uint32_t
StateOfRows::getWeightedNode(double cand) const
{
double sum = 0;
for (const RowState & rs : _rows) {
sum += rs.getAverageSearchTimeInverse();
}
double accum(0.0);
for (size_t rowId(0); (rowId + 1) < _rows.size(); rowId++) {
accum += _rows[rowId].getAverageSearchTimeInverse();
if (cand < accum/sum) {
return rowId;
}
}
return _rows.size() - 1;
}
void calculateImpactScoresForAllSamples(vector< list<Module> >* modulesListOfAllSamples,
vector< vector<Driver> >* driversOfAllSamples, TDoubleMatrix* originalGeneExpressionMatrix, vector<int>* GenesEx,
int totalGenes, double F, vector<string>* geneIdToSymbol, string filename){
int totalSamples = modulesListOfAllSamples->size();
//map the gene id to the row id in the gene expression matrix
vector<int> rowId(totalGenes);
for (int i = 0; i < totalGenes; ++i) {
rowId[i] = findIndex(GenesEx, i);
}
//OUTPUT: print drivers and impact scores for all samples
vector<string>* outputDrivers = new vector<string>;
outputDrivers->push_back("SAMPLE_ID\tDRIVER\tIMPACT_SCORE\tIS_DEREGULATED\tMODULE_SIZE\tNUM_DRIVERS");
//for each sample i
for (int i = 0; i < totalSamples; ++i) {
list<Module> modules = modulesListOfAllSamples->at(i);
//for each module
for (list<Module>::iterator it = modules.begin(); it != modules.end(); it++) {
Module module = *it;
double score = 0;
int moduleSize = 0;
int numDrivers = module.driverGeneIds.size();
//sum up the fold change
for(list<int>::iterator git = module.explainedGeneIdsUpDown.begin(); git != module.explainedGeneIdsUpDown.end(); git++){
int currentExplainedGeneId = *git;
if(currentExplainedGeneId < totalGenes){ //up
score += fabs(originalGeneExpressionMatrix->at(rowId[currentExplainedGeneId])[i]);
}else{ //down
score += fabs(originalGeneExpressionMatrix->at(rowId[currentExplainedGeneId - totalGenes])[i]);
}
moduleSize++;
}
for(list<int>::iterator git = module.phenotypeGeneIdsUpDown.begin(); git != module.phenotypeGeneIdsUpDown.end(); git++){
int currentPhenotypeGeneId = *git;
if(currentPhenotypeGeneId < totalGenes){ //up
score += fabs(originalGeneExpressionMatrix->at(rowId[currentPhenotypeGeneId])[i]);
}else{ //down
score += fabs(originalGeneExpressionMatrix->at(rowId[currentPhenotypeGeneId - totalGenes])[i]);
}
moduleSize++;
}
//save the drivers and their scores
for(list<int>::iterator git = module.driverGeneIds.begin(); git != module.driverGeneIds.end(); git++){
Driver driver;
driver.geneId = *git;
driver.sampleId = i;
driver.impactScore = score;
//OUTPUT: print drivers and impact scores for all samples (cont.)
string str = intToStr(i) + "\t" + geneIdToSymbol->at(*git) + "\t" + doubleToStr(score, 3) + "\t";
//check if the driver gene is also a deregulated gene
if(rowId[*git] != -1 and fabs(originalGeneExpressionMatrix->at(rowId[*git])[i]) >= F){
driver.isDeregulated = true;
str = str + "1\t" + intToStr(moduleSize) + "\t" + intToStr(numDrivers);
}else{
driver.isDeregulated = false;
str = str + "0\t" + intToStr(moduleSize) + "\t" + intToStr(numDrivers);
}
driversOfAllSamples->at(i).push_back(driver);
outputDrivers->push_back(str);
}
} //end for each module
}
//OUTPUT: print drivers and impact scores for all samples (cont.)
writeStrVector(filename.c_str(), outputDrivers);
delete outputDrivers;
}
