Data const &Phenotype::solve(Data const &input) {
Data biased(input);
biased.push_back(1.0f);
assert(inputNeurons.size() == biased.size());
// clear
for (Neuron *const &n : hiddenNeurons)
n->state = 0.0f;
for (Neuron *const &n : outputNeurons)
n->state = 0.0f;
for (size_t i = 0; i < inputNeurons.size(); ++i) {
Neuron *const &n = inputNeurons[i];
for (size_t j = 0; j < n->synapses.size(); ++j)
n->synapses[j].to->state += biased[i] * n->synapses[j].weight;
}
for (size_t i = 0; i < hiddenNeurons.size(); ++i) {
Neuron *const &n = hiddenNeurons[i];
n->activate();
for (size_t j = 0; j < n->synapses.size(); ++j)
n->synapses[j].to->state += n->state * n->synapses[j].weight;
}
for (size_t i = 0; i < outputNeurons.size(); ++i) {
outputNeurons[i]->activate();
output[i] = outputNeurons[i]->state;
}
return output;
}
void runAnalysis(int *dic, Args args, std::istream *in, profile &pro)
{
int s, fieldc=0, linec=0;
char consensus;
std::string name, column, number, line;
std::vector <std::string> field;
getline(*in, line);
field=split(line, args.cdel);
count_t samples=(field.size()-3)/3;
count_t ** count=new count_t* [samples];
for(count_t i = 0 ; i < samples; ++i)
count[i] = new count_t[9];
site_t site(samples);
pro.setsamples(samples);
pro.setcolumns(args.c);
pro.set_delim_column(args.cdel);
pro.set_delim_quartet(args.qdel);
if (not (args.noheader) ) pro.writeheader();
/* If multiple files are given as arguments, the default behavoir should be to merge the files,
* an option should be given to append the files, but in the event that duplicat ids exist in
* files being appended the program should throw an error or warning.
*/
while( std::getline(*in, line)!=NULL){
field=split(line, args.cdel);
++linec;
fieldc = field.size();
if(fieldc < 5){
printf("WARNING[mapgd proview]: Skipping line %d with only %d fields.\n", linec, fieldc);
continue;
}
for(count_t i = 0 ; i < samples; ++i)
memset(count[i], 0, sizeof(count_t)*9 );
site.id0 = field[0];
consensus = field[2][0];
if (site.extra_ids.size()==0) site.extra_ids.push_back(std::string(1, consensus));
else site.extra_ids[0] = consensus;
for (count_t i = 0; i < samples; ++i){
column = field[4+i*3];
scanCol(column, dic, count[i], consensus);
}
bool run=false;
for (count_t x=0 ; x<samples ; ++x){
s=0;
for(count_t y=0; y<8; ++y) s += count[x][y];
if (s>=args.min) run=true;
}
if (!args.notrim){
for (count_t x=0 ; x<samples ; ++x)
for (count_t y=0 ; y<4 ; y++)
if (biased(count[x][y], count[x][y+4], args.pvalue) ) run=false;
}
if(run){
site.id1 = field[1];
for(count_t x=0 ; x<samples ; x++){
for (count_t y=0 ; y<4 ; y++){
site.sample[x].base[y]=count[x][y]+count[x][y+4];
}
}
pro.write(site);
}
}
}
