void SiteCounts::add_sequence( std::string const& sites, CountsIntType weight )
{
if( num_seqs_ >= std::numeric_limits< CountsIntType >::max() - weight ) {
throw std::runtime_error(
"Cannot add Sequence to SiteCounts as it might lead to an overflow in the counts."
);
}
if( sites.size() != counts_.rows() ) {
throw std::runtime_error(
"Cannot add Sequence to SiteCounts if it has different number of sites: Expected "
+ std::to_string( counts_.rows() ) + " sites, but sequence has "
+ std::to_string( sites.size() ) + " sites."
);
}
for( size_t site_idx = 0; site_idx < sites.size(); ++site_idx ) {
// Get the index of the char. If not found, this char is not to be counted, so continue.
auto char_idx = lookup_[ static_cast< size_t >( sites[ site_idx ] ) ];
if( char_idx == characters_.size() ) {
continue;
}
// Increase the count at that index.
counts_( site_idx, char_idx ) += weight;
}
// We finished a sequence. Add to the counter.
num_seqs_ += weight;
}
void DiscreteFrustum::interpolatedUndistort(double* z) const
{
int idx = index(*z);
double start = bin_depth_ * idx;
int idx1;
if(*z - start < bin_depth_ / 2)
idx1 = idx;
else
idx1 = idx + 1;
int idx0 = idx1 - 1;
if(idx0 < 0 || idx1 >= num_bins_ || counts_(idx0) < 50 || counts_(idx1) < 50) {
undistort(z);
return;
}
double z0 = (idx0 + 1) * bin_depth_ - bin_depth_ * 0.5;
double coeff1 = (*z - z0) / bin_depth_;
double coeff0 = 1.0 - coeff1;
double mult = coeff0 * multipliers_.coeffRef(idx0) + coeff1 * multipliers_.coeffRef(idx1);
*z *= mult;
}
void DiscreteFrustum::addExample(double ground_truth, double measurement)
{
double mult = ground_truth / measurement;
if(mult > MAX_MULT || mult < MIN_MULT)
return;
int idx = min(num_bins_ - 1, (int)floor(measurement / bin_depth_));
UASSERT(idx >= 0);
total_numerators_(idx) += ground_truth * ground_truth;
total_denominators_(idx) += ground_truth * measurement;
++counts_(idx);
multipliers_(idx) = total_numerators_(idx) / total_denominators_(idx);
}
void DiscreteFrustum::addExample(double ground_truth, double measurement)
{
boost::unique_lock<boost::shared_mutex> ul(shared_mutex_);
double mult = ground_truth / measurement;
if(mult > MAX_MULT || mult < MIN_MULT)
return;
int idx = min(num_bins_ - 1, (int)floor(measurement / bin_depth_));
assert(idx >= 0);
total_numerators_(idx) += ground_truth * ground_truth;
total_denominators_(idx) += ground_truth * measurement;
++counts_(idx);
multipliers_(idx) = total_numerators_(idx) / total_denominators_(idx);
}
SiteCounts::CountsIntType SiteCounts::count_of( char character, size_t site_index ) const
{
if( site_index >= length() ) {
throw std::runtime_error(
"Invalid site index for retrieving count: " + std::to_string( site_index ) + "."
);
}
auto char_idx = lookup_[ character ];
if( char_idx == characters_.size() ) {
throw std::runtime_error(
"Invalid character for retrieving count: '" + std::string( 1, character ) + "'."
);
}
return counts_( site_index, char_idx );
}
SiteCounts::CountsIntType SiteCounts::count_at(
size_t character_index,
size_t site_index
) const {
if( site_index >= counts_.rows() ) {
throw std::runtime_error(
"Invalid site index for retrieving count: " + std::to_string( site_index ) + "."
);
}
if( character_index > counts_.cols() ) {
throw std::runtime_error(
"Invalid character index for retrieving count: " + std::to_string( character_index ) + "."
);
}
return counts_( site_index, character_index );
}
//! Processes a single weighted data point.
void process(const uint_vector& values, T weight) {
assert(values.size() == 2);
counts_(values[0], values[1]) += weight;
}
