Clover icon

jalviewX

  1. Project Clover database Wed Oct 31 2018 15:13:58 GMT
  2. Package jalview.analysis

File AlignmentUtils.java

 

Coverage histogram

../../img/srcFileCovDistChart9.png
12% of files have more coverage

Code metrics

438
889
48
2
3,089
1,868
319
0.36
18.52
24
6.65

Classes

Class Line # Actions
AlignmentUtils 75 883 313 214
0.8427626584.3%
AlignmentUtils.DnaVariant 88 6 6 5
0.6428571364.3%
 

Contributing tests

This file is covered by 89 tests. .

Source view

1    /*
2    * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3    * Copyright (C) $$Year-Rel$$ The Jalview Authors
4    *
5    * This file is part of Jalview.
6    *
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.
11    *
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.
16    *
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.
20    */
21    package jalview.analysis;
22   
23    import static jalview.io.gff.GffConstants.CLINICAL_SIGNIFICANCE;
24   
25    import jalview.datamodel.AlignedCodon;
26    import jalview.datamodel.AlignedCodonFrame;
27    import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
28    import jalview.datamodel.Alignment;
29    import jalview.datamodel.AlignmentAnnotation;
30    import jalview.datamodel.AlignmentI;
31    import jalview.datamodel.DBRefEntry;
32    import jalview.datamodel.GeneLociI;
33    import jalview.datamodel.IncompleteCodonException;
34    import jalview.datamodel.Mapping;
35    import jalview.datamodel.Sequence;
36    import jalview.datamodel.SequenceFeature;
37    import jalview.datamodel.SequenceGroup;
38    import jalview.datamodel.SequenceI;
39    import jalview.datamodel.features.SequenceFeatures;
40    import jalview.io.gff.Gff3Helper;
41    import jalview.io.gff.SequenceOntologyI;
42    import jalview.schemes.ResidueProperties;
43    import jalview.util.Comparison;
44    import jalview.util.DBRefUtils;
45    import jalview.util.IntRangeComparator;
46    import jalview.util.MapList;
47    import jalview.util.MappingUtils;
48    import jalview.util.StringUtils;
49   
50    import java.io.UnsupportedEncodingException;
51    import java.net.URLEncoder;
52    import java.util.ArrayList;
53    import java.util.Arrays;
54    import java.util.Collection;
55    import java.util.Collections;
56    import java.util.HashMap;
57    import java.util.HashSet;
58    import java.util.Iterator;
59    import java.util.LinkedHashMap;
60    import java.util.List;
61    import java.util.Map;
62    import java.util.Map.Entry;
63    import java.util.NoSuchElementException;
64    import java.util.Set;
65    import java.util.SortedMap;
66    import java.util.TreeMap;
67   
68    /**
69    * grab bag of useful alignment manipulation operations Expect these to be
70    * refactored elsewhere at some point.
71    *
72    * @author jimp
73    *
74    */
 
75    public class AlignmentUtils
76    {
77   
78    private static final int CODON_LENGTH = 3;
79   
80    private static final String SEQUENCE_VARIANT = "sequence_variant:";
81   
82    private static final String ID = "ID";
83   
84    /**
85    * A data model to hold the 'normal' base value at a position, and an optional
86    * sequence variant feature
87    */
 
88    static final class DnaVariant
89    {
90    final String base;
91   
92    SequenceFeature variant;
93   
 
94  25 toggle DnaVariant(String nuc)
95    {
96  25 base = nuc;
97  25 variant = null;
98    }
99   
 
100  17 toggle DnaVariant(String nuc, SequenceFeature var)
101    {
102  17 base = nuc;
103  17 variant = var;
104    }
105   
 
106  14 toggle public String getSource()
107    {
108  14 return variant == null ? null : variant.getFeatureGroup();
109    }
110   
111    /**
112    * toString for aid in the debugger only
113    */
 
114  0 toggle @Override
115    public String toString()
116    {
117  0 return base + ":" + (variant == null ? "" : variant.getDescription());
118    }
119    }
120   
121    /**
122    * given an existing alignment, create a new alignment including all, or up to
123    * flankSize additional symbols from each sequence's dataset sequence
124    *
125    * @param core
126    * @param flankSize
127    * @return AlignmentI
128    */
 
129  27 toggle public static AlignmentI expandContext(AlignmentI core, int flankSize)
130    {
131  27 List<SequenceI> sq = new ArrayList<>();
132  27 int maxoffset = 0;
133  27 for (SequenceI s : core.getSequences())
134    {
135  131 SequenceI newSeq = s.deriveSequence();
136  131 final int newSeqStart = newSeq.getStart() - 1;
137  131 if (newSeqStart > maxoffset
138    && newSeq.getDatasetSequence().getStart() < s.getStart())
139    {
140  131 maxoffset = newSeqStart;
141    }
142  131 sq.add(newSeq);
143    }
144  27 if (flankSize > -1)
145    {
146  25 maxoffset = Math.min(maxoffset, flankSize);
147    }
148   
149    /*
150    * now add offset left and right to create an expanded alignment
151    */
152  27 for (SequenceI s : sq)
153    {
154  131 SequenceI ds = s;
155  262 while (ds.getDatasetSequence() != null)
156    {
157  131 ds = ds.getDatasetSequence();
158    }
159  131 int s_end = s.findPosition(s.getStart() + s.getLength());
160    // find available flanking residues for sequence
161  131 int ustream_ds = s.getStart() - ds.getStart();
162  131 int dstream_ds = ds.getEnd() - s_end;
163   
164    // build new flanked sequence
165   
166    // compute gap padding to start of flanking sequence
167  131 int offset = maxoffset - ustream_ds;
168   
169    // padding is gapChar x ( maxoffset - min(ustream_ds, flank)
170  131 if (flankSize >= 0)
171    {
172  125 if (flankSize < ustream_ds)
173    {
174    // take up to flankSize residues
175  40 offset = maxoffset - flankSize;
176  40 ustream_ds = flankSize;
177    }
178  125 if (flankSize <= dstream_ds)
179    {
180  116 dstream_ds = flankSize - 1;
181    }
182    }
183    // TODO use Character.toLowerCase to avoid creating String objects?
184  131 char[] upstream = new String(ds
185    .getSequence(s.getStart() - 1 - ustream_ds, s.getStart() - 1))
186    .toLowerCase().toCharArray();
187  131 char[] downstream = new String(
188    ds.getSequence(s_end - 1, s_end + dstream_ds)).toLowerCase()
189    .toCharArray();
190  131 char[] coreseq = s.getSequence();
191  131 char[] nseq = new char[offset + upstream.length + downstream.length
192    + coreseq.length];
193  131 char c = core.getGapCharacter();
194   
195  131 int p = 0;
196  461 for (; p < offset; p++)
197    {
198  330 nseq[p] = c;
199    }
200   
201  131 System.arraycopy(upstream, 0, nseq, p, upstream.length);
202  131 System.arraycopy(coreseq, 0, nseq, p + upstream.length,
203    coreseq.length);
204  131 System.arraycopy(downstream, 0, nseq,
205    p + coreseq.length + upstream.length, downstream.length);
206  131 s.setSequence(new String(nseq));
207  131 s.setStart(s.getStart() - ustream_ds);
208  131 s.setEnd(s_end + downstream.length);
209    }
210  27 AlignmentI newAl = new jalview.datamodel.Alignment(
211    sq.toArray(new SequenceI[0]));
212  27 for (SequenceI s : sq)
213    {
214  131 if (s.getAnnotation() != null)
215    {
216  1 for (AlignmentAnnotation aa : s.getAnnotation())
217    {
218  1 aa.adjustForAlignment(); // JAL-1712 fix
219  1 newAl.addAnnotation(aa);
220    }
221    }
222    }
223  27 newAl.setDataset(core.getDataset());
224  27 return newAl;
225    }
226   
227    /**
228    * Returns the index (zero-based position) of a sequence in an alignment, or
229    * -1 if not found.
230    *
231    * @param al
232    * @param seq
233    * @return
234    */
 
235  57660 toggle public static int getSequenceIndex(AlignmentI al, SequenceI seq)
236    {
237  57660 int result = -1;
238  57660 int pos = 0;
239  57660 for (SequenceI alSeq : al.getSequences())
240    {
241  125897562 if (alSeq == seq)
242    {
243  57660 result = pos;
244  57660 break;
245    }
246  125839902 pos++;
247    }
248  57660 return result;
249    }
250   
251    /**
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
254    * sequences.
255    *
256    * @see jalview.datamodel.AlignmentI#getSequencesByName()
257    */
 
258  1 toggle public static Map<String, List<SequenceI>> getSequencesByName(
259    AlignmentI al)
260    {
261  1 Map<String, List<SequenceI>> theMap = new LinkedHashMap<>();
262  1 for (SequenceI seq : al.getSequences())
263    {
264  3 String name = seq.getName();
265  3 if (name != null)
266    {
267  3 List<SequenceI> seqs = theMap.get(name);
268  3 if (seqs == null)
269    {
270  2 seqs = new ArrayList<>();
271  2 theMap.put(name, seqs);
272    }
273  3 seqs.add(seq);
274    }
275    }
276  1 return theMap;
277    }
278   
279    /**
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.
284    *
285    * @param proteinAlignment
286    * @param cdnaAlignment
287    * @return
288    */
 
289  4 toggle public static boolean mapProteinAlignmentToCdna(
290    final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment)
291    {
292  4 if (proteinAlignment == null || cdnaAlignment == null)
293    {
294  0 return false;
295    }
296   
297  4 Set<SequenceI> mappedDna = new HashSet<>();
298  4 Set<SequenceI> mappedProtein = new HashSet<>();
299   
300    /*
301    * First pass - map sequences where cross-references exist. This include
302    * 1-to-many mappings to support, for example, variant cDNA.
303    */
304  4 boolean mappingPerformed = mapProteinToCdna(proteinAlignment,
305    cdnaAlignment, mappedDna, mappedProtein, true);
306   
307    /*
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.
311    */
312  4 mappingPerformed |= mapProteinToCdna(proteinAlignment, cdnaAlignment,
313    mappedDna, mappedProtein, false);
314  4 return mappingPerformed;
315    }
316   
317    /**
318    * Make mappings between compatible sequences (where the cDNA translation
319    * matches the protein).
320    *
321    * @param proteinAlignment
322    * @param cdnaAlignment
323    * @param mappedDna
324    * a set of mapped DNA sequences (to add to)
325    * @param mappedProtein
326    * a set of mapped Protein sequences (to add to)
327    * @param xrefsOnly
328    * if true, only map sequences where xrefs exist
329    * @return
330    */
 
331  8 toggle protected static boolean mapProteinToCdna(
332    final AlignmentI proteinAlignment, final AlignmentI cdnaAlignment,
333    Set<SequenceI> mappedDna, Set<SequenceI> mappedProtein,
334    boolean xrefsOnly)
335    {
336  8 boolean mappingExistsOrAdded = false;
337  8 List<SequenceI> thisSeqs = proteinAlignment.getSequences();
338  8 for (SequenceI aaSeq : thisSeqs)
339    {
340  22 boolean proteinMapped = false;
341  22 AlignedCodonFrame acf = new AlignedCodonFrame();
342   
343  22 for (SequenceI cdnaSeq : cdnaAlignment.getSequences())
344    {
345    /*
346    * Always try to map if sequences have xref to each other; this supports
347    * variant cDNA or alternative splicing for a protein sequence.
348    *
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
352    * sequences.
353    */
354  86 if (xrefsOnly && !AlignmentUtils.haveCrossRef(aaSeq, cdnaSeq))
355    {
356  39 continue;
357    }
358   
359    /*
360    * Don't map non-xrefd sequences more than once each. This heuristic
361    * allows us to pair up similar sequences in ordered alignments.
362    */
363  47 if (!xrefsOnly && (mappedProtein.contains(aaSeq)
364    || mappedDna.contains(cdnaSeq)))
365    {
366  29 continue;
367    }
368  18 if (mappingExists(proteinAlignment.getCodonFrames(),
369    aaSeq.getDatasetSequence(), cdnaSeq.getDatasetSequence()))
370    {
371  0 mappingExistsOrAdded = true;
372    }
373    else
374    {
375  18 MapList map = mapCdnaToProtein(aaSeq, cdnaSeq);
376  18 if (map != null)
377    {
378  12 acf.addMap(cdnaSeq, aaSeq, map);
379  12 mappingExistsOrAdded = true;
380  12 proteinMapped = true;
381  12 mappedDna.add(cdnaSeq);
382  12 mappedProtein.add(aaSeq);
383    }
384    }
385    }
386  22 if (proteinMapped)
387    {
388  11 proteinAlignment.addCodonFrame(acf);
389    }
390    }
391  8 return mappingExistsOrAdded;
392    }
393   
394    /**
395    * Answers true if the mappings include one between the given (dataset)
396    * sequences.
397    */
 
398  18 toggle protected static boolean mappingExists(List<AlignedCodonFrame> mappings,
399    SequenceI aaSeq, SequenceI cdnaSeq)
400    {
401  18 if (mappings != null)
402    {
403  18 for (AlignedCodonFrame acf : mappings)
404    {
405  14 if (cdnaSeq == acf.getDnaForAaSeq(aaSeq))
406    {
407  0 return true;
408    }
409    }
410    }
411  18 return false;
412    }
413   
414    /**
415    * Builds a mapping (if possible) of a cDNA to a protein sequence.
416    * <ul>
417    * <li>first checks if the cdna translates exactly to the protein
418    * sequence</li>
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>
422    * </ul>
423    * Returns null if no mapping is determined.
424    *
425    * @param proteinSeq
426    * the aligned protein sequence
427    * @param cdnaSeq
428    * the aligned cdna sequence
429    * @return
430    */
 
431  25 toggle public static MapList mapCdnaToProtein(SequenceI proteinSeq,
432    SequenceI cdnaSeq)
433    {
434    /*
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
437    * String objects.
438    */
439  25 final SequenceI proteinDataset = proteinSeq.getDatasetSequence();
440  25 char[] aaSeqChars = proteinDataset != null
441    ? proteinDataset.getSequence()
442    : proteinSeq.getSequence();
443  25 final SequenceI cdnaDataset = cdnaSeq.getDatasetSequence();
444  25 char[] cdnaSeqChars = cdnaDataset != null ? cdnaDataset.getSequence()
445    : cdnaSeq.getSequence();
446  25 if (aaSeqChars == null || cdnaSeqChars == null)
447    {
448  0 return null;
449    }
450   
451    /*
452    * cdnaStart/End, proteinStartEnd are base 1 (for dataset sequence mapping)
453    */
454  25 final int mappedLength = CODON_LENGTH * aaSeqChars.length;
455  25 int cdnaLength = cdnaSeqChars.length;
456  25 int cdnaStart = cdnaSeq.getStart();
457  25 int cdnaEnd = cdnaSeq.getEnd();
458  25 final int proteinStart = proteinSeq.getStart();
459  25 final int proteinEnd = proteinSeq.getEnd();
460   
461    /*
462    * If lengths don't match, try ignoring stop codon (if present)
463    */
464  25 if (cdnaLength != mappedLength && cdnaLength > 2)
465    {
466  10 String lastCodon = String.valueOf(cdnaSeqChars,
467    cdnaLength - CODON_LENGTH, CODON_LENGTH).toUpperCase();
468  10 for (String stop : ResidueProperties.STOP_CODONS)
469    {
470  27 if (lastCodon.equals(stop))
471    {
472  3 cdnaEnd -= CODON_LENGTH;
473  3 cdnaLength -= CODON_LENGTH;
474  3 break;
475    }
476    }
477    }
478   
479    /*
480    * If lengths still don't match, try ignoring start codon.
481    */
482  25 int startOffset = 0;
483  25 if (cdnaLength != mappedLength && cdnaLength > 2
484    && String.valueOf(cdnaSeqChars, 0, CODON_LENGTH).toUpperCase()
485    .equals(ResidueProperties.START))
486    {
487  5 startOffset += CODON_LENGTH;
488  5 cdnaStart += CODON_LENGTH;
489  5 cdnaLength -= CODON_LENGTH;
490    }
491   
492  25 if (translatesAs(cdnaSeqChars, startOffset, aaSeqChars))
493    {
494    /*
495    * protein is translation of dna (+/- start/stop codons)
496    */
497  15 MapList map = new MapList(new int[] { cdnaStart, cdnaEnd },
498    new int[]
499    { proteinStart, proteinEnd }, CODON_LENGTH, 1);
500  15 return map;
501    }
502   
503    /*
504    * translation failed - try mapping CDS annotated regions of dna
505    */
506  10 return mapCdsToProtein(cdnaSeq, proteinSeq);
507    }
508   
509    /**
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.
513    *
514    * @param cdnaSeqChars
515    * @param cdnaStart
516    * @param aaSeqChars
517    * @return
518    */
 
519  45 toggle protected static boolean translatesAs(char[] cdnaSeqChars, int cdnaStart,
520    char[] aaSeqChars)
521    {
522  45 if (cdnaSeqChars == null || aaSeqChars == null)
523    {
524  3 return false;
525    }
526   
527  42 int aaPos = 0;
528  42 int dnaPos = cdnaStart;
529  155 for (; dnaPos < cdnaSeqChars.length - 2
530    && aaPos < aaSeqChars.length; dnaPos += CODON_LENGTH, aaPos++)
531    {
532  124 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
533  124 final String translated = ResidueProperties.codonTranslate(codon);
534   
535    /*
536    * allow * in protein to match untranslatable in dna
537    */
538  124 final char aaRes = aaSeqChars[aaPos];
539  124 if ((translated == null || ResidueProperties.STOP.equals(translated))
540    && aaRes == '*')
541    {
542  4 continue;
543    }
544  120 if (translated == null || !(aaRes == translated.charAt(0)))
545    {
546    // debug
547    // System.out.println(("Mismatch at " + i + "/" + aaResidue + ": "
548    // + codon + "(" + translated + ") != " + aaRes));
549  11 return false;
550    }
551    }
552   
553    /*
554    * check we matched all of the protein sequence
555    */
556  31 if (aaPos != aaSeqChars.length)
557    {
558  2 return false;
559    }
560   
561    /*
562    * check we matched all of the dna except
563    * for optional trailing STOP codon
564    */
565  29 if (dnaPos == cdnaSeqChars.length)
566    {
567  17 return true;
568    }
569  12 if (dnaPos == cdnaSeqChars.length - CODON_LENGTH)
570    {
571  11 String codon = String.valueOf(cdnaSeqChars, dnaPos, CODON_LENGTH);
572  11 if (ResidueProperties.STOP
573    .equals(ResidueProperties.codonTranslate(codon)))
574    {
575  9 return true;
576    }
577    }
578  3 return false;
579    }
580   
581    /**
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.
584    *
585    * @param seq
586    * the sequence to be realigned
587    * @param al
588    * the alignment whose sequence alignment is to be 'copied'
589    * @param gap
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
594    */
 
595  8 toggle public static boolean alignSequenceAs(SequenceI seq, AlignmentI al,
596    String gap, boolean preserveMappedGaps,
597    boolean preserveUnmappedGaps)
598    {
599    /*
600    * Get any mappings from the source alignment to the target (dataset)
601    * sequence.
602    */
603    // TODO there may be one AlignedCodonFrame per dataset sequence, or one with
604    // all mappings. Would it help to constrain this?
605  8 List<AlignedCodonFrame> mappings = al.getCodonFrame(seq);
606  8 if (mappings == null || mappings.isEmpty())
607    {
608  0 return false;
609    }
610   
611    /*
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
614    * sequence).
615    */
616  8 SequenceI alignFrom = null;
617  8 AlignedCodonFrame mapping = null;
618  8 for (AlignedCodonFrame mp : mappings)
619    {
620  8 alignFrom = mp.findAlignedSequence(seq, al);
621  8 if (alignFrom != null)
622    {
623  4 mapping = mp;
624  4 break;
625    }
626    }
627   
628  8 if (alignFrom == null)
629    {
630  4 return false;
631    }
632  4 alignSequenceAs(seq, alignFrom, mapping, gap, al.getGapCharacter(),
633    preserveMappedGaps, preserveUnmappedGaps);
634  4 return true;
635    }
636   
637    /**
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.
642    *
643    * @param alignTo
644    * @param alignFrom
645    * @param mapping
646    * @param myGap
647    * @param sourceGap
648    * @param preserveUnmappedGaps
649    * @param preserveMappedGaps
650    */
 
651  19 toggle public static void alignSequenceAs(SequenceI alignTo, SequenceI alignFrom,
652    AlignedCodonFrame mapping, String myGap, char sourceGap,
653    boolean preserveMappedGaps, boolean preserveUnmappedGaps)
654    {
655    // TODO generalise to work for Protein-Protein, dna-dna, dna-protein
656   
657    // aligned and dataset sequence positions, all base zero
658  19 int thisSeqPos = 0;
659  19 int sourceDsPos = 0;
660   
661  19 int basesWritten = 0;
662  19 char myGapChar = myGap.charAt(0);
663  19 int ratio = myGap.length();
664   
665  19 int fromOffset = alignFrom.getStart() - 1;
666  19 int toOffset = alignTo.getStart() - 1;
667  19 int sourceGapMappedLength = 0;
668  19 boolean inExon = false;
669  19 final int toLength = alignTo.getLength();
670  19 final int fromLength = alignFrom.getLength();
671  19 StringBuilder thisAligned = new StringBuilder(2 * toLength);
672   
673    /*
674    * Traverse the 'model' aligned sequence
675    */
676  205 for (int i = 0; i < fromLength; i++)
677    {
678  186 char sourceChar = alignFrom.getCharAt(i);
679  186 if (sourceChar == sourceGap)
680    {
681  44 sourceGapMappedLength += ratio;
682  44 continue;
683    }
684   
685    /*
686    * Found a non-gap character. Locate its mapped region if any.
687    */
688  142 sourceDsPos++;
689    // Note mapping positions are base 1, our sequence positions base 0
690  142 int[] mappedPos = mapping.getMappedRegion(alignTo, alignFrom,
691    sourceDsPos + fromOffset);
692  142 if (mappedPos == null)
693    {
694    /*
695    * unmapped position; treat like a gap
696    */
697  94 sourceGapMappedLength += ratio;
698    // System.err.println("Can't align: no codon mapping to residue "
699    // + sourceDsPos + "(" + sourceChar + ")");
700    // return;
701  94 continue;
702    }
703   
704  48 int mappedCodonStart = mappedPos[0]; // position (1...) of codon start
705  48 int mappedCodonEnd = mappedPos[mappedPos.length - 1]; // codon end pos
706  48 StringBuilder trailingCopiedGap = new StringBuilder();
707   
708    /*
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
711    * (in exons).
712    *
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.
715    */
716  48 int intronLength = 0;
717  294 while (basesWritten + toOffset < mappedCodonEnd
718    && thisSeqPos < toLength)
719    {
720  246 final char c = alignTo.getCharAt(thisSeqPos++);
721  246 if (c != myGapChar)
722    {
723  146 basesWritten++;
724  146 int sourcePosition = basesWritten + toOffset;
725  146 if (sourcePosition < mappedCodonStart)
726    {
727    /*
728    * Found an unmapped (intron) base. First add in any preceding gaps
729    * (if wanted).
730    */
731  48 if (preserveUnmappedGaps && trailingCopiedGap.length() > 0)
732    {
733  17 thisAligned.append(trailingCopiedGap.toString());
734  17 intronLength += trailingCopiedGap.length();
735  17 trailingCopiedGap = new StringBuilder();
736    }
737  48 intronLength++;
738  48 inExon = false;
739    }
740    else
741    {
742  98 final boolean startOfCodon = sourcePosition == mappedCodonStart;
743  98 int gapsToAdd = calculateGapsToInsert(preserveMappedGaps,
744    preserveUnmappedGaps, sourceGapMappedLength, inExon,
745    trailingCopiedGap.length(), intronLength, startOfCodon);
746  215 for (int k = 0; k < gapsToAdd; k++)
747    {
748  117 thisAligned.append(myGapChar);
749    }
750  98 sourceGapMappedLength = 0;
751  98 inExon = true;
752    }
753  146 thisAligned.append(c);
754  146 trailingCopiedGap = new StringBuilder();
755    }
756    else
757    {
758  100 if (inExon && preserveMappedGaps)
759    {
760  32 trailingCopiedGap.append(myGapChar);
761    }
762  68 else if (!inExon && preserveUnmappedGaps)
763    {
764  27 trailingCopiedGap.append(myGapChar);
765    }
766    }
767    }
768    }
769   
770    /*
771    * At end of model aligned sequence. Copy any remaining target sequence, optionally
772    * including (intron) gaps.
773    */
774  129 while (thisSeqPos < toLength)
775    {
776  110 final char c = alignTo.getCharAt(thisSeqPos++);
777  110 if (c != myGapChar || preserveUnmappedGaps)
778    {
779  102 thisAligned.append(c);
780    }
781  110 sourceGapMappedLength--;
782    }
783   
784    /*
785    * finally add gaps to pad for any trailing source gaps or
786    * unmapped characters
787    */
788  19 if (preserveUnmappedGaps)
789    {
790  24 while (sourceGapMappedLength > 0)
791    {
792  12 thisAligned.append(myGapChar);
793  12 sourceGapMappedLength--;
794    }
795    }
796   
797    /*
798    * All done aligning, set the aligned sequence.
799    */
800  19 alignTo.setSequence(new String(thisAligned));
801    }
802   
803    /**
804    * Helper method to work out how many gaps to insert when realigning.
805    *
806    * @param preserveMappedGaps
807    * @param preserveUnmappedGaps
808    * @param sourceGapMappedLength
809    * @param inExon
810    * @param trailingCopiedGap
811    * @param intronLength
812    * @param startOfCodon
813    * @return
814    */
 
815  98 toggle protected static int calculateGapsToInsert(boolean preserveMappedGaps,
816    boolean preserveUnmappedGaps, int sourceGapMappedLength,
817    boolean inExon, int trailingGapLength, int intronLength,
818    final boolean startOfCodon)
819    {
820  98 int gapsToAdd = 0;
821  98 if (startOfCodon)
822    {
823    /*
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
827    * region.
828    */
829  40 if (inExon && !preserveMappedGaps)
830    {
831  4 trailingGapLength = 0;
832    }
833  40 if (!inExon && !(preserveMappedGaps && preserveUnmappedGaps))
834    {
835  19 trailingGapLength = 0;
836    }
837  40 if (inExon)
838    {
839  14 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
840    }
841    else
842    {
843  26 if (intronLength + trailingGapLength <= sourceGapMappedLength)
844    {
845  20 gapsToAdd = sourceGapMappedLength - intronLength;
846    }
847    else
848    {
849  6 gapsToAdd = Math.min(
850    intronLength + trailingGapLength - sourceGapMappedLength,
851    trailingGapLength);
852    }
853    }
854    }
855    else
856    {
857    /*
858    * second or third base of codon; check for any gaps in dna
859    */
860  58 if (!preserveMappedGaps)
861    {
862  32 trailingGapLength = 0;
863    }
864  58 gapsToAdd = Math.max(sourceGapMappedLength, trailingGapLength);
865    }
866  98 return gapsToAdd;
867    }
868   
869    /**
870    * Realigns the given protein to match the alignment of the dna, using codon
871    * mappings to translate aligned codon positions to protein residues.
872    *
873    * @param protein
874    * the alignment whose sequences are realigned by this method
875    * @param dna
876    * the dna alignment whose alignment we are 'copying'
877    * @return the number of sequences that were realigned
878    */
 
879  3 toggle public static int alignProteinAsDna(AlignmentI protein, AlignmentI dna)
880    {
881  3 if (protein.isNucleotide() || !dna.isNucleotide())
882    {
883  0 System.err.println("Wrong alignment type in alignProteinAsDna");
884  0 return 0;
885    }
886  3 List<SequenceI> unmappedProtein = new ArrayList<>();
887  3 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = buildCodonColumnsMap(
888    protein, dna, unmappedProtein);
889  3 return alignProteinAs(protein, alignedCodons, unmappedProtein);
890    }
891   
892    /**
893    * Realigns the given dna to match the alignment of the protein, using codon
894    * mappings to translate aligned peptide positions to codons.
895    *
896    * Always produces a padded CDS alignment.
897    *
898    * @param dna
899    * the alignment whose sequences are realigned by this method
900    * @param protein
901    * the protein alignment whose alignment we are 'copying'
902    * @return the number of sequences that were realigned
903    */
 
904  4 toggle public static int alignCdsAsProtein(AlignmentI dna, AlignmentI protein)
905    {
906  4 if (protein.isNucleotide() || !dna.isNucleotide())
907    {
908  0 System.err.println("Wrong alignment type in alignProteinAsDna");
909  0 return 0;
910    }
911    // todo: implement this
912  4 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
913  4 int alignedCount = 0;
914  4 int width = 0; // alignment width for padding CDS
915  4 for (SequenceI dnaSeq : dna.getSequences())
916    {
917  5 if (alignCdsSequenceAsProtein(dnaSeq, protein, mappings,
918    dna.getGapCharacter()))
919    {
920  5 alignedCount++;
921    }
922  5 width = Math.max(dnaSeq.getLength(), width);
923    }
924  4 int oldwidth;
925  4 int diff;
926  4 for (SequenceI dnaSeq : dna.getSequences())
927    {
928  5 oldwidth = dnaSeq.getLength();
929  5 diff = width - oldwidth;
930  5 if (diff > 0)
931    {
932  1 dnaSeq.insertCharAt(oldwidth, diff, dna.getGapCharacter());
933    }
934    }
935  4 return alignedCount;
936    }
937   
938    /**
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.
942    *
943    * @param cdsSeq
944    * @param protein
945    * @param mappings
946    * @param gapChar
947    * @return
948    */
 
949  5 toggle static boolean alignCdsSequenceAsProtein(SequenceI cdsSeq,
950    AlignmentI protein, List<AlignedCodonFrame> mappings,
951    char gapChar)
952    {
953  5 SequenceI cdsDss = cdsSeq.getDatasetSequence();
954  5 if (cdsDss == null)
955    {
956  0 System.err
957    .println("alignCdsSequenceAsProtein needs aligned sequence!");
958  0 return false;
959    }
960   
961  5 List<AlignedCodonFrame> dnaMappings = MappingUtils
962    .findMappingsForSequence(cdsSeq, mappings);
963  5 for (AlignedCodonFrame mapping : dnaMappings)
964    {
965  5 SequenceI peptide = mapping.findAlignedSequence(cdsSeq, protein);
966  5 if (peptide != null)
967    {
968  5 final int peptideLength = peptide.getLength();
969  5 Mapping map = mapping.getMappingBetween(cdsSeq, peptide);
970  5 if (map != null)
971    {
972  5 MapList mapList = map.getMap();
973  5 if (map.getTo() == peptide.getDatasetSequence())
974    {
975  5 mapList = mapList.getInverse();
976    }
977  5 final int cdsLength = cdsDss.getLength();
978  5 int mappedFromLength = MappingUtils.getLength(mapList
979    .getFromRanges());
980  5 int mappedToLength = MappingUtils
981    .getLength(mapList.getToRanges());
982  5 boolean addStopCodon = (cdsLength == mappedFromLength
983    * CODON_LENGTH + CODON_LENGTH)
984    || (peptide.getDatasetSequence()
985    .getLength() == mappedFromLength - 1);
986  5 if (cdsLength != mappedToLength && !addStopCodon)
987    {
988  0 System.err.println(String.format(
989    "Can't align cds as protein (length mismatch %d/%d): %s",
990    cdsLength, mappedToLength, cdsSeq.getName()));
991    }
992   
993    /*
994    * pre-fill the aligned cds sequence with gaps
995    */
996  5 char[] alignedCds = new char[peptideLength * CODON_LENGTH
997  5 + (addStopCodon ? CODON_LENGTH : 0)];
998  5 Arrays.fill(alignedCds, gapChar);
999   
1000    /*
1001    * walk over the aligned peptide sequence and insert mapped
1002    * codons for residues in the aligned cds sequence
1003    */
1004  5 int copiedBases = 0;
1005  5 int cdsStart = cdsDss.getStart();
1006  5 int proteinPos = peptide.getStart() - 1;
1007  5 int cdsCol = 0;
1008   
1009  33 for (int col = 0; col < peptideLength; col++)
1010    {
1011  28 char residue = peptide.getCharAt(col);
1012   
1013  28 if (Comparison.isGap(residue))
1014    {
1015  13 cdsCol += CODON_LENGTH;
1016    }
1017    else
1018    {
1019  15 proteinPos++;
1020  15 int[] codon = mapList.locateInTo(proteinPos, proteinPos);
1021  15 if (codon == null)
1022    {
1023    // e.g. incomplete start codon, X in peptide
1024  0 cdsCol += CODON_LENGTH;
1025    }
1026    else
1027    {
1028  60 for (int j = codon[0]; j <= codon[1]; j++)
1029    {
1030  45 char mappedBase = cdsDss.getCharAt(j - cdsStart);
1031  45 alignedCds[cdsCol++] = mappedBase;
1032  45 copiedBases++;
1033    }
1034    }
1035    }
1036    }
1037   
1038    /*
1039    * append stop codon if not mapped from protein,
1040    * closing it up to the end of the mapped sequence
1041    */
1042  5 if (copiedBases == cdsLength - CODON_LENGTH)
1043    {
1044  0 for (int i = alignedCds.length - 1; i >= 0; i--)
1045    {
1046  0 if (!Comparison.isGap(alignedCds[i]))
1047    {
1048  0 cdsCol = i + 1; // gap just after end of sequence
1049  0 break;
1050    }
1051    }
1052  0 for (int i = cdsLength - CODON_LENGTH; i < cdsLength; i++)
1053    {
1054  0 alignedCds[cdsCol++] = cdsDss.getCharAt(i);
1055    }
1056    }
1057  5 cdsSeq.setSequence(new String(alignedCds));
1058  5 return true;
1059    }
1060    }
1061    }
1062  0 return false;
1063    }
1064   
1065    /**
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.
1071    *
1072    * @param protein
1073    * the protein alignment
1074    * @param dna
1075    * the coding dna alignment
1076    * @param unmappedProtein
1077    * any unmapped proteins are added to this list
1078    * @return
1079    */
 
1080  3 toggle protected static Map<AlignedCodon, Map<SequenceI, AlignedCodon>> buildCodonColumnsMap(
1081    AlignmentI protein, AlignmentI dna,
1082    List<SequenceI> unmappedProtein)
1083    {
1084    /*
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
1087    */
1088  3 unmappedProtein.addAll(protein.getSequences());
1089   
1090  3 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1091   
1092    /*
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.
1096    */
1097  3 Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons = new TreeMap<>(
1098    new CodonComparator());
1099   
1100  3 for (SequenceI dnaSeq : dna.getSequences())
1101    {
1102  8 for (AlignedCodonFrame mapping : mappings)
1103    {
1104  10 SequenceI prot = mapping.findAlignedSequence(dnaSeq, protein);
1105  10 if (prot != null)
1106    {
1107  10 Mapping seqMap = mapping.getMappingForSequence(dnaSeq);
1108  10 addCodonPositions(dnaSeq, prot, protein.getGapCharacter(), seqMap,
1109    alignedCodons);
1110  10 unmappedProtein.remove(prot);
1111    }
1112    }
1113    }
1114   
1115    /*
1116    * Finally add any unmapped peptide start residues (e.g. for incomplete
1117    * codons) as if at the codon position before the second residue
1118    */
1119    // TODO resolve JAL-2022 so this fudge can be removed
1120  3 int mappedSequenceCount = protein.getHeight() - unmappedProtein.size();
1121  3 addUnmappedPeptideStarts(alignedCodons, mappedSequenceCount);
1122   
1123  3 return alignedCodons;
1124    }
1125   
1126    /**
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
1130    *
1131    * @param alignedCodons
1132    * the codon-to-peptide map
1133    * @param mappedSequenceCount
1134    * the number of distinct sequences in the map
1135    */
 
1136  3 toggle protected static void addUnmappedPeptideStarts(
1137    Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1138    int mappedSequenceCount)
1139    {
1140    // TODO delete this ugly hack once JAL-2022 is resolved
1141    // i.e. we can model startPhase > 0 (incomplete start codon)
1142   
1143  3 List<SequenceI> sequencesChecked = new ArrayList<>();
1144  3 AlignedCodon lastCodon = null;
1145  3 Map<SequenceI, AlignedCodon> toAdd = new HashMap<>();
1146   
1147  3 for (Entry<AlignedCodon, Map<SequenceI, AlignedCodon>> entry : alignedCodons
1148    .entrySet())
1149    {
1150  21 for (Entry<SequenceI, AlignedCodon> sequenceCodon : entry.getValue()
1151    .entrySet())
1152    {
1153  25 SequenceI seq = sequenceCodon.getKey();
1154  25 if (sequencesChecked.contains(seq))
1155    {
1156  17 continue;
1157    }
1158  8 sequencesChecked.add(seq);
1159  8 AlignedCodon codon = sequenceCodon.getValue();
1160  8 if (codon.peptideCol > 1)
1161    {
1162  0 System.err.println(
1163    "Problem mapping protein with >1 unmapped start positions: "
1164    + seq.getName());
1165    }
1166  8 else if (codon.peptideCol == 1)
1167    {
1168    /*
1169    * first position (peptideCol == 0) was unmapped - add it
1170    */
1171  2 if (lastCodon != null)
1172    {
1173  1 AlignedCodon firstPeptide = new AlignedCodon(lastCodon.pos1,
1174    lastCodon.pos2, lastCodon.pos3,
1175    String.valueOf(seq.getCharAt(0)), 0);
1176  1 toAdd.put(seq, firstPeptide);
1177    }
1178    else
1179    {
1180    /*
1181    * unmapped residue at start of alignment (no prior column) -
1182    * 'insert' at nominal codon [0, 0, 0]
1183    */
1184  1 AlignedCodon firstPeptide = new AlignedCodon(0, 0, 0,
1185    String.valueOf(seq.getCharAt(0)), 0);
1186  1 toAdd.put(seq, firstPeptide);
1187    }
1188    }
1189  8 if (sequencesChecked.size() == mappedSequenceCount)
1190    {
1191    // no need to check past first mapped position in all sequences
1192  3 break;
1193    }
1194    }
1195  21 lastCodon = entry.getKey();
1196    }
1197   
1198    /*
1199    * add any new codons safely after iterating over the map
1200    */
1201  3 for (Entry<SequenceI, AlignedCodon> startCodon : toAdd.entrySet())
1202    {
1203  2 addCodonToMap(alignedCodons, startCodon.getValue(),
1204    startCodon.getKey());
1205    }
1206    }
1207   
1208    /**
1209    * Update the aligned protein sequences to match the codon alignments given in
1210    * the map.
1211    *
1212    * @param protein
1213    * @param alignedCodons
1214    * an ordered map of codon positions (columns), with sequence/peptide
1215    * values present in each column
1216    * @param unmappedProtein
1217    * @return
1218    */
 
1219  3 toggle protected static int alignProteinAs(AlignmentI protein,
1220    Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1221    List<SequenceI> unmappedProtein)
1222    {
1223    /*
1224    * prefill peptide sequences with gaps
1225    */
1226  3 int alignedWidth = alignedCodons.size();
1227  3 char[] gaps = new char[alignedWidth];
1228  3 Arrays.fill(gaps, protein.getGapCharacter());
1229  3 Map<SequenceI, char[]> peptides = new HashMap<>();
1230  3 for (SequenceI seq : protein.getSequences())
1231    {
1232  9 if (!unmappedProtein.contains(seq))
1233    {
1234  8 peptides.put(seq, Arrays.copyOf(gaps, gaps.length));
1235    }
1236    }
1237   
1238    /*
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.
1243    */
1244  3 int column = 0;
1245  3 for (AlignedCodon codon : alignedCodons.keySet())
1246    {
1247  22 final Map<SequenceI, AlignedCodon> columnResidues = alignedCodons
1248    .get(codon);
1249  22 for (Entry<SequenceI, AlignedCodon> entry : columnResidues.entrySet())
1250    {
1251  28 char residue = entry.getValue().product.charAt(0);
1252  28 peptides.get(entry.getKey())[column] = residue;
1253    }
1254  22 column++;
1255    }
1256   
1257    /*
1258    * and finally set the constructed sequences
1259    */
1260  3 for (Entry<SequenceI, char[]> entry : peptides.entrySet())
1261    {
1262  8 entry.getKey().setSequence(new String(entry.getValue()));
1263    }
1264   
1265  3 return 0;
1266    }
1267   
1268    /**
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.
1272    *
1273    * @param dna
1274    * the aligned sequence we are mapping from
1275    * @param protein
1276    * the sequence to be aligned to the codons
1277    * @param gapChar
1278    * the gap character in the dna sequence
1279    * @param seqMap
1280    * a mapping to a sequence translation
1281    * @param alignedCodons
1282    * the map we are building up
1283    */
 
1284  10 toggle static void addCodonPositions(SequenceI dna, SequenceI protein,
1285    char gapChar, Mapping seqMap,
1286    Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons)
1287    {
1288  10 Iterator<AlignedCodon> codons = seqMap.getCodonIterator(dna, gapChar);
1289   
1290    /*
1291    * add codon positions, and their peptide translations, to the alignment
1292    * map, while remembering the first codon mapped
1293    */
1294  44 while (codons.hasNext())
1295    {
1296  34 try
1297    {
1298  34 AlignedCodon codon = codons.next();
1299  34 addCodonToMap(alignedCodons, codon, protein);
1300    } catch (IncompleteCodonException e)
1301    {
1302    // possible incomplete trailing codon - ignore
1303    } catch (NoSuchElementException e)
1304    {
1305    // possibly peptide lacking STOP
1306    }
1307    }
1308    }
1309   
1310    /**
1311    * Helper method to add a codon-to-peptide entry to the aligned codons map
1312    *
1313    * @param alignedCodons
1314    * @param codon
1315    * @param protein
1316    */
 
1317  36 toggle protected static void addCodonToMap(
1318    Map<AlignedCodon, Map<SequenceI, AlignedCodon>> alignedCodons,
1319    AlignedCodon codon, SequenceI protein)
1320    {
1321  36 Map<SequenceI, AlignedCodon> seqProduct = alignedCodons.get(codon);
1322  36 if (seqProduct == null)
1323    {
1324  22 seqProduct = new HashMap<>();
1325  22 alignedCodons.put(codon, seqProduct);
1326    }
1327  36 seqProduct.put(protein, codon);
1328    }
1329   
1330    /**
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,
1333    * the logic is:
1334    * <ul>
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
1338    * sequence</li>
1339    * <li>The translation may ignore start and stop codons if present in the
1340    * nucleotide</li>
1341    * </ul>
1342    *
1343    * @param al1
1344    * @param al2
1345    * @return
1346    */
 
1347  8 toggle public static boolean isMappable(AlignmentI al1, AlignmentI al2)
1348    {
1349  8 if (al1 == null || al2 == null)
1350    {
1351  3 return false;
1352    }
1353   
1354    /*
1355    * Require one nucleotide and one protein
1356    */
1357  5 if (al1.isNucleotide() == al2.isNucleotide())
1358    {
1359  3 return false;
1360    }
1361  2 AlignmentI dna = al1.isNucleotide() ? al1 : al2;
1362  2 AlignmentI protein = dna == al1 ? al2 : al1;
1363  2 List<AlignedCodonFrame> mappings = protein.getCodonFrames();
1364  2 for (SequenceI dnaSeq : dna.getSequences())
1365    {
1366  2 for (SequenceI proteinSeq : protein.getSequences())
1367    {
1368  2 if (isMappable(dnaSeq, proteinSeq, mappings))
1369    {
1370  2 return true;
1371    }
1372    }
1373    }
1374  0 return false;
1375    }
1376   
1377    /**
1378    * Returns true if the dna sequence is mapped, or could be mapped, to the
1379    * protein sequence.
1380    *
1381    * @param dnaSeq
1382    * @param proteinSeq
1383    * @param mappings
1384    * @return
1385    */
 
1386  2 toggle protected static boolean isMappable(SequenceI dnaSeq,
1387    SequenceI proteinSeq, List<AlignedCodonFrame> mappings)
1388    {
1389  2 if (dnaSeq == null || proteinSeq == null)
1390    {
1391  0 return false;
1392    }
1393   
1394  2 SequenceI dnaDs = dnaSeq.getDatasetSequence() == null ? dnaSeq
1395    : dnaSeq.getDatasetSequence();
1396  2 SequenceI proteinDs = proteinSeq.getDatasetSequence() == null
1397    ? proteinSeq
1398    : proteinSeq.getDatasetSequence();
1399   
1400  2 for (AlignedCodonFrame mapping : mappings)
1401    {
1402  0 if (proteinDs == mapping.getAaForDnaSeq(dnaDs))
1403    {
1404    /*
1405    * already mapped
1406    */
1407  0 return true;
1408    }
1409    }
1410   
1411    /*
1412    * Just try to make a mapping (it is not yet stored), test whether
1413    * successful.
1414    */
1415  2 return mapCdnaToProtein(proteinDs, dnaDs) != null;
1416    }
1417   
1418    /**
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.
1423    *
1424    * @param sequenceScope
1425    * the sequences to scan for reference annotations
1426    * @param labelForCalcId
1427    * (optional) map to populate with label for calcId
1428    * @param candidates
1429    * map to populate with annotations for sequence
1430    * @param al
1431    * the alignment to check for presence of annotations
1432    */
 
1433  42 toggle public static void findAddableReferenceAnnotations(
1434    List<SequenceI> sequenceScope, Map<String, String> labelForCalcId,
1435    final Map<SequenceI, List<AlignmentAnnotation>> candidates,
1436    AlignmentI al)
1437    {
1438  42 if (sequenceScope == null)
1439    {
1440  1 return;
1441    }
1442   
1443    /*
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.
1446    *
1447    * Add to a map of { alignmentSequence, <List of annotations to add> }
1448    */
1449  41 for (SequenceI seq : sequenceScope)
1450    {
1451  25 SequenceI dataset = seq.getDatasetSequence();
1452  25 if (dataset == null)
1453    {
1454  0 continue;
1455    }
1456  25 AlignmentAnnotation[] datasetAnnotations = dataset.getAnnotation();
1457  25 if (datasetAnnotations == null)
1458    {
1459  19 continue;
1460    }
1461  6 final List<AlignmentAnnotation> result = new ArrayList<>();
1462  6 for (AlignmentAnnotation dsann : datasetAnnotations)
1463    {
1464    /*
1465    * Find matching annotations on the alignment. If none is found, then
1466    * add this annotation to the list of 'addable' annotations for this
1467    * sequence.
1468    */
1469  9 final Iterable<AlignmentAnnotation> matchedAlignmentAnnotations = al
1470    .findAnnotations(seq, dsann.getCalcId(), dsann.label);
1471  9 if (!matchedAlignmentAnnotations.iterator().hasNext())
1472    {
1473  6 result.add(dsann);
1474  6 if (labelForCalcId != null)
1475    {
1476  6 labelForCalcId.put(dsann.getCalcId(), dsann.label);
1477    }
1478    }
1479    }
1480    /*
1481    * Save any addable annotations for this sequence
1482    */
1483  6 if (!result.isEmpty())
1484    {
1485  4 candidates.put(seq, result);
1486    }
1487    }
1488    }
1489   
1490    /**
1491    * Adds annotations to the top of the alignment annotations, in the same order
1492    * as their related sequences.
1493    *
1494    * @param annotations
1495    * the annotations to add
1496    * @param alignment
1497    * the alignment to add them to
1498    * @param selectionGroup
1499    * current selection group (or null if none)
1500    */
 
1501  0 toggle public static void addReferenceAnnotations(
1502    Map<SequenceI, List<AlignmentAnnotation>> annotations,
1503    final AlignmentI alignment, final SequenceGroup selectionGroup)
1504    {
1505  0 for (SequenceI seq : annotations.keySet())
1506    {
1507  0 for (AlignmentAnnotation ann : annotations.get(seq))
1508    {
1509  0 AlignmentAnnotation copyAnn = new AlignmentAnnotation(ann);
1510  0 int startRes = 0;
1511  0 int endRes = ann.annotations.length;
1512  0 if (selectionGroup != null)
1513    {
1514  0 startRes = selectionGroup.getStartRes();
1515  0 endRes = selectionGroup.getEndRes();
1516    }
1517  0 copyAnn.restrict(startRes, endRes);
1518   
1519    /*
1520    * Add to the sequence (sets copyAnn.datasetSequence), unless the
1521    * original annotation is already on the sequence.
1522    */
1523  0 if (!seq.hasAnnotation(ann))
1524    {
1525  0 seq.addAlignmentAnnotation(copyAnn);
1526    }
1527    // adjust for gaps
1528  0 copyAnn.adjustForAlignment();
1529    // add to the alignment and set visible
1530  0 alignment.addAnnotation(copyAnn);
1531  0 copyAnn.visible = true;
1532    }
1533    }
1534    }
1535   
1536    /**
1537    * Set visibility of alignment annotations of specified types (labels), for
1538    * specified sequences. This supports controls like "Show all secondary
1539    * structure", "Hide all Temp factor", etc.
1540    *
1541    * @al the alignment to scan for annotations
1542    * @param types
1543    * the types (labels) of annotations to be updated
1544    * @param forSequences
1545    * if not null, only annotations linked to one of these sequences are
1546    * in scope for update; if null, acts on all sequence annotations
1547    * @param anyType
1548    * if this flag is true, 'types' is ignored (label not checked)
1549    * @param doShow
1550    * if true, set visibility on, else set off
1551    */
 
1552  5 toggle public static void showOrHideSequenceAnnotations(AlignmentI al,
1553    Collection<String> types, List<SequenceI> forSequences,
1554    boolean anyType, boolean doShow)
1555    {
1556  5 AlignmentAnnotation[] anns = al.getAlignmentAnnotation();
1557  5 if (anns != null)
1558    {
1559  5 for (AlignmentAnnotation aa : anns)
1560    {
1561  30 if (anyType || types.contains(aa.label))
1562    {
1563  21 if ((aa.sequenceRef != null) && (forSequences == null
1564    || forSequences.contains(aa.sequenceRef)))
1565    {
1566  11 aa.visible = doShow;
1567    }
1568    }
1569    }
1570    }
1571    }
1572   
1573    /**
1574    * Returns true if either sequence has a cross-reference to the other
1575    *
1576    * @param seq1
1577    * @param seq2
1578    * @return
1579    */
 
1580  52 toggle public static boolean haveCrossRef(SequenceI seq1, SequenceI seq2)
1581    {
1582    // Note: moved here from class CrossRef as the latter class has dependencies
1583    // not availability to the applet's classpath
1584  52 return hasCrossRef(seq1, seq2) || hasCrossRef(seq2, seq1);
1585    }
1586   
1587    /**
1588    * Returns true if seq1 has a cross-reference to seq2. Currently this assumes
1589    * that sequence name is structured as Source|AccessionId.
1590    *
1591    * @param seq1
1592    * @param seq2
1593    * @return
1594    */
 
1595  108 toggle public static boolean hasCrossRef(SequenceI seq1, SequenceI seq2)
1596    {
1597  108 if (seq1 == null || seq2 == null)
1598    {
1599  8 return false;
1600    }
1601  100 String name = seq2.getName();
1602  100 final DBRefEntry[] xrefs = seq1.getDBRefs();
1603  100 if (xrefs != null)
1604    {
1605  22 for (DBRefEntry xref : xrefs)
1606    {
1607  24 String xrefName = xref.getSource() + "|" + xref.getAccessionId();
1608    // case-insensitive test, consistent with DBRefEntry.equalRef()
1609  24 if (xrefName.equalsIgnoreCase(name))
1610    {
1611  12 return true;
1612    }
1613    }
1614    }
1615  88 return false;
1616    }
1617   
1618    /**
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.
1624    *
1625    * @param dna
1626    * aligned nucleotide (dna or cds) sequences
1627    * @param dataset
1628    * the alignment dataset the sequences belong to
1629    * @param products
1630    * (optional) to restrict results to CDS that map to specified
1631    * protein products
1632    * @return an alignment whose sequences are the cds-only parts of the dna
1633    * sequences (or null if no mappings are found)
1634    */
 
1635  4 toggle public static AlignmentI makeCdsAlignment(SequenceI[] dna,
1636    AlignmentI dataset, SequenceI[] products)
1637    {
1638  4 if (dataset == null || dataset.getDataset() != null)
1639    {
1640  0 throw new IllegalArgumentException(
1641    "IMPLEMENTATION ERROR: dataset.getDataset() must be null!");
1642    }
1643  4 List<SequenceI> foundSeqs = new ArrayList<>();
1644  4 List<SequenceI> cdsSeqs = new ArrayList<>();
1645  4 List<AlignedCodonFrame> mappings = dataset.getCodonFrames();
1646  4 HashSet<SequenceI> productSeqs = null;
1647  4 if (products != null)
1648    {
1649  1 productSeqs = new HashSet<>();
1650  1 for (SequenceI seq : products)
1651    {
1652  2 productSeqs.add(seq.getDatasetSequence() == null ? seq : seq
1653    .getDatasetSequence());
1654    }
1655    }
1656   
1657    /*
1658    * Construct CDS sequences from mappings on the alignment dataset.
1659    * The logic is:
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)
1667    */
1668  4 for (SequenceI dnaSeq : dna)
1669    {
1670  8 SequenceI dnaDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1671    : dnaSeq.getDatasetSequence();
1672   
1673  8 List<AlignedCodonFrame> seqMappings = MappingUtils
1674    .findMappingsForSequence(dnaSeq, mappings);
1675  8 for (AlignedCodonFrame mapping : seqMappings)
1676    {
1677  9 List<Mapping> mappingsFromSequence = mapping
1678    .getMappingsFromSequence(dnaSeq);
1679   
1680  9 for (Mapping aMapping : mappingsFromSequence)
1681    {
1682  11 MapList mapList = aMapping.getMap();
1683  11 if (mapList.getFromRatio() == 1)
1684    {
1685    /*
1686    * not a dna-to-protein mapping (likely dna-to-cds)
1687    */
1688  0 continue;
1689    }
1690   
1691    /*
1692    * skip if mapping is not to one of the target set of proteins
1693    */
1694  11 SequenceI proteinProduct = aMapping.getTo();
1695  11 if (productSeqs != null && !productSeqs.contains(proteinProduct))
1696    {
1697  2 continue;
1698    }
1699   
1700    /*
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)
1704    */
1705  9 SequenceI cdsSeq = findCdsForProtein(mappings, dnaSeq,
1706    seqMappings, aMapping);
1707  9 if (cdsSeq != null)
1708    {
1709  0 if (!foundSeqs.contains(cdsSeq))
1710    {
1711  0 foundSeqs.add(cdsSeq);
1712  0 SequenceI derivedSequence = cdsSeq.deriveSequence();
1713  0 cdsSeqs.add(derivedSequence);
1714  0 if (!dataset.getSequences().contains(cdsSeq))
1715    {
1716  0 dataset.addSequence(cdsSeq);
1717    }
1718    }
1719  0 continue;
1720    }
1721   
1722    /*
1723    * didn't find mapped CDS sequence - construct it and add
1724    * its dataset sequence to the dataset
1725    */
1726  9 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  9 SequenceI cdsSeqDss = cdsSeq.getDatasetSequence();
1733   
1734  9 cdsSeqs.add(cdsSeq);
1735   
1736  9 if (!dataset.getSequences().contains(cdsSeqDss))
1737    {
1738    // check if this sequence is a newly created one
1739    // so needs adding to the dataset
1740  9 dataset.addSequence(cdsSeqDss);
1741    }
1742   
1743    /*
1744    * add a mapping from CDS to the (unchanged) mapped to range
1745    */
1746  9 List<int[]> cdsRange = Collections.singletonList(new int[] { 1,
1747    cdsSeq.getLength() });
1748  9 MapList cdsToProteinMap = new MapList(cdsRange,
1749    mapList.getToRanges(), mapList.getFromRatio(),
1750    mapList.getToRatio());
1751  9 AlignedCodonFrame cdsToProteinMapping = new AlignedCodonFrame();
1752  9 cdsToProteinMapping.addMap(cdsSeqDss, proteinProduct,
1753    cdsToProteinMap);
1754   
1755    /*
1756    * guard against duplicating the mapping if repeating this action
1757    */
1758  9 if (!mappings.contains(cdsToProteinMapping))
1759    {
1760  9 mappings.add(cdsToProteinMapping);
1761    }
1762   
1763  9 propagateDBRefsToCDS(cdsSeqDss, dnaSeq.getDatasetSequence(),
1764    proteinProduct, aMapping);
1765    /*
1766    * add another mapping from original 'from' range to CDS
1767    */
1768  9 AlignedCodonFrame dnaToCdsMapping = new AlignedCodonFrame();
1769  9 final MapList dnaToCdsMap = new MapList(mapList.getFromRanges(),
1770    cdsRange, 1, 1);
1771  9 dnaToCdsMapping.addMap(dnaSeq.getDatasetSequence(), cdsSeqDss,
1772    dnaToCdsMap);
1773  9 if (!mappings.contains(dnaToCdsMapping))
1774    {
1775  9 mappings.add(dnaToCdsMapping);
1776    }
1777   
1778    /*
1779    * transfer dna chromosomal loci (if known) to the CDS
1780    * sequence (via the mapping)
1781    */
1782  9 final MapList cdsToDnaMap = dnaToCdsMap.getInverse();
1783  9 transferGeneLoci(dnaSeq, cdsToDnaMap, cdsSeq);
1784   
1785    /*
1786    * add DBRef with mapping from protein to CDS
1787    * (this enables Get Cross-References from protein alignment)
1788    * This is tricky because we can't have two DBRefs with the
1789    * same source and accession, so need a different accession for
1790    * the CDS from the dna sequence
1791    */
1792   
1793    // specific use case:
1794    // Genomic contig ENSCHR:1, contains coding regions for ENSG01,
1795    // ENSG02, ENSG03, with transcripts and products similarly named.
1796    // cannot add distinct dbrefs mapping location on ENSCHR:1 to ENSG01
1797   
1798    // JBPNote: ?? can't actually create an example that demonstrates we
1799    // need to
1800    // synthesize an xref.
1801   
1802  9 for (DBRefEntry primRef : dnaDss.getPrimaryDBRefs())
1803    {
1804    /*
1805    * create a cross-reference from CDS to the source sequence's
1806    * primary reference and vice versa
1807    */
1808  2 String source = primRef.getSource();
1809  2 String version = primRef.getVersion();
1810  2 DBRefEntry cdsCrossRef = new DBRefEntry(source, source + ":"
1811    + version, primRef.getAccessionId());
1812  2 cdsCrossRef.setMap(new Mapping(dnaDss, new MapList(cdsToDnaMap)));
1813  2 cdsSeqDss.addDBRef(cdsCrossRef);
1814   
1815  2 dnaSeq.addDBRef(new DBRefEntry(source, version, cdsSeq
1816    .getName(), new Mapping(cdsSeqDss, dnaToCdsMap)));
1817   
1818    // problem here is that the cross-reference is synthesized -
1819    // cdsSeq.getName() may be like 'CDS|dnaaccession' or
1820    // 'CDS|emblcdsacc'
1821    // assuming cds version same as dna ?!?
1822   
1823  2 DBRefEntry proteinToCdsRef = new DBRefEntry(source, version,
1824    cdsSeq.getName());
1825    //
1826  2 proteinToCdsRef.setMap(new Mapping(cdsSeqDss, cdsToProteinMap
1827    .getInverse()));
1828  2 proteinProduct.addDBRef(proteinToCdsRef);
1829    }
1830   
1831    /*
1832    * transfer any features on dna that overlap the CDS
1833    */
1834  9 transferFeatures(dnaSeq, cdsSeq, dnaToCdsMap, null,
1835    SequenceOntologyI.CDS);
1836    }
1837    }
1838    }
1839   
1840  4 AlignmentI cds = new Alignment(cdsSeqs.toArray(new SequenceI[cdsSeqs
1841    .size()]));
1842  4 cds.setDataset(dataset);
1843   
1844  4 return cds;
1845    }
1846   
1847    /**
1848    * Tries to transfer gene loci (dbref to chromosome positions) from fromSeq to
1849    * toSeq, mediated by the given mapping between the sequences
1850    *
1851    * @param fromSeq
1852    * @param targetToFrom
1853    * Map
1854    * @param targetSeq
1855    */
 
1856  12 toggle protected static void transferGeneLoci(SequenceI fromSeq,
1857    MapList targetToFrom, SequenceI targetSeq)
1858    {
1859  12 if (targetSeq.getGeneLoci() != null)
1860    {
1861    // already have - don't override
1862  1 return;
1863    }
1864  11 GeneLociI fromLoci = fromSeq.getGeneLoci();
1865  11 if (fromLoci == null)
1866    {
1867  10 return;
1868    }
1869   
1870  1 MapList newMap = targetToFrom.traverse(fromLoci.getMap());
1871   
1872  1 if (newMap != null)
1873    {
1874  1 targetSeq.setGeneLoci(fromLoci.getSpeciesId(),
1875    fromLoci.getAssemblyId(), fromLoci.getChromosomeId(), newMap);
1876    }
1877    }
1878   
1879    /**
1880    * A helper method that finds a CDS sequence in the alignment dataset that is
1881    * mapped to the given protein sequence, and either is, or has a mapping from,
1882    * the given dna sequence.
1883    *
1884    * @param mappings
1885    * set of all mappings on the dataset
1886    * @param dnaSeq
1887    * a dna (or cds) sequence we are searching from
1888    * @param seqMappings
1889    * the set of mappings involving dnaSeq
1890    * @param aMapping
1891    * a transcript-to-peptide mapping
1892    * @return
1893    */
 
1894  9 toggle static SequenceI findCdsForProtein(List<AlignedCodonFrame> mappings,
1895    SequenceI dnaSeq, List<AlignedCodonFrame> seqMappings,
1896    Mapping aMapping)
1897    {
1898    /*
1899    * TODO a better dna-cds-protein mapping data representation to allow easy
1900    * navigation; until then this clunky looping around lists of mappings
1901    */
1902  9 SequenceI seqDss = dnaSeq.getDatasetSequence() == null ? dnaSeq
1903    : dnaSeq.getDatasetSequence();
1904  9 SequenceI proteinProduct = aMapping.getTo();
1905   
1906    /*
1907    * is this mapping from the whole dna sequence (i.e. CDS)?
1908    * allowing for possible stop codon on dna but not peptide
1909    */
1910  9 int mappedFromLength = MappingUtils
1911    .getLength(aMapping.getMap().getFromRanges());
1912  9 int dnaLength = seqDss.getLength();
1913  9 if (mappedFromLength == dnaLength
1914    || mappedFromLength == dnaLength - CODON_LENGTH)
1915    {
1916    /*
1917    * if sequence has CDS features, this is a transcript with no UTR
1918    * - do not take this as the CDS sequence! (JAL-2789)
1919    */
1920  0 if (seqDss.getFeatures().getFeaturesByOntology(SequenceOntologyI.CDS)
1921    .isEmpty())
1922    {
1923  0 return seqDss;
1924    }
1925    }
1926   
1927    /*
1928    * looks like we found the dna-to-protein mapping; search for the
1929    * corresponding cds-to-protein mapping
1930    */
1931  9 List<AlignedCodonFrame> mappingsToPeptide = MappingUtils
1932    .findMappingsForSequence(proteinProduct, mappings);
1933  9 for (AlignedCodonFrame acf : mappingsToPeptide)
1934    {
1935  9 for (SequenceToSequenceMapping map : acf.getMappings())
1936    {
1937  11 Mapping mapping = map.getMapping();
1938  11 if (mapping != aMapping
1939    && mapping.getMap().getFromRatio() == CODON_LENGTH
1940    && proteinProduct == mapping.getTo()
1941    && seqDss != map.getFromSeq())
1942    {
1943  0 mappedFromLength = MappingUtils
1944    .getLength(mapping.getMap().getFromRanges());
1945  0 if (mappedFromLength == map.getFromSeq().getLength())
1946    {
1947    /*
1948    * found a 3:1 mapping to the protein product which covers
1949    * the whole dna sequence i.e. is from CDS; finally check the CDS
1950    * is mapped from the given dna start sequence
1951    */
1952  0 SequenceI cdsSeq = map.getFromSeq();
1953    // todo this test is weak if seqMappings contains multiple mappings;
1954    // we get away with it if transcript:cds relationship is 1:1
1955  0 List<AlignedCodonFrame> dnaToCdsMaps = MappingUtils
1956    .findMappingsForSequence(cdsSeq, seqMappings);
1957  0 if (!dnaToCdsMaps.isEmpty())
1958    {
1959  0 return cdsSeq;
1960    }
1961    }
1962    }
1963    }
1964    }
1965  9 return null;
1966    }
1967   
1968    /**
1969    * Helper method that makes a CDS sequence as defined by the mappings from the
1970    * given sequence i.e. extracts the 'mapped from' ranges (which may be on
1971    * forward or reverse strand).
1972    *
1973    * @param seq
1974    * @param mapping
1975    * @param dataset
1976    * - existing dataset. We check for sequences that look like the CDS
1977    * we are about to construct, if one exists already, then we will
1978    * just return that one.
1979    * @return CDS sequence (as a dataset sequence)
1980    */
 
1981  9 toggle static SequenceI makeCdsSequence(SequenceI seq, Mapping mapping,
1982    AlignmentI dataset)
1983    {
1984  9 char[] seqChars = seq.getSequence();
1985  9 List<int[]> fromRanges = mapping.getMap().getFromRanges();
1986  9 int cdsWidth = MappingUtils.getLength(fromRanges);
1987  9 char[] newSeqChars = new char[cdsWidth];
1988   
1989  9 int newPos = 0;
1990  9 for (int[] range : fromRanges)
1991    {
1992  21 if (range[0] <= range[1])
1993    {
1994    // forward strand mapping - just copy the range
1995  21 int length = range[1] - range[0] + 1;
1996  21 System.arraycopy(seqChars, range[0] - 1, newSeqChars, newPos,
1997    length);
1998  21 newPos += length;
1999    }
2000    else
2001    {
2002    // reverse strand mapping - copy and complement one by one
2003  0 for (int i = range[0]; i >= range[1]; i--)
2004    {
2005  0 newSeqChars[newPos++] = Dna.getComplement(seqChars[i - 1]);
2006    }
2007    }
2008    }
2009   
2010    /*
2011    * assign 'from id' held in the mapping if set (e.g. EMBL protein_id),
2012    * else generate a sequence name
2013    */
2014  9 String mapFromId = mapping.getMappedFromId();
2015  9 String seqId = "CDS|" + (mapFromId != null ? mapFromId : seq.getName());
2016  9 SequenceI newSeq = new Sequence(seqId, newSeqChars, 1, newPos);
2017  9 if (dataset != null)
2018    {
2019  9 SequenceI[] matches = dataset.findSequenceMatch(newSeq.getName());
2020  9 if (matches != null)
2021    {
2022  9 boolean matched = false;
2023  9 for (SequenceI mtch : matches)
2024    {
2025  3 if (mtch.getStart() != newSeq.getStart())
2026    {
2027  0 continue;
2028    }
2029  3 if (mtch.getEnd() != newSeq.getEnd())
2030    {
2031  0 continue;
2032    }
2033  3 if (!Arrays.equals(mtch.getSequence(), newSeq.getSequence()))
2034    {
2035  3 continue;
2036    }
2037  0 if (!matched)
2038    {
2039  0 matched = true;
2040  0 newSeq = mtch;
2041    }
2042    else
2043    {
2044  0 System.err.println(
2045    "JAL-2154 regression: warning - found (and ignnored a duplicate CDS sequence):"
2046    + mtch.toString());
2047    }
2048    }
2049    }
2050    }
2051    // newSeq.setDescription(mapFromId);
2052   
2053  9 return newSeq;
2054    }
2055   
2056    /**
2057    * Adds any DBRefEntrys to cdsSeq from contig that have a Mapping congruent to
2058    * the given mapping.
2059    *
2060    * @param cdsSeq
2061    * @param contig
2062    * @param proteinProduct
2063    * @param mapping
2064    * @return list of DBRefEntrys added
2065    */
 
2066  9 toggle protected static List<DBRefEntry> propagateDBRefsToCDS(SequenceI cdsSeq,
2067    SequenceI contig, SequenceI proteinProduct, Mapping mapping)
2068    {
2069   
2070    // gather direct refs from contig congruent with mapping
2071  9 List<DBRefEntry> direct = new ArrayList<>();
2072  9 HashSet<String> directSources = new HashSet<>();
2073   
2074  9 if (contig.getDBRefs() != null)
2075    {
2076  2 for (DBRefEntry dbr : contig.getDBRefs())
2077    {
2078  4 if (dbr.hasMap() && dbr.getMap().getMap().isTripletMap())
2079    {
2080  2 MapList map = dbr.getMap().getMap();
2081    // check if map is the CDS mapping
2082  2 if (mapping.getMap().equals(map))
2083    {
2084  2 direct.add(dbr);
2085  2 directSources.add(dbr.getSource());
2086    }
2087    }
2088    }
2089    }
2090  9 DBRefEntry[] onSource = DBRefUtils.selectRefs(
2091    proteinProduct.getDBRefs(),
2092    directSources.toArray(new String[0]));
2093  9 List<DBRefEntry> propagated = new ArrayList<>();
2094   
2095    // and generate appropriate mappings
2096  9 for (DBRefEntry cdsref : direct)
2097    {
2098    // clone maplist and mapping
2099  2 MapList cdsposmap = new MapList(
2100    Arrays.asList(new int[][]
2101    { new int[] { cdsSeq.getStart(), cdsSeq.getEnd() } }),
2102    cdsref.getMap().getMap().getToRanges(), 3, 1);
2103  2 Mapping cdsmap = new Mapping(cdsref.getMap().getTo(),
2104    cdsref.getMap().getMap());
2105   
2106    // create dbref
2107  2 DBRefEntry newref = new DBRefEntry(cdsref.getSource(),
2108    cdsref.getVersion(), cdsref.getAccessionId(),
2109    new Mapping(cdsmap.getTo(), cdsposmap));
2110   
2111    // and see if we can map to the protein product for this mapping.
2112    // onSource is the filtered set of accessions on protein that we are
2113    // tranferring, so we assume accession is the same.
2114  2 if (cdsmap.getTo() == null && onSource != null)
2115    {
2116  2 List<DBRefEntry> sourceRefs = DBRefUtils.searchRefs(onSource,
2117    cdsref.getAccessionId());
2118  2 if (sourceRefs != null)
2119    {
2120  2 for (DBRefEntry srcref : sourceRefs)
2121    {
2122  2 if (srcref.getSource().equalsIgnoreCase(cdsref.getSource()))
2123    {
2124    // we have found a complementary dbref on the protein product, so
2125    // update mapping's getTo
2126  2 newref.getMap().setTo(proteinProduct);
2127    }
2128    }
2129    }
2130    }
2131  2 cdsSeq.addDBRef(newref);
2132  2 propagated.add(newref);
2133    }
2134  9 return propagated;
2135    }
2136   
2137    /**
2138    * Transfers co-located features on 'fromSeq' to 'toSeq', adjusting the
2139    * feature start/end ranges, optionally omitting specified feature types.
2140    * Returns the number of features copied.
2141    *
2142    * @param fromSeq
2143    * @param toSeq
2144    * @param mapping
2145    * the mapping from 'fromSeq' to 'toSeq'
2146    * @param select
2147    * if not null, only features of this type are copied (including
2148    * subtypes in the Sequence Ontology)
2149    * @param omitting
2150    */
 
2151  19 toggle protected static int transferFeatures(SequenceI fromSeq, SequenceI toSeq,
2152    MapList mapping, String select, String... omitting)
2153    {
2154  19 SequenceI copyTo = toSeq;
2155  28 while (copyTo.getDatasetSequence() != null)
2156    {
2157  9 copyTo = copyTo.getDatasetSequence();
2158    }
2159   
2160    /*
2161    * get features, optionally restricted by an ontology term
2162    */
2163  19 List<SequenceFeature> sfs = select == null ? fromSeq.getFeatures()
2164    .getPositionalFeatures() : fromSeq.getFeatures()
2165    .getFeaturesByOntology(select);
2166   
2167  19 int count = 0;
2168  19 for (SequenceFeature sf : sfs)
2169    {
2170  14 String type = sf.getType();
2171  14 boolean omit = false;
2172  14 for (String toOmit : omitting)
2173    {
2174  7 if (type.equals(toOmit))
2175    {
2176  2 omit = true;
2177    }
2178    }
2179  14 if (omit)
2180    {
2181  2 continue;
2182    }
2183   
2184    /*
2185    * locate the mapped range - null if either start or end is
2186    * not mapped (no partial overlaps are calculated)
2187    */
2188  12 int start = sf.getBegin();
2189  12 int end = sf.getEnd();
2190  12 int[] mappedTo = mapping.locateInTo(start, end);
2191    /*
2192    * if whole exon range doesn't map, try interpreting it
2193    * as 5' or 3' exon overlapping the CDS range
2194    */
2195  12 if (mappedTo == null)
2196    {
2197  5 mappedTo = mapping.locateInTo(end, end);
2198  5 if (mappedTo != null)
2199    {
2200    /*
2201    * end of exon is in CDS range - 5' overlap
2202    * to a range from the start of the peptide
2203    */
2204  1 mappedTo[0] = 1;
2205    }
2206    }
2207  12 if (mappedTo == null)
2208    {
2209  4 mappedTo = mapping.locateInTo(start, start);
2210  4 if (mappedTo != null)
2211    {
2212    /*
2213    * start of exon is in CDS range - 3' overlap
2214    * to a range up to the end of the peptide
2215    */
2216  1 mappedTo[1] = toSeq.getLength();
2217    }
2218    }
2219  12 if (mappedTo != null)
2220    {
2221  9 int newBegin = Math.min(mappedTo[0], mappedTo[1]);
2222  9 int newEnd = Math.max(mappedTo[0], mappedTo[1]);
2223  9 SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
2224    sf.getFeatureGroup(), sf.getScore());
2225  9 copyTo.addSequenceFeature(copy);
2226  9 count++;
2227    }
2228    }
2229  19 return count;
2230    }
2231   
2232    /**
2233    * Returns a mapping from dna to protein by inspecting sequence features of
2234    * type "CDS" on the dna. A mapping is constructed if the total CDS feature
2235    * length is 3 times the peptide length (optionally after dropping a trailing
2236    * stop codon). This method does not check whether the CDS nucleotide sequence
2237    * translates to the peptide sequence.
2238    *
2239    * @param dnaSeq
2240    * @param proteinSeq
2241    * @return
2242    */
 
2243  16 toggle public static MapList mapCdsToProtein(SequenceI dnaSeq,
2244    SequenceI proteinSeq)
2245    {
2246  16 List<int[]> ranges = findCdsPositions(dnaSeq);
2247  16 int mappedDnaLength = MappingUtils.getLength(ranges);
2248   
2249    /*
2250    * if not a whole number of codons, truncate mapping
2251    */
2252  16 int codonRemainder = mappedDnaLength % CODON_LENGTH;
2253  16 if (codonRemainder > 0)
2254    {
2255  2 mappedDnaLength -= codonRemainder;
2256  2 MappingUtils.removeEndPositions(codonRemainder, ranges);
2257    }
2258   
2259  16 int proteinLength = proteinSeq.getLength();
2260  16 int proteinStart = proteinSeq.getStart();
2261  16 int proteinEnd = proteinSeq.getEnd();
2262   
2263    /*
2264    * incomplete start codon may mean X at start of peptide
2265    * we ignore both for mapping purposes
2266    */
2267  16 if (proteinSeq.getCharAt(0) == 'X')
2268    {
2269    // todo JAL-2022 support startPhase > 0
2270  1 proteinStart++;
2271  1 proteinLength--;
2272    }
2273  16 List<int[]> proteinRange = new ArrayList<>();
2274   
2275    /*
2276    * dna length should map to protein (or protein plus stop codon)
2277    */
2278  16 int codesForResidues = mappedDnaLength / CODON_LENGTH;
2279  16 if (codesForResidues == (proteinLength + 1))
2280    {
2281    // assuming extra codon is for STOP and not in peptide
2282    // todo: check trailing codon is indeed a STOP codon
2283  2 codesForResidues--;
2284  2 mappedDnaLength -= CODON_LENGTH;
2285  2 MappingUtils.removeEndPositions(CODON_LENGTH, ranges);
2286    }
2287   
2288  16 if (codesForResidues == proteinLength)
2289    {
2290  6 proteinRange.add(new int[] { proteinStart, proteinEnd });
2291  6 return new MapList(ranges, proteinRange, CODON_LENGTH, 1);
2292    }
2293  10 return null;
2294    }
2295   
2296    /**
2297    * Returns a list of CDS ranges found (as sequence positions base 1), i.e. of
2298    * [start, end] positions of sequence features of type "CDS" (or a sub-type of
2299    * CDS in the Sequence Ontology). The ranges are sorted into ascending start
2300    * position order, so this method is only valid for linear CDS in the same
2301    * sense as the protein product.
2302    *
2303    * @param dnaSeq
2304    * @return
2305    */
 
2306  18 toggle protected static List<int[]> findCdsPositions(SequenceI dnaSeq)
2307    {
2308  18 List<int[]> result = new ArrayList<>();
2309   
2310  18 List<SequenceFeature> sfs = dnaSeq.getFeatures().getFeaturesByOntology(
2311    SequenceOntologyI.CDS);
2312  18 if (sfs.isEmpty())
2313    {
2314  8 return result;
2315    }
2316  10 SequenceFeatures.sortFeatures(sfs, true);
2317   
2318  10 for (SequenceFeature sf : sfs)
2319    {
2320  19 int phase = 0;
2321  19 try
2322    {
2323  19 phase = Integer.parseInt(sf.getPhase());
2324    } catch (NumberFormatException e)
2325    {
2326    // ignore
2327    }
2328    /*
2329    * phase > 0 on first codon means 5' incomplete - skip to the start
2330    * of the next codon; example ENST00000496384
2331    */
2332  19 int begin = sf.getBegin();
2333  19 int end = sf.getEnd();
2334  19 if (result.isEmpty() && phase > 0)
2335    {
2336  2 begin += phase;
2337  2 if (begin > end)
2338    {
2339    // shouldn't happen!
2340  0 System.err
2341    .println("Error: start phase extends beyond start CDS in "
2342    + dnaSeq.getName());
2343    }
2344    }
2345  19 result.add(new int[] { begin, end });
2346    }
2347   
2348    /*
2349    * Finally sort ranges by start position. This avoids a dependency on
2350    * keeping features in order on the sequence (if they are in order anyway,
2351    * the sort will have almost no work to do). The implicit assumption is CDS
2352    * ranges are assembled in order. Other cases should not use this method,
2353    * but instead construct an explicit mapping for CDS (e.g. EMBL parsing).
2354    */
2355  10 Collections.sort(result, IntRangeComparator.ASCENDING);
2356  10 return result;
2357    }
2358   
2359    /**
2360    * Maps exon features from dna to protein, and computes variants in peptide
2361    * product generated by variants in dna, and adds them as sequence_variant
2362    * features on the protein sequence. Returns the number of variant features
2363    * added.
2364    *
2365    * @param dnaSeq
2366    * @param peptide
2367    * @param dnaToProtein
2368    */
 
2369  7 toggle public static int computeProteinFeatures(SequenceI dnaSeq,
2370    SequenceI peptide, MapList dnaToProtein)
2371    {
2372  12 while (dnaSeq.getDatasetSequence() != null)
2373    {
2374  5 dnaSeq = dnaSeq.getDatasetSequence();
2375    }
2376  7 while (peptide.getDatasetSequence() != null)
2377    {
2378  0 peptide = peptide.getDatasetSequence();
2379    }
2380   
2381  7 transferFeatures(dnaSeq, peptide, dnaToProtein, SequenceOntologyI.EXON);
2382   
2383    /*
2384    * compute protein variants from dna variants and codon mappings;
2385    * NB - alternatively we could retrieve this using the REST service e.g.
2386    * http://rest.ensembl.org/overlap/translation
2387    * /ENSP00000288602?feature=transcript_variation;content-type=text/xml
2388    * which would be a bit slower but possibly more reliable
2389    */
2390   
2391    /*
2392    * build a map with codon variations for each potentially varying peptide
2393    */
2394  7 LinkedHashMap<Integer, List<DnaVariant>[]> variants = buildDnaVariantsMap(
2395    dnaSeq, dnaToProtein);
2396   
2397    /*
2398    * scan codon variations, compute peptide variants and add to peptide sequence
2399    */
2400  7 int count = 0;
2401  7 for (Entry<Integer, List<DnaVariant>[]> variant : variants.entrySet())
2402    {
2403  6 int peptidePos = variant.getKey();
2404  6 List<DnaVariant>[] codonVariants = variant.getValue();
2405  6 count += computePeptideVariants(peptide, peptidePos, codonVariants);
2406    }
2407   
2408  7 return count;
2409    }
2410   
2411    /**
2412    * Computes non-synonymous peptide variants from codon variants and adds them
2413    * as sequence_variant features on the protein sequence (one feature per
2414    * allele variant). Selected attributes (variant id, clinical significance)
2415    * are copied over to the new features.
2416    *
2417    * @param peptide
2418    * the protein sequence
2419    * @param peptidePos
2420    * the position to compute peptide variants for
2421    * @param codonVariants
2422    * a list of dna variants per codon position
2423    * @return the number of features added
2424    */
 
2425  9 toggle static int computePeptideVariants(SequenceI peptide, int peptidePos,
2426    List<DnaVariant>[] codonVariants)
2427    {
2428  9 String residue = String.valueOf(peptide.getCharAt(peptidePos - 1));
2429  9 int count = 0;
2430  9 String base1 = codonVariants[0].get(0).base;
2431  9 String base2 = codonVariants[1].get(0).base;
2432  9 String base3 = codonVariants[2].get(0).base;
2433   
2434    /*
2435    * variants in first codon base
2436    */
2437  9 for (DnaVariant dnavar : codonVariants[0])
2438    {
2439  11 if (dnavar.variant != null)
2440    {
2441  3 String alleles = (String) dnavar.variant.getValue(Gff3Helper.ALLELES);
2442  3 if (alleles != null)
2443    {
2444  3 for (String base : alleles.split(","))
2445    {
2446  6 if (!base1.equalsIgnoreCase(base))
2447    {
2448  3 String codon = base.toUpperCase() + base2.toLowerCase()
2449    + base3.toLowerCase();
2450  3 String canonical = base1.toUpperCase() + base2.toLowerCase()
2451    + base3.toLowerCase();
2452  3 if (addPeptideVariant(peptide, peptidePos, residue, dnavar,
2453    codon, canonical))
2454    {
2455  3 count++;
2456    }
2457    }
2458    }
2459    }
2460    }
2461    }
2462   
2463    /*
2464    * variants in second codon base
2465    */
2466  9 for (DnaVariant var : codonVariants[1])
2467    {
2468  9 if (var.variant != null)
2469    {
2470  6 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2471  6 if (alleles != null)
2472    {
2473  6 for (String base : alleles.split(","))
2474    {
2475  13 if (!base2.equalsIgnoreCase(base))
2476    {
2477  7 String codon = base1.toLowerCase() + base.toUpperCase()
2478    + base3.toLowerCase();
2479  7 String canonical = base1.toLowerCase() + base2.toUpperCase()
2480    + base3.toLowerCase();
2481  7 if (addPeptideVariant(peptide, peptidePos, residue, var,
2482    codon, canonical))
2483    {
2484  7 count++;
2485    }
2486    }
2487    }
2488    }
2489    }
2490    }
2491   
2492    /*
2493    * variants in third codon base
2494    */
2495  9 for (DnaVariant var : codonVariants[2])
2496    {
2497  10 if (var.variant != null)
2498    {
2499  4 String alleles = (String) var.variant.getValue(Gff3Helper.ALLELES);
2500  4 if (alleles != null)
2501    {
2502  4 for (String base : alleles.split(","))
2503    {
2504  8 if (!base3.equalsIgnoreCase(base))
2505    {
2506  4 String codon = base1.toLowerCase() + base2.toLowerCase()
2507    + base.toUpperCase();
2508  4 String canonical = base1.toLowerCase() + base2.toLowerCase()
2509    + base3.toUpperCase();
2510  4 if (addPeptideVariant(peptide, peptidePos, residue, var,
2511    codon, canonical))
2512    {
2513  4 count++;
2514    }
2515    }
2516    }
2517    }
2518    }
2519    }
2520   
2521  9 return count;
2522    }
2523   
2524    /**
2525    * Helper method that adds a peptide variant feature. ID and
2526    * clinical_significance attributes of the dna variant (if present) are copied
2527    * to the new feature.
2528    *
2529    * @param peptide
2530    * @param peptidePos
2531    * @param residue
2532    * @param var
2533    * @param codon
2534    * the variant codon e.g. aCg
2535    * @param canonical
2536    * the 'normal' codon e.g. aTg
2537    * @return true if a feature was added, else false
2538    */
 
2539  14 toggle static boolean addPeptideVariant(SequenceI peptide, int peptidePos,
2540    String residue, DnaVariant var, String codon, String canonical)
2541    {
2542    /*
2543    * get peptide translation of codon e.g. GAT -> D
2544    * note that variants which are not single alleles,
2545    * e.g. multibase variants or HGMD_MUTATION etc
2546    * are currently ignored here
2547    */
2548  14 String trans = codon.contains("-") ? null
2549  14 : (codon.length() > CODON_LENGTH ? null
2550    : ResidueProperties.codonTranslate(codon));
2551  14 if (trans == null)
2552    {
2553  0 return false;
2554    }
2555  14 String desc = canonical + "/" + codon;
2556  14 String featureType = "";
2557  14 if (trans.equals(residue))
2558    {
2559  3 featureType = SequenceOntologyI.SYNONYMOUS_VARIANT;
2560    }
2561  11 else if (ResidueProperties.STOP.equals(trans))
2562    {
2563  1 featureType = SequenceOntologyI.STOP_GAINED;
2564    }
2565    else
2566    {
2567  10 String residue3Char = StringUtils
2568    .toSentenceCase(ResidueProperties.aa2Triplet.get(residue));
2569  10 String trans3Char = StringUtils
2570    .toSentenceCase(ResidueProperties.aa2Triplet.get(trans));
2571  10 desc = "p." + residue3Char + peptidePos + trans3Char;
2572  10 featureType = SequenceOntologyI.NONSYNONYMOUS_VARIANT;
2573    }
2574  14 SequenceFeature sf = new SequenceFeature(featureType, desc, peptidePos,
2575    peptidePos, var.getSource());
2576   
2577  14 StringBuilder attributes = new StringBuilder(32);
2578  14 String id = (String) var.variant.getValue(ID);
2579  14 if (id != null)
2580    {
2581  8 if (id.startsWith(SEQUENCE_VARIANT))
2582    {
2583  1 id = id.substring(SEQUENCE_VARIANT.length());
2584    }
2585  8 sf.setValue(ID, id);
2586  8 attributes.append(ID).append("=").append(id);
2587    // TODO handle other species variants JAL-2064
2588  8 StringBuilder link = new StringBuilder(32);
2589  8 try
2590    {
2591  8 link.append(desc).append(" ").append(id).append(
2592    "|http://www.ensembl.org/Homo_sapiens/Variation/Summary?v=")
2593    .append(URLEncoder.encode(id, "UTF-8"));
2594  8 sf.addLink(link.toString());
2595    } catch (UnsupportedEncodingException e)
2596    {
2597    // as if
2598    }
2599    }
2600  14 String clinSig = (String) var.variant.getValue(CLINICAL_SIGNIFICANCE);
2601  14 if (clinSig != null)
2602    {
2603  6 sf.setValue(CLINICAL_SIGNIFICANCE, clinSig);
2604  6 attributes.append(";").append(CLINICAL_SIGNIFICANCE).append("=")
2605    .append(clinSig);
2606    }
2607  14 peptide.addSequenceFeature(sf);
2608  14 if (attributes.length() > 0)
2609    {
2610  8 sf.setAttributes(attributes.toString());
2611    }
2612  14 return true;
2613    }
2614   
2615    /**
2616    * Builds a map whose key is position in the protein sequence, and value is a
2617    * list of the base and all variants for each corresponding codon position.
2618    * <p>
2619    * This depends on dna variants being held as a comma-separated list as
2620    * property "alleles" on variant features.
2621    *
2622    * @param dnaSeq
2623    * @param dnaToProtein
2624    * @return
2625    */
 
2626  9 toggle @SuppressWarnings("unchecked")
2627    static LinkedHashMap<Integer, List<DnaVariant>[]> buildDnaVariantsMap(
2628    SequenceI dnaSeq, MapList dnaToProtein)
2629    {
2630    /*
2631    * map from peptide position to all variants of the codon which codes for it
2632    * LinkedHashMap ensures we keep the peptide features in sequence order
2633    */
2634  9 LinkedHashMap<Integer, List<DnaVariant>[]> variants = new LinkedHashMap<>();
2635   
2636  9 List<SequenceFeature> dnaFeatures = dnaSeq.getFeatures()
2637    .getFeaturesByOntology(SequenceOntologyI.SEQUENCE_VARIANT);
2638  9 if (dnaFeatures.isEmpty())
2639    {
2640  3 return variants;
2641    }
2642   
2643  6 int dnaStart = dnaSeq.getStart();
2644  6 int[] lastCodon = null;
2645  6 int lastPeptidePostion = 0;
2646   
2647    /*
2648    * build a map of codon variations for peptides
2649    */
2650  6 for (SequenceFeature sf : dnaFeatures)
2651    {
2652  18 int dnaCol = sf.getBegin();
2653  18 if (dnaCol != sf.getEnd())
2654    {
2655    // not handling multi-locus variant features
2656  0 continue;
2657    }
2658   
2659    /*
2660    * ignore variant if not a SNP
2661    */
2662  18 String alls = (String) sf.getValue(Gff3Helper.ALLELES);
2663  18 if (alls == null)
2664    {
2665  0 continue; // non-SNP VCF variant perhaps - can't process this
2666    }
2667   
2668  18 String[] alleles = alls.toUpperCase().split(",");
2669  18 boolean isSnp = true;
2670  18 for (String allele : alleles)
2671    {
2672  33 if (allele.trim().length() > 1)
2673    {
2674  5 isSnp = false;
2675    }
2676    }
2677  18 if (!isSnp)
2678    {
2679  5 continue;
2680    }
2681   
2682  13 int[] mapsTo = dnaToProtein.locateInTo(dnaCol, dnaCol);
2683  13 if (mapsTo == null)
2684    {
2685    // feature doesn't lie within coding region
2686  0 continue;
2687    }
2688  13 int peptidePosition = mapsTo[0];
2689  13 List<DnaVariant>[] codonVariants = variants.get(peptidePosition);
2690  13 if (codonVariants == null)
2691    {
2692  10 codonVariants = new ArrayList[CODON_LENGTH];
2693  10 codonVariants[0] = new ArrayList<>();
2694  10 codonVariants[1] = new ArrayList<>();
2695  10 codonVariants[2] = new ArrayList<>();
2696  10 variants.put(peptidePosition, codonVariants);
2697    }
2698   
2699    /*
2700    * get this peptide's codon positions e.g. [3, 4, 5] or [4, 7, 10]
2701    */
2702  13 int[] codon = peptidePosition == lastPeptidePostion ? lastCodon
2703    : MappingUtils.flattenRanges(dnaToProtein.locateInFrom(
2704    peptidePosition, peptidePosition));
2705  13 lastPeptidePostion = peptidePosition;
2706  13 lastCodon = codon;
2707   
2708    /*
2709    * save nucleotide (and any variant) for each codon position
2710    */
2711  52 for (int codonPos = 0; codonPos < CODON_LENGTH; codonPos++)
2712    {
2713  39 String nucleotide = String.valueOf(
2714    dnaSeq.getCharAt(codon[codonPos] - dnaStart)).toUpperCase();
2715  39 List<DnaVariant> codonVariant = codonVariants[codonPos];
2716  39 if (codon[codonPos] == dnaCol)
2717    {
2718  13 if (!codonVariant.isEmpty()
2719    && codonVariant.get(0).variant == null)
2720    {
2721    /*
2722    * already recorded base value, add this variant
2723    */
2724  3 codonVariant.get(0).variant = sf;
2725    }
2726    else
2727    {
2728    /*
2729    * add variant with base value
2730    */
2731  10 codonVariant.add(new DnaVariant(nucleotide, sf));
2732    }
2733    }
2734  26 else if (codonVariant.isEmpty())
2735    {
2736    /*
2737    * record (possibly non-varying) base value
2738    */
2739  20 codonVariant.add(new DnaVariant(nucleotide));
2740    }
2741    }
2742    }
2743  6 return variants;
2744    }
2745   
2746    /**
2747    * Makes an alignment with a copy of the given sequences, adding in any
2748    * non-redundant sequences which are mapped to by the cross-referenced
2749    * sequences.
2750    *
2751    * @param seqs
2752    * @param xrefs
2753    * @param dataset
2754    * the alignment dataset shared by the new copy
2755    * @return
2756    */
 
2757  0 toggle public static AlignmentI makeCopyAlignment(SequenceI[] seqs,
2758    SequenceI[] xrefs, AlignmentI dataset)
2759    {
2760  0 AlignmentI copy = new Alignment(new Alignment(seqs));
2761  0 copy.setDataset(dataset);
2762  0 boolean isProtein = !copy.isNucleotide();
2763  0 SequenceIdMatcher matcher = new SequenceIdMatcher(seqs);
2764  0 if (xrefs != null)
2765    {
2766  0 for (SequenceI xref : xrefs)
2767    {
2768  0 DBRefEntry[] dbrefs = xref.getDBRefs();
2769  0 if (dbrefs != null)
2770    {
2771  0 for (DBRefEntry dbref : dbrefs)
2772    {
2773  0 if (dbref.getMap() == null || dbref.getMap().getTo() == null
2774    || dbref.getMap().getTo().isProtein() != isProtein)
2775    {
2776  0 continue;
2777    }
2778  0 SequenceI mappedTo = dbref.getMap().getTo();
2779  0 SequenceI match = matcher.findIdMatch(mappedTo);
2780  0 if (match == null)
2781    {
2782  0 matcher.add(mappedTo);
2783  0 copy.addSequence(mappedTo);
2784    }
2785    }
2786    }
2787    }
2788    }
2789  0 return copy;
2790    }
2791   
2792    /**
2793    * Try to align sequences in 'unaligned' to match the alignment of their
2794    * mapped regions in 'aligned'. For example, could use this to align CDS
2795    * sequences which are mapped to their parent cDNA sequences.
2796    *
2797    * This method handles 1:1 mappings (dna-to-dna or protein-to-protein). For
2798    * dna-to-protein or protein-to-dna use alternative methods.
2799    *
2800    * @param unaligned
2801    * sequences to be aligned
2802    * @param aligned
2803    * holds aligned sequences and their mappings
2804    * @return
2805    */
 
2806  2 toggle public static int alignAs(AlignmentI unaligned, AlignmentI aligned)
2807    {
2808    /*
2809    * easy case - aligning a copy of aligned sequences
2810    */
2811  2 if (alignAsSameSequences(unaligned, aligned))
2812    {
2813  0 return unaligned.getHeight();
2814    }
2815   
2816    /*
2817    * fancy case - aligning via mappings between sequences
2818    */
2819  2 List<SequenceI> unmapped = new ArrayList<>();
2820  2 Map<Integer, Map<SequenceI, Character>> columnMap = buildMappedColumnsMap(
2821    unaligned, aligned, unmapped);
2822  2 int width = columnMap.size();
2823  2 char gap = unaligned.getGapCharacter();
2824  2 int realignedCount = 0;
2825    // TODO: verify this loop scales sensibly for very wide/high alignments
2826   
2827  2 for (SequenceI seq : unaligned.getSequences())
2828    {
2829  4 if (!unmapped.contains(seq))
2830    {
2831  4 char[] newSeq = new char[width];
2832  4 Arrays.fill(newSeq, gap); // JBPComment - doubt this is faster than the
2833    // Integer iteration below
2834  4 int newCol = 0;
2835  4 int lastCol = 0;
2836   
2837    /*
2838    * traverse the map to find columns populated
2839    * by our sequence
2840    */
2841  4 for (Integer column : columnMap.keySet())
2842    {
2843  60 Character c = columnMap.get(column).get(seq);
2844  60 if (c != null)
2845    {
2846    /*
2847    * sequence has a character at this position
2848    *
2849    */
2850  36 newSeq[newCol] = c;
2851  36 lastCol = newCol;
2852    }
2853  60 newCol++;
2854    }
2855   
2856    /*
2857    * trim trailing gaps
2858    */
2859  4 if (lastCol < width)
2860    {
2861  4 char[] tmp = new char[lastCol + 1];
2862  4 System.arraycopy(newSeq, 0, tmp, 0, lastCol + 1);
2863  4 newSeq = tmp;
2864    }
2865    // TODO: optimise SequenceI to avoid char[]->String->char[]
2866  4 seq.setSequence(String.valueOf(newSeq));
2867  4 realignedCount++;
2868    }
2869    }
2870  2 return realignedCount;
2871    }
2872   
2873    /**
2874    * If unaligned and aligned sequences share the same dataset sequences, then
2875    * simply copies the aligned sequences to the unaligned sequences and returns
2876    * true; else returns false
2877    *
2878    * @param unaligned
2879    * - sequences to be aligned based on aligned
2880    * @param aligned
2881    * - 'guide' alignment containing sequences derived from same dataset
2882    * as unaligned
2883    * @return
2884    */
 
2885  6 toggle static boolean alignAsSameSequences(AlignmentI unaligned,
2886    AlignmentI aligned)
2887    {
2888  6 if (aligned.getDataset() == null || unaligned.getDataset() == null)
2889    {
2890  0 return false; // should only pass alignments with datasets here
2891    }
2892   
2893    // map from dataset sequence to alignment sequence(s)
2894  6 Map<SequenceI, List<SequenceI>> alignedDatasets = new HashMap<>();
2895  6 for (SequenceI seq : aligned.getSequences())
2896    {
2897  15 SequenceI ds = seq.getDatasetSequence();
2898  15 if (alignedDatasets.get(ds) == null)
2899    {
2900  14 alignedDatasets.put(ds, new ArrayList<SequenceI>());
2901    }
2902  15 alignedDatasets.get(ds).add(seq);
2903    }
2904   
2905    /*
2906    * first pass - check whether all sequences to be aligned share a dataset
2907    * sequence with an aligned sequence
2908    */
2909  6 for (SequenceI seq : unaligned.getSequences())
2910    {
2911  12 if (!alignedDatasets.containsKey(seq.getDatasetSequence()))
2912    {
2913  3 return false;
2914    }
2915    }
2916   
2917    /*
2918    * second pass - copy aligned sequences;
2919    * heuristic rule: pair off sequences in order for the case where
2920    * more than one shares the same dataset sequence
2921    */
2922  3 for (SequenceI seq : unaligned.getSequences())
2923    {
2924  7 List<SequenceI> alignedSequences = alignedDatasets
2925    .get(seq.getDatasetSequence());
2926    // TODO: getSequenceAsString() will be deprecated in the future
2927    // TODO: need to leave to SequenceI implementor to update gaps
2928  7 seq.setSequence(alignedSequences.get(0).getSequenceAsString());
2929  7 if (alignedSequences.size() > 0)
2930    {
2931    // pop off aligned sequences (except the last one)
2932  7 alignedSequences.remove(0);
2933    }
2934    }
2935   
2936  3 return true;
2937    }
2938   
2939    /**
2940    * Returns a map whose key is alignment column number (base 1), and whose
2941    * values are a map of sequence characters in that column.
2942    *
2943    * @param unaligned
2944    * @param aligned
2945    * @param unmapped
2946    * @return
2947    */
 
2948  2 toggle static SortedMap<Integer, Map<SequenceI, Character>> buildMappedColumnsMap(
2949    AlignmentI unaligned, AlignmentI aligned,
2950    List<SequenceI> unmapped)
2951    {
2952    /*
2953    * Map will hold, for each aligned column position, a map of
2954    * {unalignedSequence, characterPerSequence} at that position.
2955    * TreeMap keeps the entries in ascending column order.
2956    */
2957  2 SortedMap<Integer, Map<SequenceI, Character>> map = new TreeMap<>();
2958   
2959    /*
2960    * record any sequences that have no mapping so can't be realigned
2961    */
2962  2 unmapped.addAll(unaligned.getSequences());
2963   
2964  2 List<AlignedCodonFrame> mappings = aligned.getCodonFrames();
2965   
2966  2 for (SequenceI seq : unaligned.getSequences())
2967    {
2968  4 for (AlignedCodonFrame mapping : mappings)
2969    {
2970  4 SequenceI fromSeq = mapping.findAlignedSequence(seq, aligned);
2971  4 if (fromSeq != null)
2972    {
2973  4 Mapping seqMap = mapping.getMappingBetween(fromSeq, seq);
2974  4 if (addMappedPositions(seq, fromSeq, seqMap, map))
2975    {
2976  4 unmapped.remove(seq);
2977    }
2978    }
2979    }
2980    }
2981  2 return map;
2982    }
2983   
2984    /**
2985    * Helper method that adds to a map the mapped column positions of a sequence.
2986    * <br>
2987    * For example if aaTT-Tg-gAAA is mapped to TTTAAA then the map should record
2988    * that columns 3,4,6,10,11,12 map to characters T,T,T,A,A,A of the mapped to
2989    * sequence.
2990    *
2991    * @param seq
2992    * the sequence whose column positions we are recording
2993    * @param fromSeq
2994    * a sequence that is mapped to the first sequence
2995    * @param seqMap
2996    * the mapping from 'fromSeq' to 'seq'
2997    * @param map
2998    * a map to add the column positions (in fromSeq) of the mapped
2999    * positions of seq
3000    * @return
3001    */
 
3002  6 toggle static boolean addMappedPositions(SequenceI seq, SequenceI fromSeq,
3003    Mapping seqMap, Map<Integer, Map<SequenceI, Character>> map)
3004    {
3005  6 if (seqMap == null)
3006    {
3007  0 return false;
3008    }
3009   
3010    /*
3011    * invert mapping if it is from unaligned to aligned sequence
3012    */
3013  6 if (seqMap.getTo() == fromSeq.getDatasetSequence())
3014    {
3015  0 seqMap = new Mapping(seq.getDatasetSequence(),
3016    seqMap.getMap().getInverse());
3017    }
3018   
3019  6 int toStart = seq.getStart();
3020   
3021    /*
3022    * traverse [start, end, start, end...] ranges in fromSeq
3023    */
3024  6 for (int[] fromRange : seqMap.getMap().getFromRanges())
3025    {
3026  18 for (int i = 0; i < fromRange.length - 1; i += 2)
3027    {
3028  9 boolean forward = fromRange[i + 1] >= fromRange[i];
3029   
3030    /*
3031    * find the range mapped to (sequence positions base 1)
3032    */
3033  9 int[] range = seqMap.locateMappedRange(fromRange[i],
3034    fromRange[i + 1]);
3035  9 if (range == null)
3036    {
3037  0 System.err.println("Error in mapping " + seqMap + " from "
3038    + fromSeq.getName());
3039  0 return false;
3040    }
3041  9 int fromCol = fromSeq.findIndex(fromRange[i]);
3042  9 int mappedCharPos = range[0];
3043   
3044    /*
3045    * walk over the 'from' aligned sequence in forward or reverse
3046    * direction; when a non-gap is found, record the column position
3047    * of the next character of the mapped-to sequence; stop when all
3048    * the characters of the range have been counted
3049    */
3050  70 while (mappedCharPos <= range[1] && fromCol <= fromSeq.getLength()
3051    && fromCol >= 0)
3052    {
3053  61 if (!Comparison.isGap(fromSeq.getCharAt(fromCol - 1)))
3054    {
3055    /*
3056    * mapped from sequence has a character in this column
3057    * record the column position for the mapped to character
3058    */
3059  48 Map<SequenceI, Character> seqsMap = map.get(fromCol);
3060  48 if (seqsMap == null)
3061    {
3062  42 seqsMap = new HashMap<>();
3063  42 map.put(fromCol, seqsMap);
3064    }
3065  48 seqsMap.put(seq, seq.getCharAt(mappedCharPos - toStart));
3066  48 mappedCharPos++;
3067    }
3068  61 fromCol += (forward ? 1 : -1);
3069    }
3070    }
3071    }
3072  6 return true;
3073    }
3074   
3075    // strictly temporary hack until proper criteria for aligning protein to cds
3076    // are in place; this is so Ensembl -> fetch xrefs Uniprot aligns the Uniprot
 
3077  0 toggle public static boolean looksLikeEnsembl(AlignmentI alignment)
3078    {
3079  0 for (SequenceI seq : alignment.getSequences())
3080    {
3081  0 String name = seq.getName();
3082  0 if (!name.startsWith("ENSG") && !name.startsWith("ENST"))
3083    {
3084  0 return false;
3085    }
3086    }
3087  0 return true;
3088    }
3089    }
  • Report generated by OpenClover v 4.2.1 on Wed Oct 31 2018 15:15:54 GMT using coverage data from Wed Oct 31 2018 15:15:47 GMT.
  • Clover free edition.