int compare_regions(void *region_1, void *region_2, char **chromosome_ordering, int num_chromosomes) {
if (region_1 == NULL || region_2 == NULL) {
return INT_MIN;
}
region_t *reg_1 = (region_t *) region_1;
region_t *reg_2 = (region_t *) region_2;
// TODO This could be avoided while inserting, because regions are classified by chromosome
int result = compare_chromosomes(reg_1->chromosome, reg_2->chromosome, chromosome_ordering, num_chromosomes);
if (result != 0) {
return result;
} else {
// return compare_position_ranges(reg_1, reg_2);
return compare_positions(reg_1->start_position, reg_2->start_position);
}
}
int merge_interval(kh_pos_t* positions_read, char *max_chromosome_merged, unsigned long max_position_merged,
char **chromosome_order, int num_chromosomes, vcf_file_t **files,
shared_options_data_t *shared_options_data, merge_options_data_t *options_data, list_t *output_list) {
int num_entries = 0;
#pragma omp parallel for num_threads(shared_options_data->num_threads) reduction(+:num_entries)
for (int k = kh_begin(positions_read); k < kh_end(positions_read); k++) {
if (kh_exist(positions_read, k)) {
array_list_t *records_in_position = kh_value(positions_read, k);
assert(records_in_position);
vcf_record_t *record = ((vcf_record_file_link*) array_list_get(0, records_in_position))->record;
vcf_record_file_link **links = NULL;
int num_links = 0;
// Remove positions prior to the last chromosome:position to merge
int cmp_chrom = compare_chromosomes(record->chromosome, max_chromosome_merged, chromosome_order, num_chromosomes);
if (cmp_chrom < 0 || (cmp_chrom == 0 && compare_positions(record->position, max_position_merged) <= 0)) {
links = records_in_position->items;
num_links = records_in_position->size;
}
// Launch merge
if (num_links > 0) {
// printf("links[0] = %s:%ld in file %s\n", links[0]->record->chromosome, links[0]->record->position, links[0]->file->filename);
int err_code = 0;
vcf_record_t *merged = merge_position(links, num_links, files, options_data->num_files, options_data, &err_code);
if (!err_code) {
list_item_t *item = list_item_new(k, MERGED_RECORD, merged);
list_insert_item(item, output_list);
num_entries += 1;
}
// Free empty nodes (lists of records in the same position)
array_list_free(records_in_position, vcf_record_file_link_free);
kh_del(pos, positions_read, k);
}
} // End kh_exist
}
return num_entries;
}
void calculate_merge_interval(vcf_record_t* current_record, char** max_chromosome_merged, long unsigned int* max_position_merged,
char **chromosome_order, int num_chromosomes) {
if (*max_chromosome_merged == NULL) {
// Max merged chrom:position not set, assign without any other consideration
*max_chromosome_merged = strndup(current_record->chromosome, current_record->chromosome_len);
*max_position_merged = current_record->position;
} else {
char *current_chromosome = strndup(current_record->chromosome, current_record->chromosome_len);
long unsigned int current_position = current_record->position;
// printf("current = %s:%ld\tmax = %s:%ld\n", current_chromosome, current_position, *max_chromosome_merged, *max_position_merged);
int chrom_comparison = compare_chromosomes(current_chromosome, *max_chromosome_merged, chromosome_order, num_chromosomes);
int position_comparison = compare_positions(current_position, *max_position_merged);
// Max merged chrom:position is posterior to the last one in this batch
if (chrom_comparison < 0 || (chrom_comparison == 0 && position_comparison < 0)) {
*max_chromosome_merged = current_chromosome;
*max_position_merged = current_position;
} else {
assert(current_chromosome);
free(current_chromosome);
}
}
}
