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.GeneLociI;
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.datamodel.features.SequenceFeatures;
40 import jalview.io.gff.Gff3Helper;
41 import jalview.io.gff.SequenceOntologyI;
42 import jalview.schemes.ResidueProperties;
43 import jalview.util.Comparison;
44 import jalview.util.DBRefUtils;
45 import jalview.util.IntRangeComparator;
46 import jalview.util.MapList;
47 import jalview.util.MappingUtils;
48 import jalview.util.StringUtils;
50 import java.io.UnsupportedEncodingException;
51 import java.net.URLEncoder;
52 import java.util.ArrayList;
53 import java.util.Arrays;
54 import java.util.Collection;
55 import java.util.Collections;
56 import java.util.HashMap;
57 import java.util.HashSet;
58 import java.util.Iterator;
59 import java.util.LinkedHashMap;
60 import java.util.List;
62 import java.util.Map.Entry;
63 import java.util.NoSuchElementException;
65 import java.util.SortedMap;
66 import java.util.TreeMap;
69 * grab bag of useful alignment manipulation operations Expect these to be
70 * refactored elsewhere at some point.
75 public class AlignmentUtils
77 private static final int CODON_LENGTH = 3;
79 private static final String SEQUENCE_VARIANT = "sequence_variant:";
82 * the 'id' attribute is provided for variant features fetched from
83 * Ensembl using its REST service with JSON format
85 public static final String VARIANT_ID = "id";
88 * A data model to hold the 'normal' base value at a position, and an optional
89 * sequence variant feature
91 static final class DnaVariant
95 SequenceFeature variant;
97 DnaVariant(String nuc)
103 DnaVariant(String nuc, SequenceFeature var)
109 public String getSource()
111 return variant == null ? null : variant.getFeatureGroup();
115 * toString for aid in the debugger only
118 public String toString()
120 return base + ":" + (variant == null ? "" : variant.getDescription());
125 * given an existing alignment, create a new alignment including all, or up to
126 * flankSize additional symbols from each sequence's dataset sequence
132 public static AlignmentI expandContext(AlignmentI core, int flankSize)
134 List<SequenceI> sq = new ArrayList<>();
136 for (SequenceI s : core.getSequences())
138 SequenceI newSeq = s.deriveSequence();
139 final int newSeqStart = newSeq.getStart() - 1;
140 if (newSeqStart > maxoffset
141 && newSeq.getDatasetSequence().getStart() < s.getStart())
143 maxoffset = newSeqStart;
149 maxoffset = Math.min(maxoffset, flankSize);
153 * now add offset left and right to create an expanded alignment
155 for (SequenceI s : sq)
158 while (ds.getDatasetSequence() != null)
160 ds = ds.getDatasetSequence();
162 int s_end = s.findPosition(s.getStart() + s.getLength());
163 // find available flanking residues for sequence
164 int ustream_ds = s.getStart() - ds.getStart();
165 int dstream_ds = ds.getEnd() - s_end;
167 // build new flanked sequence
169 // compute gap padding to start of flanking sequence
170 int offset = maxoffset - ustream_ds;
172 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
175 if (flankSize < ustream_ds)
177 // take up to flankSize residues
178 offset = maxoffset - flankSize;
179 ustream_ds = flankSize;
181 if (flankSize <= dstream_ds)
183 dstream_ds = flankSize - 1;
186 // TODO use Character.toLowerCase to avoid creating String objects?
187 char[] upstream = new String(ds
188 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
189 .toLowerCase().toCharArray();
190 char[] downstream = new String(
191 ds.getSequence(s_end - 1, s_end + dstream_ds)).toLowerCase()
193 char[] coreseq = s.getSequence();
194 char[] nseq = new char[offset + upstream.length + downstream.length
196 char c = core.getGapCharacter();
199 for (; p < offset; p++)
204 System.arraycopy(upstream, 0, nseq, p, upstream.length);
205 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
207 System.arraycopy(downstream, 0, nseq,
208 p + coreseq.length + upstream.length, downstream.length);
209 s.setSequence(new String(nseq));
210 s.setStart(s.getStart() - ustream_ds);
211 s.setEnd(s_end + downstream.length);
213 AlignmentI newAl = new jalview.datamodel.Alignment(
214 sq.toArray(new SequenceI[0]));
215 for (SequenceI s : sq)
217 if (s.getAnnotation() != null)
219 for (AlignmentAnnotation aa : s.getAnnotation())
221 aa.adjustForAlignment(); // JAL-1712 fix
222 newAl.addAnnotation(aa);
226 newAl.setDataset(core.getDataset());
231 * Returns a map of lists of sequences in the alignment, keyed by sequence
232 * name. For use in mapping between different alignment views of the same
235 * @see jalview.datamodel.AlignmentI#getSequencesByName()
237 public static Map<String, List<SequenceI>> getSequencesByName(
240 Map<String, List<SequenceI>> theMap = new LinkedHashMap<>();
241 for (SequenceI seq : al.getSequences())
243 String name = seq.getName();
246 List<SequenceI> seqs = theMap.get(name);
249 seqs = new ArrayList<>();
250 theMap.put(name, seqs);
259 * Build mapping of protein to cDNA alignment. Mappings are made between
260 * sequences where the cDNA translates to the protein sequence. Any new
261 * mappings are added to the protein alignment. Returns true if any mappings
262 * either already exist or were added, else false.
264 * @param proteinAlignment
265 * @param cdnaAlignment
268 public static boolean mapProteinAlignmentToCdna(
269 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
271 if (proteinAlignment == null || cdnaAlignment == null)
276 Set<SequenceI> mappedDna = new HashSet<>();
277 Set<SequenceI> mappedProtein = new HashSet<>();
280 * First pass - map sequences where cross-references exist. This include
281 * 1-to-many mappings to support, for example, variant cDNA.
283 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
284 cdnaAlignment, mappedDna, mappedProtein, true);
287 * Second pass - map sequences where no cross-references exist. This only
288 * does 1-to-1 mappings and assumes corresponding sequences are in the same
289 * order in the alignments.
291 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
292 mappedDna, mappedProtein, false);
293 return mappingPerformed;
297 * Make mappings between compatible sequences (where the cDNA translation
298 * matches the protein).
300 * @param proteinAlignment
301 * @param cdnaAlignment
303 * a set of mapped DNA sequences (to add to)
304 * @param mappedProtein
305 * a set of mapped Protein sequences (to add to)
307 * if true, only map sequences where xrefs exist
310 protected static boolean mapProteinToCdna(
311 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
312 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
315 boolean mappingExistsOrAdded = false;
316 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
317 for (SequenceI aaSeq : thisSeqs)
319 boolean proteinMapped = false;
320 AlignedCodonFrame acf = new AlignedCodonFrame();
322 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
325 * Always try to map if sequences have xref to each other; this supports
326 * variant cDNA or alternative splicing for a protein sequence.
328 * If no xrefs, try to map progressively, assuming that alignments have
329 * mappable sequences in corresponding order. These are not
330 * many-to-many, as that would risk mixing species with similar cDNA
333 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
339 * Don't map non-xrefd sequences more than once each. This heuristic
340 * allows us to pair up similar sequences in ordered alignments.
342 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
343 || mappedDna.contains(cdnaSeq)))
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 protected 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
398 * <li>else checks for translation after removing a STOP codon</li>
399 * <li>else checks for translation after removing a START codon</li>
400 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
402 * Returns null if no mapping is determined.
405 * the aligned protein sequence
407 * the aligned cdna sequence
410 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
414 * Here we handle either dataset sequence set (desktop) or absent (applet).
415 * Use only the char[] form of the sequence to avoid creating possibly large
418 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
419 char[] aaSeqChars = proteinDataset != null
420 ? proteinDataset.getSequence()
421 : proteinSeq.getSequence();
422 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
423 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
424 : cdnaSeq.getSequence();
425 if (aaSeqChars == null || cdnaSeqChars == null)
431 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
433 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
434 int cdnaLength = cdnaSeqChars.length;
435 int cdnaStart = cdnaSeq.getStart();
436 int cdnaEnd = cdnaSeq.getEnd();
437 final int proteinStart = proteinSeq.getStart();
438 final int proteinEnd = proteinSeq.getEnd();
441 * If lengths don't match, try ignoring stop codon (if present)
443 if (cdnaLength != mappedLength && cdnaLength > 2)
445 String lastCodon = String.valueOf(cdnaSeqChars,
446 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
447 for (String stop : ResidueProperties.STOP_CODONS)
449 if (lastCodon.equals(stop))
451 cdnaEnd -= CODON_LENGTH;
452 cdnaLength -= CODON_LENGTH;
459 * If lengths still don't match, try ignoring start codon.
462 if (cdnaLength != mappedLength && cdnaLength > 2
463 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
464 .equals(ResidueProperties.START))
466 startOffset += CODON_LENGTH;
467 cdnaStart += CODON_LENGTH;
468 cdnaLength -= CODON_LENGTH;
471 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
474 * protein is translation of dna (+/- start/stop codons)
476 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
478 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
483 * translation failed - try mapping CDS annotated regions of dna
485 return mapCdsToProtein(cdnaSeq, proteinSeq);
489 * Test whether the given cdna sequence, starting at the given offset,
490 * translates to the given amino acid sequence, using the standard translation
491 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
493 * @param cdnaSeqChars
498 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
501 if (cdnaSeqChars == null || aaSeqChars == null)
507 int dnaPos = cdnaStart;
508 for (; dnaPos < cdnaSeqChars.length - 2
509 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
511 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
512 final String translated = ResidueProperties.codonTranslate(codon);
515 * allow * in protein to match untranslatable in dna
517 final char aaRes = aaSeqChars[aaPos];
518 if ((translated == null || ResidueProperties.STOP.equals(translated))
523 if (translated == null || !(aaRes == translated.charAt(0)))
526 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
527 // + codon + "(" + translated + ") != " + aaRes));
533 * check we matched all of the protein sequence
535 if (aaPos != aaSeqChars.length)
541 * check we matched all of the dna except
542 * for optional trailing STOP codon
544 if (dnaPos == cdnaSeqChars.length)
548 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
550 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
551 if (ResidueProperties.STOP
552 .equals(ResidueProperties.codonTranslate(codon)))
561 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
562 * currently assumes that we are aligning cDNA to match protein.
565 * the sequence to be realigned
567 * the alignment whose sequence alignment is to be 'copied'
569 * character string represent a gap in the realigned sequence
570 * @param preserveUnmappedGaps
571 * @param preserveMappedGaps
572 * @return true if the sequence was realigned, false if it could not be
574 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
575 String gap, boolean preserveMappedGaps,
576 boolean preserveUnmappedGaps)
579 * Get any mappings from the source alignment to the target (dataset)
582 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
583 // all mappings. Would it help to constrain this?
584 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
585 if (mappings == null || mappings.isEmpty())
591 * Locate the aligned source sequence whose dataset sequence is mapped. We
592 * just take the first match here (as we can't align like more than one
595 SequenceI alignFrom = null;
596 AlignedCodonFrame mapping = null;
597 for (AlignedCodonFrame mp : mappings)
599 alignFrom = mp.findAlignedSequence(seq, al);
600 if (alignFrom != null)
607 if (alignFrom == null)
611 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
612 preserveMappedGaps, preserveUnmappedGaps);
617 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
618 * match residues and codons. Flags control whether existing gaps in unmapped
619 * (intron) and mapped (exon) regions are preserved or not. Gaps between
620 * intron and exon are only retained if both flags are set.
627 * @param preserveUnmappedGaps
628 * @param preserveMappedGaps
630 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
631 AlignedCodonFrame mapping, String myGap, char sourceGap,
632 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
634 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
636 // aligned and dataset sequence positions, all base zero
640 int basesWritten = 0;
641 char myGapChar = myGap.charAt(0);
642 int ratio = myGap.length();
644 int fromOffset = alignFrom.getStart() - 1;
645 int toOffset = alignTo.getStart() - 1;
646 int sourceGapMappedLength = 0;
647 boolean inExon = false;
648 final int toLength = alignTo.getLength();
649 final int fromLength = alignFrom.getLength();
650 StringBuilder thisAligned = new StringBuilder(2 * toLength);
653 * Traverse the 'model' aligned sequence
655 for (int i = 0; i < fromLength; i++)
657 char sourceChar = alignFrom.getCharAt(i);
658 if (sourceChar == sourceGap)
660 sourceGapMappedLength += ratio;
665 * Found a non-gap character. Locate its mapped region if any.
668 // Note mapping positions are base 1, our sequence positions base 0
669 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
670 sourceDsPos + fromOffset);
671 if (mappedPos == null)
674 * unmapped position; treat like a gap
676 sourceGapMappedLength += ratio;
677 // System.err.println("Can't align: no codon mapping to residue "
678 // + sourceDsPos + "(" + sourceChar + ")");
683 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
684 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
685 StringBuilder trailingCopiedGap = new StringBuilder();
688 * Copy dna sequence up to and including this codon. Optionally, include
689 * gaps before the codon starts (in introns) and/or after the codon starts
692 * Note this only works for 'linear' splicing, not reverse or interleaved.
693 * But then 'align dna as protein' doesn't make much sense otherwise.
695 int intronLength = 0;
696 while (basesWritten + toOffset < mappedCodonEnd
697 && thisSeqPos < toLength)
699 final char c = alignTo.getCharAt(thisSeqPos++);
703 int sourcePosition = basesWritten + toOffset;
704 if (sourcePosition < mappedCodonStart)
707 * Found an unmapped (intron) base. First add in any preceding gaps
710 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
712 thisAligned.append(trailingCopiedGap.toString());
713 intronLength += trailingCopiedGap.length();
714 trailingCopiedGap = new StringBuilder();
721 final boolean startOfCodon = sourcePosition == mappedCodonStart;
722 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
723 preserveUnmappedGaps, sourceGapMappedLength, inExon,
724 trailingCopiedGap.length(), intronLength, startOfCodon);
725 for (int k = 0; k < gapsToAdd; k++)
727 thisAligned.append(myGapChar);
729 sourceGapMappedLength = 0;
732 thisAligned.append(c);
733 trailingCopiedGap = new StringBuilder();
737 if (inExon && preserveMappedGaps)
739 trailingCopiedGap.append(myGapChar);
741 else if (!inExon && preserveUnmappedGaps)
743 trailingCopiedGap.append(myGapChar);
750 * At end of model aligned sequence. Copy any remaining target sequence, optionally
751 * including (intron) gaps.
753 while (thisSeqPos < toLength)
755 final char c = alignTo.getCharAt(thisSeqPos++);
756 if (c != myGapChar || preserveUnmappedGaps)
758 thisAligned.append(c);
760 sourceGapMappedLength--;
764 * finally add gaps to pad for any trailing source gaps or
765 * unmapped characters
767 if (preserveUnmappedGaps)
769 while (sourceGapMappedLength > 0)
771 thisAligned.append(myGapChar);
772 sourceGapMappedLength--;
777 * All done aligning, set the aligned sequence.
779 alignTo.setSequence(new String(thisAligned));
783 * Helper method to work out how many gaps to insert when realigning.
785 * @param preserveMappedGaps
786 * @param preserveUnmappedGaps
787 * @param sourceGapMappedLength
789 * @param trailingCopiedGap
790 * @param intronLength
791 * @param startOfCodon
794 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
795 boolean preserveUnmappedGaps, int sourceGapMappedLength,
796 boolean inExon, int trailingGapLength, int intronLength,
797 final boolean startOfCodon)
803 * Reached start of codon. Ignore trailing gaps in intron unless we are
804 * preserving gaps in both exon and intron. Ignore them anyway if the
805 * protein alignment introduces a gap at least as large as the intronic
808 if (inExon && !preserveMappedGaps)
810 trailingGapLength = 0;
812 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
814 trailingGapLength = 0;
818 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
822 if (intronLength + trailingGapLength <= sourceGapMappedLength)
824 gapsToAdd = sourceGapMappedLength - intronLength;
828 gapsToAdd = Math.min(
829 intronLength + trailingGapLength - sourceGapMappedLength,
837 * second or third base of codon; check for any gaps in dna
839 if (!preserveMappedGaps)
841 trailingGapLength = 0;
843 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
849 * Realigns the given protein to match the alignment of the dna, using codon
850 * mappings to translate aligned codon positions to protein residues.
853 * the alignment whose sequences are realigned by this method
855 * the dna alignment whose alignment we are 'copying'
856 * @return the number of sequences that were realigned
858 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
860 if (protein.isNucleotide() || !dna.isNucleotide())
862 System.err.println("Wrong alignment type in alignProteinAsDna");
865 List<SequenceI> unmappedProtein = new ArrayList<>();
866 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
867 protein, dna, unmappedProtein);
868 return alignProteinAs(protein, alignedCodons, unmappedProtein);
872 * Realigns the given dna to match the alignment of the protein, using codon
873 * mappings to translate aligned peptide positions to codons.
875 * Always produces a padded CDS alignment.
878 * the alignment whose sequences are realigned by this method
880 * the protein alignment whose alignment we are 'copying'
881 * @return the number of sequences that were realigned
883 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
885 if (protein.isNucleotide() || !dna.isNucleotide())
887 System.err.println("Wrong alignment type in alignProteinAsDna");
890 // todo: implement this
891 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
892 int alignedCount = 0;
893 int width = 0; // alignment width for padding CDS
894 for (SequenceI dnaSeq : dna.getSequences())
896 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
897 dna.getGapCharacter()))
901 width = Math.max(dnaSeq.getLength(), width);
905 for (SequenceI dnaSeq : dna.getSequences())
907 oldwidth = dnaSeq.getLength();
908 diff = width - oldwidth;
911 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
918 * Helper method to align (if possible) the dna sequence to match the
919 * alignment of a mapped protein sequence. This is currently limited to
920 * handling coding sequence only.
928 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
929 AlignmentI protein, List<AlignedCodonFrame> mappings,
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 final 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 final int cdsLength = cdsDss.getLength();
957 int mappedFromLength = MappingUtils.getLength(mapList
959 int mappedToLength = MappingUtils
960 .getLength(mapList.getToRanges());
961 boolean addStopCodon = (cdsLength == mappedFromLength
962 * CODON_LENGTH + CODON_LENGTH)
963 || (peptide.getDatasetSequence()
964 .getLength() == mappedFromLength - 1);
965 if (cdsLength != mappedToLength && !addStopCodon)
967 System.err.println(String.format(
968 "Can't align cds as protein (length mismatch %d/%d): %s",
969 cdsLength, mappedToLength, cdsSeq.getName()));
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
984 int cdsStart = cdsDss.getStart();
985 int proteinPos = peptide.getStart() - 1;
988 for (int col = 0; col < peptideLength; col++)
990 char residue = peptide.getCharAt(col);
992 if (Comparison.isGap(residue))
994 cdsCol += CODON_LENGTH;
999 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1002 // e.g. incomplete start codon, X in peptide
1003 cdsCol += CODON_LENGTH;
1007 for (int j = codon[0]; j <= codon[1]; j++)
1009 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1010 alignedCds[cdsCol++] = mappedBase;
1018 * append stop codon if not mapped from protein,
1019 * closing it up to the end of the mapped sequence
1021 if (copiedBases == cdsLength - CODON_LENGTH)
1023 for (int i = alignedCds.length - 1; i >= 0; i--)
1025 if (!Comparison.isGap(alignedCds[i]))
1027 cdsCol = i + 1; // gap just after end of sequence
1031 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1033 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1036 cdsSeq.setSequence(new String(alignedCds));
1045 * Builds a map whose key is an aligned codon position (3 alignment column
1046 * numbers base 0), and whose value is a map from protein sequence to each
1047 * protein's peptide residue for that codon. The map generates an ordering of
1048 * the codons, and allows us to read off the peptides at each position in
1049 * order to assemble 'aligned' protein sequences.
1052 * the protein alignment
1054 * the coding dna alignment
1055 * @param unmappedProtein
1056 * any unmapped proteins are added to this list
1059 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1060 AlignmentI protein, AlignmentI dna,
1061 List<SequenceI> unmappedProtein)
1064 * maintain a list of any proteins with no mappings - these will be
1065 * rendered 'as is' in the protein alignment as we can't align them
1067 unmappedProtein.addAll(protein.getSequences());
1069 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1072 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1073 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1074 * comparator keeps the codon positions ordered.
1076 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<>(
1077 new CodonComparator());
1079 for (SequenceI dnaSeq : dna.getSequences())
1081 for (AlignedCodonFrame mapping : mappings)
1083 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1086 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1087 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1089 unmappedProtein.remove(prot);
1095 * Finally add any unmapped peptide start residues (e.g. for incomplete
1096 * codons) as if at the codon position before the second residue
1098 // TODO resolve JAL-2022 so this fudge can be removed
1099 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1100 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1102 return alignedCodons;
1106 * Scans for any protein mapped from position 2 (meaning unmapped start
1107 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1108 * preceding position in the alignment
1110 * @param alignedCodons
1111 * the codon-to-peptide map
1112 * @param mappedSequenceCount
1113 * the number of distinct sequences in the map
1115 protected static void addUnmappedPeptideStarts(
1116 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1117 int mappedSequenceCount)
1119 // TODO delete this ugly hack once JAL-2022 is resolved
1120 // i.e. we can model startPhase > 0 (incomplete start codon)
1122 List<SequenceI> sequencesChecked = new ArrayList<>();
1123 AlignedCodon lastCodon = null;
1124 Map<SequenceI, AlignedCodon> toAdd = new HashMap<>();
1126 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1129 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1132 SequenceI seq = sequenceCodon.getKey();
1133 if (sequencesChecked.contains(seq))
1137 sequencesChecked.add(seq);
1138 AlignedCodon codon = sequenceCodon.getValue();
1139 if (codon.peptideCol > 1)
1142 "Problem mapping protein with >1 unmapped start positions: "
1145 else if (codon.peptideCol == 1)
1148 * first position (peptideCol == 0) was unmapped - add it
1150 if (lastCodon != null)
1152 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1153 lastCodon.pos2, lastCodon.pos3,
1154 String.valueOf(seq.getCharAt(0)), 0);
1155 toAdd.put(seq, firstPeptide);
1160 * unmapped residue at start of alignment (no prior column) -
1161 * 'insert' at nominal codon [0, 0, 0]
1163 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1164 String.valueOf(seq.getCharAt(0)), 0);
1165 toAdd.put(seq, firstPeptide);
1168 if (sequencesChecked.size() == mappedSequenceCount)
1170 // no need to check past first mapped position in all sequences
1174 lastCodon = entry.getKey();
1178 * add any new codons safely after iterating over the map
1180 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1182 addCodonToMap(alignedCodons, startCodon.getValue(),
1183 startCodon.getKey());
1188 * Update the aligned protein sequences to match the codon alignments given in
1192 * @param alignedCodons
1193 * an ordered map of codon positions (columns), with sequence/peptide
1194 * values present in each column
1195 * @param unmappedProtein
1198 protected static int alignProteinAs(AlignmentI protein,
1199 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1200 List<SequenceI> unmappedProtein)
1203 * prefill peptide sequences with gaps
1205 int alignedWidth = alignedCodons.size();
1206 char[] gaps = new char[alignedWidth];
1207 Arrays.fill(gaps, protein.getGapCharacter());
1208 Map<SequenceI, char[]> peptides = new HashMap<>();
1209 for (SequenceI seq : protein.getSequences())
1211 if (!unmappedProtein.contains(seq))
1213 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1218 * Traverse the codons left to right (as defined by CodonComparator)
1219 * and insert peptides in each column where the sequence is mapped.
1220 * This gives a peptide 'alignment' where residues are aligned if their
1221 * corresponding codons occupy the same columns in the cdna alignment.
1224 for (AlignedCodon codon : alignedCodons.keySet())
1226 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1228 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1230 char residue = entry.getValue().product.charAt(0);
1231 peptides.get(entry.getKey())[column] = residue;
1237 * and finally set the constructed sequences
1239 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1241 entry.getKey().setSequence(new String(entry.getValue()));
1248 * Populate the map of aligned codons by traversing the given sequence
1249 * mapping, locating the aligned positions of mapped codons, and adding those
1250 * positions and their translation products to the map.
1253 * the aligned sequence we are mapping from
1255 * the sequence to be aligned to the codons
1257 * the gap character in the dna sequence
1259 * a mapping to a sequence translation
1260 * @param alignedCodons
1261 * the map we are building up
1263 static void addCodonPositions(SequenceI dna, SequenceI protein,
1264 char gapChar, Mapping seqMap,
1265 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1267 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1270 * add codon positions, and their peptide translations, to the alignment
1271 * map, while remembering the first codon mapped
1273 while (codons.hasNext())
1277 AlignedCodon codon = codons.next();
1278 addCodonToMap(alignedCodons, codon, protein);
1279 } catch (IncompleteCodonException e)
1281 // possible incomplete trailing codon - ignore
1282 } catch (NoSuchElementException e)
1284 // possibly peptide lacking STOP
1290 * Helper method to add a codon-to-peptide entry to the aligned codons map
1292 * @param alignedCodons
1296 protected static void addCodonToMap(
1297 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1298 AlignedCodon codon, SequenceI protein)
1300 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1301 if (seqProduct == null)
1303 seqProduct = new HashMap<>();
1304 alignedCodons.put(codon, seqProduct);
1306 seqProduct.put(protein, codon);
1310 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1311 * between at least one pair of sequences in the two alignments. Currently,
1314 * <li>One alignment must be nucleotide, and the other protein</li>
1315 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1316 * <li>Mappable means the nucleotide translation matches the protein
1318 * <li>The translation may ignore start and stop codons if present in the
1326 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1328 if (al1 == null || al2 == null)
1334 * Require one nucleotide and one protein
1336 if (al1.isNucleotide() == al2.isNucleotide())
1340 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1341 AlignmentI protein = dna == al1 ? al2 : al1;
1342 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1343 for (SequenceI dnaSeq : dna.getSequences())
1345 for (SequenceI proteinSeq : protein.getSequences())
1347 if (isMappable(dnaSeq, proteinSeq, mappings))
1357 * Returns true if the dna sequence is mapped, or could be mapped, to the
1365 protected static boolean isMappable(SequenceI dnaSeq,
1366 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1368 if (dnaSeq == null || proteinSeq == null)
1373 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1374 : dnaSeq.getDatasetSequence();
1375 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1377 : proteinSeq.getDatasetSequence();
1379 for (AlignedCodonFrame mapping : mappings)
1381 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1391 * Just try to make a mapping (it is not yet stored), test whether
1394 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1398 * Finds any reference annotations associated with the sequences in
1399 * sequenceScope, that are not already added to the alignment, and adds them
1400 * to the 'candidates' map. Also populates a lookup table of annotation
1401 * labels, keyed by calcId, for use in constructing tooltips or the like.
1403 * @param sequenceScope
1404 * the sequences to scan for reference annotations
1405 * @param labelForCalcId
1406 * (optional) map to populate with label for calcId
1408 * map to populate with annotations for sequence
1410 * the alignment to check for presence of annotations
1412 public static void findAddableReferenceAnnotations(
1413 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1414 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1417 if (sequenceScope == null)
1423 * For each sequence in scope, make a list of any annotations on the
1424 * underlying dataset sequence which are not already on the alignment.
1426 * Add to a map of { alignmentSequence, <List of annotations to add> }
1428 for (SequenceI seq : sequenceScope)
1430 SequenceI dataset = seq.getDatasetSequence();
1431 if (dataset == null)
1435 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1436 if (datasetAnnotations == null)
1440 final List<AlignmentAnnotation> result = new ArrayList<>();
1441 for (AlignmentAnnotation dsann : datasetAnnotations)
1444 * Find matching annotations on the alignment. If none is found, then
1445 * add this annotation to the list of 'addable' annotations for this
1448 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1449 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1450 if (!matchedAlignmentAnnotations.iterator().hasNext())
1453 if (labelForCalcId != null)
1455 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1460 * Save any addable annotations for this sequence
1462 if (!result.isEmpty())
1464 candidates.put(seq, result);
1470 * Adds annotations to the top of the alignment annotations, in the same order
1471 * as their related sequences.
1473 * @param annotations
1474 * the annotations to add
1476 * the alignment to add them to
1477 * @param selectionGroup
1478 * current selection group (or null if none)
1480 public static void addReferenceAnnotations(
1481 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1482 final AlignmentI alignment, final SequenceGroup selectionGroup)
1484 for (SequenceI seq : annotations.keySet())
1486 for (AlignmentAnnotation ann : annotations.get(seq))
1488 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1490 int endRes = ann.annotations.length;
1491 if (selectionGroup != null)
1493 startRes = selectionGroup.getStartRes();
1494 endRes = selectionGroup.getEndRes();
1496 copyAnn.restrict(startRes, endRes);
1499 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1500 * original annotation is already on the sequence.
1502 if (!seq.hasAnnotation(ann))
1504 seq.addAlignmentAnnotation(copyAnn);
1507 copyAnn.adjustForAlignment();
1508 // add to the alignment and set visible
1509 alignment.addAnnotation(copyAnn);
1510 copyAnn.visible = true;
1516 * Set visibility of alignment annotations of specified types (labels), for
1517 * specified sequences. This supports controls like "Show all secondary
1518 * structure", "Hide all Temp factor", etc.
1520 * @al the alignment to scan for annotations
1522 * the types (labels) of annotations to be updated
1523 * @param forSequences
1524 * if not null, only annotations linked to one of these sequences are
1525 * in scope for update; if null, acts on all sequence annotations
1527 * if this flag is true, 'types' is ignored (label not checked)
1529 * if true, set visibility on, else set off
1531 public static void showOrHideSequenceAnnotations(AlignmentI al,
1532 Collection<String> types, List<SequenceI> forSequences,
1533 boolean anyType, boolean doShow)
1535 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1538 for (AlignmentAnnotation aa : anns)
1540 if (anyType || types.contains(aa.label))
1542 if ((aa.sequenceRef != null) && (forSequences == null
1543 || forSequences.contains(aa.sequenceRef)))
1545 aa.visible = doShow;
1553 * Returns true if either sequence has a cross-reference to the other
1559 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1561 // Note: moved here from class CrossRef as the latter class has dependencies
1562 // not availability to the applet's classpath
1563 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1567 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1568 * that sequence name is structured as Source|AccessionId.
1574 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1576 if (seq1 == null || seq2 == null)
1580 String name = seq2.getName();
1581 final DBRefEntry[] xrefs = seq1.getDBRefs();
1584 for (DBRefEntry xref : xrefs)
1586 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1587 // case-insensitive test, consistent with DBRefEntry.equalRef()
1588 if (xrefName.equalsIgnoreCase(name))
1598 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1599 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1600 * added to the original alignment's dataset, which is shared by the new
1601 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1602 * added to the alignment dataset.
1605 * aligned nucleotide (dna or cds) sequences
1607 * the alignment dataset the sequences belong to
1609 * (optional) to restrict results to CDS that map to specified
1611 * @return an alignment whose sequences are the cds-only parts of the dna
1612 * sequences (or null if no mappings are found)
1614 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1615 AlignmentI dataset, SequenceI[] products)
1617 if (dataset == null || dataset.getDataset() != null)
1619 throw new IllegalArgumentException(
1620 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1622 List<SequenceI> foundSeqs = new ArrayList<>();
1623 List<SequenceI> cdsSeqs = new ArrayList<>();
1624 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1625 HashSet<SequenceI> productSeqs = null;
1626 if (products != null)
1628 productSeqs = new HashSet<>();
1629 for (SequenceI seq : products)
1631 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1632 .getDatasetSequence());
1637 * Construct CDS sequences from mappings on the alignment dataset.
1639 * - find the protein product(s) mapped to from each dna sequence
1640 * - if the mapping covers the whole dna sequence (give or take start/stop
1641 * codon), take the dna as the CDS sequence
1642 * - else search dataset mappings for a suitable dna sequence, i.e. one
1643 * whose whole sequence is mapped to the protein
1644 * - if no sequence found, construct one from the dna sequence and mapping
1645 * (and add it to dataset so it is found if this is repeated)
1647 for (SequenceI dnaSeq : dna)
1649 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1650 : dnaSeq.getDatasetSequence();
1652 List<AlignedCodonFrame> seqMappings = MappingUtils
1653 .findMappingsForSequence(dnaSeq, mappings);
1654 for (AlignedCodonFrame mapping : seqMappings)
1656 List<Mapping> mappingsFromSequence = mapping
1657 .getMappingsFromSequence(dnaSeq);
1659 for (Mapping aMapping : mappingsFromSequence)
1661 MapList mapList = aMapping.getMap();
1662 if (mapList.getFromRatio() == 1)
1665 * not a dna-to-protein mapping (likely dna-to-cds)
1671 * skip if mapping is not to one of the target set of proteins
1673 SequenceI proteinProduct = aMapping.getTo();
1674 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1680 * try to locate the CDS from the dataset mappings;
1681 * guard against duplicate results (for the case that protein has
1682 * dbrefs to both dna and cds sequences)
1684 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1685 seqMappings, aMapping);
1688 if (!foundSeqs.contains(cdsSeq))
1690 foundSeqs.add(cdsSeq);
1691 SequenceI derivedSequence = cdsSeq.deriveSequence();
1692 cdsSeqs.add(derivedSequence);
1693 if (!dataset.getSequences().contains(cdsSeq))
1695 dataset.addSequence(cdsSeq);
1702 * didn't find mapped CDS sequence - construct it and add
1703 * its dataset sequence to the dataset
1705 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1706 dataset).deriveSequence();
1707 // cdsSeq has a name constructed as CDS|<dbref>
1708 // <dbref> will be either the accession for the coding sequence,
1709 // marked in the /via/ dbref to the protein product accession
1710 // or it will be the original nucleotide accession.
1711 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1713 cdsSeqs.add(cdsSeq);
1715 if (!dataset.getSequences().contains(cdsSeqDss))
1717 // check if this sequence is a newly created one
1718 // so needs adding to the dataset
1719 dataset.addSequence(cdsSeqDss);
1723 * add a mapping from CDS to the (unchanged) mapped to range
1725 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1726 cdsSeq.getLength() });
1727 MapList cdsToProteinMap = new MapList(cdsRange,
1728 mapList.getToRanges(), mapList.getFromRatio(),
1729 mapList.getToRatio());
1730 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1731 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1735 * guard against duplicating the mapping if repeating this action
1737 if (!mappings.contains(cdsToProteinMapping))
1739 mappings.add(cdsToProteinMapping);
1742 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1743 proteinProduct, aMapping);
1745 * add another mapping from original 'from' range to CDS
1747 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1748 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1750 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1752 if (!mappings.contains(dnaToCdsMapping))
1754 mappings.add(dnaToCdsMapping);
1758 * transfer dna chromosomal loci (if known) to the CDS
1759 * sequence (via the mapping)
1761 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1762 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1765 * add DBRef with mapping from protein to CDS
1766 * (this enables Get Cross-References from protein alignment)
1767 * This is tricky because we can't have two DBRefs with the
1768 * same source and accession, so need a different accession for
1769 * the CDS from the dna sequence
1772 // specific use case:
1773 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1774 // ENSG02, ENSG03, with transcripts and products similarly named.
1775 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1777 // JBPNote: ?? can't actually create an example that demonstrates we
1779 // synthesize an xref.
1781 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1784 * create a cross-reference from CDS to the source sequence's
1785 * primary reference and vice versa
1787 String source = primRef.getSource();
1788 String version = primRef.getVersion();
1789 DBRefEntry cdsCrossRef = new DBRefEntry(source, source + ":"
1790 + version, primRef.getAccessionId());
1791 cdsCrossRef.setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
1792 cdsSeqDss.addDBRef(cdsCrossRef);
1794 dnaSeq.addDBRef(new DBRefEntry(source, version, cdsSeq
1795 .getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
1797 // problem here is that the cross-reference is synthesized -
1798 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1800 // assuming cds version same as dna ?!?
1802 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
1805 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1807 proteinProduct.addDBRef(proteinToCdsRef);
1811 * transfer any features on dna that overlap the CDS
1813 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1814 SequenceOntologyI.CDS);
1819 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1821 cds.setDataset(dataset);
1827 * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
1828 * toSeq, mediated by the given mapping between the sequences
1831 * @param targetToFrom
1835 protected static void transferGeneLoci(SequenceI fromSeq,
1836 MapList targetToFrom, SequenceI targetSeq)
1838 if (targetSeq.getGeneLoci() != null)
1840 // already have - don't override
1843 GeneLociI fromLoci = fromSeq.getGeneLoci();
1844 if (fromLoci == null)
1849 MapList newMap = targetToFrom.traverse(fromLoci.getMap());
1853 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
1854 fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
1859 * A helper method that finds a CDS sequence in the alignment dataset that is
1860 * mapped to the given protein sequence, and either is, or has a mapping from,
1861 * the given dna sequence.
1864 * set of all mappings on the dataset
1866 * a dna (or cds) sequence we are searching from
1867 * @param seqMappings
1868 * the set of mappings involving dnaSeq
1870 * a transcript-to-peptide mapping
1873 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1874 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1878 * TODO a better dna-cds-protein mapping data representation to allow easy
1879 * navigation; until then this clunky looping around lists of mappings
1881 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1882 : dnaSeq.getDatasetSequence();
1883 SequenceI proteinProduct = aMapping.getTo();
1886 * is this mapping from the whole dna sequence (i.e. CDS)?
1887 * allowing for possible stop codon on dna but not peptide
1889 int mappedFromLength = MappingUtils
1890 .getLength(aMapping.getMap().getFromRanges());
1891 int dnaLength = seqDss.getLength();
1892 if (mappedFromLength == dnaLength
1893 || mappedFromLength == dnaLength - CODON_LENGTH)
1896 * if sequence has CDS features, this is a transcript with no UTR
1897 * - do not take this as the CDS sequence! (JAL-2789)
1899 if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
1907 * looks like we found the dna-to-protein mapping; search for the
1908 * corresponding cds-to-protein mapping
1910 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1911 .findMappingsForSequence(proteinProduct, mappings);
1912 for (AlignedCodonFrame acf : mappingsToPeptide)
1914 for (SequenceToSequenceMapping map : acf.getMappings())
1916 Mapping mapping = map.getMapping();
1917 if (mapping != aMapping
1918 && mapping.getMap().getFromRatio() == CODON_LENGTH
1919 && proteinProduct == mapping.getTo()
1920 && seqDss != map.getFromSeq())
1922 mappedFromLength = MappingUtils
1923 .getLength(mapping.getMap().getFromRanges());
1924 if (mappedFromLength == map.getFromSeq().getLength())
1927 * found a 3:1 mapping to the protein product which covers
1928 * the whole dna sequence i.e. is from CDS; finally check the CDS
1929 * is mapped from the given dna start sequence
1931 SequenceI cdsSeq = map.getFromSeq();
1932 // todo this test is weak if seqMappings contains multiple mappings;
1933 // we get away with it if transcript:cds relationship is 1:1
1934 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1935 .findMappingsForSequence(cdsSeq, seqMappings);
1936 if (!dnaToCdsMaps.isEmpty())
1948 * Helper method that makes a CDS sequence as defined by the mappings from the
1949 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1950 * forward or reverse strand).
1955 * - existing dataset. We check for sequences that look like the CDS
1956 * we are about to construct, if one exists already, then we will
1957 * just return that one.
1958 * @return CDS sequence (as a dataset sequence)
1960 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1963 char[] seqChars = seq.getSequence();
1964 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1965 int cdsWidth = MappingUtils.getLength(fromRanges);
1966 char[] newSeqChars = new char[cdsWidth];
1969 for (int[] range : fromRanges)
1971 if (range[0] <= range[1])
1973 // forward strand mapping - just copy the range
1974 int length = range[1] - range[0] + 1;
1975 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1981 // reverse strand mapping - copy and complement one by one
1982 for (int i = range[0]; i >= range[1]; i--)
1984 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
1990 * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
1991 * else generate a sequence name
1993 String mapFromId = mapping.getMappedFromId();
1994 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
1995 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
1996 if (dataset != null)
1998 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
1999 if (matches != null)
2001 boolean matched = false;
2002 for (SequenceI mtch : matches)
2004 if (mtch.getStart() != newSeq.getStart())
2008 if (mtch.getEnd() != newSeq.getEnd())
2012 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2024 "JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
2030 // newSeq.setDescription(mapFromId);
2036 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2037 * the given mapping.
2041 * @param proteinProduct
2043 * @return list of DBRefEntrys added
2045 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2046 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2049 // gather direct refs from contig congruent with mapping
2050 List<DBRefEntry> direct = new ArrayList<>();
2051 HashSet<String> directSources = new HashSet<>();
2053 if (contig.getDBRefs() != null)
2055 for (DBRefEntry dbr : contig.getDBRefs())
2057 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
2059 MapList map = dbr.getMap().getMap();
2060 // check if map is the CDS mapping
2061 if (mapping.getMap().equals(map))
2064 directSources.add(dbr.getSource());
2069 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2070 proteinProduct.getDBRefs(),
2071 directSources.toArray(new String[0]));
2072 List<DBRefEntry> propagated = new ArrayList<>();
2074 // and generate appropriate mappings
2075 for (DBRefEntry cdsref : direct)
2077 // clone maplist and mapping
2078 MapList cdsposmap = new MapList(
2079 Arrays.asList(new int[][]
2080 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2081 cdsref.getMap().getMap().getToRanges(), 3, 1);
2082 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(),
2083 cdsref.getMap().getMap());
2086 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2087 cdsref.getVersion(), cdsref.getAccessionId(),
2088 new Mapping(cdsmap.getTo(), cdsposmap));
2090 // and see if we can map to the protein product for this mapping.
2091 // onSource is the filtered set of accessions on protein that we are
2092 // tranferring, so we assume accession is the same.
2093 if (cdsmap.getTo() == null && onSource != null)
2095 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2096 cdsref.getAccessionId());
2097 if (sourceRefs != null)
2099 for (DBRefEntry srcref : sourceRefs)
2101 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2103 // we have found a complementary dbref on the protein product, so
2104 // update mapping's getTo
2105 newref.getMap().setTo(proteinProduct);
2110 cdsSeq.addDBRef(newref);
2111 propagated.add(newref);
2117 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2118 * feature start/end ranges, optionally omitting specified feature types.
2119 * Returns the number of features copied.
2124 * the mapping from 'fromSeq' to 'toSeq'
2126 * if not null, only features of this type are copied (including
2127 * subtypes in the Sequence Ontology)
2130 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2131 MapList mapping, String select, String... omitting)
2133 SequenceI copyTo = toSeq;
2134 while (copyTo.getDatasetSequence() != null)
2136 copyTo = copyTo.getDatasetSequence();
2140 * get features, optionally restricted by an ontology term
2142 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2143 .getPositionalFeatures() : fromSeq.getFeatures()
2144 .getFeaturesByOntology(select);
2147 for (SequenceFeature sf : sfs)
2149 String type = sf.getType();
2150 boolean omit = false;
2151 for (String toOmit : omitting)
2153 if (type.equals(toOmit))
2164 * locate the mapped range - null if either start or end is
2165 * not mapped (no partial overlaps are calculated)
2167 int start = sf.getBegin();
2168 int end = sf.getEnd();
2169 int[] mappedTo = mapping.locateInTo(start, end);
2171 * if whole exon range doesn't map, try interpreting it
2172 * as 5' or 3' exon overlapping the CDS range
2174 if (mappedTo == null)
2176 mappedTo = mapping.locateInTo(end, end);
2177 if (mappedTo != null)
2180 * end of exon is in CDS range - 5' overlap
2181 * to a range from the start of the peptide
2186 if (mappedTo == null)
2188 mappedTo = mapping.locateInTo(start, start);
2189 if (mappedTo != null)
2192 * start of exon is in CDS range - 3' overlap
2193 * to a range up to the end of the peptide
2195 mappedTo[1] = toSeq.getLength();
2198 if (mappedTo != null)
2200 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2201 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2202 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2203 sf.getFeatureGroup(), sf.getScore());
2204 copyTo.addSequenceFeature(copy);
2212 * Returns a mapping from dna to protein by inspecting sequence features of
2213 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2214 * length is 3 times the peptide length (optionally after dropping a trailing
2215 * stop codon). This method does not check whether the CDS nucleotide sequence
2216 * translates to the peptide sequence.
2222 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2223 SequenceI proteinSeq)
2225 List<int[]> ranges = findCdsPositions(dnaSeq);
2226 int mappedDnaLength = MappingUtils.getLength(ranges);
2229 * if not a whole number of codons, truncate mapping
2231 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2232 if (codonRemainder > 0)
2234 mappedDnaLength -= codonRemainder;
2235 MappingUtils.removeEndPositions(codonRemainder, ranges);
2238 int proteinLength = proteinSeq.getLength();
2239 int proteinStart = proteinSeq.getStart();
2240 int proteinEnd = proteinSeq.getEnd();
2243 * incomplete start codon may mean X at start of peptide
2244 * we ignore both for mapping purposes
2246 if (proteinSeq.getCharAt(0) == 'X')
2248 // todo JAL-2022 support startPhase > 0
2252 List<int[]> proteinRange = new ArrayList<>();
2255 * dna length should map to protein (or protein plus stop codon)
2257 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2258 if (codesForResidues == (proteinLength + 1))
2260 // assuming extra codon is for STOP and not in peptide
2261 // todo: check trailing codon is indeed a STOP codon
2263 mappedDnaLength -= CODON_LENGTH;
2264 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2267 if (codesForResidues == proteinLength)
2269 proteinRange.add(new int[] { proteinStart, proteinEnd });
2270 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2276 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2277 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2278 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2279 * position order, so this method is only valid for linear CDS in the same
2280 * sense as the protein product.
2285 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2287 List<int[]> result = new ArrayList<>();
2289 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2290 SequenceOntologyI.CDS);
2295 SequenceFeatures.sortFeatures(sfs, true);
2297 for (SequenceFeature sf : sfs)
2302 phase = Integer.parseInt(sf.getPhase());
2303 } catch (NumberFormatException e)
2308 * phase > 0 on first codon means 5' incomplete - skip to the start
2309 * of the next codon; example ENST00000496384
2311 int begin = sf.getBegin();
2312 int end = sf.getEnd();
2313 if (result.isEmpty() && phase > 0)
2318 // shouldn't happen!
2320 .println("Error: start phase extends beyond start CDS in "
2321 + dnaSeq.getName());
2324 result.add(new int[] { begin, end });
2328 * Finally sort ranges by start position. This avoids a dependency on
2329 * keeping features in order on the sequence (if they are in order anyway,
2330 * the sort will have almost no work to do). The implicit assumption is CDS
2331 * ranges are assembled in order. Other cases should not use this method,
2332 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2334 Collections.sort(result, IntRangeComparator.ASCENDING);
2339 * Maps exon features from dna to protein, and computes variants in peptide
2340 * product generated by variants in dna, and adds them as sequence_variant
2341 * features on the protein sequence. Returns the number of variant features
2346 * @param dnaToProtein
2348 public static int computeProteinFeatures(SequenceI dnaSeq,
2349 SequenceI peptide, MapList dnaToProtein)
2351 while (dnaSeq.getDatasetSequence() != null)
2353 dnaSeq = dnaSeq.getDatasetSequence();
2355 while (peptide.getDatasetSequence() != null)
2357 peptide = peptide.getDatasetSequence();
2360 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2363 * compute protein variants from dna variants and codon mappings;
2364 * NB - alternatively we could retrieve this using the REST service e.g.
2365 * http://rest.ensembl.org/overlap/translation
2366 * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2367 * which would be a bit slower but possibly more reliable
2371 * build a map with codon variations for each potentially varying peptide
2373 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2374 dnaSeq, dnaToProtein);
2377 * scan codon variations, compute peptide variants and add to peptide sequence
2380 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2382 int peptidePos = variant.getKey();
2383 List<DnaVariant>[] codonVariants = variant.getValue();
2384 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2391 * Computes non-synonymous peptide variants from codon variants and adds them
2392 * as sequence_variant features on the protein sequence (one feature per
2393 * allele variant). Selected attributes (variant id, clinical significance)
2394 * are copied over to the new features.
2397 * the protein sequence
2399 * the position to compute peptide variants for
2400 * @param codonVariants
2401 * a list of dna variants per codon position
2402 * @return the number of features added
2404 static int computePeptideVariants(SequenceI peptide, int peptidePos,
2405 List<DnaVariant>[] codonVariants)
2407 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2409 String base1 = codonVariants[0].get(0).base;
2410 String base2 = codonVariants[1].get(0).base;
2411 String base3 = codonVariants[2].get(0).base;
2414 * variants in first codon base
2416 for (DnaVariant var : codonVariants[0])
2418 if (var.variant != null)
2420 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2421 if (alleles != null)
2423 for (String base : alleles.split(","))
2425 if (!base1.equalsIgnoreCase(base))
2427 String codon = base.toUpperCase() + base2.toLowerCase()
2428 + base3.toLowerCase();
2429 String canonical = base1.toUpperCase() + base2.toLowerCase()
2430 + base3.toLowerCase();
2431 if (addPeptideVariant(peptide, peptidePos, residue, var,
2443 * variants in second codon base
2445 for (DnaVariant var : codonVariants[1])
2447 if (var.variant != null)
2449 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2450 if (alleles != null)
2452 for (String base : alleles.split(","))
2454 if (!base2.equalsIgnoreCase(base))
2456 String codon = base1.toLowerCase() + base.toUpperCase()
2457 + base3.toLowerCase();
2458 String canonical = base1.toLowerCase() + base2.toUpperCase()
2459 + base3.toLowerCase();
2460 if (addPeptideVariant(peptide, peptidePos, residue, var,
2472 * variants in third codon base
2474 for (DnaVariant var : codonVariants[2])
2476 if (var.variant != null)
2478 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2479 if (alleles != null)
2481 for (String base : alleles.split(","))
2483 if (!base3.equalsIgnoreCase(base))
2485 String codon = base1.toLowerCase() + base2.toLowerCase()
2486 + base.toUpperCase();
2487 String canonical = base1.toLowerCase() + base2.toLowerCase()
2488 + base3.toUpperCase();
2489 if (addPeptideVariant(peptide, peptidePos, residue, var,
2504 * Helper method that adds a peptide variant feature. ID and
2505 * clinical_significance attributes of the dna variant (if present) are copied
2506 * to the new feature.
2513 * the variant codon e.g. aCg
2515 * the 'normal' codon e.g. aTg
2516 * @return true if a feature was added, else false
2518 static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2519 String residue, DnaVariant var, String codon, String canonical)
2522 * get peptide translation of codon e.g. GAT -> D
2523 * note that variants which are not single alleles,
2524 * e.g. multibase variants or HGMD_MUTATION etc
2525 * are currently ignored here
2527 String trans = codon.contains("-") ? null
2528 : (codon.length() > CODON_LENGTH ? null
2529 : ResidueProperties.codonTranslate(codon));
2534 String desc = canonical + "/" + codon;
2535 String featureType = "";
2536 if (trans.equals(residue))
2538 featureType = SequenceOntologyI.SYNONYMOUS_VARIANT;
2540 else if (ResidueProperties.STOP.equals(trans))
2542 featureType = SequenceOntologyI.STOP_GAINED;
2546 String residue3Char = StringUtils
2547 .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2548 String trans3Char = StringUtils
2549 .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2550 desc = "p." + residue3Char + peptidePos + trans3Char;
2551 featureType = SequenceOntologyI.NONSYNONYMOUS_VARIANT;
2553 SequenceFeature sf = new SequenceFeature(featureType, desc, peptidePos,
2554 peptidePos, var.getSource());
2556 StringBuilder attributes = new StringBuilder(32);
2557 String id = (String) var.variant.getValue(VARIANT_ID);
2560 if (id.startsWith(SEQUENCE_VARIANT))
2562 id = id.substring(SEQUENCE_VARIANT.length());
2564 sf.setValue(VARIANT_ID, id);
2565 attributes.append(VARIANT_ID).append("=").append(id);
2566 // TODO handle other species variants JAL-2064
2567 StringBuilder link = new StringBuilder(32);
2570 link.append(desc).append(" ").append(id).append(
2571 "|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2572 .append(URLEncoder.encode(id, "UTF-8"));
2573 sf.addLink(link.toString());
2574 } catch (UnsupportedEncodingException e)
2579 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2580 if (clinSig != null)
2582 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2583 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2586 peptide.addSequenceFeature(sf);
2587 if (attributes.length() > 0)
2589 sf.setAttributes(attributes.toString());
2595 * Builds a map whose key is position in the protein sequence, and value is a
2596 * list of the base and all variants for each corresponding codon position.
2598 * This depends on dna variants being held as a comma-separated list as
2599 * property "alleles" on variant features.
2602 * @param dnaToProtein
2605 @SuppressWarnings("unchecked")
2606 static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2607 SequenceI dnaSeq, MapList dnaToProtein)
2610 * map from peptide position to all variants of the codon which codes for it
2611 * LinkedHashMap ensures we keep the peptide features in sequence order
2613 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<>();
2615 List<SequenceFeature> dnaFeatures = dnaSeq.getFeatures()
2616 .getFeaturesByOntology(SequenceOntologyI.SEQUENCE_VARIANT);
2617 if (dnaFeatures.isEmpty())
2622 int dnaStart = dnaSeq.getStart();
2623 int[] lastCodon = null;
2624 int lastPeptidePostion = 0;
2627 * build a map of codon variations for peptides
2629 for (SequenceFeature sf : dnaFeatures)
2631 int dnaCol = sf.getBegin();
2632 if (dnaCol != sf.getEnd())
2634 // not handling multi-locus variant features
2639 * ignore variant if not a SNP
2641 String alls = (String) sf.getValue(Gff3Helper.ALLELES);
2644 continue; // non-SNP VCF variant perhaps - can't process this
2647 String[] alleles = alls.toUpperCase().split(",");
2648 boolean isSnp = true;
2649 for (String allele : alleles)
2651 if (allele.trim().length() > 1)
2661 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2664 // feature doesn't lie within coding region
2667 int peptidePosition = mapsTo[0];
2668 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2669 if (codonVariants == null)
2671 codonVariants = new ArrayList[CODON_LENGTH];
2672 codonVariants[0] = new ArrayList<>();
2673 codonVariants[1] = new ArrayList<>();
2674 codonVariants[2] = new ArrayList<>();
2675 variants.put(peptidePosition, codonVariants);
2679 * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2681 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2682 : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2683 peptidePosition, peptidePosition));
2684 lastPeptidePostion = peptidePosition;
2688 * save nucleotide (and any variant) for each codon position
2690 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2692 String nucleotide = String.valueOf(
2693 dnaSeq.getCharAt(codon[codonPos] - dnaStart)).toUpperCase();
2694 List<DnaVariant> codonVariant = codonVariants[codonPos];
2695 if (codon[codonPos] == dnaCol)
2697 if (!codonVariant.isEmpty()
2698 && codonVariant.get(0).variant == null)
2701 * already recorded base value, add this variant
2703 codonVariant.get(0).variant = sf;
2708 * add variant with base value
2710 codonVariant.add(new DnaVariant(nucleotide, sf));
2713 else if (codonVariant.isEmpty())
2716 * record (possibly non-varying) base value
2718 codonVariant.add(new DnaVariant(nucleotide));
2726 * Makes an alignment with a copy of the given sequences, adding in any
2727 * non-redundant sequences which are mapped to by the cross-referenced
2733 * the alignment dataset shared by the new copy
2736 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2737 SequenceI[] xrefs, AlignmentI dataset)
2739 AlignmentI copy = new Alignment(new Alignment(seqs));
2740 copy.setDataset(dataset);
2741 boolean isProtein = !copy.isNucleotide();
2742 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2745 for (SequenceI xref : xrefs)
2747 DBRefEntry[] dbrefs = xref.getDBRefs();
2750 for (DBRefEntry dbref : dbrefs)
2752 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2753 || dbref.getMap().getTo().isProtein() != isProtein)
2757 SequenceI mappedTo = dbref.getMap().getTo();
2758 SequenceI match = matcher.findIdMatch(mappedTo);
2761 matcher.add(mappedTo);
2762 copy.addSequence(mappedTo);
2772 * Try to align sequences in 'unaligned' to match the alignment of their
2773 * mapped regions in 'aligned'. For example, could use this to align CDS
2774 * sequences which are mapped to their parent cDNA sequences.
2776 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2777 * dna-to-protein or protein-to-dna use alternative methods.
2780 * sequences to be aligned
2782 * holds aligned sequences and their mappings
2785 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2788 * easy case - aligning a copy of aligned sequences
2790 if (alignAsSameSequences(unaligned, aligned))
2792 return unaligned.getHeight();
2796 * fancy case - aligning via mappings between sequences
2798 List<SequenceI> unmapped = new ArrayList<>();
2799 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2800 unaligned, aligned, unmapped);
2801 int width = columnMap.size();
2802 char gap = unaligned.getGapCharacter();
2803 int realignedCount = 0;
2804 // TODO: verify this loop scales sensibly for very wide/high alignments
2806 for (SequenceI seq : unaligned.getSequences())
2808 if (!unmapped.contains(seq))
2810 char[] newSeq = new char[width];
2811 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2812 // Integer iteration below
2817 * traverse the map to find columns populated
2820 for (Integer column : columnMap.keySet())
2822 Character c = columnMap.get(column).get(seq);
2826 * sequence has a character at this position
2836 * trim trailing gaps
2838 if (lastCol < width)
2840 char[] tmp = new char[lastCol + 1];
2841 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2844 // TODO: optimise SequenceI to avoid char[]->String->char[]
2845 seq.setSequence(String.valueOf(newSeq));
2849 return realignedCount;
2853 * If unaligned and aligned sequences share the same dataset sequences, then
2854 * simply copies the aligned sequences to the unaligned sequences and returns
2855 * true; else returns false
2858 * - sequences to be aligned based on aligned
2860 * - 'guide' alignment containing sequences derived from same dataset
2864 static boolean alignAsSameSequences(AlignmentI unaligned,
2867 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2869 return false; // should only pass alignments with datasets here
2872 // map from dataset sequence to alignment sequence(s)
2873 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2874 for (SequenceI seq : aligned.getSequences())
2876 SequenceI ds = seq.getDatasetSequence();
2877 if (alignedDatasets.get(ds) == null)
2879 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2881 alignedDatasets.get(ds).add(seq);
2885 * first pass - check whether all sequences to be aligned share a dataset
2886 * sequence with an aligned sequence
2888 for (SequenceI seq : unaligned.getSequences())
2890 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2897 * second pass - copy aligned sequences;
2898 * heuristic rule: pair off sequences in order for the case where
2899 * more than one shares the same dataset sequence
2901 for (SequenceI seq : unaligned.getSequences())
2903 List<SequenceI> alignedSequences = alignedDatasets
2904 .get(seq.getDatasetSequence());
2905 // TODO: getSequenceAsString() will be deprecated in the future
2906 // TODO: need to leave to SequenceI implementor to update gaps
2907 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2908 if (alignedSequences.size() > 0)
2910 // pop off aligned sequences (except the last one)
2911 alignedSequences.remove(0);
2919 * Returns a map whose key is alignment column number (base 1), and whose
2920 * values are a map of sequence characters in that column.
2927 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2928 AlignmentI unaligned, AlignmentI aligned,
2929 List<SequenceI> unmapped)
2932 * Map will hold, for each aligned column position, a map of
2933 * {unalignedSequence, characterPerSequence} at that position.
2934 * TreeMap keeps the entries in ascending column order.
2936 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2939 * record any sequences that have no mapping so can't be realigned
2941 unmapped.addAll(unaligned.getSequences());
2943 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2945 for (SequenceI seq : unaligned.getSequences())
2947 for (AlignedCodonFrame mapping : mappings)
2949 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2950 if (fromSeq != null)
2952 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2953 if (addMappedPositions(seq, fromSeq, seqMap, map))
2955 unmapped.remove(seq);
2964 * Helper method that adds to a map the mapped column positions of a sequence.
2966 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2967 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2971 * the sequence whose column positions we are recording
2973 * a sequence that is mapped to the first sequence
2975 * the mapping from 'fromSeq' to 'seq'
2977 * a map to add the column positions (in fromSeq) of the mapped
2981 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2982 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2990 * invert mapping if it is from unaligned to aligned sequence
2992 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2994 seqMap = new Mapping(seq.getDatasetSequence(),
2995 seqMap.getMap().getInverse());
2998 int toStart = seq.getStart();
3001 * traverse [start, end, start, end...] ranges in fromSeq
3003 for (int[] fromRange : seqMap.getMap().getFromRanges())
3005 for (int i = 0; i < fromRange.length - 1; i += 2)
3007 boolean forward = fromRange[i + 1] >= fromRange[i];
3010 * find the range mapped to (sequence positions base 1)
3012 int[] range = seqMap.locateMappedRange(fromRange[i],
3016 System.err.println("Error in mapping " + seqMap + " from "
3017 + fromSeq.getName());
3020 int fromCol = fromSeq.findIndex(fromRange[i]);
3021 int mappedCharPos = range[0];
3024 * walk over the 'from' aligned sequence in forward or reverse
3025 * direction; when a non-gap is found, record the column position
3026 * of the next character of the mapped-to sequence; stop when all
3027 * the characters of the range have been counted
3029 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
3032 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
3035 * mapped from sequence has a character in this column
3036 * record the column position for the mapped to character
3038 Map<SequenceI, Character> seqsMap = map.get(fromCol);
3039 if (seqsMap == null)
3041 seqsMap = new HashMap<>();
3042 map.put(fromCol, seqsMap);
3044 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
3047 fromCol += (forward ? 1 : -1);
3054 // strictly temporary hack until proper criteria for aligning protein to cds
3055 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
3056 public static boolean looksLikeEnsembl(AlignmentI alignment)
3058 for (SequenceI seq : alignment.getSequences())
3060 String name = seq.getName();
3061 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))