// See documentation in header file.
void searchMEIBreakpoints(MEI_data& currentState, std::vector<bam_info>& bam_sources, const Chromosome* chromosome,
UserDefinedSettings* userSettings) {
LOG_DEBUG(*logStream << time_log() << "Start searching for breakpoints..." << std::endl);
std::vector<std::vector<simple_read*> > clusters;
cluster_reads(currentState.discordant_reads, currentState.current_insert_size, clusters, userSettings);
// Find breakpoints per cluster.
int bp_count = 0;
for (size_t i = 0; i < clusters.size(); i++) {
// print cluster debug info
std::vector<simple_read*> cluster = clusters.at(i);
if (cluster.size() < ((size_t) userSettings->MIN_DD_CLUSTER_SIZE)) {
// Fluke cluster, skip it. (If there are very few reads in the cluster,
// we likely won't find enough split reads supporting an insertion)
continue;
}
// Find breakpoint for this cluster
char cluster_strand = cluster.at(0)->strand;
int cluster_tid = cluster.at(0)->tid;
std::vector<MEI_breakpoint> MEI_bps;
std::vector<MEI_breakpoint>::iterator MEI_iter;
get_breakpoints(cluster, bam_sources, currentState.current_insert_size, cluster_tid, cluster_strand, chromosome,
currentState.sample_names, MEI_bps, userSettings);
if (MEI_bps.size() > 1) {
// More than one breakpoints found for current cluster. Select only the one with the
// most split reads supporting it.
size_t best_support = 0;
MEI_breakpoint best_bp;
for (MEI_iter = MEI_bps.begin(); MEI_iter != MEI_bps.end(); ++MEI_iter) {
if (MEI_iter->associated_split_reads.size() > best_support) {
best_bp = *MEI_iter;
best_support = MEI_iter->associated_split_reads.size();
}
}
MEI_bps.clear();
MEI_bps.push_back(best_bp);
} else if (MEI_bps.size() == 0) {
// No breakpoint found with split read support. Estimate breakpoint from cluster reads.
get_breakpoint_estimation(cluster, currentState.current_insert_size, cluster_tid, cluster_strand, MEI_bps);
}
// Check breakpoint validity.
for (MEI_iter = MEI_bps.begin(); MEI_iter != MEI_bps.end(); ++MEI_iter) {
currentState.MEI_breakpoints.push_back(*MEI_iter);
bp_count += 1;
LOG_INFO(*logStream << "Found potential DD breakpoint: " << (*MEI_iter).breakpoint_tid << ", " <<
(*MEI_iter).breakpoint_pos << ", " << (*MEI_iter).cluster_strand << ", " <<
(*MEI_iter).associated_reads.size() << ", " <<
(*MEI_iter).associated_split_reads.size() << std::endl);
}
}
LOG_DEBUG(*logStream << time_log() << "Found " << bp_count << " breakpoints for " << clusters.size() <<
" clusters." << std::endl);
}
void process_chunk(struct alignedread** readlist, int s, int e, FRAGMENT* flist, VARIANT* varlist, REFLIST* reflist) {
find_matepair(readlist, s, e);
fprintf(stderr, "e %d s %d \n", e, s);
//int cl = init_clusters(readlist,s,e); // cluster using a maximum intra-cluster distance value
//estimate_readdistance_distribution(readlist,s,e,cluster); // estimate distances between start positions of adjacent reads within same cluster
// cluster size distribution from data | probability of a read being a singleton read
cluster_reads(readlist, s, e, flist, varlist, reflist);
fprintf(stdout, "\n\n");
//print_clusters(readlist,s,e,flist,varlist); // print clusters
}
int main(int argc, char **argv)
{
long int prev_cluster = 0, prev_total_cluster=-1, exit_count = 0;
printf("Yet to cluster = %ld\n", yet_to_cluster);
while(yet_to_cluster > 0)
{
if(prev_cluster-yet_to_cluster==total_clusters-prev_total_cluster)
{
exit_count++;
if(exit_count >= 3*log(NO_READS))
{
printf("No more clustering ... exiting ...\n\n");
store_clusters();
return 0;
}
}
else
exit_count = 0;
hash_comparisions = 0;
false_collisions = 0;
prev_cluster = yet_to_cluster;
prev_total_cluster = total_clusters;
// copy reads from reads_rem.txt to reads_err.txt
char sequence[READ_LENGTH] = {0};
long int read_number;
FILE *f1, *f2;
f1 = fopen("reads_rem.txt","r");
f2 = fopen("reads_err.txt","w");
while(fscanf(f1,"%ld",&read_number) != EOF)
{
fscanf(f1, "%s", sequence);
fprintf(f2, "%ld\n", read_number);
fprintf(f2, "%s\n", sequence);
}
fclose(f1);
fclose(f2);
// pickup random reads and setup the cluster centers
pick_cluster_centers();
if(total_clusters > prev_total_cluster)
{
// cluster_reads
cluster_reads();
base_comparisions = hash_comparisions + (false_collisions * 30) + ((prev_cluster-yet_to_cluster) * 30);
// print clusters
print_clusters();
}
}
store_clusters();
return 0;
}
