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.Locale;
25 import jalview.commands.RemoveGapColCommand;
26 import jalview.datamodel.AlignedCodon;
27 import jalview.datamodel.AlignedCodonFrame;
28 import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
29 import jalview.datamodel.Alignment;
30 import jalview.datamodel.AlignmentAnnotation;
31 import jalview.datamodel.AlignmentI;
32 import jalview.datamodel.DBRefEntry;
33 import jalview.datamodel.GeneLociI;
34 import jalview.datamodel.IncompleteCodonException;
35 import jalview.datamodel.Mapping;
36 import jalview.datamodel.Sequence;
37 import jalview.datamodel.SequenceFeature;
38 import jalview.datamodel.SequenceGroup;
39 import jalview.datamodel.SequenceI;
40 import jalview.datamodel.features.SequenceFeatures;
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;
49 import java.util.ArrayList;
50 import java.util.Arrays;
51 import java.util.Collection;
52 import java.util.Collections;
53 import java.util.HashMap;
54 import java.util.HashSet;
55 import java.util.Iterator;
56 import java.util.LinkedHashMap;
57 import java.util.List;
59 import java.util.Map.Entry;
60 import java.util.NoSuchElementException;
62 import java.util.SortedMap;
63 import java.util.TreeMap;
66 * grab bag of useful alignment manipulation operations Expect these to be
67 * refactored elsewhere at some point.
72 public class AlignmentUtils
74 private static final int CODON_LENGTH = 3;
76 private static final String SEQUENCE_VARIANT = "sequence_variant:";
79 * the 'id' attribute is provided for variant features fetched from
80 * Ensembl using its REST service with JSON format
82 public static final String VARIANT_ID = "id";
85 * A data model to hold the 'normal' base value at a position, and an optional
86 * sequence variant feature
88 static final class DnaVariant
92 SequenceFeature variant;
94 DnaVariant(String nuc)
100 DnaVariant(String nuc, SequenceFeature var)
106 public String getSource()
108 return variant == null ? null : variant.getFeatureGroup();
112 * toString for aid in the debugger only
115 public String toString()
117 return base + ":" + (variant == null ? "" : variant.getDescription());
122 * given an existing alignment, create a new alignment including all, or up to
123 * flankSize additional symbols from each sequence's dataset sequence
129 public static AlignmentI expandContext(AlignmentI core, int flankSize)
131 List<SequenceI> sq = new ArrayList<>();
133 for (SequenceI s : core.getSequences())
135 SequenceI newSeq = s.deriveSequence();
136 final int newSeqStart = newSeq.getStart() - 1;
137 if (newSeqStart > maxoffset
138 && newSeq.getDatasetSequence().getStart() < s.getStart())
140 maxoffset = newSeqStart;
146 maxoffset = Math.min(maxoffset, flankSize);
150 * now add offset left and right to create an expanded alignment
152 for (SequenceI s : sq)
155 while (ds.getDatasetSequence() != null)
157 ds = ds.getDatasetSequence();
159 int s_end = s.findPosition(s.getStart() + s.getLength());
160 // find available flanking residues for sequence
161 int ustream_ds = s.getStart() - ds.getStart();
162 int dstream_ds = ds.getEnd() - s_end;
164 // build new flanked sequence
166 // compute gap padding to start of flanking sequence
167 int offset = maxoffset - ustream_ds;
169 // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
172 if (flankSize < ustream_ds)
174 // take up to flankSize residues
175 offset = maxoffset - flankSize;
176 ustream_ds = flankSize;
178 if (flankSize <= dstream_ds)
180 dstream_ds = flankSize - 1;
183 // TODO use Character.toLowerCase to avoid creating String objects?
184 char[] upstream = new String(ds
185 .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
186 .toLowerCase(Locale.ROOT).toCharArray();
187 char[] downstream = new String(
188 ds.getSequence(s_end - 1, s_end + dstream_ds)).toLowerCase(Locale.ROOT)
190 char[] coreseq = s.getSequence();
191 char[] nseq = new char[offset + upstream.length + downstream.length
193 char c = core.getGapCharacter();
196 for (; p < offset; p++)
201 System.arraycopy(upstream, 0, nseq, p, upstream.length);
202 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
204 System.arraycopy(downstream, 0, nseq,
205 p + coreseq.length + upstream.length, downstream.length);
206 s.setSequence(new String(nseq));
207 s.setStart(s.getStart() - ustream_ds);
208 s.setEnd(s_end + downstream.length);
210 AlignmentI newAl = new jalview.datamodel.Alignment(
211 sq.toArray(new SequenceI[0]));
212 for (SequenceI s : sq)
214 if (s.getAnnotation() != null)
216 for (AlignmentAnnotation aa : s.getAnnotation())
218 aa.adjustForAlignment(); // JAL-1712 fix
219 newAl.addAnnotation(aa);
223 newAl.setDataset(core.getDataset());
228 * Returns the index (zero-based position) of a sequence in an alignment, or
235 public static int getSequenceIndex(AlignmentI al, SequenceI seq)
239 for (SequenceI alSeq : al.getSequences())
252 * Returns a map of lists of sequences in the alignment, keyed by sequence
253 * name. For use in mapping between different alignment views of the same
256 * @see jalview.datamodel.AlignmentI#getSequencesByName()
258 public static Map<String, List<SequenceI>> getSequencesByName(
261 Map<String, List<SequenceI>> theMap = new LinkedHashMap<>();
262 for (SequenceI seq : al.getSequences())
264 String name = seq.getName();
267 List<SequenceI> seqs = theMap.get(name);
270 seqs = new ArrayList<>();
271 theMap.put(name, seqs);
280 * Build mapping of protein to cDNA alignment. Mappings are made between
281 * sequences where the cDNA translates to the protein sequence. Any new
282 * mappings are added to the protein alignment. Returns true if any mappings
283 * either already exist or were added, else false.
285 * @param proteinAlignment
286 * @param cdnaAlignment
289 public static boolean mapProteinAlignmentToCdna(
290 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
292 if (proteinAlignment == null || cdnaAlignment == null)
297 Set<SequenceI> mappedDna = new HashSet<>();
298 Set<SequenceI> mappedProtein = new HashSet<>();
301 * First pass - map sequences where cross-references exist. This include
302 * 1-to-many mappings to support, for example, variant cDNA.
304 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
305 cdnaAlignment, mappedDna, mappedProtein, true);
308 * Second pass - map sequences where no cross-references exist. This only
309 * does 1-to-1 mappings and assumes corresponding sequences are in the same
310 * order in the alignments.
312 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
313 mappedDna, mappedProtein, false);
314 return mappingPerformed;
318 * Make mappings between compatible sequences (where the cDNA translation
319 * matches the protein).
321 * @param proteinAlignment
322 * @param cdnaAlignment
324 * a set of mapped DNA sequences (to add to)
325 * @param mappedProtein
326 * a set of mapped Protein sequences (to add to)
328 * if true, only map sequences where xrefs exist
331 protected static boolean mapProteinToCdna(
332 final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
333 Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
336 boolean mappingExistsOrAdded = false;
337 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
338 for (SequenceI aaSeq : thisSeqs)
340 boolean proteinMapped = false;
341 AlignedCodonFrame acf = new AlignedCodonFrame();
343 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
346 * Always try to map if sequences have xref to each other; this supports
347 * variant cDNA or alternative splicing for a protein sequence.
349 * If no xrefs, try to map progressively, assuming that alignments have
350 * mappable sequences in corresponding order. These are not
351 * many-to-many, as that would risk mixing species with similar cDNA
354 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
360 * Don't map non-xrefd sequences more than once each. This heuristic
361 * allows us to pair up similar sequences in ordered alignments.
363 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
364 || mappedDna.contains(cdnaSeq)))
368 if (mappingExists(proteinAlignment.getCodonFrames(),
369 aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
371 mappingExistsOrAdded = true;
375 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
378 acf.addMap(cdnaSeq, aaSeq, map);
379 mappingExistsOrAdded = true;
380 proteinMapped = true;
381 mappedDna.add(cdnaSeq);
382 mappedProtein.add(aaSeq);
388 proteinAlignment.addCodonFrame(acf);
391 return mappingExistsOrAdded;
395 * Answers true if the mappings include one between the given (dataset)
398 protected static boolean mappingExists(List<AlignedCodonFrame> mappings,
399 SequenceI aaSeq, SequenceI cdnaSeq)
401 if (mappings != null)
403 for (AlignedCodonFrame acf : mappings)
405 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
415 * Builds a mapping (if possible) of a cDNA to a protein sequence.
417 * <li>first checks if the cdna translates exactly to the protein
419 * <li>else checks for translation after removing a STOP codon</li>
420 * <li>else checks for translation after removing a START codon</li>
421 * <li>if that fails, inspect CDS features on the cDNA sequence</li>
423 * Returns null if no mapping is determined.
426 * the aligned protein sequence
428 * the aligned cdna sequence
431 public static MapList mapCdnaToProtein(SequenceI proteinSeq,
435 * Here we handle either dataset sequence set (desktop) or absent (applet).
436 * Use only the char[] form of the sequence to avoid creating possibly large
439 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
440 char[] aaSeqChars = proteinDataset != null
441 ? proteinDataset.getSequence()
442 : proteinSeq.getSequence();
443 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
444 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
445 : cdnaSeq.getSequence();
446 if (aaSeqChars == null || cdnaSeqChars == null)
452 * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
454 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
455 int cdnaLength = cdnaSeqChars.length;
456 int cdnaStart = cdnaSeq.getStart();
457 int cdnaEnd = cdnaSeq.getEnd();
458 final int proteinStart = proteinSeq.getStart();
459 final int proteinEnd = proteinSeq.getEnd();
462 * If lengths don't match, try ignoring stop codon (if present)
464 if (cdnaLength != mappedLength && cdnaLength > 2)
466 String lastCodon = String.valueOf(cdnaSeqChars,
467 cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase(Locale.ROOT);
468 for (String stop : ResidueProperties.STOP_CODONS)
470 if (lastCodon.equals(stop))
472 cdnaEnd -= CODON_LENGTH;
473 cdnaLength -= CODON_LENGTH;
480 * If lengths still don't match, try ignoring start codon.
483 if (cdnaLength != mappedLength && cdnaLength > 2
484 && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase(Locale.ROOT)
485 .equals(ResidueProperties.START))
487 startOffset += CODON_LENGTH;
488 cdnaStart += CODON_LENGTH;
489 cdnaLength -= CODON_LENGTH;
492 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
495 * protein is translation of dna (+/- start/stop codons)
497 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
499 { proteinStart, proteinEnd }, CODON_LENGTH, 1);
504 * translation failed - try mapping CDS annotated regions of dna
506 return mapCdsToProtein(cdnaSeq, proteinSeq);
510 * Test whether the given cdna sequence, starting at the given offset,
511 * translates to the given amino acid sequence, using the standard translation
512 * table. Designed to fail fast i.e. as soon as a mismatch position is found.
514 * @param cdnaSeqChars
519 protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
522 if (cdnaSeqChars == null || aaSeqChars == null)
528 int dnaPos = cdnaStart;
529 for (; dnaPos < cdnaSeqChars.length - 2
530 && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
532 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
533 final String translated = ResidueProperties.codonTranslate(codon);
536 * allow * in protein to match untranslatable in dna
538 final char aaRes = aaSeqChars[aaPos];
539 if ((translated == null || ResidueProperties.STOP.equals(translated))
544 if (translated == null || !(aaRes == translated.charAt(0)))
547 // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
548 // + codon + "(" + translated + ") != " + aaRes));
554 * check we matched all of the protein sequence
556 if (aaPos != aaSeqChars.length)
562 * check we matched all of the dna except
563 * for optional trailing STOP codon
565 if (dnaPos == cdnaSeqChars.length)
569 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
571 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
572 if (ResidueProperties.STOP
573 .equals(ResidueProperties.codonTranslate(codon)))
582 * Align sequence 'seq' to match the alignment of a mapped sequence. Note this
583 * currently assumes that we are aligning cDNA to match protein.
586 * the sequence to be realigned
588 * the alignment whose sequence alignment is to be 'copied'
590 * character string represent a gap in the realigned sequence
591 * @param preserveUnmappedGaps
592 * @param preserveMappedGaps
593 * @return true if the sequence was realigned, false if it could not be
595 public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
596 String gap, boolean preserveMappedGaps,
597 boolean preserveUnmappedGaps)
600 * Get any mappings from the source alignment to the target (dataset)
603 // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
604 // all mappings. Would it help to constrain this?
605 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
606 if (mappings == null || mappings.isEmpty())
612 * Locate the aligned source sequence whose dataset sequence is mapped. We
613 * just take the first match here (as we can't align like more than one
616 SequenceI alignFrom = null;
617 AlignedCodonFrame mapping = null;
618 for (AlignedCodonFrame mp : mappings)
620 alignFrom = mp.findAlignedSequence(seq, al);
621 if (alignFrom != null)
628 if (alignFrom == null)
632 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
633 preserveMappedGaps, preserveUnmappedGaps);
638 * Align sequence 'alignTo' the same way as 'alignFrom', using the mapping to
639 * match residues and codons. Flags control whether existing gaps in unmapped
640 * (intron) and mapped (exon) regions are preserved or not. Gaps between
641 * intron and exon are only retained if both flags are set.
648 * @param preserveUnmappedGaps
649 * @param preserveMappedGaps
651 public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
652 AlignedCodonFrame mapping, String myGap, char sourceGap,
653 boolean preserveMappedGaps, boolean preserveUnmappedGaps)
655 // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
657 // aligned and dataset sequence positions, all base zero
661 int basesWritten = 0;
662 char myGapChar = myGap.charAt(0);
663 int ratio = myGap.length();
665 int fromOffset = alignFrom.getStart() - 1;
666 int toOffset = alignTo.getStart() - 1;
667 int sourceGapMappedLength = 0;
668 boolean inExon = false;
669 final int toLength = alignTo.getLength();
670 final int fromLength = alignFrom.getLength();
671 StringBuilder thisAligned = new StringBuilder(2 * toLength);
674 * Traverse the 'model' aligned sequence
676 for (int i = 0; i < fromLength; i++)
678 char sourceChar = alignFrom.getCharAt(i);
679 if (sourceChar == sourceGap)
681 sourceGapMappedLength += ratio;
686 * Found a non-gap character. Locate its mapped region if any.
689 // Note mapping positions are base 1, our sequence positions base 0
690 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
691 sourceDsPos + fromOffset);
692 if (mappedPos == null)
695 * unmapped position; treat like a gap
697 sourceGapMappedLength += ratio;
698 // System.err.println("Can't align: no codon mapping to residue "
699 // + sourceDsPos + "(" + sourceChar + ")");
704 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
705 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
706 StringBuilder trailingCopiedGap = new StringBuilder();
709 * Copy dna sequence up to and including this codon. Optionally, include
710 * gaps before the codon starts (in introns) and/or after the codon starts
713 * Note this only works for 'linear' splicing, not reverse or interleaved.
714 * But then 'align dna as protein' doesn't make much sense otherwise.
716 int intronLength = 0;
717 while (basesWritten + toOffset < mappedCodonEnd
718 && thisSeqPos < toLength)
720 final char c = alignTo.getCharAt(thisSeqPos++);
724 int sourcePosition = basesWritten + toOffset;
725 if (sourcePosition < mappedCodonStart)
728 * Found an unmapped (intron) base. First add in any preceding gaps
731 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
733 thisAligned.append(trailingCopiedGap.toString());
734 intronLength += trailingCopiedGap.length();
735 trailingCopiedGap = new StringBuilder();
742 final boolean startOfCodon = sourcePosition == mappedCodonStart;
743 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
744 preserveUnmappedGaps, sourceGapMappedLength, inExon,
745 trailingCopiedGap.length(), intronLength, startOfCodon);
746 for (int k = 0; k < gapsToAdd; k++)
748 thisAligned.append(myGapChar);
750 sourceGapMappedLength = 0;
753 thisAligned.append(c);
754 trailingCopiedGap = new StringBuilder();
758 if (inExon && preserveMappedGaps)
760 trailingCopiedGap.append(myGapChar);
762 else if (!inExon && preserveUnmappedGaps)
764 trailingCopiedGap.append(myGapChar);
771 * At end of model aligned sequence. Copy any remaining target sequence, optionally
772 * including (intron) gaps.
774 while (thisSeqPos < toLength)
776 final char c = alignTo.getCharAt(thisSeqPos++);
777 if (c != myGapChar || preserveUnmappedGaps)
779 thisAligned.append(c);
781 sourceGapMappedLength--;
785 * finally add gaps to pad for any trailing source gaps or
786 * unmapped characters
788 if (preserveUnmappedGaps)
790 while (sourceGapMappedLength > 0)
792 thisAligned.append(myGapChar);
793 sourceGapMappedLength--;
798 * All done aligning, set the aligned sequence.
800 alignTo.setSequence(new String(thisAligned));
804 * Helper method to work out how many gaps to insert when realigning.
806 * @param preserveMappedGaps
807 * @param preserveUnmappedGaps
808 * @param sourceGapMappedLength
810 * @param trailingCopiedGap
811 * @param intronLength
812 * @param startOfCodon
815 protected static int calculateGapsToInsert(boolean preserveMappedGaps,
816 boolean preserveUnmappedGaps, int sourceGapMappedLength,
817 boolean inExon, int trailingGapLength, int intronLength,
818 final boolean startOfCodon)
824 * Reached start of codon. Ignore trailing gaps in intron unless we are
825 * preserving gaps in both exon and intron. Ignore them anyway if the
826 * protein alignment introduces a gap at least as large as the intronic
829 if (inExon && !preserveMappedGaps)
831 trailingGapLength = 0;
833 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
835 trailingGapLength = 0;
839 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
843 if (intronLength + trailingGapLength <= sourceGapMappedLength)
845 gapsToAdd = sourceGapMappedLength - intronLength;
849 gapsToAdd = Math.min(
850 intronLength + trailingGapLength - sourceGapMappedLength,
858 * second or third base of codon; check for any gaps in dna
860 if (!preserveMappedGaps)
862 trailingGapLength = 0;
864 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
870 * Realigns the given protein to match the alignment of the dna, using codon
871 * mappings to translate aligned codon positions to protein residues.
874 * the alignment whose sequences are realigned by this method
876 * the dna alignment whose alignment we are 'copying'
877 * @return the number of sequences that were realigned
879 public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
881 if (protein.isNucleotide() || !dna.isNucleotide())
883 System.err.println("Wrong alignment type in alignProteinAsDna");
886 List<SequenceI> unmappedProtein = new ArrayList<>();
887 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
888 protein, dna, unmappedProtein);
889 return alignProteinAs(protein, alignedCodons, unmappedProtein);
893 * Realigns the given dna to match the alignment of the protein, using codon
894 * mappings to translate aligned peptide positions to codons.
896 * Always produces a padded CDS alignment.
899 * the alignment whose sequences are realigned by this method
901 * the protein alignment whose alignment we are 'copying'
902 * @return the number of sequences that were realigned
904 public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
906 if (protein.isNucleotide() || !dna.isNucleotide())
908 System.err.println("Wrong alignment type in alignProteinAsDna");
911 // todo: implement this
912 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
913 int alignedCount = 0;
914 int width = 0; // alignment width for padding CDS
915 for (SequenceI dnaSeq : dna.getSequences())
917 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
918 dna.getGapCharacter()))
922 width = Math.max(dnaSeq.getLength(), width);
926 for (SequenceI dnaSeq : dna.getSequences())
928 oldwidth = dnaSeq.getLength();
929 diff = width - oldwidth;
932 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
939 * Helper method to align (if possible) the dna sequence to match the
940 * alignment of a mapped protein sequence. This is currently limited to
941 * handling coding sequence only.
949 static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
950 AlignmentI protein, List<AlignedCodonFrame> mappings,
953 SequenceI cdsDss = cdsSeq.getDatasetSequence();
957 .println("alignCdsSequenceAsProtein needs aligned sequence!");
961 List<AlignedCodonFrame> dnaMappings = MappingUtils
962 .findMappingsForSequence(cdsSeq, mappings);
963 for (AlignedCodonFrame mapping : dnaMappings)
965 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
968 final int peptideLength = peptide.getLength();
969 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
972 MapList mapList = map.getMap();
973 if (map.getTo() == peptide.getDatasetSequence())
975 mapList = mapList.getInverse();
977 final int cdsLength = cdsDss.getLength();
978 int mappedFromLength = MappingUtils.getLength(mapList
980 int mappedToLength = MappingUtils
981 .getLength(mapList.getToRanges());
982 boolean addStopCodon = (cdsLength == mappedFromLength
983 * CODON_LENGTH + CODON_LENGTH)
984 || (peptide.getDatasetSequence()
985 .getLength() == mappedFromLength - 1);
986 if (cdsLength != mappedToLength && !addStopCodon)
988 System.err.println(String.format(
989 "Can't align cds as protein (length mismatch %d/%d): %s",
990 cdsLength, mappedToLength, cdsSeq.getName()));
994 * pre-fill the aligned cds sequence with gaps
996 char[] alignedCds = new char[peptideLength * CODON_LENGTH
997 + (addStopCodon ? CODON_LENGTH : 0)];
998 Arrays.fill(alignedCds, gapChar);
1001 * walk over the aligned peptide sequence and insert mapped
1002 * codons for residues in the aligned cds sequence
1004 int copiedBases = 0;
1005 int cdsStart = cdsDss.getStart();
1006 int proteinPos = peptide.getStart() - 1;
1009 for (int col = 0; col < peptideLength; col++)
1011 char residue = peptide.getCharAt(col);
1013 if (Comparison.isGap(residue))
1015 cdsCol += CODON_LENGTH;
1020 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1023 // e.g. incomplete start codon, X in peptide
1024 cdsCol += CODON_LENGTH;
1028 for (int j = codon[0]; j <= codon[1]; j++)
1030 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1031 alignedCds[cdsCol++] = mappedBase;
1039 * append stop codon if not mapped from protein,
1040 * closing it up to the end of the mapped sequence
1042 if (copiedBases == cdsLength - CODON_LENGTH)
1044 for (int i = alignedCds.length - 1; i >= 0; i--)
1046 if (!Comparison.isGap(alignedCds[i]))
1048 cdsCol = i + 1; // gap just after end of sequence
1052 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1054 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1057 cdsSeq.setSequence(new String(alignedCds));
1066 * Builds a map whose key is an aligned codon position (3 alignment column
1067 * numbers base 0), and whose value is a map from protein sequence to each
1068 * protein's peptide residue for that codon. The map generates an ordering of
1069 * the codons, and allows us to read off the peptides at each position in
1070 * order to assemble 'aligned' protein sequences.
1073 * the protein alignment
1075 * the coding dna alignment
1076 * @param unmappedProtein
1077 * any unmapped proteins are added to this list
1080 protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1081 AlignmentI protein, AlignmentI dna,
1082 List<SequenceI> unmappedProtein)
1085 * maintain a list of any proteins with no mappings - these will be
1086 * rendered 'as is' in the protein alignment as we can't align them
1088 unmappedProtein.addAll(protein.getSequences());
1090 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1093 * Map will hold, for each aligned codon position e.g. [3, 5, 6], a map of
1094 * {dnaSequence, {proteinSequence, codonProduct}} at that position. The
1095 * comparator keeps the codon positions ordered.
1097 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<>(
1098 new CodonComparator());
1100 for (SequenceI dnaSeq : dna.getSequences())
1102 for (AlignedCodonFrame mapping : mappings)
1104 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1107 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1108 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1110 unmappedProtein.remove(prot);
1116 * Finally add any unmapped peptide start residues (e.g. for incomplete
1117 * codons) as if at the codon position before the second residue
1119 // TODO resolve JAL-2022 so this fudge can be removed
1120 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1121 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1123 return alignedCodons;
1127 * Scans for any protein mapped from position 2 (meaning unmapped start
1128 * position e.g. an incomplete codon), and synthesizes a 'codon' for it at the
1129 * preceding position in the alignment
1131 * @param alignedCodons
1132 * the codon-to-peptide map
1133 * @param mappedSequenceCount
1134 * the number of distinct sequences in the map
1136 protected static void addUnmappedPeptideStarts(
1137 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1138 int mappedSequenceCount)
1140 // TODO delete this ugly hack once JAL-2022 is resolved
1141 // i.e. we can model startPhase > 0 (incomplete start codon)
1143 List<SequenceI> sequencesChecked = new ArrayList<>();
1144 AlignedCodon lastCodon = null;
1145 Map<SequenceI, AlignedCodon> toAdd = new HashMap<>();
1147 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1150 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1153 SequenceI seq = sequenceCodon.getKey();
1154 if (sequencesChecked.contains(seq))
1158 sequencesChecked.add(seq);
1159 AlignedCodon codon = sequenceCodon.getValue();
1160 if (codon.peptideCol > 1)
1163 "Problem mapping protein with >1 unmapped start positions: "
1166 else if (codon.peptideCol == 1)
1169 * first position (peptideCol == 0) was unmapped - add it
1171 if (lastCodon != null)
1173 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1174 lastCodon.pos2, lastCodon.pos3,
1175 String.valueOf(seq.getCharAt(0)), 0);
1176 toAdd.put(seq, firstPeptide);
1181 * unmapped residue at start of alignment (no prior column) -
1182 * 'insert' at nominal codon [0, 0, 0]
1184 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1185 String.valueOf(seq.getCharAt(0)), 0);
1186 toAdd.put(seq, firstPeptide);
1189 if (sequencesChecked.size() == mappedSequenceCount)
1191 // no need to check past first mapped position in all sequences
1195 lastCodon = entry.getKey();
1199 * add any new codons safely after iterating over the map
1201 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1203 addCodonToMap(alignedCodons, startCodon.getValue(),
1204 startCodon.getKey());
1209 * Update the aligned protein sequences to match the codon alignments given in
1213 * @param alignedCodons
1214 * an ordered map of codon positions (columns), with sequence/peptide
1215 * values present in each column
1216 * @param unmappedProtein
1219 protected static int alignProteinAs(AlignmentI protein,
1220 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1221 List<SequenceI> unmappedProtein)
1224 * prefill peptide sequences with gaps
1226 int alignedWidth = alignedCodons.size();
1227 char[] gaps = new char[alignedWidth];
1228 Arrays.fill(gaps, protein.getGapCharacter());
1229 Map<SequenceI, char[]> peptides = new HashMap<>();
1230 for (SequenceI seq : protein.getSequences())
1232 if (!unmappedProtein.contains(seq))
1234 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1239 * Traverse the codons left to right (as defined by CodonComparator)
1240 * and insert peptides in each column where the sequence is mapped.
1241 * This gives a peptide 'alignment' where residues are aligned if their
1242 * corresponding codons occupy the same columns in the cdna alignment.
1245 for (AlignedCodon codon : alignedCodons.keySet())
1247 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1249 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1251 char residue = entry.getValue().product.charAt(0);
1252 peptides.get(entry.getKey())[column] = residue;
1258 * and finally set the constructed sequences
1260 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1262 entry.getKey().setSequence(new String(entry.getValue()));
1269 * Populate the map of aligned codons by traversing the given sequence
1270 * mapping, locating the aligned positions of mapped codons, and adding those
1271 * positions and their translation products to the map.
1274 * the aligned sequence we are mapping from
1276 * the sequence to be aligned to the codons
1278 * the gap character in the dna sequence
1280 * a mapping to a sequence translation
1281 * @param alignedCodons
1282 * the map we are building up
1284 static void addCodonPositions(SequenceI dna, SequenceI protein,
1285 char gapChar, Mapping seqMap,
1286 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1288 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1291 * add codon positions, and their peptide translations, to the alignment
1292 * map, while remembering the first codon mapped
1294 while (codons.hasNext())
1298 AlignedCodon codon = codons.next();
1299 addCodonToMap(alignedCodons, codon, protein);
1300 } catch (IncompleteCodonException e)
1302 // possible incomplete trailing codon - ignore
1303 } catch (NoSuchElementException e)
1305 // possibly peptide lacking STOP
1311 * Helper method to add a codon-to-peptide entry to the aligned codons map
1313 * @param alignedCodons
1317 protected static void addCodonToMap(
1318 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1319 AlignedCodon codon, SequenceI protein)
1321 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1322 if (seqProduct == null)
1324 seqProduct = new HashMap<>();
1325 alignedCodons.put(codon, seqProduct);
1327 seqProduct.put(protein, codon);
1331 * Returns true if a cDNA/Protein mapping either exists, or could be made,
1332 * between at least one pair of sequences in the two alignments. Currently,
1335 * <li>One alignment must be nucleotide, and the other protein</li>
1336 * <li>At least one pair of sequences must be already mapped, or mappable</li>
1337 * <li>Mappable means the nucleotide translation matches the protein
1339 * <li>The translation may ignore start and stop codons if present in the
1347 public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1349 if (al1 == null || al2 == null)
1355 * Require one nucleotide and one protein
1357 if (al1.isNucleotide() == al2.isNucleotide())
1361 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1362 AlignmentI protein = dna == al1 ? al2 : al1;
1363 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1364 for (SequenceI dnaSeq : dna.getSequences())
1366 for (SequenceI proteinSeq : protein.getSequences())
1368 if (isMappable(dnaSeq, proteinSeq, mappings))
1378 * Returns true if the dna sequence is mapped, or could be mapped, to the
1386 protected static boolean isMappable(SequenceI dnaSeq,
1387 SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1389 if (dnaSeq == null || proteinSeq == null)
1394 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1395 : dnaSeq.getDatasetSequence();
1396 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1398 : proteinSeq.getDatasetSequence();
1400 for (AlignedCodonFrame mapping : mappings)
1402 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1412 * Just try to make a mapping (it is not yet stored), test whether
1415 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1419 * Finds any reference annotations associated with the sequences in
1420 * sequenceScope, that are not already added to the alignment, and adds them
1421 * to the 'candidates' map. Also populates a lookup table of annotation
1422 * labels, keyed by calcId, for use in constructing tooltips or the like.
1424 * @param sequenceScope
1425 * the sequences to scan for reference annotations
1426 * @param labelForCalcId
1427 * (optional) map to populate with label for calcId
1429 * map to populate with annotations for sequence
1431 * the alignment to check for presence of annotations
1433 public static void findAddableReferenceAnnotations(
1434 List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1435 final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1438 if (sequenceScope == null)
1444 * For each sequence in scope, make a list of any annotations on the
1445 * underlying dataset sequence which are not already on the alignment.
1447 * Add to a map of { alignmentSequence, <List of annotations to add> }
1449 for (SequenceI seq : sequenceScope)
1451 SequenceI dataset = seq.getDatasetSequence();
1452 if (dataset == null)
1456 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1457 if (datasetAnnotations == null)
1461 final List<AlignmentAnnotation> result = new ArrayList<>();
1462 for (AlignmentAnnotation dsann : datasetAnnotations)
1464 if (dsann.annotations != null) // ignore non-positional annotation
1467 * Find matching annotations on the alignment. If none is found, then
1468 * add this annotation to the list of 'addable' annotations for this
1471 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1472 .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1473 if (!matchedAlignmentAnnotations.iterator().hasNext())
1476 if (labelForCalcId != null)
1478 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1483 * Save any addable annotations for this sequence
1485 if (!result.isEmpty())
1487 candidates.put(seq, result);
1494 * Adds annotations to the top of the alignment annotations, in the same order
1495 * as their related sequences.
1497 * @param annotations
1498 * the annotations to add
1500 * the alignment to add them to
1501 * @param selectionGroup
1502 * current selection group (or null if none)
1504 public static void addReferenceAnnotations(
1505 Map<SequenceI, List<AlignmentAnnotation>> annotations,
1506 final AlignmentI alignment, final SequenceGroup selectionGroup)
1508 for (SequenceI seq : annotations.keySet())
1510 for (AlignmentAnnotation ann : annotations.get(seq))
1512 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1514 int endRes = ann.annotations.length;
1515 if (selectionGroup != null)
1517 startRes = selectionGroup.getStartRes();
1518 endRes = selectionGroup.getEndRes();
1520 copyAnn.restrict(startRes, endRes);
1523 * Add to the sequence (sets copyAnn.datasetSequence), unless the
1524 * original annotation is already on the sequence.
1526 if (!seq.hasAnnotation(ann))
1528 seq.addAlignmentAnnotation(copyAnn);
1531 copyAnn.adjustForAlignment();
1532 // add to the alignment and set visible
1533 alignment.addAnnotation(copyAnn);
1534 copyAnn.visible = true;
1540 * Set visibility of alignment annotations of specified types (labels), for
1541 * specified sequences. This supports controls like "Show all secondary
1542 * structure", "Hide all Temp factor", etc.
1544 * @al the alignment to scan for annotations
1546 * the types (labels) of annotations to be updated
1547 * @param forSequences
1548 * if not null, only annotations linked to one of these sequences are
1549 * in scope for update; if null, acts on all sequence annotations
1551 * if this flag is true, 'types' is ignored (label not checked)
1553 * if true, set visibility on, else set off
1555 public static void showOrHideSequenceAnnotations(AlignmentI al,
1556 Collection<String> types, List<SequenceI> forSequences,
1557 boolean anyType, boolean doShow)
1559 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1562 for (AlignmentAnnotation aa : anns)
1564 if (anyType || types.contains(aa.label))
1566 if ((aa.sequenceRef != null) && (forSequences == null
1567 || forSequences.contains(aa.sequenceRef)))
1569 aa.visible = doShow;
1577 * Returns true if either sequence has a cross-reference to the other
1583 public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1585 // Note: moved here from class CrossRef as the latter class has dependencies
1586 // not availability to the applet's classpath
1587 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1591 * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1592 * that sequence name is structured as Source|AccessionId.
1598 public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1600 if (seq1 == null || seq2 == null)
1604 String name = seq2.getName();
1605 final List<DBRefEntry> xrefs = seq1.getDBRefs();
1608 for (int ix = 0, nx = xrefs.size(); ix < nx; ix++)
1610 DBRefEntry xref = xrefs.get(ix);
1611 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1612 // case-insensitive test, consistent with DBRefEntry.equalRef()
1613 if (xrefName.equalsIgnoreCase(name))
1623 * Constructs an alignment consisting of the mapped (CDS) regions in the given
1624 * nucleotide sequences, and updates mappings to match. The CDS sequences are
1625 * added to the original alignment's dataset, which is shared by the new
1626 * alignment. Mappings from nucleotide to CDS, and from CDS to protein, are
1627 * added to the alignment dataset.
1630 * aligned nucleotide (dna or cds) sequences
1632 * the alignment dataset the sequences belong to
1634 * (optional) to restrict results to CDS that map to specified
1636 * @return an alignment whose sequences are the cds-only parts of the dna
1637 * sequences (or null if no mappings are found)
1639 public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1640 AlignmentI dataset, SequenceI[] products)
1642 if (dataset == null || dataset.getDataset() != null)
1644 throw new IllegalArgumentException(
1645 "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1647 List<SequenceI> foundSeqs = new ArrayList<>();
1648 List<SequenceI> cdsSeqs = new ArrayList<>();
1649 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1650 HashSet<SequenceI> productSeqs = null;
1651 if (products != null)
1653 productSeqs = new HashSet<>();
1654 for (SequenceI seq : products)
1656 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1657 .getDatasetSequence());
1662 * Construct CDS sequences from mappings on the alignment dataset.
1664 * - find the protein product(s) mapped to from each dna sequence
1665 * - if the mapping covers the whole dna sequence (give or take start/stop
1666 * codon), take the dna as the CDS sequence
1667 * - else search dataset mappings for a suitable dna sequence, i.e. one
1668 * whose whole sequence is mapped to the protein
1669 * - if no sequence found, construct one from the dna sequence and mapping
1670 * (and add it to dataset so it is found if this is repeated)
1672 for (SequenceI dnaSeq : dna)
1674 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1675 : dnaSeq.getDatasetSequence();
1677 List<AlignedCodonFrame> seqMappings = MappingUtils
1678 .findMappingsForSequence(dnaSeq, mappings);
1679 for (AlignedCodonFrame mapping : seqMappings)
1681 List<Mapping> mappingsFromSequence = mapping
1682 .getMappingsFromSequence(dnaSeq);
1684 for (Mapping aMapping : mappingsFromSequence)
1686 MapList mapList = aMapping.getMap();
1687 if (mapList.getFromRatio() == 1)
1690 * not a dna-to-protein mapping (likely dna-to-cds)
1696 * skip if mapping is not to one of the target set of proteins
1698 SequenceI proteinProduct = aMapping.getTo();
1699 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1705 * try to locate the CDS from the dataset mappings;
1706 * guard against duplicate results (for the case that protein has
1707 * dbrefs to both dna and cds sequences)
1709 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1710 seqMappings, aMapping);
1713 if (!foundSeqs.contains(cdsSeq))
1715 foundSeqs.add(cdsSeq);
1716 SequenceI derivedSequence = cdsSeq.deriveSequence();
1717 cdsSeqs.add(derivedSequence);
1718 if (!dataset.getSequences().contains(cdsSeq))
1720 dataset.addSequence(cdsSeq);
1727 * didn't find mapped CDS sequence - construct it and add
1728 * its dataset sequence to the dataset
1730 cdsSeq = makeCdsSequence(dnaSeq.getDatasetSequence(), aMapping,
1731 dataset).deriveSequence();
1732 // cdsSeq has a name constructed as CDS|<dbref>
1733 // <dbref> will be either the accession for the coding sequence,
1734 // marked in the /via/ dbref to the protein product accession
1735 // or it will be the original nucleotide accession.
1736 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1738 cdsSeqs.add(cdsSeq);
1741 * build the mapping from CDS to protein
1743 List<int[]> cdsRange = Collections
1744 .singletonList(new int[]
1745 { cdsSeq.getStart(),
1746 cdsSeq.getLength() + cdsSeq.getStart() - 1 });
1747 MapList cdsToProteinMap = new MapList(cdsRange,
1748 mapList.getToRanges(), mapList.getFromRatio(),
1749 mapList.getToRatio());
1751 if (!dataset.getSequences().contains(cdsSeqDss))
1754 * if this sequence is a newly created one, add it to the dataset
1755 * and made a CDS to protein mapping (if sequence already exists,
1756 * CDS-to-protein mapping _is_ the transcript-to-protein mapping)
1758 dataset.addSequence(cdsSeqDss);
1759 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1760 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1764 * guard against duplicating the mapping if repeating this action
1766 if (!mappings.contains(cdsToProteinMapping))
1768 mappings.add(cdsToProteinMapping);
1772 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1773 proteinProduct, aMapping);
1775 * add another mapping from original 'from' range to CDS
1777 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1778 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1780 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1782 if (!mappings.contains(dnaToCdsMapping))
1784 mappings.add(dnaToCdsMapping);
1788 * transfer dna chromosomal loci (if known) to the CDS
1789 * sequence (via the mapping)
1791 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1792 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1795 * add DBRef with mapping from protein to CDS
1796 * (this enables Get Cross-References from protein alignment)
1797 * This is tricky because we can't have two DBRefs with the
1798 * same source and accession, so need a different accession for
1799 * the CDS from the dna sequence
1802 // specific use case:
1803 // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1804 // ENSG02, ENSG03, with transcripts and products similarly named.
1805 // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1807 // JBPNote: ?? can't actually create an example that demonstrates we
1809 // synthesize an xref.
1811 List<DBRefEntry> primrefs = dnaDss.getPrimaryDBRefs();
1812 for (int ip = 0, np = primrefs.size(); ip < np; ip++)
1814 DBRefEntry primRef = primrefs.get(ip);
1816 * create a cross-reference from CDS to the source sequence's
1817 * primary reference and vice versa
1819 String source = primRef.getSource();
1820 String version = primRef.getVersion();
1821 DBRefEntry cdsCrossRef = new DBRefEntry(source, source + ":"
1822 + version, primRef.getAccessionId());
1823 cdsCrossRef.setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
1824 cdsSeqDss.addDBRef(cdsCrossRef);
1826 dnaSeq.addDBRef(new DBRefEntry(source, version, cdsSeq
1827 .getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
1828 // problem here is that the cross-reference is synthesized -
1829 // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1831 // assuming cds version same as dna ?!?
1833 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
1836 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1838 proteinProduct.addDBRef(proteinToCdsRef);
1841 * transfer any features on dna that overlap the CDS
1843 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1844 SequenceOntologyI.CDS);
1849 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1851 cds.setDataset(dataset);
1857 * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
1858 * toSeq, mediated by the given mapping between the sequences
1861 * @param targetToFrom
1865 protected static void transferGeneLoci(SequenceI fromSeq,
1866 MapList targetToFrom, SequenceI targetSeq)
1868 if (targetSeq.getGeneLoci() != null)
1870 // already have - don't override
1873 GeneLociI fromLoci = fromSeq.getGeneLoci();
1874 if (fromLoci == null)
1879 MapList newMap = targetToFrom.traverse(fromLoci.getMapping());
1883 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
1884 fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
1889 * A helper method that finds a CDS sequence in the alignment dataset that is
1890 * mapped to the given protein sequence, and either is, or has a mapping from,
1891 * the given dna sequence.
1894 * set of all mappings on the dataset
1896 * a dna (or cds) sequence we are searching from
1897 * @param seqMappings
1898 * the set of mappings involving dnaSeq
1900 * a transcript-to-peptide mapping
1903 static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1904 SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1908 * TODO a better dna-cds-protein mapping data representation to allow easy
1909 * navigation; until then this clunky looping around lists of mappings
1911 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1912 : dnaSeq.getDatasetSequence();
1913 SequenceI proteinProduct = aMapping.getTo();
1916 * is this mapping from the whole dna sequence (i.e. CDS)?
1917 * allowing for possible stop codon on dna but not peptide
1919 int mappedFromLength = MappingUtils
1920 .getLength(aMapping.getMap().getFromRanges());
1921 int dnaLength = seqDss.getLength();
1922 if (mappedFromLength == dnaLength
1923 || mappedFromLength == dnaLength - CODON_LENGTH)
1926 * if sequence has CDS features, this is a transcript with no UTR
1927 * - do not take this as the CDS sequence! (JAL-2789)
1929 if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
1937 * looks like we found the dna-to-protein mapping; search for the
1938 * corresponding cds-to-protein mapping
1940 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1941 .findMappingsForSequence(proteinProduct, mappings);
1942 for (AlignedCodonFrame acf : mappingsToPeptide)
1944 for (SequenceToSequenceMapping map : acf.getMappings())
1946 Mapping mapping = map.getMapping();
1947 if (mapping != aMapping
1948 && mapping.getMap().getFromRatio() == CODON_LENGTH
1949 && proteinProduct == mapping.getTo()
1950 && seqDss != map.getFromSeq())
1952 mappedFromLength = MappingUtils
1953 .getLength(mapping.getMap().getFromRanges());
1954 if (mappedFromLength == map.getFromSeq().getLength())
1957 * found a 3:1 mapping to the protein product which covers
1958 * the whole dna sequence i.e. is from CDS; finally check the CDS
1959 * is mapped from the given dna start sequence
1961 SequenceI cdsSeq = map.getFromSeq();
1962 // todo this test is weak if seqMappings contains multiple mappings;
1963 // we get away with it if transcript:cds relationship is 1:1
1964 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1965 .findMappingsForSequence(cdsSeq, seqMappings);
1966 if (!dnaToCdsMaps.isEmpty())
1978 * Helper method that makes a CDS sequence as defined by the mappings from the
1979 * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1980 * forward or reverse strand).
1985 * - existing dataset. We check for sequences that look like the CDS
1986 * we are about to construct, if one exists already, then we will
1987 * just return that one.
1988 * @return CDS sequence (as a dataset sequence)
1990 static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1994 * construct CDS sequence name as "CDS|" with 'from id' held in the mapping
1995 * if set (e.g. EMBL protein_id), else sequence name appended
1997 String mapFromId = mapping.getMappedFromId();
1998 final String seqId = "CDS|"
1999 + (mapFromId != null ? mapFromId : seq.getName());
2001 SequenceI newSeq = null;
2003 final MapList maplist = mapping.getMap();
2004 if (maplist.isContiguous() && maplist.isFromForwardStrand())
2007 * just a subsequence, keep same dataset sequence
2009 int start = maplist.getFromLowest();
2010 int end = maplist.getFromHighest();
2011 newSeq = seq.getSubSequence(start - 1, end);
2012 newSeq.setName(seqId);
2017 * construct by splicing mapped from ranges
2019 char[] seqChars = seq.getSequence();
2020 List<int[]> fromRanges = maplist.getFromRanges();
2021 int cdsWidth = MappingUtils.getLength(fromRanges);
2022 char[] newSeqChars = new char[cdsWidth];
2025 for (int[] range : fromRanges)
2027 if (range[0] <= range[1])
2029 // forward strand mapping - just copy the range
2030 int length = range[1] - range[0] + 1;
2031 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
2037 // reverse strand mapping - copy and complement one by one
2038 for (int i = range[0]; i >= range[1]; i--)
2040 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
2045 newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2048 if (dataset != null)
2050 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2051 if (matches != null)
2053 boolean matched = false;
2054 for (SequenceI mtch : matches)
2056 if (mtch.getStart() != newSeq.getStart())
2060 if (mtch.getEnd() != newSeq.getEnd())
2064 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2076 "JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
2082 // newSeq.setDescription(mapFromId);
2088 * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2089 * the given mapping.
2093 * @param proteinProduct
2095 * @return list of DBRefEntrys added
2097 protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2098 SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2101 // gather direct refs from contig congruent with mapping
2102 List<DBRefEntry> direct = new ArrayList<>();
2103 HashSet<String> directSources = new HashSet<>();
2105 List<DBRefEntry> refs = contig.getDBRefs();
2108 for (int ib = 0, nb = refs.size(); ib < nb; ib++)
2110 DBRefEntry dbr = refs.get(ib);
2112 if (dbr.hasMap() && (map = dbr.getMap().getMap()).isTripletMap())
2114 // check if map is the CDS mapping
2115 if (mapping.getMap().equals(map))
2118 directSources.add(dbr.getSource());
2123 List<DBRefEntry> onSource = DBRefUtils.selectRefs(
2124 proteinProduct.getDBRefs(),
2125 directSources.toArray(new String[0]));
2126 List<DBRefEntry> propagated = new ArrayList<>();
2128 // and generate appropriate mappings
2129 for (int ic = 0, nc = direct.size(); ic < nc; ic++)
2131 DBRefEntry cdsref = direct.get(ic);
2132 Mapping m = cdsref.getMap();
2133 // clone maplist and mapping
2134 MapList cdsposmap = new MapList(
2135 Arrays.asList(new int[][]
2136 { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2137 m.getMap().getToRanges(), 3, 1);
2138 Mapping cdsmap = new Mapping(m.getTo(), m.getMap());
2141 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2142 cdsref.getVersion(), cdsref.getAccessionId(),
2143 new Mapping(cdsmap.getTo(), cdsposmap));
2145 // and see if we can map to the protein product for this mapping.
2146 // onSource is the filtered set of accessions on protein that we are
2147 // tranferring, so we assume accession is the same.
2148 if (cdsmap.getTo() == null && onSource != null)
2150 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2151 cdsref.getAccessionId());
2152 if (sourceRefs != null)
2154 for (DBRefEntry srcref : sourceRefs)
2156 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2158 // we have found a complementary dbref on the protein product, so
2159 // update mapping's getTo
2160 newref.getMap().setTo(proteinProduct);
2165 cdsSeq.addDBRef(newref);
2166 propagated.add(newref);
2172 * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2173 * feature start/end ranges, optionally omitting specified feature types.
2174 * Returns the number of features copied.
2179 * the mapping from 'fromSeq' to 'toSeq'
2181 * if not null, only features of this type are copied (including
2182 * subtypes in the Sequence Ontology)
2185 protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2186 MapList mapping, String select, String... omitting)
2188 SequenceI copyTo = toSeq;
2189 while (copyTo.getDatasetSequence() != null)
2191 copyTo = copyTo.getDatasetSequence();
2193 if (fromSeq == copyTo || fromSeq.getDatasetSequence() == copyTo)
2195 return 0; // shared dataset sequence
2199 * get features, optionally restricted by an ontology term
2201 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2202 .getPositionalFeatures() : fromSeq.getFeatures()
2203 .getFeaturesByOntology(select);
2206 for (SequenceFeature sf : sfs)
2208 String type = sf.getType();
2209 boolean omit = false;
2210 for (String toOmit : omitting)
2212 if (type.equals(toOmit))
2223 * locate the mapped range - null if either start or end is
2224 * not mapped (no partial overlaps are calculated)
2226 int start = sf.getBegin();
2227 int end = sf.getEnd();
2228 int[] mappedTo = mapping.locateInTo(start, end);
2230 * if whole exon range doesn't map, try interpreting it
2231 * as 5' or 3' exon overlapping the CDS range
2233 if (mappedTo == null)
2235 mappedTo = mapping.locateInTo(end, end);
2236 if (mappedTo != null)
2239 * end of exon is in CDS range - 5' overlap
2240 * to a range from the start of the peptide
2245 if (mappedTo == null)
2247 mappedTo = mapping.locateInTo(start, start);
2248 if (mappedTo != null)
2251 * start of exon is in CDS range - 3' overlap
2252 * to a range up to the end of the peptide
2254 mappedTo[1] = toSeq.getLength();
2257 if (mappedTo != null)
2259 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2260 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2261 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2262 sf.getFeatureGroup(), sf.getScore());
2263 copyTo.addSequenceFeature(copy);
2271 * Returns a mapping from dna to protein by inspecting sequence features of
2272 * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2273 * length is 3 times the peptide length (optionally after dropping a trailing
2274 * stop codon). This method does not check whether the CDS nucleotide sequence
2275 * translates to the peptide sequence.
2281 public static MapList mapCdsToProtein(SequenceI dnaSeq,
2282 SequenceI proteinSeq)
2284 List<int[]> ranges = findCdsPositions(dnaSeq);
2285 int mappedDnaLength = MappingUtils.getLength(ranges);
2288 * if not a whole number of codons, truncate mapping
2290 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2291 if (codonRemainder > 0)
2293 mappedDnaLength -= codonRemainder;
2294 MappingUtils.removeEndPositions(codonRemainder, ranges);
2297 int proteinLength = proteinSeq.getLength();
2298 int proteinStart = proteinSeq.getStart();
2299 int proteinEnd = proteinSeq.getEnd();
2302 * incomplete start codon may mean X at start of peptide
2303 * we ignore both for mapping purposes
2305 if (proteinSeq.getCharAt(0) == 'X')
2307 // todo JAL-2022 support startPhase > 0
2311 List<int[]> proteinRange = new ArrayList<>();
2314 * dna length should map to protein (or protein plus stop codon)
2316 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2317 if (codesForResidues == (proteinLength + 1))
2319 // assuming extra codon is for STOP and not in peptide
2320 // todo: check trailing codon is indeed a STOP codon
2322 mappedDnaLength -= CODON_LENGTH;
2323 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2326 if (codesForResidues == proteinLength)
2328 proteinRange.add(new int[] { proteinStart, proteinEnd });
2329 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2335 * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2336 * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2337 * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2338 * position order, so this method is only valid for linear CDS in the same
2339 * sense as the protein product.
2344 protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2346 List<int[]> result = new ArrayList<>();
2348 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2349 SequenceOntologyI.CDS);
2354 SequenceFeatures.sortFeatures(sfs, true);
2356 for (SequenceFeature sf : sfs)
2361 String s = sf.getPhase();
2364 phase = Integer.parseInt(s);
2366 } catch (NumberFormatException e)
2371 * phase > 0 on first codon means 5' incomplete - skip to the start
2372 * of the next codon; example ENST00000496384
2374 int begin = sf.getBegin();
2375 int end = sf.getEnd();
2376 if (result.isEmpty() && phase > 0)
2381 // shouldn't happen!
2383 .println("Error: start phase extends beyond start CDS in "
2384 + dnaSeq.getName());
2387 result.add(new int[] { begin, end });
2391 * Finally sort ranges by start position. This avoids a dependency on
2392 * keeping features in order on the sequence (if they are in order anyway,
2393 * the sort will have almost no work to do). The implicit assumption is CDS
2394 * ranges are assembled in order. Other cases should not use this method,
2395 * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2397 Collections.sort(result, IntRangeComparator.ASCENDING);
2402 * Makes an alignment with a copy of the given sequences, adding in any
2403 * non-redundant sequences which are mapped to by the cross-referenced
2409 * the alignment dataset shared by the new copy
2412 public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2413 SequenceI[] xrefs, AlignmentI dataset)
2415 AlignmentI copy = new Alignment(new Alignment(seqs));
2416 copy.setDataset(dataset);
2417 boolean isProtein = !copy.isNucleotide();
2418 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2421 // BH 2019.01.25 recoded to remove iterators
2423 for (int ix = 0, nx = xrefs.length; ix < nx; ix++)
2425 SequenceI xref = xrefs[ix];
2426 List<DBRefEntry> dbrefs = xref.getDBRefs();
2429 for (int ir = 0, nir = dbrefs.size(); ir < nir; ir++)
2431 DBRefEntry dbref = dbrefs.get(ir);
2432 Mapping map = dbref.getMap();
2434 if (map == null || (mto = map.getTo()) == null
2435 || mto.isProtein() != isProtein)
2439 SequenceI mappedTo = mto;
2440 SequenceI match = matcher.findIdMatch(mappedTo);
2443 matcher.add(mappedTo);
2444 copy.addSequence(mappedTo);
2454 * Try to align sequences in 'unaligned' to match the alignment of their
2455 * mapped regions in 'aligned'. For example, could use this to align CDS
2456 * sequences which are mapped to their parent cDNA sequences.
2458 * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2459 * dna-to-protein or protein-to-dna use alternative methods.
2462 * sequences to be aligned
2464 * holds aligned sequences and their mappings
2467 public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2470 * easy case - aligning a copy of aligned sequences
2472 if (alignAsSameSequences(unaligned, aligned))
2474 return unaligned.getHeight();
2478 * fancy case - aligning via mappings between sequences
2480 List<SequenceI> unmapped = new ArrayList<>();
2481 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2482 unaligned, aligned, unmapped);
2483 int width = columnMap.size();
2484 char gap = unaligned.getGapCharacter();
2485 int realignedCount = 0;
2486 // TODO: verify this loop scales sensibly for very wide/high alignments
2488 for (SequenceI seq : unaligned.getSequences())
2490 if (!unmapped.contains(seq))
2492 char[] newSeq = new char[width];
2493 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2494 // Integer iteration below
2499 * traverse the map to find columns populated
2502 for (Integer column : columnMap.keySet())
2504 Character c = columnMap.get(column).get(seq);
2508 * sequence has a character at this position
2518 * trim trailing gaps
2520 if (lastCol < width)
2522 char[] tmp = new char[lastCol + 1];
2523 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2526 // TODO: optimise SequenceI to avoid char[]->String->char[]
2527 seq.setSequence(String.valueOf(newSeq));
2531 return realignedCount;
2535 * If unaligned and aligned sequences share the same dataset sequences, then
2536 * simply copies the aligned sequences to the unaligned sequences and returns
2537 * true; else returns false
2540 * - sequences to be aligned based on aligned
2542 * - 'guide' alignment containing sequences derived from same
2543 * dataset as unaligned
2546 static boolean alignAsSameSequences(AlignmentI unaligned,
2549 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2551 return false; // should only pass alignments with datasets here
2554 // map from dataset sequence to alignment sequence(s)
2555 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2556 for (SequenceI seq : aligned.getSequences())
2558 SequenceI ds = seq.getDatasetSequence();
2559 if (alignedDatasets.get(ds) == null)
2561 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2563 alignedDatasets.get(ds).add(seq);
2567 * first pass - check whether all sequences to be aligned share a
2568 * dataset sequence with an aligned sequence; also note the leftmost
2569 * ungapped column from which to copy
2571 int leftmost = Integer.MAX_VALUE;
2572 for (SequenceI seq : unaligned.getSequences())
2574 final SequenceI ds = seq.getDatasetSequence();
2575 if (!alignedDatasets.containsKey(ds))
2579 SequenceI alignedSeq = alignedDatasets.get(ds)
2581 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2582 leftmost = Math.min(leftmost, startCol);
2586 * second pass - copy aligned sequences;
2587 * heuristic rule: pair off sequences in order for the case where
2588 * more than one shares the same dataset sequence
2590 final char gapCharacter = aligned.getGapCharacter();
2591 for (SequenceI seq : unaligned.getSequences())
2593 List<SequenceI> alignedSequences = alignedDatasets
2594 .get(seq.getDatasetSequence());
2595 if (alignedSequences.isEmpty())
2598 * defensive check - shouldn't happen! (JAL-3536)
2602 SequenceI alignedSeq = alignedSequences.get(0);
2605 * gap fill for leading (5') UTR if any
2607 // TODO this copies intron columns - wrong!
2608 int startCol = alignedSeq.findIndex(seq.getStart()); // 1..
2609 int endCol = alignedSeq.findIndex(seq.getEnd());
2610 char[] seqchars = new char[endCol - leftmost + 1];
2611 Arrays.fill(seqchars, gapCharacter);
2612 char[] toCopy = alignedSeq.getSequence(startCol - 1, endCol);
2613 System.arraycopy(toCopy, 0, seqchars, startCol - leftmost,
2615 seq.setSequence(String.valueOf(seqchars));
2616 if (alignedSequences.size() > 0)
2618 // pop off aligned sequences (except the last one)
2619 alignedSequences.remove(0);
2624 * finally remove gapped columns (e.g. introns)
2626 new RemoveGapColCommand("", unaligned.getSequencesArray(), 0,
2627 unaligned.getWidth() - 1, unaligned);
2633 * Returns a map whose key is alignment column number (base 1), and whose
2634 * values are a map of sequence characters in that column.
2641 static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2642 AlignmentI unaligned, AlignmentI aligned,
2643 List<SequenceI> unmapped)
2646 * Map will hold, for each aligned column position, a map of
2647 * {unalignedSequence, characterPerSequence} at that position.
2648 * TreeMap keeps the entries in ascending column order.
2650 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2653 * record any sequences that have no mapping so can't be realigned
2655 unmapped.addAll(unaligned.getSequences());
2657 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2659 for (SequenceI seq : unaligned.getSequences())
2661 for (AlignedCodonFrame mapping : mappings)
2663 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2664 if (fromSeq != null)
2666 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2667 if (addMappedPositions(seq, fromSeq, seqMap, map))
2669 unmapped.remove(seq);
2678 * Helper method that adds to a map the mapped column positions of a sequence.
2680 * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2681 * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2685 * the sequence whose column positions we are recording
2687 * a sequence that is mapped to the first sequence
2689 * the mapping from 'fromSeq' to 'seq'
2691 * a map to add the column positions (in fromSeq) of the mapped
2695 static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
2696 Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
2704 * invert mapping if it is from unaligned to aligned sequence
2706 if (seqMap.getTo() == fromSeq.getDatasetSequence())
2708 seqMap = new Mapping(seq.getDatasetSequence(),
2709 seqMap.getMap().getInverse());
2712 int toStart = seq.getStart();
2715 * traverse [start, end, start, end...] ranges in fromSeq
2717 for (int[] fromRange : seqMap.getMap().getFromRanges())
2719 for (int i = 0; i < fromRange.length - 1; i += 2)
2721 boolean forward = fromRange[i + 1] >= fromRange[i];
2724 * find the range mapped to (sequence positions base 1)
2726 int[] range = seqMap.locateMappedRange(fromRange[i],
2730 System.err.println("Error in mapping " + seqMap + " from "
2731 + fromSeq.getName());
2734 int fromCol = fromSeq.findIndex(fromRange[i]);
2735 int mappedCharPos = range[0];
2738 * walk over the 'from' aligned sequence in forward or reverse
2739 * direction; when a non-gap is found, record the column position
2740 * of the next character of the mapped-to sequence; stop when all
2741 * the characters of the range have been counted
2743 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
2746 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
2749 * mapped from sequence has a character in this column
2750 * record the column position for the mapped to character
2752 Map<SequenceI, Character> seqsMap = map.get(fromCol);
2753 if (seqsMap == null)
2755 seqsMap = new HashMap<>();
2756 map.put(fromCol, seqsMap);
2758 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
2761 fromCol += (forward ? 1 : -1);
2768 // strictly temporary hack until proper criteria for aligning protein to cds
2769 // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
2770 public static boolean looksLikeEnsembl(AlignmentI alignment)
2772 for (SequenceI seq : alignment.getSequences())
2774 String name = seq.getName();
2775 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))