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;
}
/**
* 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;
}
