/**
* 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;
}
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 information content of this position weith matris
*
* @return
*/
public double ic() {
double ic = 0;
for (PositionWeightColumn col : this) {
ic += col.getInformationContent();
}
return ic;
}
private List<Integer> gcDinucleotidesFistIdxs() {
List<Integer> dinucleotidesToPreserveFirstIdx = new ArrayList<Integer>();
for (int i = 0; i < size() - 1; i++) {
PositionWeightColumn thisCol = get(i);
PositionWeightColumn nextCol = get(i + 1);
if (thisCol.getLogRatio('G') > 0 && nextCol.getLogRatio('C') > 0) {
dinucleotidesToPreserveFirstIdx.add(i);
}
}
return dinucleotidesToPreserveFirstIdx;
}
public void write(BufferedWriter bw, NumberFormat formatter) throws IOException {
bw.write(">" + name);
bw.newLine();
if (size() > 0) {
int rows = get(0).getAlphabetSize();
for (int i = 0; i < rows; i++) {
Iterator<PositionWeightColumn> colIt = iterator();
while (colIt.hasNext()) {
PositionWeightColumn col = colIt.next();
bw.write(formatter.format(col.getWeight(i)));
if (colIt.hasNext()) {
bw.write("\t");
}
}
bw.newLine();
}
}
}
public void addPseudoCounts() {
for (PositionWeightColumn pwc : this) {
pwc.addPseudoCounts();
}
}