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.api.DBRefEntryI;
26 import jalview.datamodel.AlignedCodon;
27 import jalview.datamodel.AlignedCodonFrame;
28 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
29 import jalview.datamodel.Alignment;
30 import jalview.datamodel.AlignmentAnnotation;
31 import jalview.datamodel.AlignmentI;
32 import jalview.datamodel.DBRefEntry;
33 import jalview.datamodel.IncompleteCodonException;
34 import jalview.datamodel.Mapping;
35 import jalview.datamodel.Sequence;
36 import jalview.datamodel.SequenceFeature;
37 import jalview.datamodel.SequenceGroup;
38 import jalview.datamodel.SequenceI;
39 import jalview.io.gff.SequenceOntologyFactory;
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.MapList;
45 import jalview.util.MappingUtils;
46 import jalview.util.StringUtils;
48 import java.io.UnsupportedEncodingException;
49 import java.net.URLEncoder;
50 import java.util.ArrayList;
51 import java.util.Arrays;
52 import java.util.Collection;
53 import java.util.Collections;
54 import java.util.Comparator;
55 import java.util.HashMap;
56 import java.util.HashSet;
57 import java.util.Iterator;
58 import java.util.LinkedHashMap;
59 import java.util.List;
61 import java.util.Map.Entry;
62 import java.util.NoSuchElementException;
64 import java.util.TreeMap;
67 * grab bag of useful alignment manipulation operations Expect these to be
68 * refactored elsewhere at some point.
73 public class AlignmentUtils
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 class DnaVariant
87 SequenceFeature variant;
89 DnaVariant(String nuc)
94 DnaVariant(String nuc, SequenceFeature var)
102 * given an existing alignment, create a new alignment including all, or up to
103 * flankSize additional symbols from each sequence's dataset sequence
109 public static AlignmentI expandContext(AlignmentI core, int flankSize)
111 List<SequenceI> sq = new ArrayList<SequenceI>();
113 for (SequenceI s : core.getSequences())
115 SequenceI newSeq = s.deriveSequence();
116 final int newSeqStart = newSeq.getStart() - 1;
117 if (newSeqStart > maxoffset
118 && newSeq.getDatasetSequence().getStart() < s.getStart())
120 maxoffset = newSeqStart;
126 maxoffset = Math.min(maxoffset, flankSize);
130 * now add offset left and right to create an expanded alignment
132 for (SequenceI s : sq)
135 while (ds.getDatasetSequence() != null)
137 ds = ds.getDatasetSequence();
139 int s_end = s.findPosition(s.getStart() + s.getLength());
140 // find available flanking residues for sequence
141 int ustream_ds = s.getStart() - ds.getStart();
142 int dstream_ds = ds.getEnd() - s_end;
144 // build new flanked sequence
146 // compute gap padding to start of flanking sequence
147 int offset = maxoffset - ustream_ds;
149 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
152 if (flankSize < ustream_ds)
154 // take up to flankSize residues
155 offset = maxoffset - flankSize;
156 ustream_ds = flankSize;
158 if (flankSize <= dstream_ds)
160 dstream_ds = flankSize - 1;
163 // TODO use Character.toLowerCase to avoid creating String objects?
164 char[] upstream = new String(ds.getSequence(s.getStart() - 1
165 - ustream_ds, s.getStart() - 1)).toLowerCase().toCharArray();
166 char[] downstream = new String(ds.getSequence(s_end - 1, s_end
167 + dstream_ds)).toLowerCase().toCharArray();
168 char[] coreseq = s.getSequence();
169 char[] nseq = new char[offset + upstream.length + downstream.length
171 char c = core.getGapCharacter();
174 for (; p < offset; p++)
179 System.arraycopy(upstream, 0, nseq, p, upstream.length);
180 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
182 System.arraycopy(downstream, 0, nseq, p + coreseq.length
183 + upstream.length, downstream.length);
184 s.setSequence(new String(nseq));
185 s.setStart(s.getStart() - ustream_ds);
186 s.setEnd(s_end + downstream.length);
188 AlignmentI newAl = new jalview.datamodel.Alignment(
189 sq.toArray(new SequenceI[0]));
190 for (SequenceI s : sq)
192 if (s.getAnnotation() != null)
194 for (AlignmentAnnotation aa : s.getAnnotation())
196 aa.adjustForAlignment(); // JAL-1712 fix
197 newAl.addAnnotation(aa);
201 newAl.setDataset(core.getDataset());
206 * Returns the index (zero-based position) of a sequence in an alignment, or
213 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
217 for (SequenceI alSeq : al.getSequences())
230 * Returns a map of lists of sequences in the alignment, keyed by sequence
231 * name. For use in mapping between different alignment views of the same
234 * @see jalview.datamodel.AlignmentI#getSequencesByName()
236 public static Map<String, List<SequenceI>> getSequencesByName(
239 Map<String, List<SequenceI>> theMap = new LinkedHashMap<String, List<SequenceI>>();
240 for (SequenceI seq : al.getSequences())
242 String name = seq.getName();
245 List<SequenceI> seqs = theMap.get(name);
248 seqs = new ArrayList<SequenceI>();
249 theMap.put(name, seqs);
258 * Build mapping of protein to cDNA alignment. Mappings are made between
259 * sequences where the cDNA translates to the protein sequence. Any new
260 * mappings are added to the protein alignment. Returns true if any mappings
261 * either already exist or were added, else false.
263 * @param proteinAlignment
264 * @param cdnaAlignment
267 public static boolean mapProteinAlignmentToCdna(
268 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
270 if (proteinAlignment == null || cdnaAlignment == null)
275 Set<SequenceI> mappedDna = new HashSet<SequenceI>();
276 Set<SequenceI> mappedProtein = new HashSet<SequenceI>();
279 * First pass - map sequences where cross-references exist. This include
280 * 1-to-many mappings to support, for example, variant cDNA.
282 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
283 cdnaAlignment, mappedDna, mappedProtein, true);
286 * Second pass - map sequences where no cross-references exist. This only
287 * does 1-to-1 mappings and assumes corresponding sequences are in the same
288 * order in the alignments.
290 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
291 mappedDna, mappedProtein, false);
292 return mappingPerformed;
296 * Make mappings between compatible sequences (where the cDNA translation
297 * matches the protein).
299 * @param proteinAlignment
300 * @param cdnaAlignment
302 * a set of mapped DNA sequences (to add to)
303 * @param mappedProtein
304 * a set of mapped Protein sequences (to add to)
306 * if true, only map sequences where xrefs exist
309 protected static boolean mapProteinToCdna(
310 final AlignmentI proteinAlignment,
311 final AlignmentI cdnaAlignment, Set<SequenceI> mappedDna,
312 Set<SequenceI> mappedProtein, boolean xrefsOnly)
314 boolean mappingExistsOrAdded = false;
315 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
316 for (SequenceI aaSeq : thisSeqs)
318 boolean proteinMapped = false;
319 AlignedCodonFrame acf = new AlignedCodonFrame();
321 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
324 * Always try to map if sequences have xref to each other; this supports
325 * variant cDNA or alternative splicing for a protein sequence.
327 * If no xrefs, try to map progressively, assuming that alignments have
328 * mappable sequences in corresponding order. These are not
329 * many-to-many, as that would risk mixing species with similar cDNA
332 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
338 * Don't map non-xrefd sequences more than once each. This heuristic
339 * allows us to pair up similar sequences in ordered alignments.
342 && (mappedProtein.contains(aaSeq) || mappedDna
347 if (mappingExists(proteinAlignment.getCodonFrames(),
348 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
350 mappingExistsOrAdded = true;
354 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
357 acf.addMap(cdnaSeq, aaSeq, map);
358 mappingExistsOrAdded = true;
359 proteinMapped = true;
360 mappedDna.add(cdnaSeq);
361 mappedProtein.add(aaSeq);
367 proteinAlignment.addCodonFrame(acf);
370 return mappingExistsOrAdded;
374 * Answers true if the mappings include one between the given (dataset)
377 public static boolean mappingExists(List<AlignedCodonFrame> mappings,
378 SequenceI aaSeq, SequenceI cdnaSeq)
380 if (mappings != null)
382 for (AlignedCodonFrame acf : mappings)
384 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
394 * Builds a mapping (if possible) of a cDNA to a protein sequence.
396 * <li>first checks if the cdna translates exactly to the protein sequence</li>
397 * <li>else checks for translation after removing a STOP codon</li>
398 * <li>else checks for translation after removing a START codon</li>
399 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
401 * Returns null if no mapping is determined.
404 * the aligned protein sequence
406 * the aligned cdna sequence
409 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
413 * Here we handle either dataset sequence set (desktop) or absent (applet).
414 * Use only the char[] form of the sequence to avoid creating possibly large
417 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
418 char[] aaSeqChars = proteinDataset != null ? proteinDataset
419 .getSequence() : proteinSeq.getSequence();
420 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
421 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
422 : cdnaSeq.getSequence();
423 if (aaSeqChars == null || cdnaSeqChars == null)
429 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
431 final int mappedLength = 3 * aaSeqChars.length;
432 int cdnaLength = cdnaSeqChars.length;
433 int cdnaStart = cdnaSeq.getStart();
434 int cdnaEnd = cdnaSeq.getEnd();
435 final int proteinStart = proteinSeq.getStart();
436 final int proteinEnd = proteinSeq.getEnd();
439 * If lengths don't match, try ignoring stop codon (if present)
441 if (cdnaLength != mappedLength && cdnaLength > 2)
443 String lastCodon = String.valueOf(cdnaSeqChars, cdnaLength - 3, 3)
445 for (String stop : ResidueProperties.STOP)
447 if (lastCodon.equals(stop))
457 * If lengths still don't match, try ignoring start codon.
460 if (cdnaLength != mappedLength
462 && String.valueOf(cdnaSeqChars, 0, 3).toUpperCase()
463 .equals(ResidueProperties.START))
470 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
473 * protein is translation of dna (+/- start/stop codons)
475 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd }, new int[]
476 { proteinStart, proteinEnd }, 3, 1);
481 * translation failed - try mapping CDS annotated regions of dna
483 return mapCdsToProtein(cdnaSeq, proteinSeq);
487 * Test whether the given cdna sequence, starting at the given offset,
488 * translates to the given amino acid sequence, using the standard translation
489 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
491 * @param cdnaSeqChars
496 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
499 if (cdnaSeqChars == null || aaSeqChars == null)
505 int dnaPos = cdnaStart;
506 for (; dnaPos < cdnaSeqChars.length - 2
507 && aaPos < aaSeqChars.length; dnaPos += 3, aaPos++)
509 String codon = String.valueOf(cdnaSeqChars, dnaPos, 3);
510 final String translated = ResidueProperties.codonTranslate(codon);
513 * allow * in protein to match untranslatable in dna
515 final char aaRes = aaSeqChars[aaPos];
516 if ((translated == null || "STOP".equals(translated)) && aaRes == '*')
520 if (translated == null || !(aaRes == translated.charAt(0)))
523 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
524 // + codon + "(" + translated + ") != " + aaRes));
530 * check we matched all of the protein sequence
532 if (aaPos != aaSeqChars.length)
538 * check we matched all of the dna except
539 * for optional trailing STOP codon
541 if (dnaPos == cdnaSeqChars.length)
545 if (dnaPos == cdnaSeqChars.length - 3)
547 String codon = String.valueOf(cdnaSeqChars, dnaPos, 3);
548 if ("STOP".equals(ResidueProperties.codonTranslate(codon)))
557 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
558 * currently assumes that we are aligning cDNA to match protein.
561 * the sequence to be realigned
563 * the alignment whose sequence alignment is to be 'copied'
565 * character string represent a gap in the realigned sequence
566 * @param preserveUnmappedGaps
567 * @param preserveMappedGaps
568 * @return true if the sequence was realigned, false if it could not be
570 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
571 String gap, boolean preserveMappedGaps,
572 boolean preserveUnmappedGaps)
575 * Get any mappings from the source alignment to the target (dataset)
578 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
579 // all mappings. Would it help to constrain this?
580 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
581 if (mappings == null || mappings.isEmpty())
587 * Locate the aligned source sequence whose dataset sequence is mapped. We
588 * just take the first match here (as we can't align like more than one
591 SequenceI alignFrom = null;
592 AlignedCodonFrame mapping = null;
593 for (AlignedCodonFrame mp : mappings)
595 alignFrom = mp.findAlignedSequence(seq, al);
596 if (alignFrom != null)
603 if (alignFrom == null)
607 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
608 preserveMappedGaps, preserveUnmappedGaps);
613 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
614 * match residues and codons. Flags control whether existing gaps in unmapped
615 * (intron) and mapped (exon) regions are preserved or not. Gaps between
616 * intron and exon are only retained if both flags are set.
623 * @param preserveUnmappedGaps
624 * @param preserveMappedGaps
626 public static void alignSequenceAs(SequenceI alignTo,
627 SequenceI alignFrom, AlignedCodonFrame mapping, String myGap,
628 char sourceGap, boolean preserveMappedGaps,
629 boolean preserveUnmappedGaps)
631 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
633 // aligned and dataset sequence positions, all base zero
637 int basesWritten = 0;
638 char myGapChar = myGap.charAt(0);
639 int ratio = myGap.length();
641 int fromOffset = alignFrom.getStart() - 1;
642 int toOffset = alignTo.getStart() - 1;
643 int sourceGapMappedLength = 0;
644 boolean inExon = false;
645 final char[] thisSeq = alignTo.getSequence();
646 final char[] thatAligned = alignFrom.getSequence();
647 StringBuilder thisAligned = new StringBuilder(2 * thisSeq.length);
650 * Traverse the 'model' aligned sequence
652 for (char sourceChar : thatAligned)
654 if (sourceChar == sourceGap)
656 sourceGapMappedLength += ratio;
661 * Found a non-gap character. Locate its mapped region if any.
664 // Note mapping positions are base 1, our sequence positions base 0
665 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
666 sourceDsPos + fromOffset);
667 if (mappedPos == null)
670 * unmapped position; treat like a gap
672 sourceGapMappedLength += ratio;
673 // System.err.println("Can't align: no codon mapping to residue "
674 // + sourceDsPos + "(" + sourceChar + ")");
679 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
680 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
681 StringBuilder trailingCopiedGap = new StringBuilder();
684 * Copy dna sequence up to and including this codon. Optionally, include
685 * gaps before the codon starts (in introns) and/or after the codon starts
688 * Note this only works for 'linear' splicing, not reverse or interleaved.
689 * But then 'align dna as protein' doesn't make much sense otherwise.
691 int intronLength = 0;
692 while (basesWritten + toOffset < mappedCodonEnd
693 && thisSeqPos < thisSeq.length)
695 final char c = thisSeq[thisSeqPos++];
699 int sourcePosition = basesWritten + toOffset;
700 if (sourcePosition < mappedCodonStart)
703 * Found an unmapped (intron) base. First add in any preceding gaps
706 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
708 thisAligned.append(trailingCopiedGap.toString());
709 intronLength += trailingCopiedGap.length();
710 trailingCopiedGap = new StringBuilder();
717 final boolean startOfCodon = sourcePosition == mappedCodonStart;
718 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
719 preserveUnmappedGaps, sourceGapMappedLength, inExon,
720 trailingCopiedGap.length(), intronLength, startOfCodon);
721 for (int i = 0; i < gapsToAdd; i++)
723 thisAligned.append(myGapChar);
725 sourceGapMappedLength = 0;
728 thisAligned.append(c);
729 trailingCopiedGap = new StringBuilder();
733 if (inExon && preserveMappedGaps)
735 trailingCopiedGap.append(myGapChar);
737 else if (!inExon && preserveUnmappedGaps)
739 trailingCopiedGap.append(myGapChar);
746 * At end of model aligned sequence. Copy any remaining target sequence, optionally
747 * including (intron) gaps.
749 while (thisSeqPos < thisSeq.length)
751 final char c = thisSeq[thisSeqPos++];
752 if (c != myGapChar || preserveUnmappedGaps)
754 thisAligned.append(c);
756 sourceGapMappedLength--;
760 * finally add gaps to pad for any trailing source gaps or
761 * unmapped characters
763 if (preserveUnmappedGaps)
765 while (sourceGapMappedLength > 0)
767 thisAligned.append(myGapChar);
768 sourceGapMappedLength--;
773 * All done aligning, set the aligned sequence.
775 alignTo.setSequence(new String(thisAligned));
779 * Helper method to work out how many gaps to insert when realigning.
781 * @param preserveMappedGaps
782 * @param preserveUnmappedGaps
783 * @param sourceGapMappedLength
785 * @param trailingCopiedGap
786 * @param intronLength
787 * @param startOfCodon
790 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
791 boolean preserveUnmappedGaps, int sourceGapMappedLength,
792 boolean inExon, int trailingGapLength, int intronLength,
793 final boolean startOfCodon)
799 * Reached start of codon. Ignore trailing gaps in intron unless we are
800 * preserving gaps in both exon and intron. Ignore them anyway if the
801 * protein alignment introduces a gap at least as large as the intronic
804 if (inExon && !preserveMappedGaps)
806 trailingGapLength = 0;
808 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
810 trailingGapLength = 0;
814 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
818 if (intronLength + trailingGapLength <= sourceGapMappedLength)
820 gapsToAdd = sourceGapMappedLength - intronLength;
824 gapsToAdd = Math.min(intronLength + trailingGapLength
825 - sourceGapMappedLength, trailingGapLength);
832 * second or third base of codon; check for any gaps in dna
834 if (!preserveMappedGaps)
836 trailingGapLength = 0;
838 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
844 * Realigns the given protein to match the alignment of the dna, using codon
845 * mappings to translate aligned codon positions to protein residues.
848 * the alignment whose sequences are realigned by this method
850 * the dna alignment whose alignment we are 'copying'
851 * @return the number of sequences that were realigned
853 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
855 if (protein.isNucleotide() || !dna.isNucleotide())
857 System.err.println("Wrong alignment type in alignProteinAsDna");
860 List<SequenceI> unmappedProtein = new ArrayList<SequenceI>();
861 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
862 protein, dna, unmappedProtein);
863 return alignProteinAs(protein, alignedCodons, unmappedProtein);
867 * Realigns the given dna to match the alignment of the protein, using codon
868 * mappings to translate aligned peptide positions to codons.
870 * Always produces a padded CDS alignment.
873 * the alignment whose sequences are realigned by this method
875 * the protein alignment whose alignment we are 'copying'
876 * @return the number of sequences that were realigned
878 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
880 if (protein.isNucleotide() || !dna.isNucleotide())
882 System.err.println("Wrong alignment type in alignProteinAsDna");
885 // todo: implement this
886 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
887 int alignedCount = 0;
888 int width = 0; // alignment width for padding CDS
889 for (SequenceI dnaSeq : dna.getSequences())
891 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
892 dna.getGapCharacter()))
896 width = Math.max(dnaSeq.getLength(), width);
899 for (SequenceI dnaSeq : dna.getSequences())
901 oldwidth = dnaSeq.getLength();
902 diff = width - oldwidth;
905 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
912 * Helper method to align (if possible) the dna sequence to match the
913 * alignment of a mapped protein sequence. This is currently limited to
914 * handling coding sequence only.
922 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
923 AlignmentI protein, List<AlignedCodonFrame> mappings, char gapChar)
925 SequenceI cdsDss = cdsSeq.getDatasetSequence();
929 .println("alignCdsSequenceAsProtein needs aligned sequence!");
933 List<AlignedCodonFrame> dnaMappings = MappingUtils
934 .findMappingsForSequence(cdsSeq, mappings);
935 for (AlignedCodonFrame mapping : dnaMappings)
937 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
938 int peptideLength = peptide.getLength();
941 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
944 MapList mapList = map.getMap();
945 if (map.getTo() == peptide.getDatasetSequence())
947 mapList = mapList.getInverse();
949 int cdsLength = cdsDss.getLength();
950 int mappedFromLength = MappingUtils.getLength(mapList
952 int mappedToLength = MappingUtils
953 .getLength(mapList.getToRanges());
954 boolean addStopCodon = (cdsLength == mappedFromLength * 3 + 3)
955 || (peptide.getDatasetSequence().getLength() == mappedFromLength - 1);
956 if (cdsLength != mappedToLength && !addStopCodon)
960 .format("Can't align cds as protein (length mismatch %d/%d): %s",
961 cdsLength, mappedToLength,
966 * pre-fill the aligned cds sequence with gaps
968 char[] alignedCds = new char[peptideLength * 3
969 + (addStopCodon ? 3 : 0)];
970 Arrays.fill(alignedCds, gapChar);
973 * walk over the aligned peptide sequence and insert mapped
974 * codons for residues in the aligned cds sequence
976 char[] alignedPeptide = peptide.getSequence();
977 char[] nucleotides = cdsDss.getSequence();
979 int cdsStart = cdsDss.getStart();
980 int proteinPos = peptide.getStart() - 1;
982 for (char residue : alignedPeptide)
984 if (Comparison.isGap(residue))
991 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
994 // e.g. incomplete start codon, X in peptide
999 for (int j = codon[0]; j <= codon[1]; j++)
1001 char mappedBase = nucleotides[j - cdsStart];
1002 alignedCds[cdsCol++] = mappedBase;
1010 * append stop codon if not mapped from protein,
1011 * closing it up to the end of the mapped sequence
1013 if (copiedBases == nucleotides.length - 3)
1015 for (int i = alignedCds.length - 1; i >= 0; i--)
1017 if (!Comparison.isGap(alignedCds[i]))
1019 cdsCol = i + 1; // gap just after end of sequence
1023 for (int i = nucleotides.length - 3; i < nucleotides.length; i++)
1025 alignedCds[cdsCol++] = nucleotides[i];
1028 cdsSeq.setSequence(new String(alignedCds));
1037 * Builds a map whose key is an aligned codon position (3 alignment column
1038 * numbers base 0), and whose value is a map from protein sequence to each
1039 * protein's peptide residue for that codon. The map generates an ordering of
1040 * the codons, and allows us to read off the peptides at each position in
1041 * order to assemble 'aligned' protein sequences.
1044 * the protein alignment
1046 * the coding dna alignment
1047 * @param unmappedProtein
1048 * any unmapped proteins are added to this list
1051 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1052 AlignmentI protein, AlignmentI dna,
1053 List<SequenceI> unmappedProtein)
1056 * maintain a list of any proteins with no mappings - these will be
1057 * rendered 'as is' in the protein alignment as we can't align them
1059 unmappedProtein.addAll(protein.getSequences());
1061 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1064 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1065 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1066 * comparator keeps the codon positions ordered.
1068 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<AlignedCodon, Map<SequenceI, AlignedCodon>>(
1069 new CodonComparator());
1071 for (SequenceI dnaSeq : dna.getSequences())
1073 for (AlignedCodonFrame mapping : mappings)
1075 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1078 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1079 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(),
1080 seqMap, alignedCodons);
1081 unmappedProtein.remove(prot);
1087 * Finally add any unmapped peptide start residues (e.g. for incomplete
1088 * codons) as if at the codon position before the second residue
1090 // TODO resolve JAL-2022 so this fudge can be removed
1091 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1092 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1094 return alignedCodons;
1098 * Scans for any protein mapped from position 2 (meaning unmapped start
1099 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1100 * preceding position in the alignment
1102 * @param alignedCodons
1103 * the codon-to-peptide map
1104 * @param mappedSequenceCount
1105 * the number of distinct sequences in the map
1107 protected static void addUnmappedPeptideStarts(
1108 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1109 int mappedSequenceCount)
1111 // TODO delete this ugly hack once JAL-2022 is resolved
1112 // i.e. we can model startPhase > 0 (incomplete start codon)
1114 List<SequenceI> sequencesChecked = new ArrayList<SequenceI>();
1115 AlignedCodon lastCodon = null;
1116 Map<SequenceI, AlignedCodon> toAdd = new HashMap<SequenceI, AlignedCodon>();
1118 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1121 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1124 SequenceI seq = sequenceCodon.getKey();
1125 if (sequencesChecked.contains(seq))
1129 sequencesChecked.add(seq);
1130 AlignedCodon codon = sequenceCodon.getValue();
1131 if (codon.peptideCol > 1)
1134 .println("Problem mapping protein with >1 unmapped start positions: "
1137 else if (codon.peptideCol == 1)
1140 * first position (peptideCol == 0) was unmapped - add it
1142 if (lastCodon != null)
1144 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1145 lastCodon.pos2, lastCodon.pos3, String.valueOf(seq
1147 toAdd.put(seq, firstPeptide);
1152 * unmapped residue at start of alignment (no prior column) -
1153 * 'insert' at nominal codon [0, 0, 0]
1155 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1156 String.valueOf(seq.getCharAt(0)), 0);
1157 toAdd.put(seq, firstPeptide);
1160 if (sequencesChecked.size() == mappedSequenceCount)
1162 // no need to check past first mapped position in all sequences
1166 lastCodon = entry.getKey();
1170 * add any new codons safely after iterating over the map
1172 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1174 addCodonToMap(alignedCodons, startCodon.getValue(),
1175 startCodon.getKey());
1180 * Update the aligned protein sequences to match the codon alignments given in
1184 * @param alignedCodons
1185 * an ordered map of codon positions (columns), with sequence/peptide
1186 * values present in each column
1187 * @param unmappedProtein
1190 protected static int alignProteinAs(AlignmentI protein,
1191 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1192 List<SequenceI> unmappedProtein)
1195 * Prefill aligned sequences with gaps before inserting aligned protein
1198 int alignedWidth = alignedCodons.size();
1199 char[] gaps = new char[alignedWidth];
1200 Arrays.fill(gaps, protein.getGapCharacter());
1201 String allGaps = String.valueOf(gaps);
1202 for (SequenceI seq : protein.getSequences())
1204 if (!unmappedProtein.contains(seq))
1206 seq.setSequence(allGaps);
1211 for (AlignedCodon codon : alignedCodons.keySet())
1213 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1215 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1217 // place translated codon at its column position in sequence
1218 entry.getKey().getSequence()[column] = entry.getValue().product
1227 * Populate the map of aligned codons by traversing the given sequence
1228 * mapping, locating the aligned positions of mapped codons, and adding those
1229 * positions and their translation products to the map.
1232 * the aligned sequence we are mapping from
1234 * the sequence to be aligned to the codons
1236 * the gap character in the dna sequence
1238 * a mapping to a sequence translation
1239 * @param alignedCodons
1240 * the map we are building up
1242 static void addCodonPositions(SequenceI dna, SequenceI protein,
1243 char gapChar, Mapping seqMap,
1244 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1246 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1249 * add codon positions, and their peptide translations, to the alignment
1250 * map, while remembering the first codon mapped
1252 while (codons.hasNext())
1256 AlignedCodon codon = codons.next();
1257 addCodonToMap(alignedCodons, codon, protein);
1258 } catch (IncompleteCodonException e)
1260 // possible incomplete trailing codon - ignore
1261 } catch (NoSuchElementException e)
1263 // possibly peptide lacking STOP
1269 * Helper method to add a codon-to-peptide entry to the aligned codons map
1271 * @param alignedCodons
1275 protected static void addCodonToMap(
1276 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1277 AlignedCodon codon, SequenceI protein)
1279 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1280 if (seqProduct == null)
1282 seqProduct = new HashMap<SequenceI, AlignedCodon>();
1283 alignedCodons.put(codon, seqProduct);
1285 seqProduct.put(protein, codon);
1289 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1290 * between at least one pair of sequences in the two alignments. Currently,
1293 * <li>One alignment must be nucleotide, and the other protein</li>
1294 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1295 * <li>Mappable means the nucleotide translation matches the protein sequence</li>
1296 * <li>The translation may ignore start and stop codons if present in the
1304 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1306 if (al1 == null || al2 == null)
1312 * Require one nucleotide and one protein
1314 if (al1.isNucleotide() == al2.isNucleotide())
1318 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1319 AlignmentI protein = dna == al1 ? al2 : al1;
1320 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1321 for (SequenceI dnaSeq : dna.getSequences())
1323 for (SequenceI proteinSeq : protein.getSequences())
1325 if (isMappable(dnaSeq, proteinSeq, mappings))
1335 * Returns true if the dna sequence is mapped, or could be mapped, to the
1343 protected static boolean isMappable(SequenceI dnaSeq,
1344 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1346 if (dnaSeq == null || proteinSeq == null)
1351 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq : dnaSeq
1352 .getDatasetSequence();
1353 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null ? proteinSeq
1354 : proteinSeq.getDatasetSequence();
1356 for (AlignedCodonFrame mapping : mappings)
1358 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1368 * Just try to make a mapping (it is not yet stored), test whether
1371 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1375 * Finds any reference annotations associated with the sequences in
1376 * sequenceScope, that are not already added to the alignment, and adds them
1377 * to the 'candidates' map. Also populates a lookup table of annotation
1378 * labels, keyed by calcId, for use in constructing tooltips or the like.
1380 * @param sequenceScope
1381 * the sequences to scan for reference annotations
1382 * @param labelForCalcId
1383 * (optional) map to populate with label for calcId
1385 * map to populate with annotations for sequence
1387 * the alignment to check for presence of annotations
1389 public static void findAddableReferenceAnnotations(
1390 List<SequenceI> sequenceScope,
1391 Map<String, String> labelForCalcId,
1392 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1395 if (sequenceScope == null)
1401 * For each sequence in scope, make a list of any annotations on the
1402 * underlying dataset sequence which are not already on the alignment.
1404 * Add to a map of { alignmentSequence, <List of annotations to add> }
1406 for (SequenceI seq : sequenceScope)
1408 SequenceI dataset = seq.getDatasetSequence();
1409 if (dataset == null)
1413 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1414 if (datasetAnnotations == null)
1418 final List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
1419 for (AlignmentAnnotation dsann : datasetAnnotations)
1422 * Find matching annotations on the alignment. If none is found, then
1423 * add this annotation to the list of 'addable' annotations for this
1426 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1427 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1428 if (!matchedAlignmentAnnotations.iterator().hasNext())
1431 if (labelForCalcId != null)
1433 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1438 * Save any addable annotations for this sequence
1440 if (!result.isEmpty())
1442 candidates.put(seq, result);
1448 * Adds annotations to the top of the alignment annotations, in the same order
1449 * as their related sequences.
1451 * @param annotations
1452 * the annotations to add
1454 * the alignment to add them to
1455 * @param selectionGroup
1456 * current selection group (or null if none)
1458 public static void addReferenceAnnotations(
1459 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1460 final AlignmentI alignment, final SequenceGroup selectionGroup)
1462 for (SequenceI seq : annotations.keySet())
1464 for (AlignmentAnnotation ann : annotations.get(seq))
1466 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1468 int endRes = ann.annotations.length;
1469 if (selectionGroup != null)
1471 startRes = selectionGroup.getStartRes();
1472 endRes = selectionGroup.getEndRes();
1474 copyAnn.restrict(startRes, endRes);
1477 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1478 * original annotation is already on the sequence.
1480 if (!seq.hasAnnotation(ann))
1482 seq.addAlignmentAnnotation(copyAnn);
1485 copyAnn.adjustForAlignment();
1486 // add to the alignment and set visible
1487 alignment.addAnnotation(copyAnn);
1488 copyAnn.visible = true;
1494 * Set visibility of alignment annotations of specified types (labels), for
1495 * specified sequences. This supports controls like
1496 * "Show all secondary structure", "Hide all Temp factor", etc.
1498 * @al the alignment to scan for annotations
1500 * the types (labels) of annotations to be updated
1501 * @param forSequences
1502 * if not null, only annotations linked to one of these sequences are
1503 * in scope for update; if null, acts on all sequence annotations
1505 * if this flag is true, 'types' is ignored (label not checked)
1507 * if true, set visibility on, else set off
1509 public static void showOrHideSequenceAnnotations(AlignmentI al,
1510 Collection<String> types, List<SequenceI> forSequences,
1511 boolean anyType, boolean doShow)
1513 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1516 for (AlignmentAnnotation aa : anns)
1518 if (anyType || types.contains(aa.label))
1520 if ((aa.sequenceRef != null)
1521 && (forSequences == null || forSequences
1522 .contains(aa.sequenceRef)))
1524 aa.visible = doShow;
1532 * Returns true if either sequence has a cross-reference to the other
1538 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1540 // Note: moved here from class CrossRef as the latter class has dependencies
1541 // not availability to the applet's classpath
1542 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1546 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1547 * that sequence name is structured as Source|AccessionId.
1553 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1555 if (seq1 == null || seq2 == null)
1559 String name = seq2.getName();
1560 final DBRefEntry[] xrefs = seq1.getDBRefs();
1563 for (DBRefEntry xref : xrefs)
1565 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1566 // case-insensitive test, consistent with DBRefEntry.equalRef()
1567 if (xrefName.equalsIgnoreCase(name))
1577 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1578 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1579 * added to the original alignment's dataset, which is shared by the new
1580 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1581 * added to the alignment dataset.
1584 * aligned nucleotide (dna or cds) sequences
1586 * the alignment dataset the sequences belong to
1588 * (optional) to restrict results to CDS that map to specified
1590 * @return an alignment whose sequences are the cds-only parts of the dna
1591 * sequences (or null if no mappings are found)
1593 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1594 AlignmentI dataset, SequenceI[] products)
1596 if (dataset == null || dataset.getDataset() != null)
1598 throw new IllegalArgumentException(
1599 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1601 List<SequenceI> foundSeqs = new ArrayList<SequenceI>();
1602 List<SequenceI> cdsSeqs = new ArrayList<SequenceI>();
1603 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1604 HashSet<SequenceI> productSeqs = null;
1605 if (products != null)
1607 productSeqs = new HashSet<SequenceI>();
1608 for (SequenceI seq : products)
1610 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1611 .getDatasetSequence());
1616 * Construct CDS sequences from mappings on the alignment dataset.
1618 * - find the protein product(s) mapped to from each dna sequence
1619 * - if the mapping covers the whole dna sequence (give or take start/stop
1620 * codon), take the dna as the CDS sequence
1621 * - else search dataset mappings for a suitable dna sequence, i.e. one
1622 * whose whole sequence is mapped to the protein
1623 * - if no sequence found, construct one from the dna sequence and mapping
1624 * (and add it to dataset so it is found if this is repeated)
1626 for (SequenceI dnaSeq : dna)
1628 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1629 : dnaSeq.getDatasetSequence();
1631 List<AlignedCodonFrame> seqMappings = MappingUtils
1632 .findMappingsForSequence(dnaSeq, mappings);
1633 for (AlignedCodonFrame mapping : seqMappings)
1635 List<Mapping> mappingsFromSequence = mapping
1636 .getMappingsFromSequence(dnaSeq);
1638 for (Mapping aMapping : mappingsFromSequence)
1640 MapList mapList = aMapping.getMap();
1641 if (mapList.getFromRatio() == 1)
1644 * not a dna-to-protein mapping (likely dna-to-cds)
1650 * skip if mapping is not to one of the target set of proteins
1652 SequenceI proteinProduct = aMapping.getTo();
1653 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1659 * try to locate the CDS from the dataset mappings;
1660 * guard against duplicate results (for the case that protein has
1661 * dbrefs to both dna and cds sequences)
1663 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1664 seqMappings, aMapping);
1667 if (!foundSeqs.contains(cdsSeq))
1669 foundSeqs.add(cdsSeq);
1670 SequenceI derivedSequence = cdsSeq.deriveSequence();
1671 cdsSeqs.add(derivedSequence);
1672 if (!dataset.getSequences().contains(cdsSeq))
1674 dataset.addSequence(cdsSeq);
1681 * didn't find mapped CDS sequence - construct it and add
1682 * its dataset sequence to the dataset
1684 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping);
1685 SequenceI cdsSeqDss = cdsSeq.createDatasetSequence();
1686 cdsSeqs.add(cdsSeq);
1687 if (!dataset.getSequences().contains(cdsSeqDss))
1689 dataset.addSequence(cdsSeqDss);
1693 * add a mapping from CDS to the (unchanged) mapped to range
1695 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1696 cdsSeq.getLength() });
1697 MapList cdsToProteinMap = new MapList(cdsRange, mapList.getToRanges(),
1698 mapList.getFromRatio(), mapList.getToRatio());
1699 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1700 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1704 * guard against duplicating the mapping if repeating this action
1706 if (!mappings.contains(cdsToProteinMapping))
1708 mappings.add(cdsToProteinMapping);
1712 * add another mapping from original 'from' range to CDS
1714 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1715 MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1718 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1720 if (!mappings.contains(dnaToCdsMapping))
1722 mappings.add(dnaToCdsMapping);
1726 * add DBRef with mapping from protein to CDS
1727 * (this enables Get Cross-References from protein alignment)
1728 * This is tricky because we can't have two DBRefs with the
1729 * same source and accession, so need a different accession for
1730 * the CDS from the dna sequence
1732 DBRefEntryI dnaRef = dnaDss.getSourceDBRef();
1735 // assuming cds version same as dna ?!?
1736 DBRefEntry proteinToCdsRef = new DBRefEntry(dnaRef.getSource(),
1737 dnaRef.getVersion(), cdsSeq.getName());
1738 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1740 proteinProduct.addDBRef(proteinToCdsRef);
1744 * transfer any features on dna that overlap the CDS
1746 transferFeatures(dnaSeq, cdsSeq, cdsToProteinMap, null,
1747 SequenceOntologyI.CDS);
1752 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1754 cds.setDataset(dataset);
1760 * A helper method that finds a CDS sequence in the alignment dataset that is
1761 * mapped to the given protein sequence, and either is, or has a mapping from,
1762 * the given dna sequence.
1765 * set of all mappings on the dataset
1767 * a dna (or cds) sequence we are searching from
1768 * @param seqMappings
1769 * the set of mappings involving dnaSeq
1771 * an initial candidate from seqMappings
1774 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1775 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1779 * TODO a better dna-cds-protein mapping data representation to allow easy
1780 * navigation; until then this clunky looping around lists of mappings
1782 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1783 : dnaSeq.getDatasetSequence();
1784 SequenceI proteinProduct = aMapping.getTo();
1787 * is this mapping from the whole dna sequence (i.e. CDS)?
1788 * allowing for possible stop codon on dna but not peptide
1790 int mappedFromLength = MappingUtils.getLength(aMapping.getMap()
1792 int dnaLength = seqDss.getLength();
1793 if (mappedFromLength == dnaLength || mappedFromLength == dnaLength - 3)
1799 * looks like we found the dna-to-protein mapping; search for the
1800 * corresponding cds-to-protein mapping
1802 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1803 .findMappingsForSequence(proteinProduct, mappings);
1804 for (AlignedCodonFrame acf : mappingsToPeptide)
1806 for (SequenceToSequenceMapping map : acf.getMappings())
1808 Mapping mapping = map.getMapping();
1809 if (mapping != aMapping && mapping.getMap().getFromRatio() == 3
1810 && proteinProduct == mapping.getTo()
1811 && seqDss != map.getFromSeq())
1813 mappedFromLength = MappingUtils.getLength(mapping.getMap()
1815 if (mappedFromLength == map.getFromSeq().getLength())
1818 * found a 3:1 mapping to the protein product which covers
1819 * the whole dna sequence i.e. is from CDS; finally check it
1820 * is from the dna start sequence
1822 SequenceI cdsSeq = map.getFromSeq();
1823 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1824 .findMappingsForSequence(cdsSeq, seqMappings);
1825 if (!dnaToCdsMaps.isEmpty())
1837 * Helper method that makes a CDS sequence as defined by the mappings from the
1838 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1839 * forward or reverse strand).
1843 * @return CDS sequence (as a dataset sequence)
1845 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping)
1847 char[] seqChars = seq.getSequence();
1848 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1849 int cdsWidth = MappingUtils.getLength(fromRanges);
1850 char[] newSeqChars = new char[cdsWidth];
1853 for (int[] range : fromRanges)
1855 if (range[0] <= range[1])
1857 // forward strand mapping - just copy the range
1858 int length = range[1] - range[0] + 1;
1859 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1865 // reverse strand mapping - copy and complement one by one
1866 for (int i = range[0]; i >= range[1]; i--)
1868 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1874 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
1875 * else generate a sequence name
1877 String mapFromId = mapping.getMappedFromId();
1878 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
1879 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
1880 // newSeq.setDescription(mapFromId);
1882 propagateDBRefsToCDS(newSeq, seq, mapping);
1888 * add any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
1889 * the given mapping.
1894 * @return list of DBRefEntrys added.
1896 public static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
1897 SequenceI contig, Mapping mapping)
1900 // gather direct refs from contig congrent with mapping
1901 List<DBRefEntry> direct = new ArrayList<DBRefEntry>();
1902 if (contig.getDBRefs() != null)
1904 for (DBRefEntry dbr : contig.getDBRefs())
1906 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
1908 MapList map = dbr.getMap().getMap();
1909 // check if map is the CDS mapping
1910 if (mapping.getMap().equals(map))
1918 List<DBRefEntry> propagated = new ArrayList<DBRefEntry>();
1920 // and generate appropriate mappings
1921 for (DBRefEntry cdsref : direct)
1923 Mapping cdsmap = cdsref.getMap();
1924 MapList cdsposmap = new MapList(Arrays.asList(new int[][] { new int[]
1925 { cdsSeq.getStart(), cdsSeq.getEnd() } }), cdsmap.getMap()
1926 .getToRanges(), 3, 1);
1928 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
1929 cdsref.getVersion(), cdsref.getAccessionId(), new Mapping(
1930 cdsmap.getTo(), cdsposmap));
1931 cdsSeq.addDBRef(newref);
1932 propagated.add(newref);
1938 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
1939 * feature start/end ranges, optionally omitting specified feature types.
1940 * Returns the number of features copied.
1945 * if not null, only features of this type are copied (including
1946 * subtypes in the Sequence Ontology)
1948 * the mapping from 'fromSeq' to 'toSeq'
1951 public static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
1952 MapList mapping, String select, String... omitting)
1954 SequenceI copyTo = toSeq;
1955 while (copyTo.getDatasetSequence() != null)
1957 copyTo = copyTo.getDatasetSequence();
1960 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
1962 SequenceFeature[] sfs = fromSeq.getSequenceFeatures();
1965 for (SequenceFeature sf : sfs)
1967 String type = sf.getType();
1968 if (select != null && !so.isA(type, select))
1972 boolean omit = false;
1973 for (String toOmit : omitting)
1975 if (type.equals(toOmit))
1986 * locate the mapped range - null if either start or end is
1987 * not mapped (no partial overlaps are calculated)
1989 int start = sf.getBegin();
1990 int end = sf.getEnd();
1991 int[] mappedTo = mapping.locateInTo(start, end);
1993 * if whole exon range doesn't map, try interpreting it
1994 * as 5' or 3' exon overlapping the CDS range
1996 if (mappedTo == null)
1998 mappedTo = mapping.locateInTo(end, end);
1999 if (mappedTo != null)
2002 * end of exon is in CDS range - 5' overlap
2003 * to a range from the start of the peptide
2008 if (mappedTo == null)
2010 mappedTo = mapping.locateInTo(start, start);
2011 if (mappedTo != null)
2014 * start of exon is in CDS range - 3' overlap
2015 * to a range up to the end of the peptide
2017 mappedTo[1] = toSeq.getLength();
2020 if (mappedTo != null)
2022 SequenceFeature copy = new SequenceFeature(sf);
2023 copy.setBegin(Math.min(mappedTo[0], mappedTo[1]));
2024 copy.setEnd(Math.max(mappedTo[0], mappedTo[1]));
2025 copyTo.addSequenceFeature(copy);
2034 * Returns a mapping from dna to protein by inspecting sequence features of
2035 * type "CDS" on the dna.
2041 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2042 SequenceI proteinSeq)
2044 List<int[]> ranges = findCdsPositions(dnaSeq);
2045 int mappedDnaLength = MappingUtils.getLength(ranges);
2047 int proteinLength = proteinSeq.getLength();
2048 int proteinStart = proteinSeq.getStart();
2049 int proteinEnd = proteinSeq.getEnd();
2052 * incomplete start codon may mean X at start of peptide
2053 * we ignore both for mapping purposes
2055 if (proteinSeq.getCharAt(0) == 'X')
2057 // todo JAL-2022 support startPhase > 0
2061 List<int[]> proteinRange = new ArrayList<int[]>();
2064 * dna length should map to protein (or protein plus stop codon)
2066 int codesForResidues = mappedDnaLength / 3;
2067 if (codesForResidues == (proteinLength + 1))
2069 // assuming extra codon is for STOP and not in peptide
2072 if (codesForResidues == proteinLength)
2074 proteinRange.add(new int[] { proteinStart, proteinEnd });
2075 return new MapList(ranges, proteinRange, 3, 1);
2081 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2082 * start/end positions of sequence features of type "CDS" (or a sub-type of
2083 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2084 * position order, so this method is only valid for linear CDS in the same
2085 * sense as the protein product.
2090 public static List<int[]> findCdsPositions(SequenceI dnaSeq)
2092 List<int[]> result = new ArrayList<int[]>();
2093 SequenceFeature[] sfs = dnaSeq.getSequenceFeatures();
2099 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2102 for (SequenceFeature sf : sfs)
2105 * process a CDS feature (or a sub-type of CDS)
2107 if (so.isA(sf.getType(), SequenceOntologyI.CDS))
2112 phase = Integer.parseInt(sf.getPhase());
2113 } catch (NumberFormatException e)
2118 * phase > 0 on first codon means 5' incomplete - skip to the start
2119 * of the next codon; example ENST00000496384
2121 int begin = sf.getBegin();
2122 int end = sf.getEnd();
2123 if (result.isEmpty())
2128 // shouldn't happen!
2130 .println("Error: start phase extends beyond start CDS in "
2131 + dnaSeq.getName());
2134 result.add(new int[] { begin, end });
2139 * remove 'startPhase' positions (usually 0) from the first range
2140 * so we begin at the start of a complete codon
2142 if (!result.isEmpty())
2144 // TODO JAL-2022 correctly model start phase > 0
2145 result.get(0)[0] += startPhase;
2149 * Finally sort ranges by start position. This avoids a dependency on
2150 * keeping features in order on the sequence (if they are in order anyway,
2151 * the sort will have almost no work to do). The implicit assumption is CDS
2152 * ranges are assembled in order. Other cases should not use this method,
2153 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2155 Collections.sort(result, new Comparator<int[]>()
2158 public int compare(int[] o1, int[] o2)
2160 return Integer.compare(o1[0], o2[0]);
2167 * Maps exon features from dna to protein, and computes variants in peptide
2168 * product generated by variants in dna, and adds them as sequence_variant
2169 * features on the protein sequence. Returns the number of variant features
2174 * @param dnaToProtein
2176 public static int computeProteinFeatures(SequenceI dnaSeq,
2177 SequenceI peptide, MapList dnaToProtein)
2179 while (dnaSeq.getDatasetSequence() != null)
2181 dnaSeq = dnaSeq.getDatasetSequence();
2183 while (peptide.getDatasetSequence() != null)
2185 peptide = peptide.getDatasetSequence();
2188 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2191 * compute protein variants from dna variants and codon mappings;
2192 * NB - alternatively we could retrieve this using the REST service e.g.
2193 * http://rest.ensembl.org/overlap/translation
2194 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2195 * which would be a bit slower but possibly more reliable
2199 * build a map with codon variations for each potentially varying peptide
2201 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2202 dnaSeq, dnaToProtein);
2205 * scan codon variations, compute peptide variants and add to peptide sequence
2208 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2210 int peptidePos = variant.getKey();
2211 List<DnaVariant>[] codonVariants = variant.getValue();
2212 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2216 * sort to get sequence features in start position order
2217 * - would be better to store in Sequence as a TreeSet or NCList?
2219 if (peptide.getSequenceFeatures() != null)
2221 Arrays.sort(peptide.getSequenceFeatures(),
2222 new Comparator<SequenceFeature>()
2225 public int compare(SequenceFeature o1, SequenceFeature o2)
2227 int c = Integer.compare(o1.getBegin(), o2.getBegin());
2228 return c == 0 ? Integer.compare(o1.getEnd(), o2.getEnd())
2237 * Computes non-synonymous peptide variants from codon variants and adds them
2238 * as sequence_variant features on the protein sequence (one feature per
2239 * allele variant). Selected attributes (variant id, clinical significance)
2240 * are copied over to the new features.
2243 * the protein sequence
2245 * the position to compute peptide variants for
2246 * @param codonVariants
2247 * a list of dna variants per codon position
2248 * @return the number of features added
2250 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2251 List<DnaVariant>[] codonVariants)
2253 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2255 String base1 = codonVariants[0].get(0).base;
2256 String base2 = codonVariants[1].get(0).base;
2257 String base3 = codonVariants[2].get(0).base;
2260 * variants in first codon base
2262 for (DnaVariant var : codonVariants[0])
2264 if (var.variant != null)
2266 String alleles = (String) var.variant.getValue("alleles");
2267 if (alleles != null)
2269 for (String base : alleles.split(","))
2271 String codon = base + base2 + base3;
2272 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2282 * variants in second codon base
2284 for (DnaVariant var : codonVariants[1])
2286 if (var.variant != null)
2288 String alleles = (String) var.variant.getValue("alleles");
2289 if (alleles != null)
2291 for (String base : alleles.split(","))
2293 String codon = base1 + base + base3;
2294 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2304 * variants in third codon base
2306 for (DnaVariant var : codonVariants[2])
2308 if (var.variant != null)
2310 String alleles = (String) var.variant.getValue("alleles");
2311 if (alleles != null)
2313 for (String base : alleles.split(","))
2315 String codon = base1 + base2 + base;
2316 if (addPeptideVariant(peptide, peptidePos, residue, var, codon))
2329 * Helper method that adds a peptide variant feature, provided the given codon
2330 * translates to a value different to the current residue (is a non-synonymous
2331 * variant). ID and clinical_significance attributes of the dna variant (if
2332 * present) are copied to the new feature.
2339 * @return true if a feature was added, else false
2341 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2342 String residue, DnaVariant var, String codon)
2345 * get peptide translation of codon e.g. GAT -> D
2346 * note that variants which are not single alleles,
2347 * e.g. multibase variants or HGMD_MUTATION etc
2348 * are currently ignored here
2350 String trans = codon.contains("-") ? "-"
2351 : (codon.length() > 3 ? null : ResidueProperties
2352 .codonTranslate(codon));
2353 if (trans != null && !trans.equals(residue))
2355 String residue3Char = StringUtils
2356 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2357 String trans3Char = StringUtils
2358 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2359 String desc = "p." + residue3Char + peptidePos + trans3Char;
2360 // set score to 0f so 'graduated colour' option is offered! JAL-2060
2361 SequenceFeature sf = new SequenceFeature(
2362 SequenceOntologyI.SEQUENCE_VARIANT, desc, peptidePos,
2363 peptidePos, 0f, "Jalview");
2364 StringBuilder attributes = new StringBuilder(32);
2365 String id = (String) var.variant.getValue(ID);
2368 if (id.startsWith(SEQUENCE_VARIANT))
2370 id = id.substring(SEQUENCE_VARIANT.length());
2372 sf.setValue(ID, id);
2373 attributes.append(ID).append("=").append(id);
2374 // TODO handle other species variants
2375 StringBuilder link = new StringBuilder(32);
2378 link.append(desc).append(" ").append(id)
2379 .append("|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2380 .append(URLEncoder.encode(id, "UTF-8"));
2381 sf.addLink(link.toString());
2382 } catch (UnsupportedEncodingException e)
2387 String clinSig = (String) var.variant
2388 .getValue(CLINICAL_SIGNIFICANCE);
2389 if (clinSig != null)
2391 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2392 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2395 peptide.addSequenceFeature(sf);
2396 if (attributes.length() > 0)
2398 sf.setAttributes(attributes.toString());
2406 * Builds a map whose key is position in the protein sequence, and value is a
2407 * list of the base and all variants for each corresponding codon position
2410 * @param dnaToProtein
2413 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2414 SequenceI dnaSeq, MapList dnaToProtein)
2417 * map from peptide position to all variants of the codon which codes for it
2418 * LinkedHashMap ensures we keep the peptide features in sequence order
2420 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<Integer, List<DnaVariant>[]>();
2421 SequenceOntologyI so = SequenceOntologyFactory.getInstance();
2423 SequenceFeature[] dnaFeatures = dnaSeq.getSequenceFeatures();
2424 if (dnaFeatures == null)
2429 int dnaStart = dnaSeq.getStart();
2430 int[] lastCodon = null;
2431 int lastPeptidePostion = 0;
2434 * build a map of codon variations for peptides
2436 for (SequenceFeature sf : dnaFeatures)
2438 int dnaCol = sf.getBegin();
2439 if (dnaCol != sf.getEnd())
2441 // not handling multi-locus variant features
2444 if (so.isA(sf.getType(), SequenceOntologyI.SEQUENCE_VARIANT))
2446 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2449 // feature doesn't lie within coding region
2452 int peptidePosition = mapsTo[0];
2453 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2454 if (codonVariants == null)
2456 codonVariants = new ArrayList[3];
2457 codonVariants[0] = new ArrayList<DnaVariant>();
2458 codonVariants[1] = new ArrayList<DnaVariant>();
2459 codonVariants[2] = new ArrayList<DnaVariant>();
2460 variants.put(peptidePosition, codonVariants);
2464 * extract dna variants to a string array
2466 String alls = (String) sf.getValue("alleles");
2471 String[] alleles = alls.toUpperCase().split(",");
2473 for (String allele : alleles)
2475 alleles[i++] = allele.trim(); // lose any space characters "A, G"
2479 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2481 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2482 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2483 peptidePosition, peptidePosition));
2484 lastPeptidePostion = peptidePosition;
2488 * save nucleotide (and any variant) for each codon position
2490 for (int codonPos = 0; codonPos < 3; codonPos++)
2492 String nucleotide = String.valueOf(
2493 dnaSeq.getCharAt(codon[codonPos] - dnaStart))
2495 List<DnaVariant> codonVariant = codonVariants[codonPos];
2496 if (codon[codonPos] == dnaCol)
2498 if (!codonVariant.isEmpty()
2499 && codonVariant.get(0).variant == null)
2502 * already recorded base value, add this variant
2504 codonVariant.get(0).variant = sf;
2509 * add variant with base value
2511 codonVariant.add(new DnaVariant(nucleotide, sf));
2514 else if (codonVariant.isEmpty())
2517 * record (possibly non-varying) base value
2519 codonVariant.add(new DnaVariant(nucleotide));
2528 * Makes an alignment with a copy of the given sequences, adding in any
2529 * non-redundant sequences which are mapped to by the cross-referenced
2535 * the alignment dataset shared by the new copy
2538 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2539 SequenceI[] xrefs, AlignmentI dataset)
2541 AlignmentI copy = new Alignment(new Alignment(seqs));
2542 copy.setDataset(dataset);
2544 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2547 for (SequenceI xref : xrefs)
2549 DBRefEntry[] dbrefs = xref.getDBRefs();
2552 for (DBRefEntry dbref : dbrefs)
2554 if (dbref.getMap() == null || dbref.getMap().getTo() == null)
2558 SequenceI mappedTo = dbref.getMap().getTo();
2559 SequenceI match = matcher.findIdMatch(mappedTo);
2562 matcher.add(mappedTo);
2563 copy.addSequence(mappedTo);
2573 * Try to align sequences in 'unaligned' to match the alignment of their
2574 * mapped regions in 'aligned'. For example, could use this to align CDS
2575 * sequences which are mapped to their parent cDNA sequences.
2577 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2578 * dna-to-protein or protein-to-dna use alternative methods.
2581 * sequences to be aligned
2583 * holds aligned sequences and their mappings
2586 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2589 * easy case - aligning a copy of aligned sequences
2591 if (alignAsSameSequences(unaligned, aligned))
2593 return unaligned.getHeight();
2597 * fancy case - aligning via mappings between sequences
2599 List<SequenceI> unmapped = new ArrayList<SequenceI>();
2600 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2601 unaligned, aligned, unmapped);
2602 int width = columnMap.size();
2603 char gap = unaligned.getGapCharacter();
2604 int realignedCount = 0;
2605 // TODO: verify this loop scales sensibly for very wide/high alignments
2607 for (SequenceI seq : unaligned.getSequences())
2609 if (!unmapped.contains(seq))
2611 char[] newSeq = new char[width];
2612 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2613 // Integer iteration below
2618 * traverse the map to find columns populated
2621 for (Integer column : columnMap.keySet())
2623 Character c = columnMap.get(column).get(seq);
2627 * sequence has a character at this position
2637 * trim trailing gaps
2639 if (lastCol < width)
2641 char[] tmp = new char[lastCol + 1];
2642 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2645 // TODO: optimise SequenceI to avoid char[]->String->char[]
2646 seq.setSequence(String.valueOf(newSeq));
2650 return realignedCount;
2654 * If unaligned and aligned sequences share the same dataset sequences, then
2655 * simply copies the aligned sequences to the unaligned sequences and returns
2656 * true; else returns false
2659 * - sequences to be aligned based on aligned
2661 * - 'guide' alignment containing sequences derived from same dataset
2665 static boolean alignAsSameSequences(AlignmentI unaligned,
2668 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2670 return false; // should only pass alignments with datasets here
2673 // map from dataset sequence to alignment sequence(s)
2674 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<SequenceI, List<SequenceI>>();
2675 for (SequenceI seq : aligned.getSequences())
2677 SequenceI ds = seq.getDatasetSequence();
2678 if (alignedDatasets.get(ds) == null)
2680 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2682 alignedDatasets.get(ds).add(seq);
2686 * first pass - check whether all sequences to be aligned share a dataset
2687 * sequence with an aligned sequence
2689 for (SequenceI seq : unaligned.getSequences())
2691 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2698 * second pass - copy aligned sequences;
2699 * heuristic rule: pair off sequences in order for the case where
2700 * more than one shares the same dataset sequence
2702 for (SequenceI seq : unaligned.getSequences())
2704 List<SequenceI> alignedSequences = alignedDatasets.get(seq
2705 .getDatasetSequence());
2706 // TODO: getSequenceAsString() will be deprecated in the future
2707 // TODO: need to leave to SequenceI implementor to update gaps
2708 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2709 if (alignedSequences.size() > 0)
2711 // pop off aligned sequences (except the last one)
2712 alignedSequences.remove(0);
2720 * Returns a map whose key is alignment column number (base 1), and whose
2721 * values are a map of sequence characters in that column.
2728 static Map<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2729 AlignmentI unaligned, AlignmentI aligned, List<SequenceI> unmapped)
2732 * Map will hold, for each aligned column position, a map of
2733 * {unalignedSequence, characterPerSequence} at that position.
2734 * TreeMap keeps the entries in ascending column order.
2736 Map<Integer, Map<SequenceI, Character>> map = new TreeMap<Integer, Map<SequenceI, Character>>();
2739 * record any sequences that have no mapping so can't be realigned
2741 unmapped.addAll(unaligned.getSequences());
2743 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2745 for (SequenceI seq : unaligned.getSequences())
2747 for (AlignedCodonFrame mapping : mappings)
2749 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2750 if (fromSeq != null)
2752 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2753 if (addMappedPositions(seq, fromSeq, seqMap, map))
2755 unmapped.remove(seq);
2764 * Helper method that adds to a map the mapped column positions of a sequence. <br>
2765 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2766 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2770 * the sequence whose column positions we are recording
2772 * a sequence that is mapped to the first sequence
2774 * the mapping from 'fromSeq' to 'seq'
2776 * a map to add the column positions (in fromSeq) of the mapped
2780 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2781 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2789 * invert mapping if it is from unaligned to aligned sequence
2791 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2793 seqMap = new Mapping(seq.getDatasetSequence(), seqMap.getMap()
2797 char[] fromChars = fromSeq.getSequence();
2798 int toStart = seq.getStart();
2799 char[] toChars = seq.getSequence();
2802 * traverse [start, end, start, end...] ranges in fromSeq
2804 for (int[] fromRange : seqMap.getMap().getFromRanges())
2806 for (int i = 0; i < fromRange.length - 1; i += 2)
2808 boolean forward = fromRange[i + 1] >= fromRange[i];
2811 * find the range mapped to (sequence positions base 1)
2813 int[] range = seqMap.locateMappedRange(fromRange[i],
2817 System.err.println("Error in mapping " + seqMap + " from "
2818 + fromSeq.getName());
2821 int fromCol = fromSeq.findIndex(fromRange[i]);
2822 int mappedCharPos = range[0];
2825 * walk over the 'from' aligned sequence in forward or reverse
2826 * direction; when a non-gap is found, record the column position
2827 * of the next character of the mapped-to sequence; stop when all
2828 * the characters of the range have been counted
2830 while (mappedCharPos <= range[1] && fromCol <= fromChars.length
2833 if (!Comparison.isGap(fromChars[fromCol - 1]))
2836 * mapped from sequence has a character in this column
2837 * record the column position for the mapped to character
2839 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2840 if (seqsMap == null)
2842 seqsMap = new HashMap<SequenceI, Character>();
2843 map.put(fromCol, seqsMap);
2845 seqsMap.put(seq, toChars[mappedCharPos - toStart]);
2848 fromCol += (forward ? 1 : -1);
2855 // strictly temporary hack until proper criteria for aligning protein to cds
2856 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2857 public static boolean looksLikeEnsembl(AlignmentI alignment)
2859 for (SequenceI seq : alignment.getSequences())
2861 String name = seq.getName();
2862 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))