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;
import jalview.util.QuickSort;
+import java.awt.Color;
import java.util.Arrays;
import java.util.Hashtable;
import java.util.List;
{
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'
*/
}
}
- public static final ProfilesI calculate(List<SequenceI> list,
- int start, int end)
+ public static final ProfilesI calculate(List<SequenceI> list, int start,
+ int end)
{
return calculate(list, start, end, false);
}
}
}
+
+
/**
* Calculate the consensus symbol(s) for each column in the given range.
*
// System.out.println(elapsed);
}
+ public static ProfilesI calculateInformation(final HiddenMarkovModel hmm,
+ int width, int start, int end, boolean saveFullProfile,
+ boolean removeBelowBackground)
+ {
+ ProfileI[] result = new ProfileI[width];
+ int symbolCount = hmm.getNumberOfSymbols();
+ String alph = hmm.getAlphabetType();
+ for (int column = start; column < end; column++)
+ {
+ ResidueCount counts = new ResidueCount();
+ for (char symbol : hmm.getSymbols())
+ {
+ int value = getAnalogueCount(hmm, column, removeBelowBackground,
+ alph, symbol);
+ 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
}
/**
+ * 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 void completeInformation(AlignmentAnnotation information,
+ ProfilesI profiles, int startCol, int endCol,
+ boolean ignoreBelowBackground,
+ boolean showSequenceLogo, long nseq)
+ {
+ // 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;
+ }
+
+ 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;
+ }
+
+ 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(" ", description,
+ ' ', value);
+ }
+ information.graphMax = max;
+ // long elapsed = System.currentTimeMillis() - now;
+ // System.out.println(-elapsed);
+ }
+
+ /**
* Derive the gap count annotation row.
*
- * @param consensus
+ * @param gaprow
* the annotation row to add annotations to
* @param profiles
* the source consensus data
* @param endCol
* end column (exclusive)
*/
- public static void completeGapAnnot(AlignmentAnnotation consensus,
+ public static void completeGapAnnot(AlignmentAnnotation gaprow,
ProfilesI profiles, int startCol, int endCol, long nseq)
{
- // long now = System.currentTimeMillis();
- if (consensus == null || consensus.annotations == null
- || consensus.annotations.length < endCol)
+ if (gaprow == null || gaprow.annotations == null
+ || gaprow.annotations.length < endCol)
{
/*
* called with a bad alignment annotation row
return;
}
// always set ranges again
- consensus.graphMax = nseq;
- consensus.graphMin = 0;
+ gaprow.graphMax = nseq;
+ gaprow.graphMin = 0;
+ double scale = 0.8/nseq;
for (int i = startCol; i < endCol; i++)
{
ProfileI profile = profiles.get(i);
* happens if sequences calculated over were
* shorter than alignment width
*/
- consensus.annotations[i] = null;
+ gaprow.annotations[i] = null;
return;
}
- final int gapped = profile.getGapped();
+ final int gapped = profile.getNonGapped();
String description = "" + gapped;
- consensus.annotations[i] = new Annotation(gapped);
+ gaprow.annotations[i] = new Annotation("", description,
+ '\0', gapped, jalview.util.ColorUtils.bleachColour(
+ Color.DARK_GRAY, (float) scale * gapped));
}
- // long elapsed = System.currentTimeMillis() - now;
- // System.out.println(-elapsed);
}
/**
* calculations
* @return
*/
- public static int[] extractProfile(ProfileI profile,
- boolean ignoreGaps)
+ public static int[] extractProfile(ProfileI profile, boolean ignoreGaps)
{
int[] rtnval = new int[64];
ResidueCount counts = profile.getCounts();
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;
+ if ("amino".equals(hmm.getAlphabetType()))
+ {
+ freq = ResidueProperties.aminoBackgroundFrequencies.get(symbol);
+ }
+ if ("DNA".equals(hmm.getAlphabetType()))
+ {
+ freq = ResidueProperties.dnaBackgroundFrequencies.get(symbol);
+ }
+ if ("RNA".equals(hmm.getAlphabetType()))
+ {
+ freq = ResidueProperties.rnaBackgroundFrequencies.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)
+ {
+
+ if (hmm != null)
+ {
+ String alph = hmm.getAlphabetType();
+ 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, removeBelowBackground,
+ alph, symbol);
+ 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;
+ }
+
+ private static int getAnalogueCount(HiddenMarkovModel hmm, int column,
+ boolean removeBelowBackground, String alph, char symbol)
+ {
+ Double value;
+
+ value = hmm.getMatchEmissionProbability(column, symbol);
+ double freq;
+
+ if (AMINO.equals(alph) && removeBelowBackground)
+ {
+ freq = ResidueProperties.aminoBackgroundFrequencies.get(symbol);
+ if (value < freq)
+ {
+ value = 0d;
+ }
+ }
+ else if (DNA.equals(alph) && removeBelowBackground)
+ {
+ freq = ResidueProperties.dnaBackgroundFrequencies.get(symbol);
+ if (value < freq)
+ {
+ value = 0d;
+ }
+ }
+ else if (RNA.equals(alph) && removeBelowBackground)
+ {
+ freq = ResidueProperties.rnaBackgroundFrequencies.get(symbol);
+ if (value < freq)
+ {
+ value = 0d;
+ }
+ }
+ value = value * 10000;
+ return Math.round(value.floatValue());
+ }
}