import jalview.datamodel.AlignmentAnnotation;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.Annotation;
+import jalview.datamodel.HiddenMarkovModel;
import jalview.datamodel.Profile;
import jalview.datamodel.ProfileI;
import jalview.datamodel.Profiles;
import jalview.datamodel.ResidueCount.SymbolCounts;
import jalview.datamodel.SequenceI;
import jalview.ext.android.SparseIntArray;
+import jalview.schemes.ResidueProperties;
import jalview.util.Comparison;
import jalview.util.Format;
import jalview.util.MappingUtils;
{
public static final String PROFILE = "P";
+ private static final String AMINO = "amino";
+
+ private static final String DNA = "DNA";
+
+ private static final String RNA = "RNA";
+
/*
* Quick look-up of String value of char 'A' to 'Z'
*/
}
}
+
+
/**
* Calculate the consensus symbol(s) for each column in the given range.
*
}
/**
+ * Returns the full set of profiles for a hidden Markov model. The underlying
+ * data is the raw probabilities of a residue being emitted at each node,
+ * however the profiles returned by this function contain the percentage
+ * chance of a residue emission.
+ *
+ * @param hmm
+ * @param width
+ * The width of the Profile array (Profiles) to be returned.
+ * @param start
+ * The alignment column on which the first profile is based.
+ * @param end
+ * The alignment column on which the last profile is based.
+ * @param saveFullProfile
+ * Flag for saving the counts for each profile
+ * @param removeBelowBackground
+ * Flag for removing any characters with a match emission probability
+ * less than its background frequency
+ * @return
+ */
+ public static ProfilesI calculateHMMProfiles(final HiddenMarkovModel hmm,
+ int width, int start, int end, boolean saveFullProfile,
+ boolean removeBelowBackground, boolean infoLetterHeight)
+ {
+ ProfileI[] result = new ProfileI[width];
+ int symbolCount = hmm.getNumberOfSymbols();
+ for (int column = start; column < end; column++)
+ {
+ ResidueCount counts = new ResidueCount();
+ for (char symbol : hmm.getSymbols())
+ {
+ int value = getAnalogueCount(hmm, column, symbol,
+ removeBelowBackground, infoLetterHeight);
+ counts.put(symbol, value);
+ }
+ int maxCount = counts.getModalCount();
+ String maxResidue = counts.getResiduesForCount(maxCount);
+ int gapCount = counts.getGapCount();
+ ProfileI profile = new Profile(symbolCount, gapCount, maxCount,
+ maxResidue);
+
+ if (saveFullProfile)
+ {
+ profile.setCounts(counts);
+ }
+
+ result[column] = profile;
+ }
+ return new Profiles(result);
+ }
+
+ /**
* Make an estimate of the profile size we are going to compute i.e. how many
* different characters may be present in it. Overestimating has a cost of
* using more memory than necessary. Underestimating has a cost of needing to
}
/**
+ * Derive the information annotations to be added to the alignment for
+ * display. This does not recompute the raw data, but may be called on a
+ * change in display options, such as 'ignore below background frequency',
+ * which may in turn result in a change in the derived values.
+ *
+ * @param information
+ * the annotation row to add annotations to
+ * @param profiles
+ * the source information data
+ * @param startCol
+ * start column (inclusive)
+ * @param endCol
+ * end column (exclusive)
+ * @param ignoreGaps
+ * if true, normalise residue percentages
+ * @param showSequenceLogo
+ * if true include all information symbols, else just show modal
+ * residue
+ * @param nseq
+ * number of sequences
+ */
+ public static float completeInformation(AlignmentAnnotation information,
+ ProfilesI profiles, int startCol, int endCol, long nseq,
+ Float currentMax)
+ {
+ // long now = System.currentTimeMillis();
+ if (information == null || information.annotations == null
+ || information.annotations.length < endCol)
+ {
+ /*
+ * called with a bad alignment annotation row
+ * wait for it to be initialised properly
+ */
+ return 0;
+ }
+
+ Float max = 0f;
+
+ for (int i = startCol; i < endCol; i++)
+ {
+ ProfileI profile = profiles.get(i);
+ if (profile == null)
+ {
+ /*
+ * happens if sequences calculated over were
+ * shorter than alignment width
+ */
+ information.annotations[i] = null;
+ return 0;
+ }
+
+ HiddenMarkovModel hmm;
+
+ SequenceI hmmSeq = information.sequenceRef;
+
+ hmm = hmmSeq.getHMM();
+
+ Float value = getInformationContent(i, hmm);
+
+ if (value > max)
+ {
+ max = value;
+ }
+
+ String description = value + " bits";
+ information.annotations[i] = new Annotation(
+ Character.toString(Character
+ .toUpperCase(hmm.getConsensusAtAlignColumn(i))),
+ description, ' ', value);
+ }
+ if (max > currentMax)
+ {
+ information.graphMax = max;
+ return max;
+ }
+ else
+ {
+ information.graphMax = currentMax;
+ return currentMax;
+ }
+ }
+
+ /**
* Derive the gap count annotation row.
*
* @param gaprow
return result;
}
+
/**
* Extract a sorted extract of cDNA codon profile data. The returned array
* contains
for (int col = 0; col < cols; col++)
{
// todo would prefer a Java bean for consensus data
- Hashtable<String, int[]> columnHash = new Hashtable<String, int[]>();
+ Hashtable<String, int[]> columnHash = new Hashtable<>();
// #seqs, #ungapped seqs, counts indexed by (codon encoded + 1)
int[] codonCounts = new int[66];
codonCounts[0] = alignment.getSequences().size();
}
return scale;
}
+
+ /**
+ * Returns the information content at a specified column.
+ *
+ * @param column
+ * Index of the column, starting from 0.
+ * @return
+ */
+ public static float getInformationContent(int column,
+ HiddenMarkovModel hmm)
+ {
+ float informationContent = 0f;
+
+ for (char symbol : hmm.getSymbols())
+ {
+ float freq = 0f;
+ freq = ResidueProperties.backgroundFrequencies
+ .get(hmm.getAlphabetType()).get(symbol);
+ Double hmmProb = hmm.getMatchEmissionProbability(column, symbol);
+ float prob = hmmProb.floatValue();
+ informationContent += prob * (Math.log(prob / freq) / Math.log(2));
+
+ }
+
+ return informationContent;
+ }
+
+ /**
+ * Produces a HMM profile for a column in an alignment
+ *
+ * @param aa
+ * Alignment annotation for which the profile is being calculated.
+ * @param column
+ * Column in the alignment the profile is being made for.
+ * @param removeBelowBackground
+ * Boolean indicating whether to ignore residues with probabilities
+ * less than their background frequencies.
+ * @return
+ */
+ public static int[] extractHMMProfile(HiddenMarkovModel hmm, int column,
+ boolean removeBelowBackground, boolean infoHeight)
+ {
+
+ if (hmm != null)
+ {
+ int size = hmm.getNumberOfSymbols();
+ char symbols[] = new char[size];
+ int values[] = new int[size];
+ List<Character> charList = hmm.getSymbols();
+ Integer totalCount = 0;
+
+ for (int i = 0; i < size; i++)
+ {
+ char symbol = charList.get(i);
+ symbols[i] = symbol;
+ int value = getAnalogueCount(hmm, column, symbol,
+ removeBelowBackground, infoHeight);
+ values[i] = value;
+ totalCount += value;
+ }
+
+ QuickSort.sort(values, symbols);
+
+ int[] profile = new int[3 + size * 2];
+
+ profile[0] = AlignmentAnnotation.SEQUENCE_PROFILE;
+ profile[1] = size;
+ profile[2] = 100;
+
+ if (totalCount != 0)
+ {
+ int arrayPos = 3;
+ for (int k = size - 1; k >= 0; k--)
+ {
+ Float percentage;
+ Integer value = values[k];
+ if (removeBelowBackground)
+ {
+ percentage = (value.floatValue() / totalCount.floatValue())
+ * 100;
+ }
+ else
+ {
+ percentage = value.floatValue() / 100f;
+ }
+ int intPercent = Math.round(percentage);
+ profile[arrayPos] = symbols[k];
+ profile[arrayPos + 1] = intPercent;
+ arrayPos += 2;
+ }
+ }
+ return profile;
+ }
+ return null;
+ }
+
+ /**
+ * Converts the emission probability of a residue at a column in the alignment
+ * to a 'count' to allow for processing by the annotation renderer.
+ *
+ * @param hmm
+ * @param column
+ * @param removeBelowBackground
+ * When true, this method returns 0 for any symbols with a match
+ * emission probability less than the background frequency.
+ * @param symbol
+ * @return
+ */
+ static int getAnalogueCount(HiddenMarkovModel hmm, int column,
+ char symbol, boolean removeBelowBackground, boolean infoHeight)
+ {
+ Double value;
+
+ value = hmm.getMatchEmissionProbability(column, symbol);
+ double freq;
+
+ freq = ResidueProperties.backgroundFrequencies
+ .get(hmm.getAlphabetType()).get(symbol);
+ if (value < freq && removeBelowBackground)
+ {
+ return 0;
+ }
+
+ if (infoHeight)
+ {
+ value = value * (Math.log(value / freq) / Math.log(2));
+ }
+
+ value = value * 10000;
+ return Math.round(value.floatValue());
+ }
}