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.StringUtils;
47 import java.io.UnsupportedEncodingException;
48 import java.net.URLEncoder;
49 import java.util.ArrayList;
50 import java.util.Arrays;
51 import java.util.Collection;
52 import java.util.Collections;
53 import java.util.Comparator;
54 import java.util.HashMap;
55 import java.util.HashSet;
56 import java.util.Iterator;
57 import java.util.LinkedHashMap;
58 import java.util.List;
60 import java.util.Map.Entry;
61 import java.util.NoSuchElementException;
63 import java.util.TreeMap;
66 * grab bag of useful alignment manipulation operations Expect these to be
67 * refactored elsewhere at some point.
72 public class AlignmentUtils
75 private static final int CODON_LENGTH = 3;
76 private static final String SEQUENCE_VARIANT = "sequence_variant:";
77 private static final String ID = "ID";
80 * A data model to hold the 'normal' base value at a position, and an optional
81 * sequence variant feature
83 static final class DnaVariant
87 SequenceFeature variant;
89 DnaVariant(String nuc)
95 DnaVariant(String nuc, SequenceFeature var)
101 public String getSource()
103 return variant == null ? null : variant.getFeatureGroup();
108 * given an existing alignment, create a new alignment including all, or up to
109 * flankSize additional symbols from each sequence's dataset sequence
115 public static AlignmentI expandContext(AlignmentI core, int flankSize)
117 List<SequenceI> sq = new ArrayList<SequenceI>();
119 for (SequenceI s : core.getSequences())
121 SequenceI newSeq = s.deriveSequence();
122 final int newSeqStart = newSeq.getStart() - 1;
123 if (newSeqStart > maxoffset
124 && newSeq.getDatasetSequence().getStart() < s.getStart())
126 maxoffset = newSeqStart;
132 maxoffset = Math.min(maxoffset, flankSize);
136 * now add offset left and right to create an expanded alignment
138 for (SequenceI s : sq)
141 while (ds.getDatasetSequence() != null)
143 ds = ds.getDatasetSequence();
145 int s_end = s.findPosition(s.getStart() + s.getLength());
146 // find available flanking residues for sequence
147 int ustream_ds = s.getStart() - ds.getStart();
148 int dstream_ds = ds.getEnd() - s_end;
150 // build new flanked sequence
152 // compute gap padding to start of flanking sequence
153 int offset = maxoffset - ustream_ds;
155 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
158 if (flankSize < ustream_ds)
160 // take up to flankSize residues
161 offset = maxoffset - flankSize;
162 ustream_ds = flankSize;
164 if (flankSize <= dstream_ds)
166 dstream_ds = flankSize - 1;
169 // TODO use Character.toLowerCase to avoid creating String objects?
170 char[] upstream = new String(ds.getSequence(s.getStart() - 1
171 - ustream_ds, s.getStart() - 1)).toLowerCase().toCharArray();
172 char[] downstream = new String(ds.getSequence(s_end - 1, s_end
173 + dstream_ds)).toLowerCase().toCharArray();
174 char[] coreseq = s.getSequence();
175 char[] nseq = new char[offset + upstream.length + downstream.length
177 char c = core.getGapCharacter();
180 for (; p < offset; p++)
185 System.arraycopy(upstream, 0, nseq, p, upstream.length);
186 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
188 System.arraycopy(downstream, 0, nseq, p + coreseq.length
189 + upstream.length, downstream.length);
190 s.setSequence(new String(nseq));
191 s.setStart(s.getStart() - ustream_ds);
192 s.setEnd(s_end + downstream.length);
194 AlignmentI newAl = new jalview.datamodel.Alignment(
195 sq.toArray(new SequenceI[0]));
196 for (SequenceI s : sq)
198 if (s.getAnnotation() != null)
200 for (AlignmentAnnotation aa : s.getAnnotation())
202 aa.adjustForAlignment(); // JAL-1712 fix
203 newAl.addAnnotation(aa);
207 newAl.setDataset(core.getDataset());
212 * Returns the index (zero-based position) of a sequence in an alignment, or
219 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
223 for (SequenceI alSeq : al.getSequences())
236 * Returns a map of lists of sequences in the alignment, keyed by sequence
237 * name. For use in mapping between different alignment views of the same
240 * @see jalview.datamodel.AlignmentI#getSequencesByName()
242 public static Map<String, List<SequenceI>> getSequencesByName(
245 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
246 for (SequenceI seq : al.getSequences())
248 String name = seq.getName();
251 List<SequenceI> seqs = theMap.get(name);
254 seqs = new ArrayList<SequenceI>();
255 theMap.put(name, seqs);
264 * Build mapping of protein to cDNA alignment. Mappings are made between
265 * sequences where the cDNA translates to the protein sequence. Any new
266 * mappings are added to the protein alignment. Returns true if any mappings
267 * either already exist or were added, else false.
269 * @param proteinAlignment
270 * @param cdnaAlignment
273 public static boolean mapProteinAlignmentToCdna(
274 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
276 if (proteinAlignment == null || cdnaAlignment == null)
281 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
282 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
285 * First pass - map sequences where cross-references exist. This include
286 * 1-to-many mappings to support, for example, variant cDNA.
288 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
289 cdnaAlignment, mappedDna, mappedProtein, true);
292 * Second pass - map sequences where no cross-references exist. This only
293 * does 1-to-1 mappings and assumes corresponding sequences are in the same
294 * order in the alignments.
296 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
297 mappedDna, mappedProtein, false);
298 return mappingPerformed;
302 * Make mappings between compatible sequences (where the cDNA translation
303 * matches the protein).
305 * @param proteinAlignment
306 * @param cdnaAlignment
308 * a set of mapped DNA sequences (to add to)
309 * @param mappedProtein
310 * a set of mapped Protein sequences (to add to)
312 * if true, only map sequences where xrefs exist
315 protected static boolean mapProteinToCdna(
316 final AlignmentI proteinAlignment,
317 final AlignmentI cdnaAlignment, Set<SequenceI> mappedDna,
318 Set<SequenceI> mappedProtein, boolean xrefsOnly)
320 boolean mappingExistsOrAdded = false;
321 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
322 for (SequenceI aaSeq : thisSeqs)
324 boolean proteinMapped = false;
325 AlignedCodonFrame acf = new AlignedCodonFrame();
327 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
330 * Always try to map if sequences have xref to each other; this supports
331 * variant cDNA or alternative splicing for a protein sequence.
333 * If no xrefs, try to map progressively, assuming that alignments have
334 * mappable sequences in corresponding order. These are not
335 * many-to-many, as that would risk mixing species with similar cDNA
338 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
344 * Don't map non-xrefd sequences more than once each. This heuristic
345 * allows us to pair up similar sequences in ordered alignments.
348 && (mappedProtein.contains(aaSeq) || mappedDna
353 if (mappingExists(proteinAlignment.getCodonFrames(),
354 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
356 mappingExistsOrAdded = true;
360 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
363 acf.addMap(cdnaSeq, aaSeq, map);
364 mappingExistsOrAdded = true;
365 proteinMapped = true;
366 mappedDna.add(cdnaSeq);
367 mappedProtein.add(aaSeq);
373 proteinAlignment.addCodonFrame(acf);
376 return mappingExistsOrAdded;
380 * Answers true if the mappings include one between the given (dataset)
383 public static boolean mappingExists(List<AlignedCodonFrame> mappings,
384 SequenceI aaSeq, SequenceI cdnaSeq)
386 if (mappings != null)
388 for (AlignedCodonFrame acf : mappings)
390 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
400 * Builds a mapping (if possible) of a cDNA to a protein sequence.
402 * <li>first checks if the cdna translates exactly to the protein sequence</li>
403 * <li>else checks for translation after removing a STOP codon</li>
404 * <li>else checks for translation after removing a START codon</li>
405 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
407 * Returns null if no mapping is determined.
410 * the aligned protein sequence
412 * the aligned cdna sequence
415 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
419 * Here we handle either dataset sequence set (desktop) or absent (applet).
420 * Use only the char[] form of the sequence to avoid creating possibly large
423 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
424 char[] aaSeqChars = proteinDataset != null ? proteinDataset
425 .getSequence() : proteinSeq.getSequence();
426 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
427 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
428 : cdnaSeq.getSequence();
429 if (aaSeqChars == null || cdnaSeqChars == null)
435 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
437 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
438 int cdnaLength = cdnaSeqChars.length;
439 int cdnaStart = cdnaSeq.getStart();
440 int cdnaEnd = cdnaSeq.getEnd();
441 final int proteinStart = proteinSeq.getStart();
442 final int proteinEnd = proteinSeq.getEnd();
445 * If lengths don't match, try ignoring stop codon (if present)
447 if (cdnaLength != mappedLength && cdnaLength > 2)
449 String lastCodon = String.valueOf(cdnaSeqChars, cdnaLength - CODON_LENGTH, CODON_LENGTH)
451 for (String stop : ResidueProperties.STOP)
453 if (lastCodon.equals(stop))
455 cdnaEnd -= CODON_LENGTH;
456 cdnaLength -= CODON_LENGTH;
463 * If lengths still don't match, try ignoring start codon.
466 if (cdnaLength != mappedLength
468 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
469 .equals(ResidueProperties.START))
471 startOffset += CODON_LENGTH;
472 cdnaStart += CODON_LENGTH;
473 cdnaLength -= CODON_LENGTH;
476 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
479 * protein is translation of dna (+/- start/stop codons)
481 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd }, new int[]
482 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
487 * translation failed - try mapping CDS annotated regions of dna
489 return mapCdsToProtein(cdnaSeq, proteinSeq);
493 * Test whether the given cdna sequence, starting at the given offset,
494 * translates to the given amino acid sequence, using the standard translation
495 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
497 * @param cdnaSeqChars
502 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
505 if (cdnaSeqChars == null || aaSeqChars == null)
511 int dnaPos = cdnaStart;
512 for (; dnaPos < cdnaSeqChars.length - 2
513 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
515 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
516 final String translated = ResidueProperties.codonTranslate(codon);
519 * allow * in protein to match untranslatable in dna
521 final char aaRes = aaSeqChars[aaPos];
522 if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
526 if (translated == null || !(aaRes == translated.charAt(0)))
529 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
530 // + codon + "(" + translated + ") != " + aaRes));
536 * check we matched all of the protein sequence
538 if (aaPos != aaSeqChars.length)
544 * check we matched all of the dna except
545 * for optional trailing STOP codon
547 if (dnaPos == cdnaSeqChars.length)
551 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
553 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
554 if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
563 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
564 * currently assumes that we are aligning cDNA to match protein.
567 * the sequence to be realigned
569 * the alignment whose sequence alignment is to be 'copied'
571 * character string represent a gap in the realigned sequence
572 * @param preserveUnmappedGaps
573 * @param preserveMappedGaps
574 * @return true if the sequence was realigned, false if it could not be
576 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
577 String gap, boolean preserveMappedGaps,
578 boolean preserveUnmappedGaps)
581 * Get any mappings from the source alignment to the target (dataset)
584 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
585 // all mappings. Would it help to constrain this?
586 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
587 if (mappings == null || mappings.isEmpty())
593 * Locate the aligned source sequence whose dataset sequence is mapped. We
594 * just take the first match here (as we can't align like more than one
597 SequenceI alignFrom = null;
598 AlignedCodonFrame mapping = null;
599 for (AlignedCodonFrame mp : mappings)
601 alignFrom = mp.findAlignedSequence(seq, al);
602 if (alignFrom != null)
609 if (alignFrom == null)
613 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
614 preserveMappedGaps, preserveUnmappedGaps);
619 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
620 * match residues and codons. Flags control whether existing gaps in unmapped
621 * (intron) and mapped (exon) regions are preserved or not. Gaps between
622 * intron and exon are only retained if both flags are set.
629 * @param preserveUnmappedGaps
630 * @param preserveMappedGaps
632 public static void alignSequenceAs(SequenceI alignTo,
633 SequenceI alignFrom, AlignedCodonFrame mapping, String myGap,
634 char sourceGap, boolean preserveMappedGaps,
635 boolean preserveUnmappedGaps)
637 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
639 // aligned and dataset sequence positions, all base zero
643 int basesWritten = 0;
644 char myGapChar = myGap.charAt(0);
645 int ratio = myGap.length();
647 int fromOffset = alignFrom.getStart() - 1;
648 int toOffset = alignTo.getStart() - 1;
649 int sourceGapMappedLength = 0;
650 boolean inExon = false;
651 final char[] thisSeq = alignTo.getSequence();
652 final char[] thatAligned = alignFrom.getSequence();
653 StringBuilder thisAligned = new StringBuilder(2 * thisSeq.length);
656 * Traverse the 'model' aligned sequence
658 for (char sourceChar : thatAligned)
660 if (sourceChar == sourceGap)
662 sourceGapMappedLength += ratio;
667 * Found a non-gap character. Locate its mapped region if any.
670 // Note mapping positions are base 1, our sequence positions base 0
671 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
672 sourceDsPos + fromOffset);
673 if (mappedPos == null)
676 * unmapped position; treat like a gap
678 sourceGapMappedLength += ratio;
679 // System.err.println("Can't align: no codon mapping to residue "
680 // + sourceDsPos + "(" + sourceChar + ")");
685 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
686 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
687 StringBuilder trailingCopiedGap = new StringBuilder();
690 * Copy dna sequence up to and including this codon. Optionally, include
691 * gaps before the codon starts (in introns) and/or after the codon starts
694 * Note this only works for 'linear' splicing, not reverse or interleaved.
695 * But then 'align dna as protein' doesn't make much sense otherwise.
697 int intronLength = 0;
698 while (basesWritten + toOffset < mappedCodonEnd
699 && thisSeqPos < thisSeq.length)
701 final char c = thisSeq[thisSeqPos++];
705 int sourcePosition = basesWritten + toOffset;
706 if (sourcePosition < mappedCodonStart)
709 * Found an unmapped (intron) base. First add in any preceding gaps
712 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
714 thisAligned.append(trailingCopiedGap.toString());
715 intronLength += trailingCopiedGap.length();
716 trailingCopiedGap = new StringBuilder();
723 final boolean startOfCodon = sourcePosition == mappedCodonStart;
724 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
725 preserveUnmappedGaps, sourceGapMappedLength, inExon,
726 trailingCopiedGap.length(), intronLength, startOfCodon);
727 for (int i = 0; i < gapsToAdd; i++)
729 thisAligned.append(myGapChar);
731 sourceGapMappedLength = 0;
734 thisAligned.append(c);
735 trailingCopiedGap = new StringBuilder();
739 if (inExon && preserveMappedGaps)
741 trailingCopiedGap.append(myGapChar);
743 else if (!inExon && preserveUnmappedGaps)
745 trailingCopiedGap.append(myGapChar);
752 * At end of model aligned sequence. Copy any remaining target sequence, optionally
753 * including (intron) gaps.
755 while (thisSeqPos < thisSeq.length)
757 final char c = thisSeq[thisSeqPos++];
758 if (c != myGapChar || preserveUnmappedGaps)
760 thisAligned.append(c);
762 sourceGapMappedLength--;
766 * finally add gaps to pad for any trailing source gaps or
767 * unmapped characters
769 if (preserveUnmappedGaps)
771 while (sourceGapMappedLength > 0)
773 thisAligned.append(myGapChar);
774 sourceGapMappedLength--;
779 * All done aligning, set the aligned sequence.
781 alignTo.setSequence(new String(thisAligned));
785 * Helper method to work out how many gaps to insert when realigning.
787 * @param preserveMappedGaps
788 * @param preserveUnmappedGaps
789 * @param sourceGapMappedLength
791 * @param trailingCopiedGap
792 * @param intronLength
793 * @param startOfCodon
796 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
797 boolean preserveUnmappedGaps, int sourceGapMappedLength,
798 boolean inExon, int trailingGapLength, int intronLength,
799 final boolean startOfCodon)
805 * Reached start of codon. Ignore trailing gaps in intron unless we are
806 * preserving gaps in both exon and intron. Ignore them anyway if the
807 * protein alignment introduces a gap at least as large as the intronic
810 if (inExon && !preserveMappedGaps)
812 trailingGapLength = 0;
814 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
816 trailingGapLength = 0;
820 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
824 if (intronLength + trailingGapLength <= sourceGapMappedLength)
826 gapsToAdd = sourceGapMappedLength - intronLength;
830 gapsToAdd = Math.min(intronLength + trailingGapLength
831 - sourceGapMappedLength, trailingGapLength);
838 * second or third base of codon; check for any gaps in dna
840 if (!preserveMappedGaps)
842 trailingGapLength = 0;
844 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
850 * Realigns the given protein to match the alignment of the dna, using codon
851 * mappings to translate aligned codon positions to protein residues.
854 * the alignment whose sequences are realigned by this method
856 * the dna alignment whose alignment we are 'copying'
857 * @return the number of sequences that were realigned
859 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
861 if (protein.isNucleotide() || !dna.isNucleotide())
863 System.err.println("Wrong alignment type in alignProteinAsDna");
866 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
867 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
868 protein, dna, unmappedProtein);
869 return alignProteinAs(protein, alignedCodons, unmappedProtein);
873 * Realigns the given dna to match the alignment of the protein, using codon
874 * mappings to translate aligned peptide positions to codons.
876 * Always produces a padded CDS alignment.
879 * the alignment whose sequences are realigned by this method
881 * the protein alignment whose alignment we are 'copying'
882 * @return the number of sequences that were realigned
884 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
886 if (protein.isNucleotide() || !dna.isNucleotide())
888 System.err.println("Wrong alignment type in alignProteinAsDna");
891 // todo: implement this
892 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
893 int alignedCount = 0;
894 int width = 0; // alignment width for padding CDS
895 for (SequenceI dnaSeq : dna.getSequences())
897 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
898 dna.getGapCharacter()))
902 width = Math.max(dnaSeq.getLength(), width);
906 for (SequenceI dnaSeq : dna.getSequences())
908 oldwidth = dnaSeq.getLength();
909 diff = width - oldwidth;
912 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
919 * Helper method to align (if possible) the dna sequence to match the
920 * alignment of a mapped protein sequence. This is currently limited to
921 * handling coding sequence only.
929 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
930 AlignmentI protein, List<AlignedCodonFrame> mappings, char gapChar)
932 SequenceI cdsDss = cdsSeq.getDatasetSequence();
936 .println("alignCdsSequenceAsProtein needs aligned sequence!");
940 List<AlignedCodonFrame> dnaMappings = MappingUtils
941 .findMappingsForSequence(cdsSeq, mappings);
942 for (AlignedCodonFrame mapping : dnaMappings)
944 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
947 int peptideLength = peptide.getLength();
948 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
951 MapList mapList = map.getMap();
952 if (map.getTo() == peptide.getDatasetSequence())
954 mapList = mapList.getInverse();
956 int cdsLength = cdsDss.getLength();
957 int mappedFromLength = MappingUtils.getLength(mapList
959 int mappedToLength = MappingUtils
960 .getLength(mapList.getToRanges());
961 boolean addStopCodon = (cdsLength == mappedFromLength * CODON_LENGTH + CODON_LENGTH)
962 || (peptide.getDatasetSequence().getLength() == mappedFromLength - 1);
963 if (cdsLength != mappedToLength && !addStopCodon)
967 .format("Can't align cds as protein (length mismatch %d/%d): %s",
968 cdsLength, mappedToLength,
973 * pre-fill the aligned cds sequence with gaps
975 char[] alignedCds = new char[peptideLength * CODON_LENGTH
976 + (addStopCodon ? CODON_LENGTH : 0)];
977 Arrays.fill(alignedCds, gapChar);
980 * walk over the aligned peptide sequence and insert mapped
981 * codons for residues in the aligned cds sequence
983 char[] alignedPeptide = peptide.getSequence();
984 char[] nucleotides = cdsDss.getSequence();
986 int cdsStart = cdsDss.getStart();
987 int proteinPos = peptide.getStart() - 1;
989 for (char residue : alignedPeptide)
991 if (Comparison.isGap(residue))
993 cdsCol += CODON_LENGTH;
998 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1001 // e.g. incomplete start codon, X in peptide
1002 cdsCol += CODON_LENGTH;
1006 for (int j = codon[0]; j <= codon[1]; j++)
1008 char mappedBase = nucleotides[j - cdsStart];
1009 alignedCds[cdsCol++] = mappedBase;
1017 * append stop codon if not mapped from protein,
1018 * closing it up to the end of the mapped sequence
1020 if (copiedBases == nucleotides.length - CODON_LENGTH)
1022 for (int i = alignedCds.length - 1; i >= 0; i--)
1024 if (!Comparison.isGap(alignedCds[i]))
1026 cdsCol = i + 1; // gap just after end of sequence
1030 for (int i = nucleotides.length - CODON_LENGTH; i < nucleotides.length; i++)
1032 alignedCds[cdsCol++] = nucleotides[i];
1035 cdsSeq.setSequence(new String(alignedCds));
1044 * Builds a map whose key is an aligned codon position (3 alignment column
1045 * numbers base 0), and whose value is a map from protein sequence to each
1046 * protein's peptide residue for that codon. The map generates an ordering of
1047 * the codons, and allows us to read off the peptides at each position in
1048 * order to assemble 'aligned' protein sequences.
1051 * the protein alignment
1053 * the coding dna alignment
1054 * @param unmappedProtein
1055 * any unmapped proteins are added to this list
1058 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1059 AlignmentI protein, AlignmentI dna,
1060 List<SequenceI> unmappedProtein)
1063 * maintain a list of any proteins with no mappings - these will be
1064 * rendered 'as is' in the protein alignment as we can't align them
1066 unmappedProtein.addAll(protein.getSequences());
1068 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1071 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1072 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1073 * comparator keeps the codon positions ordered.
1075 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
1076 new CodonComparator());
1078 for (SequenceI dnaSeq : dna.getSequences())
1080 for (AlignedCodonFrame mapping : mappings)
1082 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1085 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1086 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
1087 seqMap, alignedCodons);
1088 unmappedProtein.remove(prot);
1094 * Finally add any unmapped peptide start residues (e.g. for incomplete
1095 * codons) as if at the codon position before the second residue
1097 // TODO resolve JAL-2022 so this fudge can be removed
1098 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1099 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1101 return alignedCodons;
1105 * Scans for any protein mapped from position 2 (meaning unmapped start
1106 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1107 * preceding position in the alignment
1109 * @param alignedCodons
1110 * the codon-to-peptide map
1111 * @param mappedSequenceCount
1112 * the number of distinct sequences in the map
1114 protected static void addUnmappedPeptideStarts(
1115 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1116 int mappedSequenceCount)
1118 // TODO delete this ugly hack once JAL-2022 is resolved
1119 // i.e. we can model startPhase > 0 (incomplete start codon)
1121 List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
1122 AlignedCodon lastCodon = null;
1123 Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
1125 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1128 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1131 SequenceI seq = sequenceCodon.getKey();
1132 if (sequencesChecked.contains(seq))
1136 sequencesChecked.add(seq);
1137 AlignedCodon codon = sequenceCodon.getValue();
1138 if (codon.peptideCol > 1)
1141 .println("Problem mapping protein with >1 unmapped start positions: "
1144 else if (codon.peptideCol == 1)
1147 * first position (peptideCol == 0) was unmapped - add it
1149 if (lastCodon != null)
1151 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1152 lastCodon.pos2, lastCodon.pos3, String.valueOf(seq
1154 toAdd.put(seq, firstPeptide);
1159 * unmapped residue at start of alignment (no prior column) -
1160 * 'insert' at nominal codon [0, 0, 0]
1162 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1163 String.valueOf(seq.getCharAt(0)), 0);
1164 toAdd.put(seq, firstPeptide);
1167 if (sequencesChecked.size() == mappedSequenceCount)
1169 // no need to check past first mapped position in all sequences
1173 lastCodon = entry.getKey();
1177 * add any new codons safely after iterating over the map
1179 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1181 addCodonToMap(alignedCodons, startCodon.getValue(),
1182 startCodon.getKey());
1187 * Update the aligned protein sequences to match the codon alignments given in
1191 * @param alignedCodons
1192 * an ordered map of codon positions (columns), with sequence/peptide
1193 * values present in each column
1194 * @param unmappedProtein
1197 protected static int alignProteinAs(AlignmentI protein,
1198 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1199 List<SequenceI> unmappedProtein)
1202 * Prefill aligned sequences with gaps before inserting aligned protein
1205 int alignedWidth = alignedCodons.size();
1206 char[] gaps = new char[alignedWidth];
1207 Arrays.fill(gaps, protein.getGapCharacter());
1208 String allGaps = String.valueOf(gaps);
1209 for (SequenceI seq : protein.getSequences())
1211 if (!unmappedProtein.contains(seq))
1213 seq.setSequence(allGaps);
1218 for (AlignedCodon codon : alignedCodons.keySet())
1220 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1222 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1224 // place translated codon at its column position in sequence
1225 entry.getKey().getSequence()[column] = entry.getValue().product
1234 * Populate the map of aligned codons by traversing the given sequence
1235 * mapping, locating the aligned positions of mapped codons, and adding those
1236 * positions and their translation products to the map.
1239 * the aligned sequence we are mapping from
1241 * the sequence to be aligned to the codons
1243 * the gap character in the dna sequence
1245 * a mapping to a sequence translation
1246 * @param alignedCodons
1247 * the map we are building up
1249 static void addCodonPositions(SequenceI dna, SequenceI protein,
1250 char gapChar, Mapping seqMap,
1251 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1253 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1256 * add codon positions, and their peptide translations, to the alignment
1257 * map, while remembering the first codon mapped
1259 while (codons.hasNext())
1263 AlignedCodon codon = codons.next();
1264 addCodonToMap(alignedCodons, codon, protein);
1265 } catch (IncompleteCodonException e)
1267 // possible incomplete trailing codon - ignore
1268 } catch (NoSuchElementException e)
1270 // possibly peptide lacking STOP
1276 * Helper method to add a codon-to-peptide entry to the aligned codons map
1278 * @param alignedCodons
1282 protected static void addCodonToMap(
1283 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1284 AlignedCodon codon, SequenceI protein)
1286 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1287 if (seqProduct == null)
1289 seqProduct = new HashMap<SequenceI, AlignedCodon>();
1290 alignedCodons.put(codon, seqProduct);
1292 seqProduct.put(protein, codon);
1296 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1297 * between at least one pair of sequences in the two alignments. Currently,
1300 * <li>One alignment must be nucleotide, and the other protein</li>
1301 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1302 * <li>Mappable means the nucleotide translation matches the protein sequence</li>
1303 * <li>The translation may ignore start and stop codons if present in the
1311 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1313 if (al1 == null || al2 == null)
1319 * Require one nucleotide and one protein
1321 if (al1.isNucleotide() == al2.isNucleotide())
1325 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1326 AlignmentI protein = dna == al1 ? al2 : al1;
1327 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1328 for (SequenceI dnaSeq : dna.getSequences())
1330 for (SequenceI proteinSeq : protein.getSequences())
1332 if (isMappable(dnaSeq, proteinSeq, mappings))
1342 * Returns true if the dna sequence is mapped, or could be mapped, to the
1350 protected static boolean isMappable(SequenceI dnaSeq,
1351 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1353 if (dnaSeq == null || proteinSeq == null)
1358 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq : dnaSeq
1359 .getDatasetSequence();
1360 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null ? proteinSeq
1361 : proteinSeq.getDatasetSequence();
1363 for (AlignedCodonFrame mapping : mappings)
1365 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1375 * Just try to make a mapping (it is not yet stored), test whether
1378 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1382 * Finds any reference annotations associated with the sequences in
1383 * sequenceScope, that are not already added to the alignment, and adds them
1384 * to the 'candidates' map. Also populates a lookup table of annotation
1385 * labels, keyed by calcId, for use in constructing tooltips or the like.
1387 * @param sequenceScope
1388 * the sequences to scan for reference annotations
1389 * @param labelForCalcId
1390 * (optional) map to populate with label for calcId
1392 * map to populate with annotations for sequence
1394 * the alignment to check for presence of annotations
1396 public static void findAddableReferenceAnnotations(
1397 List<SequenceI> sequenceScope,
1398 Map<String, String> labelForCalcId,
1399 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1402 if (sequenceScope == null)
1408 * For each sequence in scope, make a list of any annotations on the
1409 * underlying dataset sequence which are not already on the alignment.
1411 * Add to a map of { alignmentSequence, <List of annotations to add> }
1413 for (SequenceI seq : sequenceScope)
1415 SequenceI dataset = seq.getDatasetSequence();
1416 if (dataset == null)
1420 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1421 if (datasetAnnotations == null)
1425 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1426 for (AlignmentAnnotation dsann : datasetAnnotations)
1429 * Find matching annotations on the alignment. If none is found, then
1430 * add this annotation to the list of 'addable' annotations for this
1433 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1434 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1435 if (!matchedAlignmentAnnotations.iterator().hasNext())
1438 if (labelForCalcId != null)
1440 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1445 * Save any addable annotations for this sequence
1447 if (!result.isEmpty())
1449 candidates.put(seq, result);
1455 * Adds annotations to the top of the alignment annotations, in the same order
1456 * as their related sequences.
1458 * @param annotations
1459 * the annotations to add
1461 * the alignment to add them to
1462 * @param selectionGroup
1463 * current selection group (or null if none)
1465 public static void addReferenceAnnotations(
1466 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1467 final AlignmentI alignment, final SequenceGroup selectionGroup)
1469 for (SequenceI seq : annotations.keySet())
1471 for (AlignmentAnnotation ann : annotations.get(seq))
1473 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1475 int endRes = ann.annotations.length;
1476 if (selectionGroup != null)
1478 startRes = selectionGroup.getStartRes();
1479 endRes = selectionGroup.getEndRes();
1481 copyAnn.restrict(startRes, endRes);
1484 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1485 * original annotation is already on the sequence.
1487 if (!seq.hasAnnotation(ann))
1489 seq.addAlignmentAnnotation(copyAnn);
1492 copyAnn.adjustForAlignment();
1493 // add to the alignment and set visible
1494 alignment.addAnnotation(copyAnn);
1495 copyAnn.visible = true;
1501 * Set visibility of alignment annotations of specified types (labels), for
1502 * specified sequences. This supports controls like
1503 * "Show all secondary structure", "Hide all Temp factor", etc.
1505 * @al the alignment to scan for annotations
1507 * the types (labels) of annotations to be updated
1508 * @param forSequences
1509 * if not null, only annotations linked to one of these sequences are
1510 * in scope for update; if null, acts on all sequence annotations
1512 * if this flag is true, 'types' is ignored (label not checked)
1514 * if true, set visibility on, else set off
1516 public static void showOrHideSequenceAnnotations(AlignmentI al,
1517 Collection<String> types, List<SequenceI> forSequences,
1518 boolean anyType, boolean doShow)
1520 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1523 for (AlignmentAnnotation aa : anns)
1525 if (anyType || types.contains(aa.label))
1527 if ((aa.sequenceRef != null)
1528 && (forSequences == null || forSequences
1529 .contains(aa.sequenceRef)))
1531 aa.visible = doShow;
1539 * Returns true if either sequence has a cross-reference to the other
1545 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1547 // Note: moved here from class CrossRef as the latter class has dependencies
1548 // not availability to the applet's classpath
1549 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1553 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1554 * that sequence name is structured as Source|AccessionId.
1560 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1562 if (seq1 == null || seq2 == null)
1566 String name = seq2.getName();
1567 final DBRefEntry[] xrefs = seq1.getDBRefs();
1570 for (DBRefEntry xref : xrefs)
1572 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1573 // case-insensitive test, consistent with DBRefEntry.equalRef()
1574 if (xrefName.equalsIgnoreCase(name))
1584 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1585 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1586 * added to the original alignment's dataset, which is shared by the new
1587 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1588 * added to the alignment dataset.
1591 * aligned nucleotide (dna or cds) sequences
1593 * the alignment dataset the sequences belong to
1595 * (optional) to restrict results to CDS that map to specified
1597 * @return an alignment whose sequences are the cds-only parts of the dna
1598 * sequences (or null if no mappings are found)
1600 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1601 AlignmentI dataset, SequenceI[] products)
1603 if (dataset == null || dataset.getDataset() != null)
1605 throw new IllegalArgumentException(
1606 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1608 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
1609 List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
1610 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1611 HashSet<SequenceI> productSeqs = null;
1612 if (products != null)
1614 productSeqs = new HashSet<SequenceI>();
1615 for (SequenceI seq : products)
1617 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1618 .getDatasetSequence());
1623 * Construct CDS sequences from mappings on the alignment dataset.
1625 * - find the protein product(s) mapped to from each dna sequence
1626 * - if the mapping covers the whole dna sequence (give or take start/stop
1627 * codon), take the dna as the CDS sequence
1628 * - else search dataset mappings for a suitable dna sequence, i.e. one
1629 * whose whole sequence is mapped to the protein
1630 * - if no sequence found, construct one from the dna sequence and mapping
1631 * (and add it to dataset so it is found if this is repeated)
1633 for (SequenceI dnaSeq : dna)
1635 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1636 : dnaSeq.getDatasetSequence();
1638 List<AlignedCodonFrame> seqMappings = MappingUtils
1639 .findMappingsForSequence(dnaSeq, mappings);
1640 for (AlignedCodonFrame mapping : seqMappings)
1642 List<Mapping> mappingsFromSequence = mapping
1643 .getMappingsFromSequence(dnaSeq);
1645 for (Mapping aMapping : mappingsFromSequence)
1647 MapList mapList = aMapping.getMap();
1648 if (mapList.getFromRatio() == 1)
1651 * not a dna-to-protein mapping (likely dna-to-cds)
1657 * skip if mapping is not to one of the target set of proteins
1659 SequenceI proteinProduct = aMapping.getTo();
1660 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1666 * try to locate the CDS from the dataset mappings;
1667 * guard against duplicate results (for the case that protein has
1668 * dbrefs to both dna and cds sequences)
1670 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1671 seqMappings, aMapping);
1674 if (!foundSeqs.contains(cdsSeq))
1676 foundSeqs.add(cdsSeq);
1677 SequenceI derivedSequence = cdsSeq.deriveSequence();
1678 cdsSeqs.add(derivedSequence);
1679 if (!dataset.getSequences().contains(cdsSeq))
1681 dataset.addSequence(cdsSeq);
1688 * didn't find mapped CDS sequence - construct it and add
1689 * its dataset sequence to the dataset
1691 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1692 dataset).deriveSequence();
1693 // cdsSeq has a name constructed as CDS|<dbref>
1694 // <dbref> will be either the accession for the coding sequence,
1695 // marked in the /via/ dbref to the protein product accession
1696 // or it will be the original nucleotide accession.
1697 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1699 cdsSeqs.add(cdsSeq);
1701 if (!dataset.getSequences().contains(cdsSeqDss))
1703 // check if this sequence is a newly created one
1704 // so needs adding to the dataset
1705 dataset.addSequence(cdsSeqDss);
1709 * add a mapping from CDS to the (unchanged) mapped to range
1711 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1712 cdsSeq.getLength() });
1713 MapList cdsToProteinMap = new MapList(cdsRange, mapList.getToRanges(),
1714 mapList.getFromRatio(), mapList.getToRatio());
1715 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1716 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1720 * guard against duplicating the mapping if repeating this action
1722 if (!mappings.contains(cdsToProteinMapping))
1724 mappings.add(cdsToProteinMapping);
1727 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1728 proteinProduct, aMapping);
1730 * add another mapping from original 'from' range to CDS
1732 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1733 MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1735 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1737 if (!mappings.contains(dnaToCdsMapping))
1739 mappings.add(dnaToCdsMapping);
1743 * add DBRef with mapping from protein to CDS
1744 * (this enables Get Cross-References from protein alignment)
1745 * This is tricky because we can't have two DBRefs with the
1746 * same source and accession, so need a different accession for
1747 * the CDS from the dna sequence
1750 // specific use case:
1751 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1752 // ENSG02, ENSG03, with transcripts and products similarly named.
1753 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1755 // JBPNote: ?? can't actually create an example that demonstrates we
1757 // synthesize an xref.
1759 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1761 // creates a complementary cross-reference to the source sequence's
1762 // primary reference.
1764 DBRefEntry cdsCrossRef = new DBRefEntry(primRef.getSource(),
1765 primRef.getSource() + ":" + primRef.getVersion(),
1766 primRef.getAccessionId());
1768 .setMap(new Mapping(dnaDss, new MapList(dnaToCdsMap)));
1769 cdsSeqDss.addDBRef(cdsCrossRef);
1771 // problem here is that the cross-reference is synthesized -
1772 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1774 // assuming cds version same as dna ?!?
1776 DBRefEntry proteinToCdsRef = new DBRefEntry(
1777 primRef.getSource(), primRef.getVersion(),
1780 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1782 proteinProduct.addDBRef(proteinToCdsRef);
1786 * transfer any features on dna that overlap the CDS
1788 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1789 SequenceOntologyI.CDS);
1794 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1796 cds.setDataset(dataset);
1802 * A helper method that finds a CDS sequence in the alignment dataset that is
1803 * mapped to the given protein sequence, and either is, or has a mapping from,
1804 * the given dna sequence.
1807 * set of all mappings on the dataset
1809 * a dna (or cds) sequence we are searching from
1810 * @param seqMappings
1811 * the set of mappings involving dnaSeq
1813 * an initial candidate from seqMappings
1816 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1817 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1821 * TODO a better dna-cds-protein mapping data representation to allow easy
1822 * navigation; until then this clunky looping around lists of mappings
1824 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1825 : dnaSeq.getDatasetSequence();
1826 SequenceI proteinProduct = aMapping.getTo();
1829 * is this mapping from the whole dna sequence (i.e. CDS)?
1830 * allowing for possible stop codon on dna but not peptide
1832 int mappedFromLength = MappingUtils.getLength(aMapping.getMap()
1834 int dnaLength = seqDss.getLength();
1835 if (mappedFromLength == dnaLength || mappedFromLength == dnaLength - CODON_LENGTH)
1841 * looks like we found the dna-to-protein mapping; search for the
1842 * corresponding cds-to-protein mapping
1844 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1845 .findMappingsForSequence(proteinProduct, mappings);
1846 for (AlignedCodonFrame acf : mappingsToPeptide)
1848 for (SequenceToSequenceMapping map : acf.getMappings())
1850 Mapping mapping = map.getMapping();
1851 if (mapping != aMapping && mapping.getMap().getFromRatio() == CODON_LENGTH
1852 && proteinProduct == mapping.getTo()
1853 && seqDss != map.getFromSeq())
1855 mappedFromLength = MappingUtils.getLength(mapping.getMap()
1857 if (mappedFromLength == map.getFromSeq().getLength())
1860 * found a 3:1 mapping to the protein product which covers
1861 * the whole dna sequence i.e. is from CDS; finally check it
1862 * is from the dna start sequence
1864 SequenceI cdsSeq = map.getFromSeq();
1865 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1866 .findMappingsForSequence(cdsSeq, seqMappings);
1867 if (!dnaToCdsMaps.isEmpty())
1879 * Helper method that makes a CDS sequence as defined by the mappings from the
1880 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1881 * forward or reverse strand).
1886 * - existing dataset. We check for sequences that look like the CDS
1887 * we are about to construct, if one exists already, then we will
1888 * just return that one.
1889 * @return CDS sequence (as a dataset sequence)
1891 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1894 char[] seqChars = seq.getSequence();
1895 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1896 int cdsWidth = MappingUtils.getLength(fromRanges);
1897 char[] newSeqChars = new char[cdsWidth];
1900 for (int[] range : fromRanges)
1902 if (range[0] <= range[1])
1904 // forward strand mapping - just copy the range
1905 int length = range[1] - range[0] + 1;
1906 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1912 // reverse strand mapping - copy and complement one by one
1913 for (int i = range[0]; i >= range[1]; i--)
1915 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1921 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
1922 * else generate a sequence name
1924 String mapFromId = mapping.getMappedFromId();
1925 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
1926 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
1927 if (dataset != null)
1929 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
1930 if (matches != null)
1932 boolean matched = false;
1933 for (SequenceI mtch : matches)
1935 if (mtch.getStart() != newSeq.getStart())
1939 if (mtch.getEnd() != newSeq.getEnd())
1943 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
1955 .println("JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
1961 // newSeq.setDescription(mapFromId);
1967 * add any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
1968 * the given mapping.
1973 * @return list of DBRefEntrys added.
1975 public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
1976 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
1979 // gather direct refs from contig congrent with mapping
1980 List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
1981 HashSet<String> directSources = new HashSet<String>();
1982 if (contig.getDBRefs() != null)
1984 for (DBRefEntry dbr : contig.getDBRefs())
1986 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
1988 MapList map = dbr.getMap().getMap();
1989 // check if map is the CDS mapping
1990 if (mapping.getMap().equals(map))
1993 directSources.add(dbr.getSource());
1998 DBRefEntry[] onSource = DBRefUtils.selectRefs(
1999 proteinProduct.getDBRefs(),
2000 directSources.toArray(new String[0]));
2001 List<DBRefEntry> propagated = new ArrayList<DBRefEntry>();
2003 // and generate appropriate mappings
2004 for (DBRefEntry cdsref : direct)
2006 // clone maplist and mapping
2007 MapList cdsposmap = new MapList(Arrays.asList(new int[][] { new int[]
2008 { cdsSeq.getStart(), cdsSeq.getEnd() } }), cdsref.getMap().getMap()
2009 .getToRanges(), 3, 1);
2010 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(), cdsref.getMap()
2014 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2015 cdsref.getVersion(), cdsref.getAccessionId(), new Mapping(
2016 cdsmap.getTo(), cdsposmap));
2018 // and see if we can map to the protein product for this mapping.
2019 // onSource is the filtered set of accessions on protein that we are
2020 // tranferring, so we assume accession is the same.
2021 if (cdsmap.getTo() == null && onSource != null)
2023 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2024 cdsref.getAccessionId());
2025 if (sourceRefs != null)
2027 for (DBRefEntry srcref : sourceRefs)
2029 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2031 // we have found a complementary dbref on the protein product, so
2032 // update mapping's getTo
2033 newref.getMap().setTo(proteinProduct);
2038 cdsSeq.addDBRef(newref);
2039 propagated.add(newref);
2045 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2046 * feature start/end ranges, optionally omitting specified feature types.
2047 * Returns the number of features copied.
2052 * if not null, only features of this type are copied (including
2053 * subtypes in the Sequence Ontology)
2055 * the mapping from 'fromSeq' to 'toSeq'
2058 public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2059 MapList mapping, String select, String... omitting)
2061 SequenceI copyTo = toSeq;
2062 while (copyTo.getDatasetSequence() != null)
2064 copyTo = copyTo.getDatasetSequence();
2067 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2069 SequenceFeature[] sfs = fromSeq.getSequenceFeatures();
2072 for (SequenceFeature sf : sfs)
2074 String type = sf.getType();
2075 if (select != null && !so.isA(type, select))
2079 boolean omit = false;
2080 for (String toOmit : omitting)
2082 if (type.equals(toOmit))
2093 * locate the mapped range - null if either start or end is
2094 * not mapped (no partial overlaps are calculated)
2096 int start = sf.getBegin();
2097 int end = sf.getEnd();
2098 int[] mappedTo = mapping.locateInTo(start, end);
2100 * if whole exon range doesn't map, try interpreting it
2101 * as 5' or 3' exon overlapping the CDS range
2103 if (mappedTo == null)
2105 mappedTo = mapping.locateInTo(end, end);
2106 if (mappedTo != null)
2109 * end of exon is in CDS range - 5' overlap
2110 * to a range from the start of the peptide
2115 if (mappedTo == null)
2117 mappedTo = mapping.locateInTo(start, start);
2118 if (mappedTo != null)
2121 * start of exon is in CDS range - 3' overlap
2122 * to a range up to the end of the peptide
2124 mappedTo[1] = toSeq.getLength();
2127 if (mappedTo != null)
2129 SequenceFeature copy = new SequenceFeature(sf);
2130 copy.setBegin(Math.min(mappedTo[0], mappedTo[1]));
2131 copy.setEnd(Math.max(mappedTo[0], mappedTo[1]));
2132 copyTo.addSequenceFeature(copy);
2141 * Returns a mapping from dna to protein by inspecting sequence features of
2142 * type "CDS" on the dna.
2148 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2149 SequenceI proteinSeq)
2151 List<int[]> ranges = findCdsPositions(dnaSeq);
2152 int mappedDnaLength = MappingUtils.getLength(ranges);
2154 int proteinLength = proteinSeq.getLength();
2155 int proteinStart = proteinSeq.getStart();
2156 int proteinEnd = proteinSeq.getEnd();
2159 * incomplete start codon may mean X at start of peptide
2160 * we ignore both for mapping purposes
2162 if (proteinSeq.getCharAt(0) == 'X')
2164 // todo JAL-2022 support startPhase > 0
2168 List<int[]> proteinRange = new ArrayList<int[]>();
2171 * dna length should map to protein (or protein plus stop codon)
2173 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2174 if (codesForResidues == (proteinLength + 1))
2176 // assuming extra codon is for STOP and not in peptide
2179 if (codesForResidues == proteinLength)
2181 proteinRange.add(new int[] { proteinStart, proteinEnd });
2182 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2188 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2189 * start/end positions of sequence features of type "CDS" (or a sub-type of
2190 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2191 * position order, so this method is only valid for linear CDS in the same
2192 * sense as the protein product.
2197 public static List<int[]> findCdsPositions(SequenceI dnaSeq)
2199 List<int[]> result = new ArrayList<int[]>();
2200 SequenceFeature[] sfs = dnaSeq.getSequenceFeatures();
2206 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2209 for (SequenceFeature sf : sfs)
2212 * process a CDS feature (or a sub-type of CDS)
2214 if (so.isA(sf.getType(), SequenceOntologyI.CDS))
2219 phase = Integer.parseInt(sf.getPhase());
2220 } catch (NumberFormatException e)
2225 * phase > 0 on first codon means 5' incomplete - skip to the start
2226 * of the next codon; example ENST00000496384
2228 int begin = sf.getBegin();
2229 int end = sf.getEnd();
2230 if (result.isEmpty())
2235 // shouldn't happen!
2237 .println("Error: start phase extends beyond start CDS in "
2238 + dnaSeq.getName());
2241 result.add(new int[] { begin, end });
2246 * remove 'startPhase' positions (usually 0) from the first range
2247 * so we begin at the start of a complete codon
2249 if (!result.isEmpty())
2251 // TODO JAL-2022 correctly model start phase > 0
2252 result.get(0)[0] += startPhase;
2256 * Finally sort ranges by start position. This avoids a dependency on
2257 * keeping features in order on the sequence (if they are in order anyway,
2258 * the sort will have almost no work to do). The implicit assumption is CDS
2259 * ranges are assembled in order. Other cases should not use this method,
2260 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2262 Collections.sort(result, new Comparator<int[]>()
2265 public int compare(int[] o1, int[] o2)
2267 return Integer.compare(o1[0], o2[0]);
2274 * Maps exon features from dna to protein, and computes variants in peptide
2275 * product generated by variants in dna, and adds them as sequence_variant
2276 * features on the protein sequence. Returns the number of variant features
2281 * @param dnaToProtein
2283 public static int computeProteinFeatures(SequenceI dnaSeq,
2284 SequenceI peptide, MapList dnaToProtein)
2286 while (dnaSeq.getDatasetSequence() != null)
2288 dnaSeq = dnaSeq.getDatasetSequence();
2290 while (peptide.getDatasetSequence() != null)
2292 peptide = peptide.getDatasetSequence();
2295 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2298 * compute protein variants from dna variants and codon mappings;
2299 * NB - alternatively we could retrieve this using the REST service e.g.
2300 * http://rest.ensembl.org/overlap/translation
2301 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2302 * which would be a bit slower but possibly more reliable
2306 * build a map with codon variations for each potentially varying peptide
2308 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2309 dnaSeq, dnaToProtein);
2312 * scan codon variations, compute peptide variants and add to peptide sequence
2315 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2317 int peptidePos = variant.getKey();
2318 List<DnaVariant>[] codonVariants = variant.getValue();
2319 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2323 * sort to get sequence features in start position order
2324 * - would be better to store in Sequence as a TreeSet or NCList?
2326 if (peptide.getSequenceFeatures() != null)
2328 Arrays.sort(peptide.getSequenceFeatures(),
2329 new Comparator<SequenceFeature>()
2332 public int compare(SequenceFeature o1, SequenceFeature o2)
2334 int c = Integer.compare(o1.getBegin(), o2.getBegin());
2335 return c == 0 ? Integer.compare(o1.getEnd(), o2.getEnd())
2344 * Computes non-synonymous peptide variants from codon variants and adds them
2345 * as sequence_variant features on the protein sequence (one feature per
2346 * allele variant). Selected attributes (variant id, clinical significance)
2347 * are copied over to the new features.
2350 * the protein sequence
2352 * the position to compute peptide variants for
2353 * @param codonVariants
2354 * a list of dna variants per codon position
2355 * @return the number of features added
2357 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2358 List<DnaVariant>[] codonVariants)
2360 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2362 String base1 = codonVariants[0].get(0).base;
2363 String base2 = codonVariants[1].get(0).base;
2364 String base3 = codonVariants[2].get(0).base;
2367 * variants in first codon base
2369 for (DnaVariant var : codonVariants[0])
2371 if (var.variant != null)
2373 String alleles = (String) var.variant.getValue("alleles");
2374 if (alleles != null)
2376 for (String base : alleles.split(","))
2378 String codon = base + base2 + base3;
2379 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2389 * variants in second codon base
2391 for (DnaVariant var : codonVariants[1])
2393 if (var.variant != null)
2395 String alleles = (String) var.variant.getValue("alleles");
2396 if (alleles != null)
2398 for (String base : alleles.split(","))
2400 String codon = base1 + base + base3;
2401 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2411 * variants in third codon base
2413 for (DnaVariant var : codonVariants[2])
2415 if (var.variant != null)
2417 String alleles = (String) var.variant.getValue("alleles");
2418 if (alleles != null)
2420 for (String base : alleles.split(","))
2422 String codon = base1 + base2 + base;
2423 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2436 * Helper method that adds a peptide variant feature, provided the given codon
2437 * translates to a value different to the current residue (is a non-synonymous
2438 * variant). ID and clinical_significance attributes of the dna variant (if
2439 * present) are copied to the new feature.
2446 * @return true if a feature was added, else false
2448 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2449 String residue, DnaVariant var, String codon)
2452 * get peptide translation of codon e.g. GAT -> D
2453 * note that variants which are not single alleles,
2454 * e.g. multibase variants or HGMD_MUTATION etc
2455 * are currently ignored here
2457 String trans = codon.contains("-") ? "-"
2458 : (codon.length() > CODON_LENGTH ? null : ResidueProperties
2459 .codonTranslate(codon));
2460 if (trans != null && !trans.equals(residue))
2462 String residue3Char = StringUtils
2463 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2464 String trans3Char = StringUtils
2465 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2466 String desc = "p." + residue3Char + peptidePos + trans3Char;
2467 // set score to 0f so 'graduated colour' option is offered! JAL-2060
2468 SequenceFeature sf = new SequenceFeature(
2469 SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
2470 peptidePos, 0f, var.getSource());
2471 StringBuilder attributes = new StringBuilder(32);
2472 String id = (String) var.variant.getValue(ID);
2475 if (id.startsWith(SEQUENCE_VARIANT))
2477 id = id.substring(SEQUENCE_VARIANT.length());
2479 sf.setValue(ID, id);
2480 attributes.append(ID).append("=").append(id);
2481 // TODO handle other species variants JAL-2064
2482 StringBuilder link = new StringBuilder(32);
2485 link.append(desc).append(" ").append(id)
2486 .append("|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2487 .append(URLEncoder.encode(id, "UTF-8"));
2488 sf.addLink(link.toString());
2489 } catch (UnsupportedEncodingException e)
2494 String clinSig = (String) var.variant
2495 .getValue(CLINICAL_SIGNIFICANCE);
2496 if (clinSig != null)
2498 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2499 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2502 peptide.addSequenceFeature(sf);
2503 if (attributes.length() > 0)
2505 sf.setAttributes(attributes.toString());
2513 * Builds a map whose key is position in the protein sequence, and value is a
2514 * list of the base and all variants for each corresponding codon position
2517 * @param dnaToProtein
2520 @SuppressWarnings("unchecked")
2521 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2522 SequenceI dnaSeq, MapList dnaToProtein)
2525 * map from peptide position to all variants of the codon which codes for it
2526 * LinkedHashMap ensures we keep the peptide features in sequence order
2528 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
2529 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2531 SequenceFeature[] dnaFeatures = dnaSeq.getSequenceFeatures();
2532 if (dnaFeatures == null)
2537 int dnaStart = dnaSeq.getStart();
2538 int[] lastCodon = null;
2539 int lastPeptidePostion = 0;
2542 * build a map of codon variations for peptides
2544 for (SequenceFeature sf : dnaFeatures)
2546 int dnaCol = sf.getBegin();
2547 if (dnaCol != sf.getEnd())
2549 // not handling multi-locus variant features
2552 if (so.isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
2554 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2557 // feature doesn't lie within coding region
2560 int peptidePosition = mapsTo[0];
2561 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2562 if (codonVariants == null)
2564 codonVariants = new ArrayList[CODON_LENGTH];
2565 codonVariants[0] = new ArrayList<DnaVariant>();
2566 codonVariants[1] = new ArrayList<DnaVariant>();
2567 codonVariants[2] = new ArrayList<DnaVariant>();
2568 variants.put(peptidePosition, codonVariants);
2572 * extract dna variants to a string array
2574 String alls = (String) sf.getValue("alleles");
2579 String[] alleles = alls.toUpperCase().split(",");
2581 for (String allele : alleles)
2583 alleles[i++] = allele.trim(); // lose any space characters "A, G"
2587 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2589 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2590 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2591 peptidePosition, peptidePosition));
2592 lastPeptidePostion = peptidePosition;
2596 * save nucleotide (and any variant) for each codon position
2598 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2600 String nucleotide = String.valueOf(
2601 dnaSeq.getCharAt(codon[codonPos] - dnaStart))
2603 List<DnaVariant> codonVariant = codonVariants[codonPos];
2604 if (codon[codonPos] == dnaCol)
2606 if (!codonVariant.isEmpty()
2607 && codonVariant.get(0).variant == null)
2610 * already recorded base value, add this variant
2612 codonVariant.get(0).variant = sf;
2617 * add variant with base value
2619 codonVariant.add(new DnaVariant(nucleotide, sf));
2622 else if (codonVariant.isEmpty())
2625 * record (possibly non-varying) base value
2627 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.get(seq
2814 .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 Map<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2838 AlignmentI unaligned, AlignmentI aligned, List<SequenceI> unmapped)
2841 * Map will hold, for each aligned column position, a map of
2842 * {unalignedSequence, characterPerSequence} at that position.
2843 * TreeMap keeps the entries in ascending column order.
2845 Map<Integer, Map<SequenceI, Character>> map = new TreeMap<Integer, Map<SequenceI, Character>>();
2848 * record any sequences that have no mapping so can't be realigned
2850 unmapped.addAll(unaligned.getSequences());
2852 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2854 for (SequenceI seq : unaligned.getSequences())
2856 for (AlignedCodonFrame mapping : mappings)
2858 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2859 if (fromSeq != null)
2861 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2862 if (addMappedPositions(seq, fromSeq, seqMap, map))
2864 unmapped.remove(seq);
2873 * Helper method that adds to a map the mapped column positions of a sequence. <br>
2874 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2875 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2879 * the sequence whose column positions we are recording
2881 * a sequence that is mapped to the first sequence
2883 * the mapping from 'fromSeq' to 'seq'
2885 * a map to add the column positions (in fromSeq) of the mapped
2889 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2890 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2898 * invert mapping if it is from unaligned to aligned sequence
2900 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2902 seqMap = new Mapping(seq.getDatasetSequence(), seqMap.getMap()
2906 char[] fromChars = fromSeq.getSequence();
2907 int toStart = seq.getStart();
2908 char[] toChars = seq.getSequence();
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 <= fromChars.length
2942 if (!Comparison.isGap(fromChars[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, toChars[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"))