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 java.util.ArrayList;
24 import java.util.Arrays;
25 import java.util.Collection;
26 import java.util.Collections;
27 import java.util.HashMap;
28 import java.util.HashSet;
29 import java.util.Iterator;
30 import java.util.LinkedHashMap;
31 import java.util.List;
33 import java.util.Map.Entry;
34 import java.util.NoSuchElementException;
36 import java.util.SortedMap;
37 import java.util.TreeMap;
39 import jalview.bin.Cache;
40 import jalview.commands.RemoveGapColCommand;
41 import jalview.datamodel.AlignedCodon;
42 import jalview.datamodel.AlignedCodonFrame;
43 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
44 import jalview.datamodel.Alignment;
45 import jalview.datamodel.AlignmentAnnotation;
46 import jalview.datamodel.AlignmentI;
47 import jalview.datamodel.DBRefEntry;
48 import jalview.datamodel.GeneLociI;
49 import jalview.datamodel.IncompleteCodonException;
50 import jalview.datamodel.Mapping;
51 import jalview.datamodel.Sequence;
52 import jalview.datamodel.SequenceFeature;
53 import jalview.datamodel.SequenceGroup;
54 import jalview.datamodel.SequenceI;
55 import jalview.datamodel.features.SequenceFeatures;
56 import jalview.io.gff.SequenceOntologyI;
57 import jalview.schemes.ResidueProperties;
58 import jalview.util.Comparison;
59 import jalview.util.DBRefUtils;
60 import jalview.util.IntRangeComparator;
61 import jalview.util.MapList;
62 import jalview.util.MappingUtils;
65 * grab bag of useful alignment manipulation operations Expect these to be
66 * refactored elsewhere at some point.
71 public class AlignmentUtils
73 private static final int CODON_LENGTH = 3;
75 private static final String SEQUENCE_VARIANT = "sequence_variant:";
78 * the 'id' attribute is provided for variant features fetched from
79 * Ensembl using its REST service with JSON format
81 public static final String VARIANT_ID = "id";
84 * A data model to hold the 'normal' base value at a position, and an optional
85 * sequence variant feature
87 static final class DnaVariant
91 SequenceFeature variant;
93 DnaVariant(String nuc)
99 DnaVariant(String nuc, SequenceFeature var)
105 public String getSource()
107 return variant == null ? null : variant.getFeatureGroup();
111 * toString for aid in the debugger only
114 public String toString()
116 return base + ":" + (variant == null ? "" : variant.getDescription());
121 * given an existing alignment, create a new alignment including all, or up to
122 * flankSize additional symbols from each sequence's dataset sequence
128 public static AlignmentI expandContext(AlignmentI core, int flankSize)
130 List<SequenceI> sq = new ArrayList<>();
132 for (SequenceI s : core.getSequences())
134 SequenceI newSeq = s.deriveSequence();
135 final int newSeqStart = newSeq.getStart() - 1;
136 if (newSeqStart > maxoffset
137 && newSeq.getDatasetSequence().getStart() < s.getStart())
139 maxoffset = newSeqStart;
145 maxoffset = Math.min(maxoffset, flankSize);
149 * now add offset left and right to create an expanded alignment
151 for (SequenceI s : sq)
154 while (ds.getDatasetSequence() != null)
156 ds = ds.getDatasetSequence();
158 int s_end = s.findPosition(s.getStart() + s.getLength());
159 // find available flanking residues for sequence
160 int ustream_ds = s.getStart() - ds.getStart();
161 int dstream_ds = ds.getEnd() - s_end;
163 // build new flanked sequence
165 // compute gap padding to start of flanking sequence
166 int offset = maxoffset - ustream_ds;
168 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
171 if (flankSize < ustream_ds)
173 // take up to flankSize residues
174 offset = maxoffset - flankSize;
175 ustream_ds = flankSize;
177 if (flankSize <= dstream_ds)
179 dstream_ds = flankSize - 1;
182 // TODO use Character.toLowerCase to avoid creating String objects?
183 char[] upstream = new String(ds
184 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
185 .toLowerCase().toCharArray();
186 char[] downstream = new String(
187 ds.getSequence(s_end - 1, s_end + dstream_ds)).toLowerCase()
189 char[] coreseq = s.getSequence();
190 char[] nseq = new char[offset + upstream.length + downstream.length
192 char c = core.getGapCharacter();
195 for (; p < offset; p++)
200 System.arraycopy(upstream, 0, nseq, p, upstream.length);
201 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
203 System.arraycopy(downstream, 0, nseq,
204 p + coreseq.length + upstream.length, downstream.length);
205 s.setSequence(new String(nseq));
206 s.setStart(s.getStart() - ustream_ds);
207 s.setEnd(s_end + downstream.length);
209 AlignmentI newAl = new jalview.datamodel.Alignment(
210 sq.toArray(new SequenceI[0]));
211 for (SequenceI s : sq)
213 if (s.getAnnotation() != null)
215 for (AlignmentAnnotation aa : s.getAnnotation())
217 aa.adjustForAlignment(); // JAL-1712 fix
218 newAl.addAnnotation(aa);
222 newAl.setDataset(core.getDataset());
227 * Returns the index (zero-based position) of a sequence in an alignment, or
234 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
238 for (SequenceI alSeq : al.getSequences())
251 * Returns a map of lists of sequences in the alignment, keyed by sequence
252 * name. For use in mapping between different alignment views of the same
255 * @see jalview.datamodel.AlignmentI#getSequencesByName()
257 public static Map<String, List<SequenceI>> getSequencesByName(
260 Map<String, List<SequenceI>> theMap = new LinkedHashMap<>();
261 for (SequenceI seq : al.getSequences())
263 String name = seq.getName();
266 List<SequenceI> seqs = theMap.get(name);
269 seqs = new ArrayList<>();
270 theMap.put(name, seqs);
279 * Build mapping of protein to cDNA alignment. Mappings are made between
280 * sequences where the cDNA translates to the protein sequence. Any new
281 * mappings are added to the protein alignment. Returns true if any mappings
282 * either already exist or were added, else false.
284 * @param proteinAlignment
285 * @param cdnaAlignment
288 public static boolean mapProteinAlignmentToCdna(
289 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
291 if (proteinAlignment == null || cdnaAlignment == null)
296 Set<SequenceI> mappedDna = new HashSet<>();
297 Set<SequenceI> mappedProtein = new HashSet<>();
300 * First pass - map sequences where cross-references exist. This include
301 * 1-to-many mappings to support, for example, variant cDNA.
303 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
304 cdnaAlignment, mappedDna, mappedProtein, true);
307 * Second pass - map sequences where no cross-references exist. This only
308 * does 1-to-1 mappings and assumes corresponding sequences are in the same
309 * order in the alignments.
311 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
312 mappedDna, mappedProtein, false);
313 return mappingPerformed;
317 * Make mappings between compatible sequences (where the cDNA translation
318 * matches the protein).
320 * @param proteinAlignment
321 * @param cdnaAlignment
323 * a set of mapped DNA sequences (to add to)
324 * @param mappedProtein
325 * a set of mapped Protein sequences (to add to)
327 * if true, only map sequences where xrefs exist
330 protected static boolean mapProteinToCdna(
331 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
332 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
335 boolean mappingExistsOrAdded = false;
336 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
337 for (SequenceI aaSeq : thisSeqs)
339 boolean proteinMapped = false;
340 AlignedCodonFrame acf = new AlignedCodonFrame();
342 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
345 * Always try to map if sequences have xref to each other; this supports
346 * variant cDNA or alternative splicing for a protein sequence.
348 * If no xrefs, try to map progressively, assuming that alignments have
349 * mappable sequences in corresponding order. These are not
350 * many-to-many, as that would risk mixing species with similar cDNA
353 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
359 * Don't map non-xrefd sequences more than once each. This heuristic
360 * allows us to pair up similar sequences in ordered alignments.
362 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
363 || mappedDna.contains(cdnaSeq)))
367 if (mappingExists(proteinAlignment.getCodonFrames(),
368 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
370 mappingExistsOrAdded = true;
374 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
377 acf.addMap(cdnaSeq, aaSeq, map);
378 mappingExistsOrAdded = true;
379 proteinMapped = true;
380 mappedDna.add(cdnaSeq);
381 mappedProtein.add(aaSeq);
387 proteinAlignment.addCodonFrame(acf);
390 return mappingExistsOrAdded;
394 * Answers true if the mappings include one between the given (dataset)
397 protected static boolean mappingExists(List<AlignedCodonFrame> mappings,
398 SequenceI aaSeq, SequenceI cdnaSeq)
400 if (mappings != null)
402 for (AlignedCodonFrame acf : mappings)
404 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
414 * Builds a mapping (if possible) of a cDNA to a protein sequence.
416 * <li>first checks if the cdna translates exactly to the protein
418 * <li>else checks for translation after removing a STOP codon</li>
419 * <li>else checks for translation after removing a START codon</li>
420 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
422 * Returns null if no mapping is determined.
425 * the aligned protein sequence
427 * the aligned cdna sequence
430 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
434 * Here we handle either dataset sequence set (desktop) or absent (applet).
435 * Use only the char[] form of the sequence to avoid creating possibly large
438 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
439 char[] aaSeqChars = proteinDataset != null
440 ? proteinDataset.getSequence()
441 : proteinSeq.getSequence();
442 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
443 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
444 : cdnaSeq.getSequence();
445 if (aaSeqChars == null || cdnaSeqChars == null)
451 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
453 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
454 int cdnaLength = cdnaSeqChars.length;
455 int cdnaStart = cdnaSeq.getStart();
456 int cdnaEnd = cdnaSeq.getEnd();
457 final int proteinStart = proteinSeq.getStart();
458 final int proteinEnd = proteinSeq.getEnd();
461 * If lengths don't match, try ignoring stop codon (if present)
463 if (cdnaLength != mappedLength && cdnaLength > 2)
465 String lastCodon = String.valueOf(cdnaSeqChars,
466 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
467 for (String stop : ResidueProperties.STOP_CODONS)
469 if (lastCodon.equals(stop))
471 cdnaEnd -= CODON_LENGTH;
472 cdnaLength -= CODON_LENGTH;
479 * If lengths still don't match, try ignoring start codon.
482 if (cdnaLength != mappedLength && cdnaLength > 2
483 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
484 .equals(ResidueProperties.START))
486 startOffset += CODON_LENGTH;
487 cdnaStart += CODON_LENGTH;
488 cdnaLength -= CODON_LENGTH;
491 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
494 * protein is translation of dna (+/- start/stop codons)
496 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
498 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
503 * translation failed - try mapping CDS annotated regions of dna
505 return mapCdsToProtein(cdnaSeq, proteinSeq);
509 * Test whether the given cdna sequence, starting at the given offset,
510 * translates to the given amino acid sequence, using the standard translation
511 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
513 * @param cdnaSeqChars
518 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
521 if (cdnaSeqChars == null || aaSeqChars == null)
527 int dnaPos = cdnaStart;
528 for (; dnaPos < cdnaSeqChars.length - 2
529 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
531 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
532 final String translated = ResidueProperties.codonTranslate(codon);
535 * allow * in protein to match untranslatable in dna
537 final char aaRes = aaSeqChars[aaPos];
538 if ((translated == null || ResidueProperties.STOP.equals(translated))
543 if (translated == null || !(aaRes == translated.charAt(0)))
546 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
547 // + codon + "(" + translated + ") != " + aaRes));
553 * check we matched all of the protein sequence
555 if (aaPos != aaSeqChars.length)
561 * check we matched all of the dna except
562 * for optional trailing STOP codon
564 if (dnaPos == cdnaSeqChars.length)
568 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
570 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
571 if (ResidueProperties.STOP
572 .equals(ResidueProperties.codonTranslate(codon)))
581 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
582 * currently assumes that we are aligning cDNA to match protein.
585 * the sequence to be realigned
587 * the alignment whose sequence alignment is to be 'copied'
589 * character string represent a gap in the realigned sequence
590 * @param preserveUnmappedGaps
591 * @param preserveMappedGaps
592 * @return true if the sequence was realigned, false if it could not be
594 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
595 String gap, boolean preserveMappedGaps,
596 boolean preserveUnmappedGaps)
599 * Get any mappings from the source alignment to the target (dataset)
602 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
603 // all mappings. Would it help to constrain this?
604 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
605 if (mappings == null || mappings.isEmpty())
611 * Locate the aligned source sequence whose dataset sequence is mapped. We
612 * just take the first match here (as we can't align like more than one
615 SequenceI alignFrom = null;
616 AlignedCodonFrame mapping = null;
617 for (AlignedCodonFrame mp : mappings)
619 alignFrom = mp.findAlignedSequence(seq, al);
620 if (alignFrom != null)
627 if (alignFrom == null)
631 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
632 preserveMappedGaps, preserveUnmappedGaps);
637 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
638 * match residues and codons. Flags control whether existing gaps in unmapped
639 * (intron) and mapped (exon) regions are preserved or not. Gaps between
640 * intron and exon are only retained if both flags are set.
647 * @param preserveUnmappedGaps
648 * @param preserveMappedGaps
650 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
651 AlignedCodonFrame mapping, String myGap, char sourceGap,
652 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
654 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
656 // aligned and dataset sequence positions, all base zero
660 int basesWritten = 0;
661 char myGapChar = myGap.charAt(0);
662 int ratio = myGap.length();
664 int fromOffset = alignFrom.getStart() - 1;
665 int toOffset = alignTo.getStart() - 1;
666 int sourceGapMappedLength = 0;
667 boolean inExon = false;
668 final int toLength = alignTo.getLength();
669 final int fromLength = alignFrom.getLength();
670 StringBuilder thisAligned = new StringBuilder(2 * toLength);
673 * Traverse the 'model' aligned sequence
675 for (int i = 0; i < fromLength; i++)
677 char sourceChar = alignFrom.getCharAt(i);
678 if (sourceChar == sourceGap)
680 sourceGapMappedLength += ratio;
685 * Found a non-gap character. Locate its mapped region if any.
688 // Note mapping positions are base 1, our sequence positions base 0
689 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
690 sourceDsPos + fromOffset);
691 if (mappedPos == null)
694 * unmapped position; treat like a gap
696 sourceGapMappedLength += ratio;
697 // System.err.println("Can't align: no codon mapping to residue "
698 // + sourceDsPos + "(" + sourceChar + ")");
703 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
704 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
705 StringBuilder trailingCopiedGap = new StringBuilder();
708 * Copy dna sequence up to and including this codon. Optionally, include
709 * gaps before the codon starts (in introns) and/or after the codon starts
712 * Note this only works for 'linear' splicing, not reverse or interleaved.
713 * But then 'align dna as protein' doesn't make much sense otherwise.
715 int intronLength = 0;
716 while (basesWritten + toOffset < mappedCodonEnd
717 && thisSeqPos < toLength)
719 final char c = alignTo.getCharAt(thisSeqPos++);
723 int sourcePosition = basesWritten + toOffset;
724 if (sourcePosition < mappedCodonStart)
727 * Found an unmapped (intron) base. First add in any preceding gaps
730 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
732 thisAligned.append(trailingCopiedGap.toString());
733 intronLength += trailingCopiedGap.length();
734 trailingCopiedGap = new StringBuilder();
741 final boolean startOfCodon = sourcePosition == mappedCodonStart;
742 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
743 preserveUnmappedGaps, sourceGapMappedLength, inExon,
744 trailingCopiedGap.length(), intronLength, startOfCodon);
745 for (int k = 0; k < gapsToAdd; k++)
747 thisAligned.append(myGapChar);
749 sourceGapMappedLength = 0;
752 thisAligned.append(c);
753 trailingCopiedGap = new StringBuilder();
757 if (inExon && preserveMappedGaps)
759 trailingCopiedGap.append(myGapChar);
761 else if (!inExon && preserveUnmappedGaps)
763 trailingCopiedGap.append(myGapChar);
770 * At end of model aligned sequence. Copy any remaining target sequence, optionally
771 * including (intron) gaps.
773 while (thisSeqPos < toLength)
775 final char c = alignTo.getCharAt(thisSeqPos++);
776 if (c != myGapChar || preserveUnmappedGaps)
778 thisAligned.append(c);
780 sourceGapMappedLength--;
784 * finally add gaps to pad for any trailing source gaps or
785 * unmapped characters
787 if (preserveUnmappedGaps)
789 while (sourceGapMappedLength > 0)
791 thisAligned.append(myGapChar);
792 sourceGapMappedLength--;
797 * All done aligning, set the aligned sequence.
799 alignTo.setSequence(new String(thisAligned));
803 * Helper method to work out how many gaps to insert when realigning.
805 * @param preserveMappedGaps
806 * @param preserveUnmappedGaps
807 * @param sourceGapMappedLength
809 * @param trailingCopiedGap
810 * @param intronLength
811 * @param startOfCodon
814 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
815 boolean preserveUnmappedGaps, int sourceGapMappedLength,
816 boolean inExon, int trailingGapLength, int intronLength,
817 final boolean startOfCodon)
823 * Reached start of codon. Ignore trailing gaps in intron unless we are
824 * preserving gaps in both exon and intron. Ignore them anyway if the
825 * protein alignment introduces a gap at least as large as the intronic
828 if (inExon && !preserveMappedGaps)
830 trailingGapLength = 0;
832 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
834 trailingGapLength = 0;
838 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
842 if (intronLength + trailingGapLength <= sourceGapMappedLength)
844 gapsToAdd = sourceGapMappedLength - intronLength;
848 gapsToAdd = Math.min(
849 intronLength + trailingGapLength - sourceGapMappedLength,
857 * second or third base of codon; check for any gaps in dna
859 if (!preserveMappedGaps)
861 trailingGapLength = 0;
863 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
869 * Realigns the given protein to match the alignment of the dna, using codon
870 * mappings to translate aligned codon positions to protein residues.
873 * the alignment whose sequences are realigned by this method
875 * the dna alignment whose alignment we are 'copying'
876 * @return the number of sequences that were realigned
878 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
880 if (protein.isNucleotide() || !dna.isNucleotide())
882 System.err.println("Wrong alignment type in alignProteinAsDna");
885 List<SequenceI> unmappedProtein = new ArrayList<>();
886 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
887 protein, dna, unmappedProtein);
888 return alignProteinAs(protein, alignedCodons, unmappedProtein);
892 * Realigns the given dna to match the alignment of the protein, using codon
893 * mappings to translate aligned peptide positions to codons.
895 * Always produces a padded CDS alignment.
898 * the alignment whose sequences are realigned by this method
900 * the protein alignment whose alignment we are 'copying'
901 * @return the number of sequences that were realigned
903 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
905 if (protein.isNucleotide() || !dna.isNucleotide())
907 System.err.println("Wrong alignment type in alignProteinAsDna");
910 // todo: implement this
911 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
912 int alignedCount = 0;
913 int width = 0; // alignment width for padding CDS
914 for (SequenceI dnaSeq : dna.getSequences())
916 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
917 dna.getGapCharacter()))
921 width = Math.max(dnaSeq.getLength(), width);
925 for (SequenceI dnaSeq : dna.getSequences())
927 oldwidth = dnaSeq.getLength();
928 diff = width - oldwidth;
931 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
938 * Helper method to align (if possible) the dna sequence to match the
939 * alignment of a mapped protein sequence. This is currently limited to
940 * handling coding sequence only.
948 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
949 AlignmentI protein, List<AlignedCodonFrame> mappings,
952 SequenceI cdsDss = cdsSeq.getDatasetSequence();
956 .println("alignCdsSequenceAsProtein needs aligned sequence!");
960 List<AlignedCodonFrame> dnaMappings = MappingUtils
961 .findMappingsForSequence(cdsSeq, mappings);
962 for (AlignedCodonFrame mapping : dnaMappings)
964 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
967 final int peptideLength = peptide.getLength();
968 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
971 MapList mapList = map.getMap();
972 if (map.getTo() == peptide.getDatasetSequence())
974 mapList = mapList.getInverse();
976 final int cdsLength = cdsDss.getLength();
977 int mappedFromLength = MappingUtils.getLength(mapList
979 int mappedToLength = MappingUtils
980 .getLength(mapList.getToRanges());
981 boolean addStopCodon = (cdsLength == mappedFromLength
982 * CODON_LENGTH + CODON_LENGTH)
983 || (peptide.getDatasetSequence()
984 .getLength() == mappedFromLength - 1);
985 if (cdsLength != mappedToLength && !addStopCodon)
987 System.err.println(String.format(
988 "Can't align cds as protein (length mismatch %d/%d): %s",
989 cdsLength, mappedToLength, cdsSeq.getName()));
993 * pre-fill the aligned cds sequence with gaps
995 char[] alignedCds = new char[peptideLength * CODON_LENGTH
996 + (addStopCodon ? CODON_LENGTH : 0)];
997 Arrays.fill(alignedCds, gapChar);
1000 * walk over the aligned peptide sequence and insert mapped
1001 * codons for residues in the aligned cds sequence
1003 int copiedBases = 0;
1004 int cdsStart = cdsDss.getStart();
1005 int proteinPos = peptide.getStart() - 1;
1008 for (int col = 0; col < peptideLength; col++)
1010 char residue = peptide.getCharAt(col);
1012 if (Comparison.isGap(residue))
1014 cdsCol += CODON_LENGTH;
1019 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1022 // e.g. incomplete start codon, X in peptide
1023 cdsCol += CODON_LENGTH;
1027 for (int j = codon[0]; j <= codon[1]; j++)
1029 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1030 alignedCds[cdsCol++] = mappedBase;
1038 * append stop codon if not mapped from protein,
1039 * closing it up to the end of the mapped sequence
1041 if (copiedBases == cdsLength - CODON_LENGTH)
1043 for (int i = alignedCds.length - 1; i >= 0; i--)
1045 if (!Comparison.isGap(alignedCds[i]))
1047 cdsCol = i + 1; // gap just after end of sequence
1051 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1053 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1056 cdsSeq.setSequence(new String(alignedCds));
1065 * Builds a map whose key is an aligned codon position (3 alignment column
1066 * numbers base 0), and whose value is a map from protein sequence to each
1067 * protein's peptide residue for that codon. The map generates an ordering of
1068 * the codons, and allows us to read off the peptides at each position in
1069 * order to assemble 'aligned' protein sequences.
1072 * the protein alignment
1074 * the coding dna alignment
1075 * @param unmappedProtein
1076 * any unmapped proteins are added to this list
1079 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1080 AlignmentI protein, AlignmentI dna,
1081 List<SequenceI> unmappedProtein)
1084 * maintain a list of any proteins with no mappings - these will be
1085 * rendered 'as is' in the protein alignment as we can't align them
1087 unmappedProtein.addAll(protein.getSequences());
1089 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1092 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1093 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1094 * comparator keeps the codon positions ordered.
1096 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<>(
1097 new CodonComparator());
1099 for (SequenceI dnaSeq : dna.getSequences())
1101 for (AlignedCodonFrame mapping : mappings)
1103 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1106 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1107 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1109 unmappedProtein.remove(prot);
1115 * Finally add any unmapped peptide start residues (e.g. for incomplete
1116 * codons) as if at the codon position before the second residue
1118 // TODO resolve JAL-2022 so this fudge can be removed
1119 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1120 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1122 return alignedCodons;
1126 * Scans for any protein mapped from position 2 (meaning unmapped start
1127 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1128 * preceding position in the alignment
1130 * @param alignedCodons
1131 * the codon-to-peptide map
1132 * @param mappedSequenceCount
1133 * the number of distinct sequences in the map
1135 protected static void addUnmappedPeptideStarts(
1136 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1137 int mappedSequenceCount)
1139 // TODO delete this ugly hack once JAL-2022 is resolved
1140 // i.e. we can model startPhase > 0 (incomplete start codon)
1142 List<SequenceI> sequencesChecked = new ArrayList<>();
1143 AlignedCodon lastCodon = null;
1144 Map<SequenceI, AlignedCodon> toAdd = new HashMap<>();
1146 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1149 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1152 SequenceI seq = sequenceCodon.getKey();
1153 if (sequencesChecked.contains(seq))
1157 sequencesChecked.add(seq);
1158 AlignedCodon codon = sequenceCodon.getValue();
1159 if (codon.peptideCol > 1)
1162 "Problem mapping protein with >1 unmapped start positions: "
1165 else if (codon.peptideCol == 1)
1168 * first position (peptideCol == 0) was unmapped - add it
1170 if (lastCodon != null)
1172 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1173 lastCodon.pos2, lastCodon.pos3,
1174 String.valueOf(seq.getCharAt(0)), 0);
1175 toAdd.put(seq, firstPeptide);
1180 * unmapped residue at start of alignment (no prior column) -
1181 * 'insert' at nominal codon [0, 0, 0]
1183 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1184 String.valueOf(seq.getCharAt(0)), 0);
1185 toAdd.put(seq, firstPeptide);
1188 if (sequencesChecked.size() == mappedSequenceCount)
1190 // no need to check past first mapped position in all sequences
1194 lastCodon = entry.getKey();
1198 * add any new codons safely after iterating over the map
1200 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1202 addCodonToMap(alignedCodons, startCodon.getValue(),
1203 startCodon.getKey());
1208 * Update the aligned protein sequences to match the codon alignments given in
1212 * @param alignedCodons
1213 * an ordered map of codon positions (columns), with sequence/peptide
1214 * values present in each column
1215 * @param unmappedProtein
1218 protected static int alignProteinAs(AlignmentI protein,
1219 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1220 List<SequenceI> unmappedProtein)
1223 * prefill peptide sequences with gaps
1225 int alignedWidth = alignedCodons.size();
1226 char[] gaps = new char[alignedWidth];
1227 Arrays.fill(gaps, protein.getGapCharacter());
1228 Map<SequenceI, char[]> peptides = new HashMap<>();
1229 for (SequenceI seq : protein.getSequences())
1231 if (!unmappedProtein.contains(seq))
1233 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1238 * Traverse the codons left to right (as defined by CodonComparator)
1239 * and insert peptides in each column where the sequence is mapped.
1240 * This gives a peptide 'alignment' where residues are aligned if their
1241 * corresponding codons occupy the same columns in the cdna alignment.
1244 for (AlignedCodon codon : alignedCodons.keySet())
1246 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1248 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1250 char residue = entry.getValue().product.charAt(0);
1251 peptides.get(entry.getKey())[column] = residue;
1257 * and finally set the constructed sequences
1259 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1261 entry.getKey().setSequence(new String(entry.getValue()));
1268 * Populate the map of aligned codons by traversing the given sequence
1269 * mapping, locating the aligned positions of mapped codons, and adding those
1270 * positions and their translation products to the map.
1273 * the aligned sequence we are mapping from
1275 * the sequence to be aligned to the codons
1277 * the gap character in the dna sequence
1279 * a mapping to a sequence translation
1280 * @param alignedCodons
1281 * the map we are building up
1283 static void addCodonPositions(SequenceI dna, SequenceI protein,
1284 char gapChar, Mapping seqMap,
1285 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1287 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1290 * add codon positions, and their peptide translations, to the alignment
1291 * map, while remembering the first codon mapped
1293 while (codons.hasNext())
1297 AlignedCodon codon = codons.next();
1298 addCodonToMap(alignedCodons, codon, protein);
1299 } catch (IncompleteCodonException e)
1301 // possible incomplete trailing codon - ignore
1302 } catch (NoSuchElementException e)
1304 // possibly peptide lacking STOP
1310 * Helper method to add a codon-to-peptide entry to the aligned codons map
1312 * @param alignedCodons
1316 protected static void addCodonToMap(
1317 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1318 AlignedCodon codon, SequenceI protein)
1320 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1321 if (seqProduct == null)
1323 seqProduct = new HashMap<>();
1324 alignedCodons.put(codon, seqProduct);
1326 seqProduct.put(protein, codon);
1330 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1331 * between at least one pair of sequences in the two alignments. Currently,
1334 * <li>One alignment must be nucleotide, and the other protein</li>
1335 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1336 * <li>Mappable means the nucleotide translation matches the protein
1338 * <li>The translation may ignore start and stop codons if present in the
1346 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1348 if (al1 == null || al2 == null)
1354 * Require one nucleotide and one protein
1356 if (al1.isNucleotide() == al2.isNucleotide())
1360 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1361 AlignmentI protein = dna == al1 ? al2 : al1;
1362 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1363 for (SequenceI dnaSeq : dna.getSequences())
1365 for (SequenceI proteinSeq : protein.getSequences())
1367 if (isMappable(dnaSeq, proteinSeq, mappings))
1377 * Returns true if the dna sequence is mapped, or could be mapped, to the
1385 protected static boolean isMappable(SequenceI dnaSeq,
1386 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1388 if (dnaSeq == null || proteinSeq == null)
1393 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1394 : dnaSeq.getDatasetSequence();
1395 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1397 : proteinSeq.getDatasetSequence();
1399 for (AlignedCodonFrame mapping : mappings)
1401 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1411 * Just try to make a mapping (it is not yet stored), test whether
1414 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1418 * Finds any reference annotations associated with the sequences in
1419 * sequenceScope, that are not already added to the alignment, and adds them
1420 * to the 'candidates' map. Also populates a lookup table of annotation
1421 * labels, keyed by calcId, for use in constructing tooltips or the like.
1423 * @param sequenceScope
1424 * the sequences to scan for reference annotations
1425 * @param labelForCalcId
1426 * (optional) map to populate with label for calcId
1428 * map to populate with annotations for sequence
1430 * the alignment to check for presence of annotations
1432 public static void findAddableReferenceAnnotations(
1433 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1434 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1437 if (sequenceScope == null)
1443 * For each sequence in scope, make a list of any annotations on the
1444 * underlying dataset sequence which are not already on the alignment.
1446 * Add to a map of { alignmentSequence, <List of annotations to add> }
1448 for (SequenceI seq : sequenceScope)
1450 SequenceI dataset = seq.getDatasetSequence();
1451 if (dataset == null)
1455 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1456 if (datasetAnnotations == null)
1460 final List<AlignmentAnnotation> result = new ArrayList<>();
1461 for (AlignmentAnnotation dsann : datasetAnnotations)
1464 * Find matching annotations on the alignment. If none is found, then
1465 * add this annotation to the list of 'addable' annotations for this
1468 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1469 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1470 if (!matchedAlignmentAnnotations.iterator().hasNext())
1473 if (labelForCalcId != null)
1475 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1480 * Save any addable annotations for this sequence
1482 if (!result.isEmpty())
1484 candidates.put(seq, result);
1490 * Adds annotations to the top of the alignment annotations, in the same order
1491 * as their related sequences.
1493 * @param annotations
1494 * the annotations to add
1496 * the alignment to add them to
1497 * @param selectionGroup
1498 * current selection group (or null if none)
1500 public static void addReferenceAnnotations(
1501 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1502 final AlignmentI alignment, final SequenceGroup selectionGroup)
1504 for (SequenceI seq : annotations.keySet())
1506 for (AlignmentAnnotation ann : annotations.get(seq))
1508 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1510 int endRes = ann.annotations.length;
1511 if (selectionGroup != null)
1513 startRes = selectionGroup.getStartRes();
1514 endRes = selectionGroup.getEndRes();
1516 copyAnn.restrict(startRes, endRes);
1519 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1520 * original annotation is already on the sequence.
1522 if (!seq.hasAnnotation(ann))
1524 seq.addAlignmentAnnotation(copyAnn);
1527 copyAnn.adjustForAlignment();
1528 // add to the alignment and set visible
1529 alignment.addAnnotation(copyAnn);
1530 copyAnn.visible = true;
1536 * Set visibility of alignment annotations of specified types (labels), for
1537 * specified sequences. This supports controls like "Show all secondary
1538 * structure", "Hide all Temp factor", etc.
1540 * @al the alignment to scan for annotations
1542 * the types (labels) of annotations to be updated
1543 * @param forSequences
1544 * if not null, only annotations linked to one of these sequences are
1545 * in scope for update; if null, acts on all sequence annotations
1547 * if this flag is true, 'types' is ignored (label not checked)
1549 * if true, set visibility on, else set off
1551 public static void showOrHideSequenceAnnotations(AlignmentI al,
1552 Collection<String> types, List<SequenceI> forSequences,
1553 boolean anyType, boolean doShow)
1555 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1558 for (AlignmentAnnotation aa : anns)
1560 if (anyType || types.contains(aa.label))
1562 if ((aa.sequenceRef != null) && (forSequences == null
1563 || forSequences.contains(aa.sequenceRef)))
1565 aa.visible = doShow;
1573 * Returns true if either sequence has a cross-reference to the other
1579 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1581 // Note: moved here from class CrossRef as the latter class has dependencies
1582 // not availability to the applet's classpath
1583 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1587 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1588 * that sequence name is structured as Source|AccessionId.
1594 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1596 if (seq1 == null || seq2 == null)
1600 String name = seq2.getName();
1601 final List<DBRefEntry> xrefs = seq1.getDBRefs();
1604 for (int ix = 0, nx = xrefs.size(); ix < nx; ix++)
1606 DBRefEntry xref = xrefs.get(ix);
1607 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1608 // case-insensitive test, consistent with DBRefEntry.equalRef()
1609 if (xrefName.equalsIgnoreCase(name))
1619 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1620 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1621 * added to the original alignment's dataset, which is shared by the new
1622 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1623 * added to the alignment dataset.
1626 * aligned nucleotide (dna or cds) sequences
1628 * the alignment dataset the sequences belong to
1630 * (optional) to restrict results to CDS that map to specified
1632 * @return an alignment whose sequences are the cds-only parts of the dna
1633 * sequences (or null if no mappings are found)
1635 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1636 AlignmentI dataset, SequenceI[] products)
1638 if (dataset == null || dataset.getDataset() != null)
1640 throw new IllegalArgumentException(
1641 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1643 List<SequenceI> foundSeqs = new ArrayList<>();
1644 List<SequenceI> cdsSeqs = new ArrayList<>();
1645 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1646 HashSet<SequenceI> productSeqs = null;
1647 if (products != null)
1649 productSeqs = new HashSet<>();
1650 for (SequenceI seq : products)
1652 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1653 .getDatasetSequence());
1658 * Construct CDS sequences from mappings on the alignment dataset.
1660 * - find the protein product(s) mapped to from each dna sequence
1661 * - if the mapping covers the whole dna sequence (give or take start/stop
1662 * codon), take the dna as the CDS sequence
1663 * - else search dataset mappings for a suitable dna sequence, i.e. one
1664 * whose whole sequence is mapped to the protein
1665 * - if no sequence found, construct one from the dna sequence and mapping
1666 * (and add it to dataset so it is found if this is repeated)
1668 for (SequenceI dnaSeq : dna)
1670 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1671 : dnaSeq.getDatasetSequence();
1673 List<AlignedCodonFrame> seqMappings = MappingUtils
1674 .findMappingsForSequence(dnaSeq, mappings);
1675 for (AlignedCodonFrame mapping : seqMappings)
1677 List<Mapping> mappingsFromSequence = mapping
1678 .getMappingsFromSequence(dnaSeq);
1680 for (Mapping aMapping : mappingsFromSequence)
1682 MapList mapList = aMapping.getMap();
1683 if (mapList.getFromRatio() == 1)
1686 * not a dna-to-protein mapping (likely dna-to-cds)
1692 * skip if mapping is not to one of the target set of proteins
1694 SequenceI proteinProduct = aMapping.getTo();
1695 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1701 * try to locate the CDS from the dataset mappings;
1702 * guard against duplicate results (for the case that protein has
1703 * dbrefs to both dna and cds sequences)
1705 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1706 seqMappings, aMapping);
1709 if (!foundSeqs.contains(cdsSeq))
1711 foundSeqs.add(cdsSeq);
1712 SequenceI derivedSequence = cdsSeq.deriveSequence();
1713 cdsSeqs.add(derivedSequence);
1714 if (!dataset.getSequences().contains(cdsSeq))
1716 dataset.addSequence(cdsSeq);
1723 * didn't find mapped CDS sequence - construct it and add
1724 * its dataset sequence to the dataset
1726 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1727 dataset).deriveSequence();
1728 // cdsSeq has a name constructed as CDS|<dbref>
1729 // <dbref> will be either the accession for the coding sequence,
1730 // marked in the /via/ dbref to the protein product accession
1731 // or it will be the original nucleotide accession.
1732 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1734 cdsSeqs.add(cdsSeq);
1737 * build the mapping from CDS to protein
1739 List<int[]> cdsRange = Collections
1740 .singletonList(new int[]
1741 { cdsSeq.getStart(),
1742 cdsSeq.getLength() + cdsSeq.getStart() - 1 });
1743 MapList cdsToProteinMap = new MapList(cdsRange,
1744 mapList.getToRanges(), mapList.getFromRatio(),
1745 mapList.getToRatio());
1747 if (!dataset.getSequences().contains(cdsSeqDss))
1750 * if this sequence is a newly created one, add it to the dataset
1751 * and made a CDS to protein mapping (if sequence already exists,
1752 * CDS-to-protein mapping _is_ the transcript-to-protein mapping)
1754 dataset.addSequence(cdsSeqDss);
1755 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1756 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1760 * guard against duplicating the mapping if repeating this action
1762 if (!mappings.contains(cdsToProteinMapping))
1764 mappings.add(cdsToProteinMapping);
1768 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1769 proteinProduct, aMapping);
1771 * add another mapping from original 'from' range to CDS
1773 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1774 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1776 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1778 if (!mappings.contains(dnaToCdsMapping))
1780 mappings.add(dnaToCdsMapping);
1784 * transfer dna chromosomal loci (if known) to the CDS
1785 * sequence (via the mapping)
1787 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1788 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1791 * add DBRef with mapping from protein to CDS
1792 * (this enables Get Cross-References from protein alignment)
1793 * This is tricky because we can't have two DBRefs with the
1794 * same source and accession, so need a different accession for
1795 * the CDS from the dna sequence
1798 // specific use case:
1799 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1800 // ENSG02, ENSG03, with transcripts and products similarly named.
1801 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1803 // JBPNote: ?? can't actually create an example that demonstrates we
1805 // synthesize an xref.
1807 List<DBRefEntry> primrefs = dnaDss.getPrimaryDBRefs();
1808 for (int ip = 0, np = primrefs.size(); ip < np; ip++)
1810 DBRefEntry primRef = primrefs.get(ip);
1812 * create a cross-reference from CDS to the source sequence's
1813 * primary reference and vice versa
1815 String source = primRef.getSource();
1816 String version = primRef.getVersion();
1817 DBRefEntry cdsCrossRef = new DBRefEntry(source, source + ":"
1818 + version, primRef.getAccessionId());
1819 cdsCrossRef.setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
1820 cdsSeqDss.addDBRef(cdsCrossRef);
1822 dnaSeq.addDBRef(new DBRefEntry(source, version, cdsSeq
1823 .getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
1824 // problem here is that the cross-reference is synthesized -
1825 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1827 // assuming cds version same as dna ?!?
1829 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
1832 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1834 proteinProduct.addDBRef(proteinToCdsRef);
1837 * transfer any features on dna that overlap the CDS
1839 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1840 SequenceOntologyI.CDS);
1845 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1847 cds.setDataset(dataset);
1853 * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
1854 * toSeq, mediated by the given mapping between the sequences
1857 * @param targetToFrom
1861 protected static void transferGeneLoci(SequenceI fromSeq,
1862 MapList targetToFrom, SequenceI targetSeq)
1864 if (targetSeq.getGeneLoci() != null)
1866 // already have - don't override
1869 GeneLociI fromLoci = fromSeq.getGeneLoci();
1870 if (fromLoci == null)
1875 MapList newMap = targetToFrom.traverse(fromLoci.getMapping());
1879 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
1880 fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
1885 * A helper method that finds a CDS sequence in the alignment dataset that is
1886 * mapped to the given protein sequence, and either is, or has a mapping from,
1887 * the given dna sequence.
1890 * set of all mappings on the dataset
1892 * a dna (or cds) sequence we are searching from
1893 * @param seqMappings
1894 * the set of mappings involving dnaSeq
1896 * a transcript-to-peptide mapping
1899 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1900 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1904 * TODO a better dna-cds-protein mapping data representation to allow easy
1905 * navigation; until then this clunky looping around lists of mappings
1907 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1908 : dnaSeq.getDatasetSequence();
1909 SequenceI proteinProduct = aMapping.getTo();
1912 * is this mapping from the whole dna sequence (i.e. CDS)?
1913 * allowing for possible stop codon on dna but not peptide
1915 int mappedFromLength = MappingUtils
1916 .getLength(aMapping.getMap().getFromRanges());
1917 int dnaLength = seqDss.getLength();
1918 if (mappedFromLength == dnaLength
1919 || mappedFromLength == dnaLength - CODON_LENGTH)
1922 * if sequence has CDS features, this is a transcript with no UTR
1923 * - do not take this as the CDS sequence! (JAL-2789)
1925 if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
1933 * looks like we found the dna-to-protein mapping; search for the
1934 * corresponding cds-to-protein mapping
1936 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1937 .findMappingsForSequence(proteinProduct, mappings);
1938 for (AlignedCodonFrame acf : mappingsToPeptide)
1940 for (SequenceToSequenceMapping map : acf.getMappings())
1942 Mapping mapping = map.getMapping();
1943 if (mapping != aMapping
1944 && mapping.getMap().getFromRatio() == CODON_LENGTH
1945 && proteinProduct == mapping.getTo()
1946 && seqDss != map.getFromSeq())
1948 mappedFromLength = MappingUtils
1949 .getLength(mapping.getMap().getFromRanges());
1950 if (mappedFromLength == map.getFromSeq().getLength())
1953 * found a 3:1 mapping to the protein product which covers
1954 * the whole dna sequence i.e. is from CDS; finally check the CDS
1955 * is mapped from the given dna start sequence
1957 SequenceI cdsSeq = map.getFromSeq();
1958 // todo this test is weak if seqMappings contains multiple mappings;
1959 // we get away with it if transcript:cds relationship is 1:1
1960 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1961 .findMappingsForSequence(cdsSeq, seqMappings);
1962 if (!dnaToCdsMaps.isEmpty())
1974 * Helper method that makes a CDS sequence as defined by the mappings from the
1975 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1976 * forward or reverse strand).
1981 * - existing dataset. We check for sequences that look like the CDS
1982 * we are about to construct, if one exists already, then we will
1983 * just return that one.
1984 * @return CDS sequence (as a dataset sequence)
1986 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1990 * construct CDS sequence name as "CDS|" with 'from id' held in the mapping
1991 * if set (e.g. EMBL protein_id), else sequence name appended
1993 String mapFromId = mapping.getMappedFromId();
1994 final String seqId = "CDS|"
1995 + (mapFromId != null ? mapFromId : seq.getName());
1997 SequenceI newSeq = null;
2000 * construct CDS sequence by splicing mapped from ranges
2002 char[] seqChars = seq.getSequence();
2003 List<int[]> fromRanges = mapping.getMap().getFromRanges();
2004 int cdsWidth = MappingUtils.getLength(fromRanges);
2005 char[] newSeqChars = new char[cdsWidth];
2008 for (int[] range : fromRanges)
2010 if (range[0] <= range[1])
2012 // forward strand mapping - just copy the range
2013 int length = range[1] - range[0] + 1;
2014 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
2020 // reverse strand mapping - copy and complement one by one
2021 for (int i = range[0]; i >= range[1]; i--)
2023 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
2027 newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2030 if (dataset != null)
2032 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2033 if (matches != null)
2035 boolean matched = false;
2036 for (SequenceI mtch : matches)
2038 if (mtch.getStart() != newSeq.getStart())
2042 if (mtch.getEnd() != newSeq.getEnd())
2046 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2058 "JAL-2154 regression: warning - found (and ignored) a duplicate CDS sequence:" + mtch.toString());
2063 // newSeq.setDescription(mapFromId);
2069 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2070 * the given mapping.
2074 * @param proteinProduct
2076 * @return list of DBRefEntrys added
2078 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2079 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2082 // gather direct refs from contig congruent with mapping
2083 List<DBRefEntry> direct = new ArrayList<>();
2084 HashSet<String> directSources = new HashSet<>();
2086 List<DBRefEntry> refs = contig.getDBRefs();
2089 for (int ib = 0, nb = refs.size(); ib < nb; ib++)
2091 DBRefEntry dbr = refs.get(ib);
2093 if (dbr.hasMap() && (map = dbr.getMap().getMap()).isTripletMap())
2095 // check if map is the CDS mapping
2096 if (mapping.getMap().equals(map))
2099 directSources.add(dbr.getSource());
2104 List<DBRefEntry> onSource = DBRefUtils.selectRefs(
2105 proteinProduct.getDBRefs(),
2106 directSources.toArray(new String[0]));
2107 List<DBRefEntry> propagated = new ArrayList<>();
2109 // and generate appropriate mappings
2110 for (int ic = 0, nc = direct.size(); ic < nc; ic++)
2112 DBRefEntry cdsref = direct.get(ic);
2113 Mapping m = cdsref.getMap();
2114 // clone maplist and mapping
2115 MapList cdsposmap = new MapList(
2116 Arrays.asList(new int[][]
2117 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2118 m.getMap().getToRanges(), 3, 1);
2119 Mapping cdsmap = new Mapping(m.getTo(), m.getMap());
2122 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2123 cdsref.getVersion(), cdsref.getAccessionId(),
2124 new Mapping(cdsmap.getTo(), cdsposmap));
2126 // and see if we can map to the protein product for this mapping.
2127 // onSource is the filtered set of accessions on protein that we are
2128 // tranferring, so we assume accession is the same.
2129 if (cdsmap.getTo() == null && onSource != null)
2131 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2132 cdsref.getAccessionId());
2133 if (sourceRefs != null)
2135 for (DBRefEntry srcref : sourceRefs)
2137 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2139 // we have found a complementary dbref on the protein product, so
2140 // update mapping's getTo
2141 newref.getMap().setTo(proteinProduct);
2146 cdsSeq.addDBRef(newref);
2147 propagated.add(newref);
2153 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2154 * feature start/end ranges, optionally omitting specified feature types.
2155 * Returns the number of features copied.
2160 * the mapping from 'fromSeq' to 'toSeq'
2162 * if not null, only features of this type are copied (including
2163 * subtypes in the Sequence Ontology)
2166 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2167 MapList mapping, String select, String... omitting)
2169 SequenceI copyTo = toSeq;
2170 while (copyTo.getDatasetSequence() != null)
2172 copyTo = copyTo.getDatasetSequence();
2174 if (fromSeq == copyTo || fromSeq.getDatasetSequence() == copyTo)
2176 return 0; // shared dataset sequence
2180 * get features, optionally restricted by an ontology term
2182 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2183 .getPositionalFeatures() : fromSeq.getFeatures()
2184 .getFeaturesByOntology(select);
2187 for (SequenceFeature sf : sfs)
2189 String type = sf.getType();
2190 boolean omit = false;
2191 for (String toOmit : omitting)
2193 if (type.equals(toOmit))
2204 * locate the mapped range - null if either start or end is
2205 * not mapped (no partial overlaps are calculated)
2207 int start = sf.getBegin();
2208 int end = sf.getEnd();
2209 int[] mappedTo = mapping.locateInTo(start, end);
2211 * if whole exon range doesn't map, try interpreting it
2212 * as 5' or 3' exon overlapping the CDS range
2214 if (mappedTo == null)
2216 mappedTo = mapping.locateInTo(end, end);
2217 if (mappedTo != null)
2220 * end of exon is in CDS range - 5' overlap
2221 * to a range from the start of the peptide
2226 if (mappedTo == null)
2228 mappedTo = mapping.locateInTo(start, start);
2229 if (mappedTo != null)
2232 * start of exon is in CDS range - 3' overlap
2233 * to a range up to the end of the peptide
2235 mappedTo[1] = toSeq.getLength();
2238 if (mappedTo != null)
2240 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2241 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2242 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2243 sf.getFeatureGroup(), sf.getScore());
2244 copyTo.addSequenceFeature(copy);
2252 * Returns a mapping from dna to protein by inspecting sequence features of
2253 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2254 * length is 3 times the peptide length (optionally after dropping a trailing
2255 * stop codon). This method does not check whether the CDS nucleotide sequence
2256 * translates to the peptide sequence.
2262 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2263 SequenceI proteinSeq)
2265 List<int[]> ranges = findCdsPositions(dnaSeq);
2266 int mappedDnaLength = MappingUtils.getLength(ranges);
2269 * if not a whole number of codons, truncate mapping
2271 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2272 if (codonRemainder > 0)
2274 mappedDnaLength -= codonRemainder;
2275 MappingUtils.removeEndPositions(codonRemainder, ranges);
2278 int proteinLength = proteinSeq.getLength();
2279 int proteinStart = proteinSeq.getStart();
2280 int proteinEnd = proteinSeq.getEnd();
2283 * incomplete start codon may mean X at start of peptide
2284 * we ignore both for mapping purposes
2286 if (proteinSeq.getCharAt(0) == 'X')
2288 // todo JAL-2022 support startPhase > 0
2292 List<int[]> proteinRange = new ArrayList<>();
2295 * dna length should map to protein (or protein plus stop codon)
2297 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2298 if (codesForResidues == (proteinLength + 1))
2300 // assuming extra codon is for STOP and not in peptide
2301 // todo: check trailing codon is indeed a STOP codon
2303 mappedDnaLength -= CODON_LENGTH;
2304 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2307 if (codesForResidues == proteinLength)
2309 proteinRange.add(new int[] { proteinStart, proteinEnd });
2310 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2316 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2317 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2318 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2319 * position order, so this method is only valid for linear CDS in the same
2320 * sense as the protein product.
2325 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2327 List<int[]> result = new ArrayList<>();
2329 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2330 SequenceOntologyI.CDS);
2335 SequenceFeatures.sortFeatures(sfs, true);
2337 for (SequenceFeature sf : sfs)
2342 String s = sf.getPhase();
2345 phase = Integer.parseInt(s);
2347 } catch (NumberFormatException e)
2352 * phase > 0 on first codon means 5' incomplete - skip to the start
2353 * of the next codon; example ENST00000496384
2355 int begin = sf.getBegin();
2356 int end = sf.getEnd();
2357 if (result.isEmpty() && phase > 0)
2362 // shouldn't happen!
2364 .println("Error: start phase extends beyond start CDS in "
2365 + dnaSeq.getName());
2368 result.add(new int[] { begin, end });
2372 * Finally sort ranges by start position. This avoids a dependency on
2373 * keeping features in order on the sequence (if they are in order anyway,
2374 * the sort will have almost no work to do). The implicit assumption is CDS
2375 * ranges are assembled in order. Other cases should not use this method,
2376 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2378 Collections.sort(result, IntRangeComparator.ASCENDING);
2383 * Makes an alignment with a copy of the given sequences, adding in any
2384 * non-redundant sequences which are mapped to by the cross-referenced
2390 * the alignment dataset shared by the new copy
2393 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2394 SequenceI[] xrefs, AlignmentI dataset)
2396 AlignmentI copy = new Alignment(new Alignment(seqs));
2397 copy.setDataset(dataset);
2398 boolean isProtein = !copy.isNucleotide();
2399 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2402 // BH 2019.01.25 recoded to remove iterators
2404 for (int ix = 0, nx = xrefs.length; ix < nx; ix++)
2406 SequenceI xref = xrefs[ix];
2407 List<DBRefEntry> dbrefs = xref.getDBRefs();
2410 for (int ir = 0, nir = dbrefs.size(); ir < nir; ir++)
2412 DBRefEntry dbref = dbrefs.get(ir);
2413 Mapping map = dbref.getMap();
2415 if (map == null || (mto = map.getTo()) == null
2416 || mto.isProtein() != isProtein)
2420 SequenceI mappedTo = mto;
2421 SequenceI match = matcher.findIdMatch(mappedTo);
2424 matcher.add(mappedTo);
2425 copy.addSequence(mappedTo);
2435 * Try to align sequences in 'unaligned' to match the alignment of their
2436 * mapped regions in 'aligned'. For example, could use this to align CDS
2437 * sequences which are mapped to their parent cDNA sequences.
2439 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2440 * dna-to-protein or protein-to-dna use alternative methods.
2443 * sequences to be aligned
2445 * holds aligned sequences and their mappings
2448 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2451 * easy case - aligning a copy of aligned sequences
2453 if (alignAsSameSequences(unaligned, aligned))
2455 return unaligned.getHeight();
2459 * fancy case - aligning via mappings between sequences
2461 List<SequenceI> unmapped = new ArrayList<>();
2462 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2463 unaligned, aligned, unmapped);
2464 int width = columnMap.size();
2465 char gap = unaligned.getGapCharacter();
2466 int realignedCount = 0;
2467 // TODO: verify this loop scales sensibly for very wide/high alignments
2469 for (SequenceI seq : unaligned.getSequences())
2471 if (!unmapped.contains(seq))
2473 char[] newSeq = new char[width];
2474 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2475 // Integer iteration below
2480 * traverse the map to find columns populated
2483 for (Integer column : columnMap.keySet())
2485 Character c = columnMap.get(column).get(seq);
2489 * sequence has a character at this position
2499 * trim trailing gaps
2501 if (lastCol < width)
2503 char[] tmp = new char[lastCol + 1];
2504 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2507 // TODO: optimise SequenceI to avoid char[]->String->char[]
2508 seq.setSequence(String.valueOf(newSeq));
2512 return realignedCount;
2516 * If unaligned and aligned sequences share the same dataset sequences, then
2517 * simply copies the aligned sequences to the unaligned sequences and returns
2518 * true; else returns false
2521 * - sequences to be aligned based on aligned
2523 * - 'guide' alignment containing sequences derived from same
2524 * dataset as unaligned
2527 static boolean alignAsSameSequences(AlignmentI unaligned,
2530 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2532 return false; // should only pass alignments with datasets here
2535 // map from dataset sequence to alignment sequence(s)
2536 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2537 for (SequenceI seq : aligned.getSequences())
2539 SequenceI ds = seq.getDatasetSequence();
2540 if (alignedDatasets.get(ds) == null)
2542 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2544 alignedDatasets.get(ds).add(seq);
2548 * first pass - check whether all sequences to be aligned share a
2549 * dataset sequence with an aligned sequence; also note the leftmost
2550 * ungapped column from which to copy
2552 int leftmost = Integer.MAX_VALUE;
2553 for (SequenceI seq : unaligned.getSequences())
2555 final SequenceI ds = seq.getDatasetSequence();
2556 if (!alignedDatasets.containsKey(ds))
2560 SequenceI alignedSeq = alignedDatasets.get(ds)
2562 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2563 leftmost = Math.min(leftmost, startCol);
2567 * second pass - copy aligned sequences;
2568 * heuristic rule: pair off sequences in order for the case where
2569 * more than one shares the same dataset sequence
2571 final char gapCharacter = aligned.getGapCharacter();
2572 for (SequenceI seq : unaligned.getSequences())
2574 List<SequenceI> alignedSequences = alignedDatasets
2575 .get(seq.getDatasetSequence());
2576 if (alignedSequences.isEmpty())
2579 * defensive check - shouldn't happen! (JAL-3536)
2583 SequenceI alignedSeq = alignedSequences.get(0);
2586 * gap fill for leading (5') UTR if any
2588 // TODO this copies intron columns - wrong!
2589 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2590 int endCol = alignedSeq.findIndex(seq.getEnd());
2591 char[] seqchars = new char[endCol - leftmost + 1];
2592 Arrays.fill(seqchars, gapCharacter);
2593 char[] toCopy = alignedSeq.getSequence(startCol - 1, endCol);
2594 System.arraycopy(toCopy, 0, seqchars, startCol - leftmost,
2596 seq.setSequence(String.valueOf(seqchars));
2597 if (alignedSequences.size() > 0)
2599 // pop off aligned sequences (except the last one)
2600 alignedSequences.remove(0);
2605 * finally remove gapped columns (e.g. introns)
2607 new RemoveGapColCommand("", unaligned.getSequencesArray(), 0,
2608 unaligned.getWidth() - 1, unaligned);
2614 * Returns a map whose key is alignment column number (base 1), and whose
2615 * values are a map of sequence characters in that column.
2622 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2623 AlignmentI unaligned, AlignmentI aligned,
2624 List<SequenceI> unmapped)
2627 * Map will hold, for each aligned column position, a map of
2628 * {unalignedSequence, characterPerSequence} at that position.
2629 * TreeMap keeps the entries in ascending column order.
2631 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2634 * record any sequences that have no mapping so can't be realigned
2636 unmapped.addAll(unaligned.getSequences());
2638 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2640 for (SequenceI seq : unaligned.getSequences())
2642 for (AlignedCodonFrame mapping : mappings)
2644 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2645 if (fromSeq != null)
2647 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2648 if (addMappedPositions(seq, fromSeq, seqMap, map))
2650 unmapped.remove(seq);
2659 * Helper method that adds to a map the mapped column positions of a sequence.
2661 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2662 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2666 * the sequence whose column positions we are recording
2668 * a sequence that is mapped to the first sequence
2670 * the mapping from 'fromSeq' to 'seq'
2672 * a map to add the column positions (in fromSeq) of the mapped
2676 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2677 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2685 * invert mapping if it is from unaligned to aligned sequence
2687 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2689 seqMap = new Mapping(seq.getDatasetSequence(),
2690 seqMap.getMap().getInverse());
2693 int toStart = seq.getStart();
2696 * traverse [start, end, start, end...] ranges in fromSeq
2698 for (int[] fromRange : seqMap.getMap().getFromRanges())
2700 for (int i = 0; i < fromRange.length - 1; i += 2)
2702 boolean forward = fromRange[i + 1] >= fromRange[i];
2705 * find the range mapped to (sequence positions base 1)
2707 int[] range = seqMap.locateMappedRange(fromRange[i],
2711 System.err.println("Error in mapping " + seqMap + " from "
2712 + fromSeq.getName());
2715 int fromCol = fromSeq.findIndex(fromRange[i]);
2716 int mappedCharPos = range[0];
2719 * walk over the 'from' aligned sequence in forward or reverse
2720 * direction; when a non-gap is found, record the column position
2721 * of the next character of the mapped-to sequence; stop when all
2722 * the characters of the range have been counted
2724 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
2727 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
2730 * mapped from sequence has a character in this column
2731 * record the column position for the mapped to character
2733 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2734 if (seqsMap == null)
2736 seqsMap = new HashMap<>();
2737 map.put(fromCol, seqsMap);
2739 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
2742 fromCol += (forward ? 1 : -1);
2749 // strictly temporary hack until proper criteria for aligning protein to cds
2750 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2751 public static boolean looksLikeEnsembl(AlignmentI alignment)
2753 for (SequenceI seq : alignment.getSequences())
2755 String name = seq.getName();
2756 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))