/**
* Collapsing @param in (people by marker) to @param out (people by marker),
* if @param freqIn is empty, then frequncy is calculated from @param in
* or according to @param freqIn to rearrange columns of @param in.
* Reordered frequency are stored in @param freqOut, in ascending order
*/
void rearrangeGenotypeByFrequency(Matrix& in, const std::vector<double>& freqIn,
Matrix* out, std::vector<double>* freqOut) {
std::map<double, std::vector<int> > freqGroup;
std::map<double, std::vector<int> >::const_iterator freqGroupIter;
if (freqIn.empty()) {
getMarkerFrequency(in, freqOut);
groupFrequency(*freqOut, &freqGroup);
} else {
groupFrequency(freqIn, &freqGroup);
}
Matrix& sortedGenotype = *out;
sortedGenotype.Dimension(in.rows, freqGroup.size());
sortedGenotype.Zero();
freqOut->clear();
int idx = 0;
for (freqGroupIter = freqGroup.begin(); freqGroupIter != freqGroup.end();
freqGroupIter++) {
freqOut->push_back(freqGroupIter->first);
const std::vector<int>& cols = freqGroupIter->second;
for (size_t j = 0; j != cols.size(); ++j) {
for (int i = 0; i < in.rows; ++i) {
sortedGenotype[i][cols[j]] += in[i][cols[j]];
}
}
++idx;
}
}
void makeVariableThreshodlGenotype(
Matrix& in, Matrix* out, std::vector<double>* freqOut,
void (*collapseFunc)(Matrix&, const std::vector<int>&, Matrix*, int)) {
std::vector<double> freqIn;
getMarkerFrequency(in, &freqIn);
makeVariableThreshodlGenotype(in, freqIn, out, freqOut, collapseFunc);
}
void fpCollapse(Matrix& in, Matrix* out) {
assert(out);
int& numPeople = in.rows;
int numMarker = in.cols;
out->Dimension(numPeople, 1);
out->Zero();
for (int m = 0; m < numMarker; m++) {
// calculate weight
double freq = getMarkerFrequency(in, m);
if (freq <= 0.0 || freq >= 1.0) continue; // avoid freq == 1.0
double weight = 1.0 / sqrt(freq * (1.0 - freq));
// fprintf(stderr, "freq = %f\n", freq);
for (int p = 0; p < numPeople; p++) {
(*out)[p][0] += in[p][m] * weight;
}
}
}
void getMarkerFrequency(Matrix& in, std::vector<double>* freq) {
freq->resize(in.cols);
for (int i = 0; i < in.cols; ++i) {
(*freq)[i] = getMarkerFrequency(in, i);
}
}
