void ComputeErrorMetrics() {
metrics_.constructed_metrics.num_errors = 0;
for(auto it = constructed_clusters_.clusters_begin(); it != constructed_clusters_.clusters_end(); it++) {
if(!IsConstructedClusterNotErroneous(it->first))
metrics_.constructed_metrics.num_errors++;
else {
size_t orig_supercluster = GetOriginalSuperCluster(it->first);
if(constructed_clusters_.ClusterSize(it->first) > original_clusters_.ClusterSize(orig_supercluster))
metrics_.constructed_metrics.num_errors++;
}
}
}
void ComputeMaxClusterMetrics() {
size_t max_orig_cluster = 0;
for(auto it = original_clusters_.clusters_begin(); it != original_clusters_.clusters_end(); it++)
if(metrics_.original_metrics.max_cluster < original_clusters_.ClusterSize(it->first)) {
metrics_.original_metrics.max_cluster = original_clusters_.ClusterSize(it->first);
max_orig_cluster = it->first;
}
for(auto it = constructed_clusters_.clusters_begin(); it != constructed_clusters_.clusters_end(); it++) {
metrics_.constructed_metrics.max_cluster = max<size_t>(metrics_.constructed_metrics.max_cluster,
constructed_clusters_.ClusterSize(it->first));
}
metrics_.max_cluster_fillin = ComputeNotMergedFillin(max_orig_cluster);
}
double ComputeNotMergedFillin(size_t orig_cluster_ind) {
vector<string> read_names = GetReadNamesFromCluster(orig_cluster_ind, original_clusters_, original_reads_);
map<size_t, vector<size_t> > constr_subclusters;
for(auto it = read_names.begin(); it != read_names.end(); it++) {
size_t cluster_ind = constructed_clusters_.GetCluster(*it);
constr_subclusters[constructed_clusters_.ClusterSize(cluster_ind)].push_back(cluster_ind);
}
for(auto it = constr_subclusters.rbegin(); it != constr_subclusters.rend(); it++) {
auto clusters = it->second;
for(auto cluster_ind = clusters.begin(); cluster_ind != clusters.end(); cluster_ind++)
if(IsConstructedClusterNotErroneous(*cluster_ind)) {
return double(constructed_clusters_.ClusterSize(*cluster_ind)) / original_clusters_.ClusterSize(orig_cluster_ind);
}
}
return 0;
}
void ComputeLostClustersMetrics() {
size_t lost_cluster_size = 0;
for(auto it = original_clusters_.clusters_begin(); it != original_clusters_.clusters_end();
it++) {
if(IsOriginalClusterLost(it->first)) {
metrics_.lost_clusters_number++;
lost_cluster_size += original_clusters_.ClusterSize(it->first);
}
}
metrics_.lost_clusters_size = double(lost_cluster_size) / original_reads_.size();
}
void ComputeNotMergedMetrics() {
double sum_fill = 0;
for(auto it = original_clusters_.clusters_begin(); it != original_clusters_.clusters_end(); it++) {
size_t orig_cluster_ind = it->first;
if(original_clusters_.IsClusterSingleton(orig_cluster_ind))
continue;
// for each non trivial cluster we compute it "not merged"
if(!IsClusterNotMerged(orig_cluster_ind))
continue;
metrics_.original_metrics.num_not_merged++;
if(IsClusterBigAndSingletons(orig_cluster_ind))
metrics_.original_metrics.num_nm_big_singletons++;
size_t orig_cluster_size = original_clusters_.ClusterSize(orig_cluster_ind);
double orig_cluster_fill = ComputeNotMergedFillin(orig_cluster_ind);
sum_fill += orig_cluster_fill;
}
metrics_.avg_fillin = sum_fill / metrics_.original_metrics.num_not_merged;
}
