public String induceRulesWithTestSet(double testSetPercent) throws DataFormatException {
SystemVariables.getInstance()
.appendToExperimentNotes("\\n* Total number of PSSM matches in positive sequences: ");
// System.out.println("Total number of PSSM matches in positive sequences: ");
ArrayList<Feature> posTrainingSet =
regulatoryElementService.getRegulatoryElements(
regulatoryRegionService.getPositiveRegulatoryRegions(),
positiveCutOffScore,
posATcomposition,
posCGcomposition);
SystemVariables.getInstance()
.appendToExperimentNotes("\\n* Total number of PSSM matches in negative sequences: ");
// System.out.println("Total number of PSSM matches in negative sequences: ");
ArrayList<Feature> negTrainingSet =
regulatoryElementService.getRegulatoryElements(
regulatoryRegionService.getNegativeRegulatoryRegions(),
negativeCutOffScore,
negATcomposition,
negCGcomposition);
IlpService formatter = new IlpService(tempIlpJobDirName);
formatter.createIlpFilesWithTestSet(
regulatoryRegionService.getPositiveRegulatoryRegions(),
regulatoryRegionService.getNegativeRegulatoryRegions(),
posTrainingSet,
negTrainingSet,
testSetPercent);
String ilpTheory = formatter.runILP();
return ilpTheory;
}
public IlpService createIlpFiles() throws DataFormatException {
ArrayList<Feature> posRegElements =
regulatoryElementService.getRegulatoryElements(
regulatoryRegionService.getPositiveRegulatoryRegions(),
positiveCutOffScore,
posATcomposition,
posCGcomposition);
if (posRegElements == null) {
posRegElements =
regulatoryElementService.getRegulatoryElements(
regulatoryRegionService.getPositiveRegulatoryRegions(),
regulatoryRegionService.getNegativeRegulatoryRegions(),
positiveCutOffScore);
}
// ***** Add statistics
addMotifMatchingStatisticsToNotes("positive");
SystemVariables.getInstance().setPosSeqRelElMatchesNum(posRegElements.size());
ArrayList<Feature> negRegElements =
regulatoryElementService.getRegulatoryElements(
regulatoryRegionService.getNegativeRegulatoryRegions(),
negativeCutOffScore,
negATcomposition,
negCGcomposition);
if (negRegElements == null) {
negRegElements =
regulatoryElementService.getRegulatoryElements(
regulatoryRegionService.getNegativeRegulatoryRegions(), null, negativeCutOffScore);
}
// ***** Add statistics
addMotifMatchingStatisticsToNotes("negative");
SystemVariables.getInstance().setNegSeqRelElMatchesNum(negRegElements.size());
/*
System.out.println("\n Number of matches of pairs of matrices in jurkat and erythroid sequences");
String[] pwmPair = new String[]{"MA0055.1", "MA0152.1"};
System.out.println(pwmPair[0] + " and " + pwmPair[1]+" :\t" +
FeaturesTools.numSequencesContainingAll(pwmPair, posRegElements) + "\t"+
FeaturesTools.numSequencesContainingAll(pwmPair, negRegElements));
System.out.println("\nDistances between Gata and Ebox for positive sequences:");
FeaturesTools.printRegElDistances("Gata", "Ebox", posRegElements);
*/
IlpService ilpService = new IlpService(tempIlpJobDirName);
ilpService.createIlpFiles(
regulatoryRegionService.getPositiveRegulatoryRegions(),
regulatoryRegionService.getNegativeRegulatoryRegions(),
posRegElements,
negRegElements);
return ilpService;
}
/* Note: unsound method
* Based on the statistics, collected when extracting motif matches by RegElementService,
* collects text note of motif matching statistics, which will be added to ILP file.
* This method has to be called immediately after RegElementService.getRegulatoryElements call,
* since every such call overrides the pssmMatchStatistics in RegElementService
*
* @param sequenceKind - can only be "positive" or "negative", depending on the kind of sequences
*/
private void addMotifMatchingStatisticsToNotes(String sequenceKind) throws DataFormatException {
Hashtable<String, Double> pssmMatchStats = regulatoryElementService.getPssmMatchingStatistics();
if (pssmMatchStats != null && !pssmMatchStats.isEmpty()) {
SystemVariables.getInstance()
.appendToExperimentNotes(
"\\n* Number of PSSM matches in " + sequenceKind + " sequences: ");
int numberOfRegions = 0;
if ("positive".equals(sequenceKind)) {
numberOfRegions = regulatoryRegionService.getPositiveRegulatoryRegions().size();
} else {
numberOfRegions = regulatoryRegionService.getNegativeRegulatoryRegions().size();
}
Enumeration<String> pssmMatchNames = pssmMatchStats.keys();
int totalNumMatches = 0;
String r_pssmNames = "";
String r_matches = "";
while (pssmMatchNames.hasMoreElements()) {
String pssmName = pssmMatchNames.nextElement();
double pssmStat = pssmMatchStats.get(pssmName);
int numMatches = (int) (pssmStat * numberOfRegions);
totalNumMatches = totalNumMatches + numMatches;
r_pssmNames = r_pssmNames + "\'" + pssmName + "\', ";
r_matches = r_matches + numMatches + ", ";
SystemVariables.getInstance()
.appendToExperimentNotes("\\n\\t" + pssmName + "\\t" + numMatches);
}
SystemVariables.getInstance()
.appendToExperimentNotes("\\n Total number of matches: " + totalNumMatches);
// System.out.println("In positive sequences: ");
// System.out.println(r_pssmNames);
// System.out.println(r_matches);
}
}
