Matrix Matrix::operator* (Matrix M)//operator+ with an other matrix
{
int i,j,k;
int nrow2, ncol2;
//test the dimension
M.Dimension(&nrow2,&ncol2);
if(nrow2!=ncol)
{
error err(OUFFF_ERROR_DIMENSIONDISAGREE, "Matrix::*(matrix)", "Try to make product between matrix of bad size");
throw err;
}
//creation of the new matrix
Matrix N(nrow,ncol2);
for(i=0; i<nrow; i++)
{
for(j=0;j<ncol2;j++)
{
//N(i,j) is the product of the ith row from this by the jth col from M
N(i,j)=0;//init value to 0
for(k=0; k<ncol; k++)
{
N(i,j) += value[i][k]*M(k,j);
}
}
}
return N;
}
void LoadMatrix(const char* fn, Matrix& m) {
LineReader lr(fn);
std::vector<std::string> s;
int lineNo = 0;
while (lr.readLineBySep(&s, " \t")) {
lineNo++;
m.Dimension(lineNo, s.size());
for (int j = 0; j < s.size(); j++) {
m(lineNo - 1, j) = atof(s[j].c_str());
}
}
};
Matrix Matrix::operator+ (Matrix M)//operator+ with an other matrix
{
int i,j;
M.Dimension(&i,&j);
if(i!=nrow || j!= ncol)
{
throw error(OUFFF_ERROR_DIMENSIONDISAGREE, "Matrix::+(matrix)", "Try to make sum with matrix of different size");;
}
Matrix N(nrow,ncol);
for(i=0; i<nrow*ncol; i++)
N(i)=value[0][i] + M(i);
return N;
}
void LinearRegressionPermutationTest::splitMatrix(Matrix& x, int col,
Matrix& xnull, Vector& xcol) {
if (x.cols < 2) {
printf("input matrix has too few cols!\n");
}
xnull.Dimension(x.rows, x.cols - 1);
xcol.Dimension(x.rows);
for (int i = 0; i < x.rows; i++) {
for (int j = 0; j < x.cols; j++) {
if (j < col) {
xnull[i][j] = x[i][j];
} else if (j == col) {
xcol[i] = x[i][j];
} else {
xnull[i][j - 1] = x[i][j];
}
}
}
};
void HashErrorModel::setDataforPrediction(Matrix & X, Vector & succ, Vector & total,bool binarizeFlag)
{
int i = 0;
BaseData data;
if(ourUseFast)
{
return;
}
for(HashMatch::const_iterator it = mismatchTable.begin();
it != mismatchTable.end();
++it)
{
data.parseKey(it->first);
Covariates cov;
cov.setCovariates(data);
if(i == 0)
{
uint32_t rows = mismatchTable.size();
succ.Dimension(rows);
total.Dimension(rows);
X.Dimension(rows, cov.covariates.size() + 1);
X.Zero();
}
// The first column of the design matrix is constant one, for the slope
X[i][0] = 1.0;
int j = 0;
//binarize a couple of co-variates
for(std::vector<uint16_t>::const_iterator itv = cov.covariates.begin();
itv != cov.covariates.end();
++itv)
{
if(binarizeFlag)
{
//hardcoded pos is 2
if(j==1){
uint16_t pos = (uint16_t)*itv;
// hardcoded
pos += 7;
X[i][pos] = 1;
}
else
{
if(j>1)
X[i][j] = (uint16_t)*itv;
else
X[i][j+1] = (uint16_t)*itv;
}
j++;
}
else
{
X[i][j+1] = (uint16_t)*itv;
j++;
}
}
total[i] = it->second.mm + it->second.m;
succ[i] = it->second.m;
i++;
}
}
int main(int argc, char** argv){
time_t currentTime = time(0);
fprintf(stderr, "Analysis started at: %s", ctime(¤tTime));
////////////////////////////////////////////////
BEGIN_PARAMETER_LIST(pl)
ADD_PARAMETER_GROUP(pl, "Input/Output")
ADD_STRING_PARAMETER(pl, inVcf, "--inVcf", "input VCF File")
ADD_STRING_PARAMETER(pl, outMerlin, "--outMerlin", "output prefix")
ADD_PARAMETER_GROUP(pl, "People Filter")
ADD_STRING_PARAMETER(pl, peopleIncludeID, "--peopleIncludeID", "give IDs of people that will be included in study")
ADD_STRING_PARAMETER(pl, peopleIncludeFile, "--peopleIncludeFile", "from given file, set IDs of people that will be included in study")
ADD_STRING_PARAMETER(pl, peopleExcludeID, "--peopleExcludeID", "give IDs of people that will be included in study")
ADD_STRING_PARAMETER(pl, peopleExcludeFile, "--peopleExcludeFile", "from given file, set IDs of people that will be included in study")
ADD_PARAMETER_GROUP(pl, "Site Filter")
ADD_STRING_PARAMETER(pl, rangeList, "--rangeList", "Specify some ranges to use, please use chr:begin-end format.")
ADD_STRING_PARAMETER(pl, rangeFile, "--rangeFile", "Specify the file containing ranges, please use chr:begin-end format.")
END_PARAMETER_LIST(pl)
;
pl.Read(argc, argv);
pl.Status();
if (FLAG_REMAIN_ARG.size() > 0){
fprintf(stderr, "Unparsed arguments: ");
for (unsigned int i = 0; i < FLAG_REMAIN_ARG.size(); i++){
fprintf(stderr, " %s", FLAG_REMAIN_ARG[i].c_str());
}
fprintf(stderr, "\n");
abort();
}
REQUIRE_STRING_PARAMETER(FLAG_inVcf, "Please provide input file using: --inVcf");
const char* fn = FLAG_inVcf.c_str();
VCFInputFile vin(fn);
// set range filters here
// e.g.
// vin.setRangeList("1:69500-69600");
vin.setRangeList(FLAG_rangeList.c_str());
vin.setRangeFile(FLAG_rangeFile.c_str());
// set people filters here
if (FLAG_peopleIncludeID.size() || FLAG_peopleIncludeFile.size()) {
vin.excludeAllPeople();
vin.includePeople(FLAG_peopleIncludeID.c_str());
vin.includePeopleFromFile(FLAG_peopleIncludeFile.c_str());
}
vin.excludePeople(FLAG_peopleExcludeID.c_str());
vin.excludePeopleFromFile(FLAG_peopleExcludeFile.c_str());
// let's write it out.
FILE* fMap; // CHROMOSOME MARKER POSITION
FILE* fDat; // A some_disease\n
// T some_trait
// M some_marker
// M another_marker
FILE* fPed; // first 5 column: FID, IID, PID, MID, SEX; then follow Dat file
FILE* fPid; // Person ID file, (extra for Merlin), including all people ID as they are in PED file.
fMap = fopen( (FLAG_outMerlin + ".map").c_str(), "wt");
fDat = fopen( (FLAG_outMerlin + ".dat").c_str(), "wt");
fPed = fopen( (FLAG_outMerlin + ".ped").c_str(), "wt");
fPid = fopen( (FLAG_outMerlin + ".pid").c_str(), "wt");
assert(fMap && fDat && fPed && fPid);
std::string marker; // marker x people
std::vector<std::string> allMarker;
Matrix geno;
fputs("CHROMOSOME\tMARKER\tPOSITION\n", fMap);
while (vin.readRecord()){
VCFRecord& r = vin.getVCFRecord();
VCFPeople& people = r.getPeople();
VCFIndividual* indv;
// write map file
marker = r.getID();
if ( marker == "." ) {
fprintf(fMap, "%s\t%s:%d\t%d\n", r.getChrom(), r.getChrom(), r.getPos(), r.getPos());
fprintf(fDat, "M\t%s:%d\n", r.getChrom(), r.getPos());
} else {
fprintf(fMap, "%s\t%s\t%d\n", r.getChrom(), marker.c_str(), r.getPos());
fprintf(fDat, "M\t%s\n", marker.c_str());
}
allMarker.push_back(marker);
geno.Dimension(allMarker.size(), people.size());
// e.g.: get TAG from INFO field
// fprintf(stderr, "%s\n", r.getInfoTag("ANNO"));
int m = allMarker.size() - 1;
// e.g.: Loop each (selected) people in the same order as in the VCF
for (int i = 0; i < people.size(); i++) {
indv = people[i];
// get GT index. if you are sure the index will not change, call this function only once!
int GTidx = r.getFormatIndex("GT");
if (GTidx >= 0) {
geno[m][i] = (*indv)[GTidx].getGenotype();
}else {
fprintf(stderr, "Cannot find GT field!\n");
abort();
}
}
};
VCFHeader* h = vin.getVCFHeader();
std::vector< std::string> peopleId;
h->getPeopleName(&peopleId);
// dump PED and PID file
for (int p = 0; p < peopleId.size(); p++){
fprintf(fPed, "%s\t%s\t0\t0\t0", peopleId[p].c_str(), peopleId[p].c_str());
for (int m = 0; m < allMarker.size(); m++){
int g = (int)geno[m][p];
switch (g){
case 0:
fputs("\t0/0", fPed);
break;
case 1:
fputs("\t0/1", fPed);
break;
case 2:
fputs("\t1/1", fPed);
break;
default:
fputs("x/x", fPed);
break;
}
}
fputs("\n", fPed);
fprintf(fPid, "%s\n", peopleId[p].c_str());
}
return 0;
};
int GenotypeExtractor::extractMultipleGenotype(Matrix* g) {
static Matrix m; // make it static to reduce memory allocation
int row = 0;
std::vector<std::string> colNames;
std::string name;
this->hemiRegion.clear();
GenotypeCounter genoCounter;
while (this->vin->readRecord()) {
VCFRecord& r = this->vin->getVCFRecord();
VCFPeople& people = r.getPeople();
VCFIndividual* indv;
m.Dimension(row + 1, people.size());
genoCounter.reset();
int genoIdx;
const bool useDosage = (!this->dosageTag.empty());
if (useDosage) {
genoIdx = r.getFormatIndex(dosageTag.c_str());
} else {
genoIdx = r.getFormatIndex("GT");
}
int GDidx = r.getFormatIndex("GD");
int GQidx = r.getFormatIndex("GQ");
assert(this->parRegion);
bool hemiRegion = this->parRegion->isHemiRegion(r.getChrom(), r.getPos());
// e.g.: Loop each (selected) people in the same order as in the VCF
const int numPeople = (int)people.size();
for (int i = 0; i < numPeople; i++) {
indv = people[i];
// get GT index. if you are sure the index will not change, call this
// function only once!
if (genoIdx >= 0) {
// printf("%s ", indv->justGet(0).toStr()); // [0] meaning the first
// field of each individual
if (useDosage) {
if (!hemiRegion) {
m[row][i] = indv->justGet(genoIdx).toDouble();
} else {
// for male hemi region, imputated dosage is usually between 0 and 1
// need to multiply by 2.0
if ((*sex)[i] == PLINK_MALE) {
m[row][i] = indv->justGet(genoIdx).toDouble() * 2.0;
}
}
} else {
if (!hemiRegion) {
m[row][i] = indv->justGet(genoIdx).getGenotype();
} else {
if ((*sex)[i] == PLINK_MALE) {
m[row][i] = indv->justGet(genoIdx).getMaleNonParGenotype02();
} else if ((*sex)[i] == PLINK_FEMALE) {
m[row][i] = indv->justGet(genoIdx).getGenotype();
} else {
m[row][i] = MISSING_GENOTYPE;
}
}
}
if (!checkGD(indv, GDidx) || !checkGQ(indv, GQidx)) {
m[row][i] = MISSING_GENOTYPE;
}
genoCounter.add(m[row][i]);
} else {
logger->error("Cannot find %s field!",
this->dosageTag.empty() ? "GT" : dosageTag.c_str());
return -1;
}
}
// check frequency cutoffs
// int numNonMissingPeople = 0;
// double maf = 0.;
// for (int i = 0; i < numPeople; ++i) {
// if (m[row][i] < 0) continue;
// maf += m[row][i];
// ++numNonMissingPeople;
// }
// if (numNonMissingPeople) {
// maf = maf / (2. * numNonMissingPeople);
// } else {
// maf = 0.0;
// }
// if (maf > .5) {
// maf = 1.0 - maf;
// }
const double maf = genoCounter.getMAF();
if (this->freqMin > 0. && this->freqMin > maf) continue;
if (this->freqMax > 0. && this->freqMax < maf) continue;
// store genotype results
name = r.getChrom();
name += ":";
name += r.getPosStr();
colNames.push_back(name);
++row;
assert(this->parRegion);
if (this->parRegion &&
this->parRegion->isHemiRegion(r.getChrom(), r.getPos())) {
this->hemiRegion.push_back(true);
} else {
this->hemiRegion.push_back(false);
}
this->counter.push_back(genoCounter);
} // end while (this->vin->readRecord())
// delete rows (ugly code here, as we may allocate extra row in previous
// loop)
m.Dimension(row, m.cols);
// now transpose (marker by people -> people by marker)
g->Transpose(m);
for (int i = 0; i < row; ++i) {
g->SetColumnLabel(i, colNames[i].c_str());
}
return SUCCEED;
} // end GenotypeExtractor