2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.analysis;
23 import jalview.analysis.ResidueCount.SymbolCounts;
24 import jalview.datamodel.AlignedCodonFrame;
25 import jalview.datamodel.AlignmentAnnotation;
26 import jalview.datamodel.AlignmentI;
27 import jalview.datamodel.Annotation;
28 import jalview.datamodel.SequenceI;
29 import jalview.ext.android.SparseIntArray;
30 import jalview.util.Comparison;
31 import jalview.util.Format;
32 import jalview.util.MappingUtils;
33 import jalview.util.QuickSort;
35 import java.util.Arrays;
36 import java.util.Hashtable;
37 import java.util.List;
40 * Takes in a vector or array of sequences and column start and column end and
41 * returns a new Hashtable[] of size maxSeqLength, if Hashtable not supplied.
42 * This class is used extensively in calculating alignment colourschemes that
43 * depend on the amount of conservation in each alignment column.
48 public class AAFrequency
50 public static final String PROFILE = "P";
53 * Quick look-up of String value of char 'A' to 'Z'
55 private static final String[] CHARS = new String['Z' - 'A' + 1];
59 for (char c = 'A'; c <= 'Z'; c++)
61 CHARS[c - 'A'] = String.valueOf(c);
65 public static final Profile[] calculate(List<SequenceI> list,
68 return calculate(list, start, end, false);
71 public static final Profile[] calculate(List<SequenceI> sequences,
72 int start, int end, boolean profile)
74 SequenceI[] seqs = new SequenceI[sequences.size()];
76 synchronized (sequences)
78 for (int i = 0; i < sequences.size(); i++)
80 seqs[i] = sequences.get(i);
81 if (seqs[i].getLength() > width)
83 width = seqs[i].getLength();
87 Profile[] reply = new Profile[width];
94 calculate(seqs, start, end, reply, profile);
100 * Calculate the consensus symbol(s) for each column in the given range.
104 * start column (inclusive, base zero)
106 * end column (exclusive)
108 * array in which to store profile per column
109 * @param saveFullProfile
110 * if true, store all symbol counts
112 public static final void calculate(final SequenceI[] sequences,
113 int start, int end, Profile[] result, boolean saveFullProfile)
115 // long now = System.currentTimeMillis();
116 int seqCount = sequences.length;
117 boolean nucleotide = false;
118 int nucleotideCount = 0;
119 int peptideCount = 0;
121 for (int column = start; column < end; column++)
124 * Apply a heuristic to detect nucleotide data (which can
125 * be counted in more compact arrays); here we test for
126 * more than 90% nucleotide; recheck every 10 columns in case
127 * of misleading data e.g. highly conserved Alanine in peptide!
128 * Mistakenly guessing nucleotide has a small performance cost,
129 * as it will result in counting in sparse arrays.
130 * Mistakenly guessing peptide has a small space cost,
131 * as it will use a larger than necessary array to hold counts.
133 if (nucleotideCount > 100 && column % 10 == 0)
135 nucleotide = (9 * peptideCount < nucleotideCount);
137 ResidueCount residueCounts = new ResidueCount(nucleotide);
139 for (int row = 0; row < seqCount; row++)
141 if (sequences[row] == null)
144 .println("WARNING: Consensus skipping null sequence - possible race condition.");
147 char[] seq = sequences[row].getSequence();
148 if (seq.length > column)
150 char c = seq[column];
151 residueCounts.add(c);
152 if (Comparison.isNucleotide(c))
156 else if (!Comparison.isGap(c))
164 * count a gap if the sequence doesn't reach this column
166 residueCounts.addGap();
170 int maxCount = residueCounts.getModalCount();
171 String maxResidue = residueCounts.getResiduesForCount(maxCount);
172 int gapCount = residueCounts.getGapCount();
173 Profile profile = new Profile(seqCount, gapCount, maxCount,
178 profile.setCounts(residueCounts);
181 result[column] = profile;
183 // long elapsed = System.currentTimeMillis() - now;
184 // System.out.println(elapsed);
188 * Make an estimate of the profile size we are going to compute i.e. how many
189 * different characters may be present in it. Overestimating has a cost of
190 * using more memory than necessary. Underestimating has a cost of needing to
191 * extend the SparseIntArray holding the profile counts.
193 * @param profileSizes
194 * counts of sizes of profiles so far encountered
197 static int estimateProfileSize(SparseIntArray profileSizes)
199 if (profileSizes.size() == 0)
205 * could do a statistical heuristic here e.g. 75%ile
206 * for now just return the largest value
208 return profileSizes.keyAt(profileSizes.size() - 1);
212 * Derive the consensus annotations to be added to the alignment for display.
213 * This does not recompute the raw data, but may be called on a change in
214 * display options, such as 'ignore gaps', which may in turn result in a
215 * change in the derived values.
218 * the annotation row to add annotations to
220 * the source consensus data
226 * if true, normalise residue percentages ignoring gaps
227 * @param showSequenceLogo
228 * if true include all consensus symbols, else just show modal
231 * number of sequences
233 public static void completeConsensus(AlignmentAnnotation consensus,
234 Profile[] profiles, int iStart, int width, boolean ignoreGaps,
235 boolean showSequenceLogo, long nseq)
237 // long now = System.currentTimeMillis();
238 if (consensus == null || consensus.annotations == null
239 || consensus.annotations.length < width)
242 * called with a bad alignment annotation row
243 * wait for it to be initialised properly
248 final int dp = getPercentageDp(nseq);
250 for (int i = iStart; i < width; i++)
253 if (i >= profiles.length || ((profile = profiles[i]) == null))
256 * happens if sequences calculated over were
257 * shorter than alignment width
259 consensus.annotations[i] = null;
263 float value = profile.getPercentageIdentity(ignoreGaps);
265 String description = getTooltip(profile, value, showSequenceLogo,
268 String modalResidue = profile.getModalResidue();
269 if ("".equals(modalResidue))
273 else if (modalResidue.length() > 1)
277 consensus.annotations[i] = new Annotation(modalResidue,
278 description, ' ', value);
280 // long elapsed = System.currentTimeMillis() - now;
281 // System.out.println(-elapsed);
285 * Returns a tooltip showing either
287 * <li>the full profile (percentages of all residues present), if
288 * showSequenceLogo is true, or</li>
289 * <li>just the modal (most common) residue(s), if showSequenceLogo is false</li>
291 * Percentages are as a fraction of all sequence, or only ungapped sequences
292 * if ignoreGaps is true.
296 * @param showSequenceLogo
299 * the number of decimal places to format percentages to
302 static String getTooltip(Profile profile, float pid,
303 boolean showSequenceLogo, boolean ignoreGaps, int dp)
305 ResidueCount counts = profile.getCounts();
307 String description = null;
308 if (counts != null && showSequenceLogo)
310 int normaliseBy = ignoreGaps ? profile.getNonGapped() : profile
312 description = counts.getTooltip(normaliseBy, dp);
316 StringBuilder sb = new StringBuilder(64);
317 String maxRes = profile.getModalResidue();
318 if (maxRes.length() > 1)
320 sb.append("[").append(maxRes).append("]");
326 if (maxRes.length() > 0)
329 Format.appendPercentage(sb, pid, dp);
332 description = sb.toString();
338 * Returns the sorted profile for the given consensus data. The returned array
342 * [profileType, numberOfValues, nonGapCount, charValue1, percentage1, charValue2, percentage2, ...]
343 * in descending order of percentage value
347 * the data object from which to extract and sort values
349 * if true, only non-gapped values are included in percentage
353 public static int[] extractProfile(Profile profile,
356 int[] rtnval = new int[64];
357 ResidueCount counts = profile.getCounts();
363 SymbolCounts symbolCounts = counts.getSymbolCounts();
364 char[] symbols = symbolCounts.symbols;
365 int[] values = symbolCounts.values;
366 QuickSort.sort(values, symbols);
367 int nextArrayPos = 2;
368 int totalPercentage = 0;
369 final int divisor = ignoreGaps ? profile.getNonGapped() : profile
373 * traverse the arrays in reverse order (highest counts first)
375 for (int i = symbols.length - 1; i >= 0; i--)
377 int theChar = symbols[i];
378 int charCount = values[i];
380 rtnval[nextArrayPos++] = theChar;
381 final int percentage = (charCount * 100) / divisor;
382 rtnval[nextArrayPos++] = percentage;
383 totalPercentage += percentage;
385 rtnval[0] = symbols.length;
386 rtnval[1] = totalPercentage;
387 int[] result = new int[rtnval.length + 1];
388 result[0] = AlignmentAnnotation.SEQUENCE_PROFILE;
389 System.arraycopy(rtnval, 0, result, 1, rtnval.length);
395 * Extract a sorted extract of cDNA codon profile data. The returned array
399 * [profileType, numberOfValues, totalCount, charValue1, percentage1, charValue2, percentage2, ...]
400 * in descending order of percentage value, where the character values encode codon triplets
406 public static int[] extractCdnaProfile(Hashtable hashtable,
409 // this holds #seqs, #ungapped, and then codon count, indexed by encoded
411 int[] codonCounts = (int[]) hashtable.get(PROFILE);
412 int[] sortedCounts = new int[codonCounts.length - 2];
413 System.arraycopy(codonCounts, 2, sortedCounts, 0,
414 codonCounts.length - 2);
416 int[] result = new int[3 + 2 * sortedCounts.length];
417 // first value is just the type of profile data
418 result[0] = AlignmentAnnotation.CDNA_PROFILE;
420 char[] codons = new char[sortedCounts.length];
421 for (int i = 0; i < codons.length; i++)
423 codons[i] = (char) i;
425 QuickSort.sort(sortedCounts, codons);
426 int totalPercentage = 0;
427 int distinctValuesCount = 0;
429 int divisor = ignoreGaps ? codonCounts[1] : codonCounts[0];
430 for (int i = codons.length - 1; i >= 0; i--)
432 final int codonCount = sortedCounts[i];
435 break; // nothing else of interest here
437 distinctValuesCount++;
438 result[j++] = codons[i];
439 final int percentage = codonCount * 100 / divisor;
440 result[j++] = percentage;
441 totalPercentage += percentage;
443 result[2] = totalPercentage;
446 * Just return the non-zero values
448 // todo next value is redundant if we limit the array to non-zero counts
449 result[1] = distinctValuesCount;
450 return Arrays.copyOfRange(result, 0, j);
454 * Compute a consensus for the cDNA coding for a protein alignment.
457 * the protein alignment (which should hold mappings to cDNA
460 * the consensus data stores to be populated (one per column)
462 public static void calculateCdna(AlignmentI alignment,
463 Hashtable[] hconsensus)
465 final char gapCharacter = alignment.getGapCharacter();
466 List<AlignedCodonFrame> mappings = alignment.getCodonFrames();
467 if (mappings == null || mappings.isEmpty())
472 int cols = alignment.getWidth();
473 for (int col = 0; col < cols; col++)
475 // todo would prefer a Java bean for consensus data
476 Hashtable<String, int[]> columnHash = new Hashtable<String, int[]>();
477 // #seqs, #ungapped seqs, counts indexed by (codon encoded + 1)
478 int[] codonCounts = new int[66];
479 codonCounts[0] = alignment.getSequences().size();
480 int ungappedCount = 0;
481 for (SequenceI seq : alignment.getSequences())
483 if (seq.getCharAt(col) == gapCharacter)
487 List<char[]> codons = MappingUtils
488 .findCodonsFor(seq, col, mappings);
489 for (char[] codon : codons)
491 int codonEncoded = CodingUtils.encodeCodon(codon);
492 if (codonEncoded >= 0)
494 codonCounts[codonEncoded + 2]++;
499 codonCounts[1] = ungappedCount;
500 // todo: sort values here, save counts and codons?
501 columnHash.put(PROFILE, codonCounts);
502 hconsensus[col] = columnHash;
507 * Derive displayable cDNA consensus annotation from computed consensus data.
509 * @param consensusAnnotation
510 * the annotation row to be populated for display
511 * @param consensusData
512 * the computed consensus data
513 * @param showProfileLogo
514 * if true show all symbols present at each position, else only the
517 * the number of sequences in the alignment
519 public static void completeCdnaConsensus(
520 AlignmentAnnotation consensusAnnotation,
521 Hashtable[] consensusData, boolean showProfileLogo, int nseqs)
523 if (consensusAnnotation == null
524 || consensusAnnotation.annotations == null
525 || consensusAnnotation.annotations.length < consensusData.length)
527 // called with a bad alignment annotation row - wait for it to be
528 // initialised properly
532 // ensure codon triplet scales with font size
533 consensusAnnotation.scaleColLabel = true;
534 for (int col = 0; col < consensusData.length; col++)
536 Hashtable hci = consensusData[col];
539 // gapped protein column?
542 // array holds #seqs, #ungapped, then codon counts indexed by codon
543 final int[] codonCounts = (int[]) hci.get(PROFILE);
547 * First pass - get total count and find the highest
549 final char[] codons = new char[codonCounts.length - 2];
550 for (int j = 2; j < codonCounts.length; j++)
552 final int codonCount = codonCounts[j];
553 codons[j - 2] = (char) (j - 2);
554 totalCount += codonCount;
558 * Sort array of encoded codons by count ascending - so the modal value
559 * goes to the end; start by copying the count (dropping the first value)
561 int[] sortedCodonCounts = new int[codonCounts.length - 2];
562 System.arraycopy(codonCounts, 2, sortedCodonCounts, 0,
563 codonCounts.length - 2);
564 QuickSort.sort(sortedCodonCounts, codons);
566 int modalCodonEncoded = codons[codons.length - 1];
567 int modalCodonCount = sortedCodonCounts[codons.length - 1];
568 String modalCodon = String.valueOf(CodingUtils
569 .decodeCodon(modalCodonEncoded));
570 if (sortedCodonCounts.length > 1
571 && sortedCodonCounts[codons.length - 2] == sortedCodonCounts[codons.length - 1])
574 * two or more codons share the modal count
578 float pid = sortedCodonCounts[sortedCodonCounts.length - 1] * 100
579 / (float) totalCount;
582 * todo ? Replace consensus hashtable with sorted arrays of codons and
583 * counts (non-zero only). Include total count in count array [0].
587 * Scan sorted array backwards for most frequent values first. Show
588 * repeated values compactly.
590 StringBuilder mouseOver = new StringBuilder(32);
591 StringBuilder samePercent = new StringBuilder();
592 String percent = null;
593 String lastPercent = null;
594 int percentDecPl = getPercentageDp(nseqs);
596 for (int j = codons.length - 1; j >= 0; j--)
598 int codonCount = sortedCodonCounts[j];
602 * remaining codons are 0% - ignore, but finish off the last one if
605 if (samePercent.length() > 0)
607 mouseOver.append(samePercent).append(": ").append(percent)
612 int codonEncoded = codons[j];
613 final int pct = codonCount * 100 / totalCount;
614 String codon = String
615 .valueOf(CodingUtils.decodeCodon(codonEncoded));
616 StringBuilder sb = new StringBuilder();
617 Format.appendPercentage(sb, pct, percentDecPl);
618 percent = sb.toString();
619 if (showProfileLogo || codonCount == modalCodonCount)
621 if (percent.equals(lastPercent) && j > 0)
623 samePercent.append(samePercent.length() == 0 ? "" : ", ");
624 samePercent.append(codon);
628 if (samePercent.length() > 0)
630 mouseOver.append(samePercent).append(": ")
631 .append(lastPercent).append("% ");
633 samePercent.setLength(0);
634 samePercent.append(codon);
636 lastPercent = percent;
640 consensusAnnotation.annotations[col] = new Annotation(modalCodon,
641 mouseOver.toString(), ' ', pid);
646 * Returns the number of decimal places to show for profile percentages. For
647 * less than 100 sequences, returns zero (the integer percentage value will be
648 * displayed). For 100-999 sequences, returns 1, for 1000-9999 returns 2, etc.
653 protected static int getPercentageDp(long nseq)