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.datamodel.AlignedCodonFrame;
24 import jalview.datamodel.AlignmentAnnotation;
25 import jalview.datamodel.AlignmentI;
26 import jalview.datamodel.Annotation;
27 import jalview.datamodel.HiddenMarkovModel;
28 import jalview.datamodel.Profile;
29 import jalview.datamodel.ProfileI;
30 import jalview.datamodel.Profiles;
31 import jalview.datamodel.ProfilesI;
32 import jalview.datamodel.ResidueCount;
33 import jalview.datamodel.ResidueCount.SymbolCounts;
34 import jalview.datamodel.SequenceI;
35 import jalview.ext.android.SparseIntArray;
36 import jalview.schemes.ResidueProperties;
37 import jalview.util.Comparison;
38 import jalview.util.Format;
39 import jalview.util.MappingUtils;
40 import jalview.util.QuickSort;
42 import java.awt.Color;
43 import java.util.Arrays;
44 import java.util.Hashtable;
45 import java.util.List;
48 * Takes in a vector or array of sequences and column start and column end and
49 * returns a new Hashtable[] of size maxSeqLength, if Hashtable not supplied.
50 * This class is used extensively in calculating alignment colourschemes that
51 * depend on the amount of conservation in each alignment column.
56 public class AAFrequency
58 public static final String PROFILE = "P";
60 private static final String AMINO = "amino";
62 private static final String DNA = "DNA";
64 private static final String RNA = "RNA";
67 * Quick look-up of String value of char 'A' to 'Z'
69 private static final String[] CHARS = new String['Z' - 'A' + 1];
73 for (char c = 'A'; c <= 'Z'; c++)
75 CHARS[c - 'A'] = String.valueOf(c);
79 public static final ProfilesI calculate(List<SequenceI> list, int start,
82 return calculate(list, start, end, false);
85 public static final ProfilesI calculate(List<SequenceI> sequences,
86 int start, int end, boolean profile)
88 SequenceI[] seqs = new SequenceI[sequences.size()];
90 synchronized (sequences)
92 for (int i = 0; i < sequences.size(); i++)
94 seqs[i] = sequences.get(i);
95 int length = seqs[i].getLength();
107 ProfilesI reply = calculate(seqs, width, start, end, profile);
115 * Calculate the consensus symbol(s) for each column in the given range.
119 * the full width of the alignment
121 * start column (inclusive, base zero)
123 * end column (exclusive)
124 * @param saveFullProfile
125 * if true, store all symbol counts
127 public static final ProfilesI calculate(final SequenceI[] sequences,
128 int width, int start, int end, boolean saveFullProfile)
130 // long now = System.currentTimeMillis();
131 int seqCount = sequences.length;
132 boolean nucleotide = false;
133 int nucleotideCount = 0;
134 int peptideCount = 0;
136 ProfileI[] result = new ProfileI[width];
138 for (int column = start; column < end; column++)
141 * Apply a heuristic to detect nucleotide data (which can
142 * be counted in more compact arrays); here we test for
143 * more than 90% nucleotide; recheck every 10 columns in case
144 * of misleading data e.g. highly conserved Alanine in peptide!
145 * Mistakenly guessing nucleotide has a small performance cost,
146 * as it will result in counting in sparse arrays.
147 * Mistakenly guessing peptide has a small space cost,
148 * as it will use a larger than necessary array to hold counts.
150 if (nucleotideCount > 100 && column % 10 == 0)
152 nucleotide = (9 * peptideCount < nucleotideCount);
154 ResidueCount residueCounts = new ResidueCount(nucleotide);
156 for (int row = 0; row < seqCount; row++)
158 if (sequences[row] == null)
161 .println("WARNING: Consensus skipping null sequence - possible race condition.");
164 char[] seq = sequences[row].getSequence();
165 if (seq.length > column)
167 char c = seq[column];
168 residueCounts.add(c);
169 if (Comparison.isNucleotide(c))
173 else if (!Comparison.isGap(c))
181 * count a gap if the sequence doesn't reach this column
183 residueCounts.addGap();
187 int maxCount = residueCounts.getModalCount();
188 String maxResidue = residueCounts.getResiduesForCount(maxCount);
189 int gapCount = residueCounts.getGapCount();
190 ProfileI profile = new Profile(seqCount, gapCount, maxCount,
195 profile.setCounts(residueCounts);
198 result[column] = profile;
200 return new Profiles(result);
201 // long elapsed = System.currentTimeMillis() - now;
202 // System.out.println(elapsed);
206 * Returns the full set of profiles for a hidden Markov model. The underlying
207 * data is the raw probabilities of a residue being emitted at each node,
208 * however the profiles returned by this function contain the percentage
209 * chance of a residue emission.
213 * The width of the Profile array (Profiles) to be returned.
215 * The alignment column on which the first profile is based.
217 * The alignment column on which the last profile is based.
218 * @param saveFullProfile
219 * Flag for saving the counts for each profile
220 * @param removeBelowBackground
221 * Flag for removing any characters with a match emission probability
222 * less than its background frequency
225 public static ProfilesI calculateHMMProfiles(final HiddenMarkovModel hmm,
226 int width, int start, int end, boolean saveFullProfile,
227 boolean removeBelowBackground)
229 ProfileI[] result = new ProfileI[width];
230 int symbolCount = hmm.getNumberOfSymbols();
231 for (int column = start; column < end; column++)
233 ResidueCount counts = new ResidueCount();
234 for (char symbol : hmm.getSymbols())
236 int value = getAnalogueCount(hmm, column, symbol,
237 removeBelowBackground);
238 counts.put(symbol, value);
240 int maxCount = counts.getModalCount();
241 String maxResidue = counts.getResiduesForCount(maxCount);
242 int gapCount = counts.getGapCount();
243 ProfileI profile = new Profile(symbolCount, gapCount, maxCount,
248 profile.setCounts(counts);
251 result[column] = profile;
253 return new Profiles(result);
257 * Make an estimate of the profile size we are going to compute i.e. how many
258 * different characters may be present in it. Overestimating has a cost of
259 * using more memory than necessary. Underestimating has a cost of needing to
260 * extend the SparseIntArray holding the profile counts.
262 * @param profileSizes
263 * counts of sizes of profiles so far encountered
266 static int estimateProfileSize(SparseIntArray profileSizes)
268 if (profileSizes.size() == 0)
274 * could do a statistical heuristic here e.g. 75%ile
275 * for now just return the largest value
277 return profileSizes.keyAt(profileSizes.size() - 1);
281 * Derive the consensus annotations to be added to the alignment for display.
282 * This does not recompute the raw data, but may be called on a change in
283 * display options, such as 'ignore gaps', which may in turn result in a
284 * change in the derived values.
287 * the annotation row to add annotations to
289 * the source consensus data
291 * start column (inclusive)
293 * end column (exclusive)
295 * if true, normalise residue percentages ignoring gaps
296 * @param showSequenceLogo
297 * if true include all consensus symbols, else just show modal
300 * number of sequences
302 public static void completeConsensus(AlignmentAnnotation consensus,
303 ProfilesI profiles, int startCol, int endCol, boolean ignoreGaps,
304 boolean showSequenceLogo, long nseq)
306 // long now = System.currentTimeMillis();
307 if (consensus == null || consensus.annotations == null
308 || consensus.annotations.length < endCol)
311 * called with a bad alignment annotation row
312 * wait for it to be initialised properly
317 for (int i = startCol; i < endCol; i++)
319 ProfileI profile = profiles.get(i);
323 * happens if sequences calculated over were
324 * shorter than alignment width
326 consensus.annotations[i] = null;
330 final int dp = getPercentageDp(nseq);
332 float value = profile.getPercentageIdentity(ignoreGaps);
334 String description = getTooltip(profile, value, showSequenceLogo,
337 String modalResidue = profile.getModalResidue();
338 if ("".equals(modalResidue))
342 else if (modalResidue.length() > 1)
346 consensus.annotations[i] = new Annotation(modalResidue, description,
349 // long elapsed = System.currentTimeMillis() - now;
350 // System.out.println(-elapsed);
354 * Derive the information annotations to be added to the alignment for
355 * display. This does not recompute the raw data, but may be called on a
356 * change in display options, such as 'ignore below background frequency',
357 * which may in turn result in a change in the derived values.
360 * the annotation row to add annotations to
362 * the source information data
364 * start column (inclusive)
366 * end column (exclusive)
368 * if true, normalise residue percentages
369 * @param showSequenceLogo
370 * if true include all information symbols, else just show modal
373 * number of sequences
375 public static void completeInformation(AlignmentAnnotation information,
376 ProfilesI profiles, int startCol, int endCol,
377 boolean ignoreBelowBackground,
378 boolean showSequenceLogo, long nseq)
380 // long now = System.currentTimeMillis();
381 if (information == null || information.annotations == null
382 || information.annotations.length < endCol)
385 * called with a bad alignment annotation row
386 * wait for it to be initialised properly
393 for (int i = startCol; i < endCol; i++)
395 ProfileI profile = profiles.get(i);
399 * happens if sequences calculated over were
400 * shorter than alignment width
402 information.annotations[i] = null;
406 HiddenMarkovModel hmm;
408 SequenceI hmmSeq = information.sequenceRef;
410 hmm = hmmSeq.getHMM();
412 Float value = getInformationContent(i, hmm);
419 String description = value + " bits";
420 information.annotations[i] = new Annotation(
421 Character.toString(hmm.getConsensusAtAlignColumn(i)),
422 description, ' ', value);
424 information.graphMax = max;
425 // long elapsed = System.currentTimeMillis() - now;
426 // System.out.println(-elapsed);
430 * Derive the gap count annotation row.
433 * the annotation row to add annotations to
435 * the source consensus data
437 * start column (inclusive)
439 * end column (exclusive)
441 public static void completeGapAnnot(AlignmentAnnotation gaprow,
442 ProfilesI profiles, int startCol, int endCol, long nseq)
444 if (gaprow == null || gaprow.annotations == null
445 || gaprow.annotations.length < endCol)
448 * called with a bad alignment annotation row
449 * wait for it to be initialised properly
453 // always set ranges again
454 gaprow.graphMax = nseq;
456 double scale = 0.8/nseq;
457 for (int i = startCol; i < endCol; i++)
459 ProfileI profile = profiles.get(i);
463 * happens if sequences calculated over were
464 * shorter than alignment width
466 gaprow.annotations[i] = null;
470 final int gapped = profile.getNonGapped();
472 String description = "" + gapped;
474 gaprow.annotations[i] = new Annotation("", description,
475 '\0', gapped, jalview.util.ColorUtils.bleachColour(
476 Color.DARK_GRAY, (float) scale * gapped));
481 * Returns a tooltip showing either
483 * <li>the full profile (percentages of all residues present), if
484 * showSequenceLogo is true, or</li>
485 * <li>just the modal (most common) residue(s), if showSequenceLogo is false</li>
487 * Percentages are as a fraction of all sequence, or only ungapped sequences
488 * if ignoreGaps is true.
492 * @param showSequenceLogo
495 * the number of decimal places to format percentages to
498 static String getTooltip(ProfileI profile, float pid,
499 boolean showSequenceLogo, boolean ignoreGaps, int dp)
501 ResidueCount counts = profile.getCounts();
503 String description = null;
504 if (counts != null && showSequenceLogo)
506 int normaliseBy = ignoreGaps ? profile.getNonGapped() : profile
508 description = counts.getTooltip(normaliseBy, dp);
512 StringBuilder sb = new StringBuilder(64);
513 String maxRes = profile.getModalResidue();
514 if (maxRes.length() > 1)
516 sb.append("[").append(maxRes).append("]");
522 if (maxRes.length() > 0)
525 Format.appendPercentage(sb, pid, dp);
528 description = sb.toString();
534 * Returns the sorted profile for the given consensus data. The returned array
538 * [profileType, numberOfValues, nonGapCount, charValue1, percentage1, charValue2, percentage2, ...]
539 * in descending order of percentage value
543 * the data object from which to extract and sort values
545 * if true, only non-gapped values are included in percentage
549 public static int[] extractProfile(ProfileI profile, boolean ignoreGaps)
551 int[] rtnval = new int[64];
552 ResidueCount counts = profile.getCounts();
558 SymbolCounts symbolCounts = counts.getSymbolCounts();
559 char[] symbols = symbolCounts.symbols;
560 int[] values = symbolCounts.values;
561 QuickSort.sort(values, symbols);
562 int nextArrayPos = 2;
563 int totalPercentage = 0;
564 final int divisor = ignoreGaps ? profile.getNonGapped() : profile
568 * traverse the arrays in reverse order (highest counts first)
570 for (int i = symbols.length - 1; i >= 0; i--)
572 int theChar = symbols[i];
573 int charCount = values[i];
575 rtnval[nextArrayPos++] = theChar;
576 final int percentage = (charCount * 100) / divisor;
577 rtnval[nextArrayPos++] = percentage;
578 totalPercentage += percentage;
580 rtnval[0] = symbols.length;
581 rtnval[1] = totalPercentage;
582 int[] result = new int[rtnval.length + 1];
583 result[0] = AlignmentAnnotation.SEQUENCE_PROFILE;
584 System.arraycopy(rtnval, 0, result, 1, rtnval.length);
591 * Extract a sorted extract of cDNA codon profile data. The returned array
595 * [profileType, numberOfValues, totalCount, charValue1, percentage1, charValue2, percentage2, ...]
596 * in descending order of percentage value, where the character values encode codon triplets
602 public static int[] extractCdnaProfile(Hashtable hashtable,
605 // this holds #seqs, #ungapped, and then codon count, indexed by encoded
607 int[] codonCounts = (int[]) hashtable.get(PROFILE);
608 int[] sortedCounts = new int[codonCounts.length - 2];
609 System.arraycopy(codonCounts, 2, sortedCounts, 0,
610 codonCounts.length - 2);
612 int[] result = new int[3 + 2 * sortedCounts.length];
613 // first value is just the type of profile data
614 result[0] = AlignmentAnnotation.CDNA_PROFILE;
616 char[] codons = new char[sortedCounts.length];
617 for (int i = 0; i < codons.length; i++)
619 codons[i] = (char) i;
621 QuickSort.sort(sortedCounts, codons);
622 int totalPercentage = 0;
623 int distinctValuesCount = 0;
625 int divisor = ignoreGaps ? codonCounts[1] : codonCounts[0];
626 for (int i = codons.length - 1; i >= 0; i--)
628 final int codonCount = sortedCounts[i];
631 break; // nothing else of interest here
633 distinctValuesCount++;
634 result[j++] = codons[i];
635 final int percentage = codonCount * 100 / divisor;
636 result[j++] = percentage;
637 totalPercentage += percentage;
639 result[2] = totalPercentage;
642 * Just return the non-zero values
644 // todo next value is redundant if we limit the array to non-zero counts
645 result[1] = distinctValuesCount;
646 return Arrays.copyOfRange(result, 0, j);
650 * Compute a consensus for the cDNA coding for a protein alignment.
653 * the protein alignment (which should hold mappings to cDNA
656 * the consensus data stores to be populated (one per column)
658 public static void calculateCdna(AlignmentI alignment,
659 Hashtable[] hconsensus)
661 final char gapCharacter = alignment.getGapCharacter();
662 List<AlignedCodonFrame> mappings = alignment.getCodonFrames();
663 if (mappings == null || mappings.isEmpty())
668 int cols = alignment.getWidth();
669 for (int col = 0; col < cols; col++)
671 // todo would prefer a Java bean for consensus data
672 Hashtable<String, int[]> columnHash = new Hashtable<>();
673 // #seqs, #ungapped seqs, counts indexed by (codon encoded + 1)
674 int[] codonCounts = new int[66];
675 codonCounts[0] = alignment.getSequences().size();
676 int ungappedCount = 0;
677 for (SequenceI seq : alignment.getSequences())
679 if (seq.getCharAt(col) == gapCharacter)
683 List<char[]> codons = MappingUtils
684 .findCodonsFor(seq, col, mappings);
685 for (char[] codon : codons)
687 int codonEncoded = CodingUtils.encodeCodon(codon);
688 if (codonEncoded >= 0)
690 codonCounts[codonEncoded + 2]++;
695 codonCounts[1] = ungappedCount;
696 // todo: sort values here, save counts and codons?
697 columnHash.put(PROFILE, codonCounts);
698 hconsensus[col] = columnHash;
703 * Derive displayable cDNA consensus annotation from computed consensus data.
705 * @param consensusAnnotation
706 * the annotation row to be populated for display
707 * @param consensusData
708 * the computed consensus data
709 * @param showProfileLogo
710 * if true show all symbols present at each position, else only the
713 * the number of sequences in the alignment
715 public static void completeCdnaConsensus(
716 AlignmentAnnotation consensusAnnotation,
717 Hashtable[] consensusData, boolean showProfileLogo, int nseqs)
719 if (consensusAnnotation == null
720 || consensusAnnotation.annotations == null
721 || consensusAnnotation.annotations.length < consensusData.length)
723 // called with a bad alignment annotation row - wait for it to be
724 // initialised properly
728 // ensure codon triplet scales with font size
729 consensusAnnotation.scaleColLabel = true;
730 for (int col = 0; col < consensusData.length; col++)
732 Hashtable hci = consensusData[col];
735 // gapped protein column?
738 // array holds #seqs, #ungapped, then codon counts indexed by codon
739 final int[] codonCounts = (int[]) hci.get(PROFILE);
743 * First pass - get total count and find the highest
745 final char[] codons = new char[codonCounts.length - 2];
746 for (int j = 2; j < codonCounts.length; j++)
748 final int codonCount = codonCounts[j];
749 codons[j - 2] = (char) (j - 2);
750 totalCount += codonCount;
754 * Sort array of encoded codons by count ascending - so the modal value
755 * goes to the end; start by copying the count (dropping the first value)
757 int[] sortedCodonCounts = new int[codonCounts.length - 2];
758 System.arraycopy(codonCounts, 2, sortedCodonCounts, 0,
759 codonCounts.length - 2);
760 QuickSort.sort(sortedCodonCounts, codons);
762 int modalCodonEncoded = codons[codons.length - 1];
763 int modalCodonCount = sortedCodonCounts[codons.length - 1];
764 String modalCodon = String.valueOf(CodingUtils
765 .decodeCodon(modalCodonEncoded));
766 if (sortedCodonCounts.length > 1
767 && sortedCodonCounts[codons.length - 2] == sortedCodonCounts[codons.length - 1])
770 * two or more codons share the modal count
774 float pid = sortedCodonCounts[sortedCodonCounts.length - 1] * 100
775 / (float) totalCount;
778 * todo ? Replace consensus hashtable with sorted arrays of codons and
779 * counts (non-zero only). Include total count in count array [0].
783 * Scan sorted array backwards for most frequent values first. Show
784 * repeated values compactly.
786 StringBuilder mouseOver = new StringBuilder(32);
787 StringBuilder samePercent = new StringBuilder();
788 String percent = null;
789 String lastPercent = null;
790 int percentDecPl = getPercentageDp(nseqs);
792 for (int j = codons.length - 1; j >= 0; j--)
794 int codonCount = sortedCodonCounts[j];
798 * remaining codons are 0% - ignore, but finish off the last one if
801 if (samePercent.length() > 0)
803 mouseOver.append(samePercent).append(": ").append(percent)
808 int codonEncoded = codons[j];
809 final int pct = codonCount * 100 / totalCount;
810 String codon = String
811 .valueOf(CodingUtils.decodeCodon(codonEncoded));
812 StringBuilder sb = new StringBuilder();
813 Format.appendPercentage(sb, pct, percentDecPl);
814 percent = sb.toString();
815 if (showProfileLogo || codonCount == modalCodonCount)
817 if (percent.equals(lastPercent) && j > 0)
819 samePercent.append(samePercent.length() == 0 ? "" : ", ");
820 samePercent.append(codon);
824 if (samePercent.length() > 0)
826 mouseOver.append(samePercent).append(": ")
827 .append(lastPercent).append("% ");
829 samePercent.setLength(0);
830 samePercent.append(codon);
832 lastPercent = percent;
836 consensusAnnotation.annotations[col] = new Annotation(modalCodon,
837 mouseOver.toString(), ' ', pid);
842 * Returns the number of decimal places to show for profile percentages. For
843 * less than 100 sequences, returns zero (the integer percentage value will be
844 * displayed). For 100-999 sequences, returns 1, for 1000-9999 returns 2, etc.
849 protected static int getPercentageDp(long nseq)
861 * Returns the information content at a specified column.
864 * Index of the column, starting from 0.
867 public static float getInformationContent(int column,
868 HiddenMarkovModel hmm)
870 float informationContent = 0f;
872 for (char symbol : hmm.getSymbols())
875 freq = ResidueProperties.backgroundFrequencies
876 .get(hmm.getAlphabetType()).get(symbol);
877 Double hmmProb = hmm.getMatchEmissionProbability(column, symbol);
878 float prob = hmmProb.floatValue();
879 informationContent += prob * (Math.log(prob / freq) / Math.log(2));
883 return informationContent;
887 * Produces a HMM profile for a column in an alignment
890 * Alignment annotation for which the profile is being calculated.
892 * Column in the alignment the profile is being made for.
893 * @param removeBelowBackground
894 * Boolean indicating whether to ignore residues with probabilities
895 * less than their background frequencies.
898 public static int[] extractHMMProfile(HiddenMarkovModel hmm, int column,
899 boolean removeBelowBackground)
904 int size = hmm.getNumberOfSymbols();
905 char symbols[] = new char[size];
906 int values[] = new int[size];
907 List<Character> charList = hmm.getSymbols();
908 Integer totalCount = 0;
910 for (int i = 0; i < size; i++)
912 char symbol = charList.get(i);
914 int value = getAnalogueCount(hmm, column, symbol,
915 removeBelowBackground);
920 QuickSort.sort(values, symbols);
922 int[] profile = new int[3 + size * 2];
924 profile[0] = AlignmentAnnotation.SEQUENCE_PROFILE;
931 for (int k = size - 1; k >= 0; k--)
934 Integer value = values[k];
935 if (removeBelowBackground)
937 percentage = (value.floatValue() / totalCount.floatValue())
942 percentage = value.floatValue() / 100f;
944 int intPercent = Math.round(percentage);
945 profile[arrayPos] = symbols[k];
946 profile[arrayPos + 1] = intPercent;
956 * Converts the emission probability of a residue at a column in the alignment
957 * to a 'count' to allow for processing by the annotation renderer.
961 * @param removeBelowBackground
962 * When true, this method returns 0 for any symbols with a match
963 * emission probability less than the background frequency.
967 static int getAnalogueCount(HiddenMarkovModel hmm, int column,
968 char symbol, boolean removeBelowBackground)
972 value = hmm.getMatchEmissionProbability(column, symbol);
975 freq = ResidueProperties.backgroundFrequencies
976 .get(hmm.getAlphabetType()).get(symbol);
977 if (value < freq && removeBelowBackground)
982 value = value * 10000;
983 return Math.round(value.floatValue());