public double kullbackLeiber(PositionWeightMatrix other) {
double kl = 0;
for (int i = 0; i < other.getNumCol(); i++) {
kl += get(i).kullbackLeiber(other.get(i));
}
return kl;
}
public List<BED> match(SequenceRegion region, double[] background, float minScore) {
PositionWeightMatrix neutralPWM = createIsoPWM(background, "neutral");
WindowSlider slider = WindowSlider.getSlider(region, size(), size() - 1);
List<BED> scoredWindows = new ArrayList<BED>();
while (slider.hasNext()) {
SequenceRegion window = slider.next();
// System.out.println(window + " seq: " + window.getSequenceBases());
char[] windowChrs = window.getSequenceBases().toCharArray();
double directScore = getLogLikelihood(windowChrs) - neutralPWM.getLogLikelihood(windowChrs);
window.reverse();
char[] reversedChrs = window.getSequenceBases().toCharArray();
double reverseScore =
getLogLikelihood(reversedChrs) - neutralPWM.getLogLikelihood(reversedChrs);
double max = Math.max(directScore, reverseScore);
if (max >= minScore) {
BED scoredWindow = new BED(window);
scoredWindow.setStart(scoredWindow.getStart() + region.getStart());
scoredWindow.setEnd(scoredWindow.getEnd() + region.getStart());
scoredWindow.setOrientation(directScore > reverseScore);
scoredWindow.setScore(max);
scoredWindow.setChromosome(region.getContainingSequenceId());
scoredWindows.add(scoredWindow);
}
}
return scoredWindows;
}
/**
* Creates a PWM of similar dimension to this PWM but with all columns set to the given vector,
* usually a neutral mutation vector.
*
* @param column
* @return
*/
public PositionWeightMatrix createIsoPWM(double[] column, String name) {
PositionWeightMatrix pwm = new PositionWeightMatrix(name);
for (int i = 0; i < size(); i++) {
pwm.addColumn(column);
}
return pwm;
}
public PositionWeightMatrix copy() {
PositionWeightMatrix copy = new PositionWeightMatrix(getName());
for (PositionWeightColumn c : this) {
copy.add(c);
}
return copy;
}
public PositionWeightMatrix reverseComplement() {
PositionWeightMatrix wm = new PositionWeightMatrix(getName());
int numCol = size();
for (int i = 0; i < numCol; i++) wm.add(get(numCol - 1 - i).getComplement());
wm.name = name;
wm.rightHighInfoStart = numCol - 1 - leftHighInfoStart;
wm.leftHighInfoStart = numCol - 1 - rightHighInfoStart;
return wm;
}
protected void addToCentroid(Matrix m, PositionWeightMatrix pwm) {
for (int i = 0; i < m.getColumnDimension(); i++) {
PositionWeightColumn c = pwm.get(i);
for (int j = 0; j < m.getRowDimension(); j++) {
m.set(j, i, m.get(j, i) + c.getWeight(j));
}
}
}
/**
* Computes the euclidean centroid, it may not be the formal centroid for different metrics but
* intuitively the average counts should provide a good cluster representative which is what this
* method intends to return
*
* @param pwmSet - Collection of pwms from which to compute the centroid0
* @return The euclidean centroid
* @throws IllegalArgumentException - When not all PWMs have the same dimension.
*/
public PositionWeightMatrix centroidOf(Collection<PositionWeightMatrix> pwmSet)
throws IllegalArgumentException {
Matrix centroidMatrix = null;
Iterator<PositionWeightMatrix> pwmIt = pwmSet.iterator();
if (pwmIt.hasNext()) {
PositionWeightMatrix first = pwmIt.next();
PositionWeightColumn firstCol = first.get(0);
centroidMatrix = new Matrix(firstCol.getAlphabetSize(), first.getNumCol());
addToCentroid(centroidMatrix, first);
}
while (pwmIt.hasNext()) {
PositionWeightMatrix pwm = pwmIt.next();
if (pwm.getNumCol() != centroidMatrix.getColumnDimension()) {
throw new IllegalArgumentException(
"Error computing centroid. All PWMs in set should have the same dimension");
}
addToCentroid(centroidMatrix, pwm);
}
centroidMatrix.times(1 / (double) pwmSet.size());
PositionWeightMatrix centroid = new PositionWeightMatrix("centroid");
for (int j = 0; j < centroidMatrix.getColumnDimension(); j++) {
centroid.addColumn(centroidMatrix.getColumn(j));
}
return centroid;
}
/**
* Compute Distribution of scores for motif.
*
* @param s
* @param isNumSeq
* @return
*/
public List<Double> computeScoreDistribution(
float backgroundA, float backgroundC, float backgroundG, float backgroundT, int sampleSize) {
double[] bg = {backgroundA, backgroundC, backgroundG, backgroundT};
PositionWeightMatrix bgPWM = createIsoPWM(bg, "bg");
// System.out.println("bg vector " + backgroundA + "," + backgroundC + "," + backgroundG + "," +
// backgroundT);
List<Double> dist = new ArrayList<Double>(sampleSize);
int[] kmer = new int[size()];
for (int i = 0; i < sampleSize; i++) {
Random r = new Random();
for (int j = 0; j < size(); j++) {
kmer[j] = r.nextInt(ALPHABET_SIZE);
}
dist.add(getLogLikelihood(kmer) - bgPWM.getLogLikelihood(kmer));
// System.out.println(printKmer(kmer)+ " -- " + (getLogLikelihood(kmer) -
// bgPWM.getLogLikelihood(kmer)));
}
return dist;
}
/**
* Trims this PWM by removing start end ending columns with information content lesser than given.
*
* @param ic - Columns below this ic at PWM edges will be trimmed.
* @return The resulting PWM.
*/
public PositionWeightMatrix trimByInformationContent(double ic) {
PositionWeightMatrix trimmed = new PositionWeightMatrix(getName());
int lastPos = size() - 1;
while (lastPos >= 0 && get(lastPos).getInformationContent() < ic) {
lastPos--;
}
int startPos = 0;
while (startPos < lastPos && get(startPos).getInformationContent() < ic) {
startPos++;
}
for (int i = startPos; i <= lastPos; i++) {
trimmed.add(get(i));
}
return trimmed;
}
public PositionWeightMatrix permuteColumns(boolean preserveGCDinucleotides) {
PositionWeightMatrix original = copy();
List<Integer> dinucleotidesToPreserveFirstIdx =
preserveGCDinucleotides ? original.gcDinucleotidesFistIdxs() : new ArrayList<Integer>();
PositionWeightMatrix permutted = new PositionWeightMatrix(getName() + "_perm");
Random r = new Random();
List<Integer> idxList = new ArrayList<Integer>(original.size());
for (int i = 0; i < original.size(); i++) {
idxList.add(i);
}
while (idxList.size() > 0) {
int idxOfIdx = r.nextInt(idxList.size());
int idx = idxList.remove(idxOfIdx);
if (dinucleotidesToPreserveFirstIdx.contains(idx - 1)) {
PositionWeightColumn col = original.get(idx - 1);
permutted.add(col);
PositionWeightColumn nextCol = original.get(idx);
idxList.remove(idxOfIdx - 1);
permutted.add(nextCol);
} else {
PositionWeightColumn col = original.get(idx);
permutted.add(col);
if (dinucleotidesToPreserveFirstIdx.contains(idx)) {
PositionWeightColumn nextCol = original.get(idx + 1);
idxList.remove(idxOfIdx);
permutted.add(nextCol);
}
}
}
return permutted;
}
public void addPWMColumn(double[] col) {
pwm.addColumn(col);
}
public String getName() {
return pwm.getName();
}