public SyntheticRegRegionService(int numOfPositiveRegRegions, int negExMultiplicationFactor)
throws DataFormatException {
modelRegRegionService = new CElegansRegRegionService(0);
File pwmDir = new File(SystemVariables.getInstance().getString("C.elegans.PWMs.dir"));
String tmpJobDir =
SystemVariables.getInstance()
.getString("temp.output.dir"); // TODO create a proper tmp job dir for experiments
modelRegElService = new PatserRegElementService(pwmDir, tmpJobDir);
if (modelRegRegionService.getPositiveRegulatoryRegions().size() > 0) {
regulatorySequenceLength =
modelRegRegionService.getPositiveRegulatoryRegions().get(0).getSequence().length();
} else {
throw new DataFormatException(
"Model regulatory regions are empty. Can not model synthetic regions based on an empty regions.");
}
// TODO: remove - just for testing
regulatorySequenceLength = 25;
negativeRegRegions =
FeaturesTools.generateSimulatedRegulatoryRegions(
negExMultiplicationFactor * numOfPositiveRegRegions,
regulatorySequenceLength,
NEGATIVE_REGION_PREFIX,
sequenceNtProbabilities);
positiveRegRegions =
FeaturesTools.generateSimulatedRegulatoryRegionsWithPositionalPSSMs(
numOfPositiveRegRegions,
regulatorySequenceLength,
sequenceNtProbabilities,
SystemVariables.getInstance().getPositivePatserCutOffScore(),
POSITIVE_REGION_PREFIX,
modelRegRegionService.getPositiveRegulatoryRegions(),
modelRegElService);
// this.generateRegulatoryRegions(numOfPositiveRegRegions, );
}
@Override
public void updateNumberOfNegativeRegRegions(int multiplicationFactor) {
// TODO check for empty regions
// TODO test this
int deltaSeqNum = positiveRegRegions.size() * multiplicationFactor - negativeRegRegions.size();
if (deltaSeqNum < 0) { // desired number of negative sequences is less than is available now
for (int i = 0; i > deltaSeqNum; i--) {
negativeRegRegions.remove(0);
}
} else { // desired number is more than available -> need to generate extra
ArrayList<Feature> extraSimulatedGenes =
FeaturesTools.generateSimulatedRegulatoryRegions(
deltaSeqNum,
positiveRegRegions.get(0).getSequence().length(),
NEGATIVE_REGION_PREFIX,
sequenceNtProbabilities);
negativeRegRegions.addAll(extraSimulatedGenes);
}
}