/*
we want to cut the first row and the fourth row.
chromosomes and positions.
But we want to check if the cromosomes are 1-22
These are the ones to include, so well return a keeplist,
and the correct pars->chromo and pars->positions.
This is rather slow because we use make a new vector on each line.
This can be optimized in future versions. */
bArray *doBimFile(functionPars* pars,const char *filename,const std::string delim){
///@param filename A filename to read.@param delim A string of delimiters.
std::vector<int> chromos;//for all lines
std::vector<double> positions;//for all lines
const int SIZE=500000;//this should only be 6 elements but lets make it big..
char buffer[SIZE];
std::ifstream pFile (filename,std::ios::in);
if(!pFile){
fileError(filename);
exit(0);
}
std::string tmp_string;
int itemsInRow;
int numRows =0;
while(!pFile.eof()){
pFile.getline(buffer,SIZE);
tmp_string = std::string(buffer);
std::vector<std::string> tokens;
itemsInRow = get_lexemes(tmp_string,tokens,delim);
if (itemsInRow!=6 && itemsInRow!=0){
printf("plink bim file:%s doesn't have 6 columns in row:%d\n",filename,numRows);
exit(0);
}else if(itemsInRow==0)
break;
chromos.push_back(atoi((tokens[0]).c_str()));
positions.push_back(atof((tokens[3]).c_str()));
numRows++;
}
// copy(chromos.begin(), chromos.end(), std::ostream_iterator<int>(std::cout, ", "));
// copy(positions.begin(), positions.end(), std::ostream_iterator<float>(std::cout, ", "));
bArray *ret = allocBoolArray(chromos.size());
int numTrue = 0;
for(unsigned int i=0;i<chromos.size();i++)
if(chromos[i]!=0 && chromos[i]<23 ){
ret->array[i] = 1;
numTrue++;
}else
ret->array[i] = 0;
ret->numTrue = numTrue;
dArray *pos = allocDoubleArray(ret->numTrue);
iArray *chr = allocIntArray(ret->numTrue);
int atPos=0;
for(int i=0;i<ret->x;i++){
if(ret->array[i]){
pos->array[atPos] = positions[i]/PLINK_POS_SCALING;
chr->array[atPos] = chromos[i];
atPos++;
}
}
pars->chr = chr;
pars->position= pos;
return ret;
}
//! constructor.
Matrices()
{
//work=new double[n];
work=allocDoubleArray(n);
lwork=n;
}
dArray *typecast_stringarray_to_dArray(const std::vector<std::string> &tokens){
dArray *ary = allocDoubleArray(tokens.size());
for (unsigned int i=0;i<tokens.size();i++)
ary->array[i] = to_double( tokens.at(i));
return ary;
}
