VectorXf ClstrSeqWeightEstimator::estimate(const vector<string>& msa) const {
const size_t rows = msa.size();
VectorXf memnum = VectorXf::Ones(rows);
for (size_t i = 0; i < rows; ++i) {
for (size_t j = i + 1; j < rows; ++j) {
if (calc_identity(msa[i], msa[j]) > maxidt) {
memnum(i) += 1;
memnum(j) += 1;
}
}
}
return VectorXf::Ones(rows).cwiseQuotient(memnum);
}
double Cluster::add_to_cluster(Node* new_read, Model_factory *mf, int read_index) {
finalise();
stringstream ss;
ss << "#" << read_index << "#";
Node* new_root = new Node();
new_root->set_name(ss.str());
new_root->add_left_child(root);
new_root->add_right_child(new_read);
new_root->set_nhx_tid(root->get_nhx_tid());
new_read->set_nhx_tid(root->get_nhx_tid());
root->set_distance_to_parent(0.001);
new_root->align_sequences_this_node(mf, true, false);
double read_identity = calc_identity(new_root, new_read);
bool cluster_success = read_identity > id_threshold;
new_root->has_left_child(false);
new_root->has_right_child(false);
if(cluster_success) {
// clustering was successful, so add the read to list of members
// and replace the old root with the new one
if(fast_cluster) {
delete root;
}
else {
prev_root = root;
}
root = new_root;
cluster_members.push_back(read_index);
}
else {
// not added to cluster, so ignore old root and read and
// delete the failed new root
delete new_root;
}
return read_identity;
}
