/**
* Returns a lexicographically sorted list of merged barcodes, each paired with
* the 0-based index of corresponding barcode in the source vector.
*/
barcode_vec sort_barcodes(const fastq_pair_vec& barcodes)
{
AR_DEBUG_ASSERT(!barcodes.empty());
barcode_vec sorted_barcodes;
const size_t max_key_1_len = barcodes.front().first.length();
const size_t max_key_2_len = barcodes.front().second.length();
for (auto it = barcodes.begin(); it != barcodes.end(); ++it) {
if (it->first.length() != max_key_1_len) {
throw barcode_error("mate 1 barcodes do not have the same length");
} else if (it->second.length() != max_key_2_len) {
throw barcode_error("mate 2 barcodes do not have the same length");
}
std::string barcode;
barcode.reserve(max_key_1_len + max_key_2_len);
barcode.append(it->first.sequence());
barcode.append(it->second.sequence());
sorted_barcodes.push_back(barcode_pair(barcode, it - barcodes.begin()));
}
std::sort(sorted_barcodes.begin(), sorted_barcodes.end());
return sorted_barcodes;
}
void adjust_sigma(Simulator *sim, int steps, double step_size,
double factor)
{
double original_sigma = sim->sigma;
double best_sigma = sim->sigma;
double best_error = barcode_error(sim);
for (int step = -steps; step <= steps; step++) {
if (step == 0)
continue;
sim->sigma = original_sigma + step * step_size;
if (sim->sigma < 1.0)
continue;
compute_kernel(sim);
convolve_1d(sim->kernel_width, sim->kernel,
sim->width, sim->guess, sim->blur, 0, sim->width);
double error = barcode_error(sim);
// printf("sigma=%.2f error=%.2f\n", sim->sigma, error);
if (error < best_error) {
best_error = error;
best_sigma = sim->sigma;
}
}
sim->sigma = best_sigma * factor + original_sigma * (1.0 - factor);
compute_kernel(sim);
convolve_1d(sim->kernel_width, sim->kernel,
sim->width, sim->guess, sim->blur, 0, sim->width);
}
/** Adds a nucleotide sequence with a given ID to a quad-tree. */
void add_sequence_to_tree(demux_node_vec& tree, const std::string& sequence,
const size_t barcode_id)
{
size_t node_idx = 0;
bool added_last_node = false;
for (auto nuc : sequence) {
auto& node = tree.at(node_idx);
// Indicate when PE barcodes can be unambigiously identified from SE
// reads
node.value = (node.value == barcode_table::no_match)
? barcode_id
: barcode_table::ambigious;
const auto nuc_idx = ACGT_TO_IDX(nuc);
auto child = node.children[nuc_idx];
added_last_node = (child == barcode_table::no_match);
if (added_last_node) {
// New nodes are added to the end of the list; as barcodes are
// added in lexicographic order, this helps ensure that a set of
// similar barcodes will be placed in mostly contiguous runs
// of the vector representation.
child = node.children[nuc_idx] = tree.size();
tree.push_back(demultiplexer_node());
}
node_idx = child;
}
if (!added_last_node) {
throw barcode_error(std::string("duplicate barcode (pair): ") +
sequence);
}
tree.at(node_idx).value = barcode_id;
}
