), 0);
1173 toAdd.put(seq, firstPeptide);
1178 * unmapped residue at start of alignment (no prior column) -
1179 * 'insert' at nominal codon [0, 0, 0]
1181 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1182 String.valueOf(seq.getCharAt(0)), 0);
1183 toAdd.put(seq, firstPeptide);
1186 if (sequencesChecked.size() == mappedSequenceCount)
1188 // no need to check past first mapped position in all sequences
1192 lastCodon = entry.getKey();
1196 * add any new codons safely after iterating over the map
1198 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1200 addCodonToMap(alignedCodons, startCodon.getValue(),
1201 startCodon.getKey());
1206 * Update the aligned protein sequences to match the codon alignments given in
1210 * @param alignedCodons
1211 * an ordered map of codon positions (columns), with sequence/peptide
1212 * values present in each column
1213 * @param unmappedProtein
1216 protected static int alignProteinAs(AlignmentI protein,
1217 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1218 List<SequenceI> unmappedProtein)
1221 * prefill peptide sequences with gaps
1223 int alignedWidth = alignedCodons.size();
1224 char[] gaps = new char[alignedWidth];
1225 Arrays.fill(gaps, protein.getGapCharacter());
1226 Map<SequenceI, char[]> peptides = new HashMap<>();
1227 for (SequenceI seq : protein.getSequences())
1229 if (!unmappedProtein.contains(seq))
1231 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1236 * Traverse the codons left to right (as defined by CodonComparator)
1237 * and insert peptides in each column where the sequence is mapped.
1238 * This gives a peptide 'alignment' where residues are aligned if their
1239 * corresponding codons occupy the same columns in the cdna alignment.
1242 for (AlignedCodon codon : alignedCodons.keySet())
1244 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1246 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1248 char residue = entry.getValue().product.charAt(0);
1249 peptides.get(entry.getKey())[column] = residue;
1255 * and finally set the constructed sequences
1257 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1259 entry.getKey().setSequence(new String(entry.getValue()));
1266 * Populate the map of aligned codons by traversing the given sequence
1267 * mapping, locating the aligned positions of mapped codons, and adding those
1268 * positions and their translation products to the map.
1271 * the aligned sequence we are mapping from
1273 * the sequence to be aligned to the codons
1275 * the gap character in the dna sequence
1277 * a mapping to a sequence translation
1278 * @param alignedCodons
1279 * the map we are building up
1281 static void addCodonPositions(SequenceI dna, SequenceI protein,
1282 char gapChar, Mapping seqMap,
1283 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1285 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1288 * add codon positions, and their peptide translations, to the alignment
1289 * map, while remembering the first codon mapped
1291 while (codons.hasNext())
1295 AlignedCodon codon = codons.next();
1296 addCodonToMap(alignedCodons, codon, protein);
1297 } catch (IncompleteCodonException e)
1299 // possible incomplete trailing codon - ignore
1300 } catch (NoSuchElementException e)
1302 // possibly peptide lacking STOP
1308 * Helper method to add a codon-to-peptide entry to the aligned codons map
1310 * @param alignedCodons
1314 protected static void addCodonToMap(
1315 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1316 AlignedCodon codon, SequenceI protein)
1318 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1319 if (seqProduct == null)
1321 seqProduct = new HashMap<SequenceI, AlignedCodon>();
1322 alignedCodons.put(codon, seqProduct);
1324 seqProduct.put(protein, codon);
1328 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1329 * between at least one pair of sequences in the two alignments. Currently,
1332 * <li>One alignment must be nucleotide, and the other protein</li>
1333 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1334 * <li>Mappable means the nucleotide translation matches the protein
1336 * <li>The translation may ignore start and stop codons if present in the
1344 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1346 if (al1 == null || al2 == null)
1352 * Require one nucleotide and one protein
1354 if (al1.isNucleotide() == al2.isNucleotide())
1358 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1359 AlignmentI protein = dna == al1 ? al2 : al1;
1360 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1361 for (SequenceI dnaSeq : dna.getSequences())
1363 for (SequenceI proteinSeq : protein.getSequences())
1365 if (isMappable(dnaSeq, proteinSeq, mappings))
1375 * Returns true if the dna sequence is mapped, or could be mapped, to the
1383 protected static boolean isMappable(SequenceI dnaSeq,
1384 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1386 if (dnaSeq == null || proteinSeq == null)
1391 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1392 : dnaSeq.getDatasetSequence();
1393 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1395 : proteinSeq.getDatasetSequence();
1397 for (AlignedCodonFrame mapping : mappings)
1399 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1409 * Just try to make a mapping (it is not yet stored), test whether
1412 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1416 * Finds any reference annotations associated with the sequences in
1417 * sequenceScope, that are not already added to the alignment, and adds them
1418 * to the 'candidates' map. Also populates a lookup table of annotation
1419 * labels, keyed by calcId, for use in constructing tooltips or the like.
1421 * @param sequenceScope
1422 * the sequences to scan for reference annotations
1423 * @param labelForCalcId
1424 * (optional) map to populate with label for calcId
1426 * map to populate with annotations for sequence
1428 * the alignment to check for presence of annotations
1430 public static void findAddableReferenceAnnotations(
1431 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1432 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1435 if (sequenceScope == null)
1441 * For each sequence in scope, make a list of any annotations on the
1442 * underlying dataset sequence which are not already on the alignment.
1444 * Add to a map of { alignmentSequence, <List of annotations to add> }
1446 for (SequenceI seq : sequenceScope)
1448 SequenceI dataset = seq.getDatasetSequence();
1449 if (dataset == null)
1453 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1454 if (datasetAnnotations == null)
1458 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1459 for (AlignmentAnnotation dsann : datasetAnnotations)
1462 * Find matching annotations on the alignment. If none is found, then
1463 * add this annotation to the list of 'addable' annotations for this
1466 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1467 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1468 if (!matchedAlignmentAnnotations.iterator().hasNext())
1471 if (labelForCalcId != null)
1473 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1478 * Save any addable annotations for this sequence
1480 if (!result.isEmpty())
1482 candidates.put(seq, result);
1488 * Adds annotations to the top of the alignment annotations, in the same order
1489 * as their related sequences.
1491 * @param annotations
1492 * the annotations to add
1494 * the alignment to add them to
1495 * @param selectionGroup
1496 * current selection group (or null if none)
1498 public static void addReferenceAnnotations(
1499 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1500 final AlignmentI alignment, final SequenceGroup selectionGroup)
1502 for (SequenceI seq : annotations.keySet())
1504 for (AlignmentAnnotation ann : annotations.get(seq))
1506 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1508 int endRes = ann.annotations.length;
1509 if (selectionGroup != null)
1511 startRes = selectionGroup.getStartRes();
1512 endRes = selectionGroup.getEndRes();
1514 copyAnn.restrict(startRes, endRes);
1517 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1518 * original annotation is already on the sequence.
1520 if (!seq.hasAnnotation(ann))
1522 seq.addAlignmentAnnotation(copyAnn);
1525 copyAnn.adjustForAlignment();
1526 // add to the alignment and set visible
1527 alignment.addAnnotation(copyAnn);
1528 copyAnn.visible = true;
1534 * Set visibility of alignment annotations of specified types (labels), for
1535 * specified sequences. This supports controls like "Show all secondary
1536 * structure", "Hide all Temp factor", etc.
1538 * @al the alignment to scan for annotations
1540 * the types (labels) of annotations to be updated
1541 * @param forSequences
1542 * if not null, only annotations linked to one of these sequences are
1543 * in scope for update; if null, acts on all sequence annotations
1545 * if this flag is true, 'types' is ignored (label not checked)
1547 * if true, set visibility on, else set off
1549 public static void showOrHideSequenceAnnotations(AlignmentI al,
1550 Collection<String> types, List<SequenceI> forSequences,
1551 boolean anyType, boolean doShow)
1553 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1556 for (AlignmentAnnotation aa : anns)
1558 if (anyType || types.contains(aa.label))
1560 if ((aa.sequenceRef != null) && (forSequences == null
1561 || forSequences.contains(aa.sequenceRef)))
1563 aa.visible = doShow;
1571 * Returns true if either sequence has a cross-reference to the other
1577 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1579 // Note: moved here from class CrossRef as the latter class has dependencies
1580 // not availability to the applet's classpath
1581 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1585 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1586 * that sequence name is structured as Source|AccessionId.
1592 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1594 if (seq1 == null || seq2 == null)
1598 String name = seq2.getName();
1599 final DBRefEntry[] xrefs = seq1.getDBRefs();
1602 for (DBRefEntry xref : xrefs)
1604 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1605 // case-insensitive test, consistent with DBRefEntry.equalRef()
1606 if (xrefName.equalsIgnoreCase(name))
1616 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1617 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1618 * added to the original alignment's dataset, which is shared by the new
1619 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1620 * added to the alignment dataset.
1623 * aligned nucleotide (dna or cds) sequences
1625 * the alignment dataset the sequences belong to
1627 * (optional) to restrict results to CDS that map to specified
1629 * @return an alignment whose sequences are the cds-only parts of the dna
1630 * sequences (or null if no mappings are found)
1632 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1633 AlignmentI dataset, SequenceI[] products)
1635 if (dataset == null || dataset.getDataset() != null)
1637 throw new IllegalArgumentException(
1638 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1640 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
1641 List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
1642 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1643 HashSet<SequenceI> productSeqs = null;
1644 if (products != null)
1646 productSeqs = new HashSet<SequenceI>();
1647 for (SequenceI seq : products)
1649 productSeqs.add(seq.getDatasetSequence() == null ? seq
1650 : seq.getDatasetSequence());
1655 * Construct CDS sequences from mappings on the alignment dataset.
1657 * - find the protein product(s) mapped to from each dna sequence
1658 * - if the mapping covers the whole dna sequence (give or take start/stop
1659 * codon), take the dna as the CDS sequence
1660 * - else search dataset mappings for a suitable dna sequence, i.e. one
1661 * whose whole sequence is mapped to the protein
1662 * - if no sequence found, construct one from the dna sequence and mapping
1663 * (and add it to dataset so it is found if this is repeated)
1665 for (SequenceI dnaSeq : dna)
1667 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1668 : dnaSeq.getDatasetSequence();
1670 List<AlignedCodonFrame> seqMappings = MappingUtils
1671 .findMappingsForSequence(dnaSeq, mappings);
1672 for (AlignedCodonFrame mapping : seqMappings)
1674 List<Mapping> mappingsFromSequence = mapping
1675 .getMappingsFromSequence(dnaSeq);
1677 for (Mapping aMapping : mappingsFromSequence)
1679 MapList mapList = aMapping.getMap();
1680 if (mapList.getFromRatio() == 1)
1683 * not a dna-to-protein mapping (likely dna-to-cds)
1689 * skip if mapping is not to one of the target set of proteins
1691 SequenceI proteinProduct = aMapping.getTo();
1692 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1698 * try to locate the CDS from the dataset mappings;
1699 * guard against duplicate results (for the case that protein has
1700 * dbrefs to both dna and cds sequences)
1702 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1703 seqMappings, aMapping);
1706 if (!foundSeqs.contains(cdsSeq))
1708 foundSeqs.add(cdsSeq);
1709 SequenceI derivedSequence = cdsSeq.deriveSequence();
1710 cdsSeqs.add(derivedSequence);
1711 if (!dataset.getSequences().contains(cdsSeq))
1713 dataset.addSequence(cdsSeq);
1720 * didn't find mapped CDS sequence - construct it and add
1721 * its dataset sequence to the dataset
1723 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1724 dataset).deriveSequence();
1725 // cdsSeq has a name constructed as CDS|<dbref>
1726 // <dbref> will be either the accession for the coding sequence,
1727 // marked in the /via/ dbref to the protein product accession
1728 // or it will be the original nucleotide accession.
1729 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1731 cdsSeqs.add(cdsSeq);
1733 if (!dataset.getSequences().contains(cdsSeqDss))
1735 // check if this sequence is a newly created one
1736 // so needs adding to the dataset
1737 dataset.addSequence(cdsSeqDss);
1741 * add a mapping from CDS to the (unchanged) mapped to range
1743 List<int[]> cdsRange = Collections
1744 .singletonList(new int[]
1745 { 1, cdsSeq.getLength() });
1746 MapList cdsToProteinMap = new MapList(cdsRange,
1747 mapList.getToRanges(), mapList.getFromRatio(),
1748 mapList.getToRatio());
1749 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1750 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1754 * guard against duplicating the mapping if repeating this action
1756 if (!mappings.contains(cdsToProteinMapping))
1758 mappings.add(cdsToProteinMapping);
1761 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1762 proteinProduct, aMapping);
1764 * add another mapping from original 'from' range to CDS
1766 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1767 MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1769 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1771 if (!mappings.contains(dnaToCdsMapping))
1773 mappings.add(dnaToCdsMapping);
1777 * add DBRef with mapping from protein to CDS
1778 * (this enables Get Cross-References from protein alignment)
1779 * This is tricky because we can't have two DBRefs with the
1780 * same source and accession, so need a different accession for
1781 * the CDS from the dna sequence
1784 // specific use case:
1785 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1786 // ENSG02, ENSG03, with transcripts and products similarly named.
1787 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1789 // JBPNote: ?? can't actually create an example that demonstrates we
1791 // synthesize an xref.
1793 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1795 // creates a complementary cross-reference to the source sequence's
1796 // primary reference.
1798 DBRefEntry cdsCrossRef = new DBRefEntry(primRef.getSource(),
1799 primRef.getSource() + ":" + primRef.getVersion(),
1800 primRef.getAccessionId());
1802 .setMap(new Mapping(dnaDss, new MapList(dnaToCdsMap)));
1803 cdsSeqDss.addDBRef(cdsCrossRef);
1805 // problem here is that the cross-reference is synthesized -
1806 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1808 // assuming cds version same as dna ?!?
1810 DBRefEntry proteinToCdsRef = new DBRefEntry(primRef.getSource(),
1811 primRef.getVersion(), cdsSeq.getName());
1813 proteinToCdsRef.setMap(
1814 new Mapping(cdsSeqDss, cdsToProteinMap.getInverse()));
1815 proteinProduct.addDBRef(proteinToCdsRef);
1819 * transfer any features on dna that overlap the CDS
1821 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1822 SequenceOntologyI.CDS);
1827 AlignmentI cds = new Alignment(
1828 cdsSeqs.toArray(new SequenceI[cdsSeqs.size()]));
1829 cds.setDataset(dataset);
1835 * A helper method that finds a CDS sequence in the alignment dataset that is
1836 * mapped to the given protein sequence, and either is, or has a mapping from,
1837 * the given dna sequence.
1840 * set of all mappings on the dataset
1842 * a dna (or cds) sequence we are searching from
1843 * @param seqMappings
1844 * the set of mappings involving dnaSeq
1846 * an initial candidate from seqMappings
1849 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1850 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1854 * TODO a better dna-cds-protein mapping data representation to allow easy
1855 * navigation; until then this clunky looping around lists of mappings
1857 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1858 : dnaSeq.getDatasetSequence();
1859 SequenceI proteinProduct = aMapping.getTo();
1862 * is this mapping from the whole dna sequence (i.e. CDS)?
1863 * allowing for possible stop codon on dna but not peptide
1865 int mappedFromLength = MappingUtils
1866 .getLength(aMapping.getMap().getFromRanges());
1867 int dnaLength = seqDss.getLength();
1868 if (mappedFromLength == dnaLength
1869 || mappedFromLength == dnaLength - CODON_LENGTH)
1875 * looks like we found the dna-to-protein mapping; search for the
1876 * corresponding cds-to-protein mapping
1878 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1879 .findMappingsForSequence(proteinProduct, mappings);
1880 for (AlignedCodonFrame acf : mappingsToPeptide)
1882 for (SequenceToSequenceMapping map : acf.getMappings())
1884 Mapping mapping = map.getMapping();
1885 if (mapping != aMapping
1886 && mapping.getMap().getFromRatio() == CODON_LENGTH
1887 && proteinProduct == mapping.getTo()
1888 && seqDss != map.getFromSeq())
1890 mappedFromLength = MappingUtils
1891 .getLength(mapping.getMap().getFromRanges());
1892 if (mappedFromLength == map.getFromSeq().getLength())
1895 * found a 3:1 mapping to the protein product which covers
1896 * the whole dna sequence i.e. is from CDS; finally check it
1897 * is from the dna start sequence
1899 SequenceI cdsSeq = map.getFromSeq();
1900 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1901 .findMappingsForSequence(cdsSeq, seqMappings);
1902 if (!dnaToCdsMaps.isEmpty())
1914 * Helper method that makes a CDS sequence as defined by the mappings from the
1915 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1916 * forward or reverse strand).
1921 * - existing dataset. We check for sequences that look like the CDS
1922 * we are about to construct, if one exists already, then we will
1923 * just return that one.
1924 * @return CDS sequence (as a dataset sequence)
1926 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1929 char[] seqChars = seq.getSequence();
1930 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1931 int cdsWidth = MappingUtils.getLength(fromRanges);
1932 char[] newSeqChars = new char[cdsWidth];
1935 for (int[] range : fromRanges)
1937 if (range[0] <= range[1])
1939 // forward strand mapping - just copy the range
1940 int length = range[1] - range[0] + 1;
1941 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1947 // reverse strand mapping - copy and complement one by one
1948 for (int i = range[0]; i >= range[1]; i--)
1950 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1956 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
1957 * else generate a sequence name
1959 String mapFromId = mapping.getMappedFromId();
1960 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
1961 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
1962 if (dataset != null)
1964 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
1965 if (matches != null)
1967 boolean matched = false;
1968 for (SequenceI mtch : matches)
1970 if (mtch.getStart() != newSeq.getStart())
1974 if (mtch.getEnd() != newSeq.getEnd())
1978 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
1990 "JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
1996 // newSeq.setDescription(mapFromId);
2002 * add any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2003 * the given mapping.
2008 * @return list of DBRefEntrys added.
2010 public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2011 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2014 // gather direct refs from contig congrent with mapping
2015 List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
2016 HashSet<String> directSources = new HashSet<String>();
2017 if (contig.getDBRefs() != null)
2019 for (DBRefEntry dbr : contig.getDBRefs())
2021 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
2023 MapList map = dbr.getMap().getMap();
2024 // check if map is the CDS mapping
2025 if (mapping.getMap().equals(map))
2028 directSources.add(dbr.getSource());
2033 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2034 proteinProduct.getDBRefs(),
2035 directSources.toArray(new String[0]));
2036 List<DBRefEntry> propagated = new ArrayList<DBRefEntry>();
2038 // and generate appropriate mappings
2039 for (DBRefEntry cdsref : direct)
2041 // clone maplist and mapping
2042 MapList cdsposmap = new MapList(
2043 Arrays.asList(new int[][]
2044 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2045 cdsref.getMap().getMap().getToRanges(), 3, 1);
2046 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(),
2047 cdsref.getMap().getMap());
2050 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2051 cdsref.getVersion(), cdsref.getAccessionId(),
2052 new Mapping(cdsmap.getTo(), cdsposmap));
2054 // and see if we can map to the protein product for this mapping.
2055 // onSource is the filtered set of accessions on protein that we are
2056 // tranferring, so we assume accession is the same.
2057 if (cdsmap.getTo() == null && onSource != null)
2059 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2060 cdsref.getAccessionId());
2061 if (sourceRefs != null)
2063 for (DBRefEntry srcref : sourceRefs)
2065 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2067 // we have found a complementary dbref on the protein product, so
2068 // update mapping's getTo
2069 newref.getMap().setTo(proteinProduct);
2074 cdsSeq.addDBRef(newref);
2075 propagated.add(newref);
2081 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2082 * feature start/end ranges, optionally omitting specified feature types.
2083 * Returns the number of features copied.
2088 * the mapping from 'fromSeq' to 'toSeq'
2090 * if not null, only features of this type are copied (including
2091 * subtypes in the Sequence Ontology)
2094 public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2095 MapList mapping, String select, String... omitting)
2097 SequenceI copyTo = toSeq;
2098 while (copyTo.getDatasetSequence() != null)
2100 copyTo = copyTo.getDatasetSequence();
2104 * get features, optionally restricted by an ontology term
2106 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2107 .getPositionalFeatures() : fromSeq.getFeatures()
2108 .getFeaturesByOntology(select);
2111 for (SequenceFeature sf : sfs)
2113 String type = sf.getType();
2114 boolean omit = false;
2115 for (String toOmit : omitting)
2117 if (type.equals(toOmit))
2128 * locate the mapped range - null if either start or end is
2129 * not mapped (no partial overlaps are calculated)
2131 int start = sf.getBegin();
2132 int end = sf.getEnd();
2133 int[] mappedTo = mapping.locateInTo(start, end);
2135 * if whole exon range doesn't map, try interpreting it
2136 * as 5' or 3' exon overlapping the CDS range
2138 if (mappedTo == null)
2140 mappedTo = mapping.locateInTo(end, end);
2141 if (mappedTo != null)
2144 * end of exon is in CDS range - 5' overlap
2145 * to a range from the start of the peptide
2150 if (mappedTo == null)
2152 mappedTo = mapping.locateInTo(start, start);
2153 if (mappedTo != null)
2156 * start of exon is in CDS range - 3' overlap
2157 * to a range up to the end of the peptide
2159 mappedTo[1] = toSeq.getLength();
2162 if (mappedTo != null)
2164 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2165 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2166 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2167 sf.getFeatureGroup(), sf.getScore());
2168 copyTo.addSequenceFeature(copy);
2176 * Returns a mapping from dna to protein by inspecting sequence features of
2177 * type "CDS" on the dna.
2183 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2184 SequenceI proteinSeq)
2186 List<int[]> ranges = findCdsPositions(dnaSeq);
2187 int mappedDnaLength = MappingUtils.getLength(ranges);
2189 int proteinLength = proteinSeq.getLength();
2190 int proteinStart = proteinSeq.getStart();
2191 int proteinEnd = proteinSeq.getEnd();
2194 * incomplete start codon may mean X at start of peptide
2195 * we ignore both for mapping purposes
2197 if (proteinSeq.getCharAt(0) == 'X')
2199 // todo JAL-2022 support startPhase > 0
2203 List<int[]> proteinRange = new ArrayList<int[]>();
2206 * dna length should map to protein (or protein plus stop codon)
2208 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2209 if (codesForResidues == (proteinLength + 1))
2211 // assuming extra codon is for STOP and not in peptide
2214 if (codesForResidues == proteinLength)
2216 proteinRange.add(new int[] { proteinStart, proteinEnd });
2217 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2223 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2224 * start/end positions of sequence features of type "CDS" (or a sub-type of
2225 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2226 * position order, so this method is only valid for linear CDS in the same
2227 * sense as the protein product.
2232 public static List<int[]> findCdsPositions(SequenceI dnaSeq)
2234 List<int[]> result = new ArrayList<int[]>();
2236 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2237 SequenceOntologyI.CDS);
2242 SequenceFeatures.sortFeatures(sfs, true);
2245 for (SequenceFeature sf : sfs)
2250 phase = Integer.parseInt(sf.getPhase());
2251 } catch (NumberFormatException e)
2256 * phase > 0 on first codon means 5' incomplete - skip to the start
2257 * of the next codon; example ENST00000496384
2259 int begin = sf.getBegin();
2260 int end = sf.getEnd();
2261 if (result.isEmpty())
2266 // shouldn't happen!
2268 .println("Error: start phase extends beyond start CDS in "
2269 + dnaSeq.getName());
2272 result.add(new int[] { begin, end });
2276 * remove 'startPhase' positions (usually 0) from the first range
2277 * so we begin at the start of a complete codon
2279 if (!result.isEmpty())
2281 // TODO JAL-2022 correctly model start phase > 0
2282 result.get(0)[0] += startPhase;
2286 * Finally sort ranges by start position. This avoids a dependency on
2287 * keeping features in order on the sequence (if they are in order anyway,
2288 * the sort will have almost no work to do). The implicit assumption is CDS
2289 * ranges are assembled in order. Other cases should not use this method,
2290 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2292 Collections.sort(result, IntRangeComparator.ASCENDING);
2297 * Maps exon features from dna to protein, and computes variants in peptide
2298 * product generated by variants in dna, and adds them as sequence_variant
2299 * features on the protein sequence. Returns the number of variant features
2304 * @param dnaToProtein
2306 public static int computeProteinFeatures(SequenceI dnaSeq,
2307 SequenceI peptide, MapList dnaToProtein)
2309 while (dnaSeq.getDatasetSequence() != null)
2311 dnaSeq = dnaSeq.getDatasetSequence();
2313 while (peptide.getDatasetSequence() != null)
2315 peptide = peptide.getDatasetSequence();
2318 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2321 * compute protein variants from dna variants and codon mappings;
2322 * NB - alternatively we could retrieve this using the REST service e.g.
2323 * http://rest.ensembl.org/overlap/translation
2324 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2325 * which would be a bit slower but possibly more reliable
2329 * build a map with codon variations for each potentially varying peptide
2331 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2332 dnaSeq, dnaToProtein);
2335 * scan codon variations, compute peptide variants and add to peptide sequence
2338 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2340 int peptidePos = variant.getKey();
2341 List<DnaVariant>[] codonVariants = variant.getValue();
2342 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2349 * Computes non-synonymous peptide variants from codon variants and adds them
2350 * as sequence_variant features on the protein sequence (one feature per
2351 * allele variant). Selected attributes (variant id, clinical significance)
2352 * are copied over to the new features.
2355 * the protein sequence
2357 * the position to compute peptide variants for
2358 * @param codonVariants
2359 * a list of dna variants per codon position
2360 * @return the number of features added
2362 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2363 List<DnaVariant>[] codonVariants)
2365 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2367 String base1 = codonVariants[0].get(0).base;
2368 String base2 = codonVariants[1].get(0).base;
2369 String base3 = codonVariants[2].get(0).base;
2372 * variants in first codon base
2374 for (DnaVariant var : codonVariants[0])
2376 if (var.variant != null)
2378 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2379 if (alleles != null)
2381 for (String base : alleles.split(","))
2383 String codon = base + base2 + base3;
2384 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2394 * variants in second codon base
2396 for (DnaVariant var : codonVariants[1])
2398 if (var.variant != null)
2400 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2401 if (alleles != null)
2403 for (String base : alleles.split(","))
2405 String codon = base1 + base + base3;
2406 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2416 * variants in third codon base
2418 for (DnaVariant var : codonVariants[2])
2420 if (var.variant != null)
2422 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2423 if (alleles != null)
2425 for (String base : alleles.split(","))
2427 String codon = base1 + base2 + base;
2428 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2441 * Helper method that adds a peptide variant feature, provided the given codon
2442 * translates to a value different to the current residue (is a non-synonymous
2443 * variant). ID and clinical_significance attributes of the dna variant (if
2444 * present) are copied to the new feature.
2451 * @return true if a feature was added, else false
2453 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2454 String residue, DnaVariant var, String codon)
2457 * get peptide translation of codon e.g. GAT -> D
2458 * note that variants which are not single alleles,
2459 * e.g. multibase variants or HGMD_MUTATION etc
2460 * are currently ignored here
2462 String trans = codon.contains("-") ? "-"
2463 : (codon.length() > CODON_LENGTH ? null
2464 : ResidueProperties.codonTranslate(codon));
2465 if (trans != null && !trans.equals(residue))
2467 String residue3Char = StringUtils
2468 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2469 String trans3Char = StringUtils
2470 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2471 String desc = "p." + residue3Char + peptidePos + trans3Char;
2472 SequenceFeature sf = new SequenceFeature(
2473 SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
2474 peptidePos, var.getSource());
2475 StringBuilder attributes = new StringBuilder(32);
2476 String id = (String) var.variant.getValue(ID);
2479 if (id.startsWith(SEQUENCE_VARIANT))
2481 id = id.substring(SEQUENCE_VARIANT.length());
2483 sf.setValue(ID, id);
2484 attributes.append(ID).append("=").append(id);
2485 // TODO handle other species variants JAL-2064
2486 StringBuilder link = new StringBuilder(32);
2489 link.append(desc).append(" ").append(id).append(
2490 "|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2491 .append(URLEncoder.encode(id, "UTF-8"));
2492 sf.addLink(link.toString());
2493 } catch (UnsupportedEncodingException e)
2498 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2499 if (clinSig != null)
2501 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2502 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2505 peptide.addSequenceFeature(sf);
2506 if (attributes.length() > 0)
2508 sf.setAttributes(attributes.toString());
2516 * Builds a map whose key is position in the protein sequence, and value is a
2517 * list of the base and all variants for each corresponding codon position
2520 * @param dnaToProtein
2523 @SuppressWarnings("unchecked")
2524 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2525 SequenceI dnaSeq, MapList dnaToProtein)
2528 * map from peptide position to all variants of the codon which codes for it
2529 * LinkedHashMap ensures we keep the peptide features in sequence order
2531 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
2533 List<SequenceFeature> dnaFeatures = dnaSeq.getFeatures()
2534 .getFeaturesByOntology(SequenceOntologyI.SEQUENCE_VARIANT);
2535 if (dnaFeatures.isEmpty())
2540 int dnaStart = dnaSeq.getStart();
2541 int[] lastCodon = null;
2542 int lastPeptidePostion = 0;
2545 * build a map of codon variations for peptides
2547 for (SequenceFeature sf : dnaFeatures)
2549 int dnaCol = sf.getBegin();
2550 if (dnaCol != sf.getEnd())
2552 // not handling multi-locus variant features
2557 * extract dna variants to a string array
2559 String alls = (String) sf.getValue(Gff3Helper.ALLELES);
2562 continue; // non-SNP VCF variant perhaps - can't process this
2564 String[] alleles = alls.toUpperCase().split(",");
2566 for (String allele : alleles)
2568 alleles[i++] = allele.trim(); // lose any space characters "A, G"
2571 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2574 // feature doesn't lie within coding region
2577 int peptidePosition = mapsTo[0];
2578 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2579 if (codonVariants == null)
2581 codonVariants = new ArrayList[CODON_LENGTH];
2582 codonVariants[0] = new ArrayList<DnaVariant>();
2583 codonVariants[1] = new ArrayList<DnaVariant>();
2584 codonVariants[2] = new ArrayList<DnaVariant>();
2585 variants.put(peptidePosition, codonVariants);
2589 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2591 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2592 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2593 peptidePosition, peptidePosition));
2594 lastPeptidePostion = peptidePosition;
2598 * save nucleotide (and any variant) for each codon position
2600 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2602 String nucleotide = String.valueOf(
2603 dnaSeq.getCharAt(codon[codonPos] - dnaStart)).toUpperCase();
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));
2636 * Makes an alignment with a copy of the given sequences, adding in any
2637 * non-redundant sequences which are mapped to by the cross-referenced
2643 * the alignment dataset shared by the new copy
2646 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2647 SequenceI[] xrefs, AlignmentI dataset)
2649 AlignmentI copy = new Alignment(new Alignment(seqs));
2650 copy.setDataset(dataset);
2651 boolean isProtein = !copy.isNucleotide();
2652 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2655 for (SequenceI xref : xrefs)
2657 DBRefEntry[] dbrefs = xref.getDBRefs();
2660 for (DBRefEntry dbref : dbrefs)
2662 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2663 || dbref.getMap().getTo().isProtein() != isProtein)
2667 SequenceI mappedTo = dbref.getMap().getTo();
2668 SequenceI match = matcher.findIdMatch(mappedTo);
2671 matcher.add(mappedTo);
2672 copy.addSequence(mappedTo);
2682 * Try to align sequences in 'unaligned' to match the alignment of their
2683 * mapped regions in 'aligned'. For example, could use this to align CDS
2684 * sequences which are mapped to their parent cDNA sequences.
2686 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2687 * dna-to-protein or protein-to-dna use alternative methods.
2690 * sequences to be aligned
2692 * holds aligned sequences and their mappings
2695 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2698 * easy case - aligning a copy of aligned sequences
2700 if (alignAsSameSequences(unaligned, aligned))
2702 return unaligned.getHeight();
2706 * fancy case - aligning via mappings between sequences
2708 List<SequenceI> unmapped = new ArrayList<SequenceI>();
2709 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2710 unaligned, aligned, unmapped);
2711 int width = columnMap.size();
2712 char gap = unaligned.getGapCharacter();
2713 int realignedCount = 0;
2714 // TODO: verify this loop scales sensibly for very wide/high alignments
2716 for (SequenceI seq : unaligned.getSequences())
2718 if (!unmapped.contains(seq))
2720 char[] newSeq = new char[width];
2721 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2722 // Integer iteration below
2727 * traverse the map to find columns populated
2730 for (Integer column : columnMap.keySet())
2732 Character c = columnMap.get(column).get(seq);
2736 * sequence has a character at this position
2746 * trim trailing gaps
2748 if (lastCol < width)
2750 char[] tmp = new char[lastCol + 1];
2751 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2754 // TODO: optimise SequenceI to avoid char[]->String->char[]
2755 seq.setSequence(String.valueOf(newSeq));
2759 return realignedCount;
2763 * If unaligned and aligned sequences share the same dataset sequences, then
2764 * simply copies the aligned sequences to the unaligned sequences and returns
2765 * true; else returns false
2768 * - sequences to be aligned based on aligned
2770 * - 'guide' alignment containing sequences derived from same dataset
2774 static boolean alignAsSameSequences(AlignmentI unaligned,
2777 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2779 return false; // should only pass alignments with datasets here
2782 // map from dataset sequence to alignment sequence(s)
2783 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<SequenceI, List<SequenceI>>();
2784 for (SequenceI seq : aligned.getSequences())
2786 SequenceI ds = seq.getDatasetSequence();
2787 if (alignedDatasets.get(ds) == null)
2789 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2791 alignedDatasets.get(ds).add(seq);
2795 * first pass - check whether all sequences to be aligned share a dataset
2796 * sequence with an aligned sequence
2798 for (SequenceI seq : unaligned.getSequences())
2800 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2807 * second pass - copy aligned sequences;
2808 * heuristic rule: pair off sequences in order for the case where
2809 * more than one shares the same dataset sequence
2811 for (SequenceI seq : unaligned.getSequences())
2813 List<SequenceI> alignedSequences = alignedDatasets
2814 .get(seq.getDatasetSequence());
2815 // TODO: getSequenceAsString() will be deprecated in the future
2816 // TODO: need to leave to SequenceI implementor to update gaps
2817 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2818 if (alignedSequences.size() > 0)
2820 // pop off aligned sequences (except the last one)
2821 alignedSequences.remove(0);
2829 * Returns a map whose key is alignment column number (base 1), and whose
2830 * values are a map of sequence characters in that column.
2837 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2838 AlignmentI unaligned, AlignmentI aligned,
2839 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.
2876 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2877 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2881 * the sequence whose column positions we are recording
2883 * a sequence that is mapped to the first sequence
2885 * the mapping from 'fromSeq' to 'seq'
2887 * a map to add the column positions (in fromSeq) of the mapped
2891 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2892 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2900 * invert mapping if it is from unaligned to aligned sequence
2902 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2904 seqMap = new Mapping(seq.getDatasetSequence(),
2905 seqMap.getMap().getInverse());
2908 int toStart = seq.getStart();
2911 * traverse [start, end, start, end...] ranges in fromSeq
2913 for (int[] fromRange : seqMap.getMap().getFromRanges())
2915 for (int i = 0; i < fromRange.length - 1; i += 2)
2917 boolean forward = fromRange[i + 1] >= fromRange[i];
2920 * find the range mapped to (sequence positions base 1)
2922 int[] range = seqMap.locateMappedRange(fromRange[i],
2926 System.err.println("Error in mapping " + seqMap + " from "
2927 + fromSeq.getName());
2930 int fromCol = fromSeq.findIndex(fromRange[i]);
2931 int mappedCharPos = range[0];
2934 * walk over the 'from' aligned sequence in forward or reverse
2935 * direction; when a non-gap is found, record the column position
2936 * of the next character of the mapped-to sequence; stop when all
2937 * the characters of the range have been counted
2939 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
2942 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
2945 * mapped from sequence has a character in this column
2946 * record the column position for the mapped to character
2948 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2949 if (seqsMap == null)
2951 seqsMap = new HashMap<SequenceI, Character>();
2952 map.put(fromCol, seqsMap);
2954 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
2957 fromCol += (forward ? 1 : -1);
2964 // strictly temporary hack until proper criteria for aligning protein to cds
2965 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2966 public static boolean looksLikeEnsembl(AlignmentI alignment)
2968 for (SequenceI seq : alignment.getSequences())
2970 String name = seq.getName();
2971 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))