/*Takes in as input the vector of iterations from coarse BLAST, an index into
*the vector (representing the index of which query we want to get the results
*from), and the database we are using for search and returns a vector of
*every original sequence section re-created from the calls to
*cb_coarse_expand for the hits in the iteration we are expanding.
*/
struct DSVector *expand_blast_hits(struct DSVector *iterations, int index,
struct cb_database_r *db){
struct DSVector *expanded_hits = ds_vector_create(),
*hits = get_blast_hits(
(xmlNode *)ds_vector_get(iterations, index));
int i = 0, j = 0, k = 0;
for (i = 0; i < hits->size; i++) {
struct hit *current_hit = (struct hit *)ds_vector_get(hits, i);
struct DSVector *hsps = current_hit->hsps;
for (j = 0; j < hsps->size; j++) {
struct DSVector *oseqs;
struct hsp *h = (struct hsp *)ds_vector_get(hsps, j);
int32_t coarse_start = h->hit_from-1, coarse_end = h->hit_to-1,
coarse_seq_id = current_hit->accession;
oseqs = cb_coarse_expand(db->coarse_db, db->com_db, coarse_seq_id,
coarse_start, coarse_end, 50);
for (k = 0; k < oseqs->size; k++)
ds_vector_append(expanded_hits, ds_vector_get(oseqs, k));
ds_vector_free_no_data(oseqs);
}
ds_vector_free(current_hit->hsps);
free(current_hit);
}
ds_vector_free_no_data(hits);
return expanded_hits;
}
static void php_ds_vector_free_object(zend_object *object)
{
php_ds_vector_t *obj = (php_ds_vector_t*) object;
zend_object_std_dtor(&obj->std);
ds_vector_free(obj->vector);
}
int main(int argc, char **argv){
FILE *query_file = NULL, *test_hits_file = NULL;
struct cb_database_r *db = NULL;
struct opt_config *conf;
struct opt_args *args;
struct DSVector *iterations = NULL, *expanded_hits = NULL, *queries = NULL,
*oseqs = ds_vector_create();
struct fasta_seq *query = NULL;
xmlDoc *doc = NULL;
xmlNode *root = NULL;
uint64_t dbsize = 0;
int i = 0, j = 0;
conf = load_search_args();
args = opt_config_parse(conf, argc, argv);
if (args->nargs < 2) {
fprintf(stderr,
"Usage: %s [flags] database-dir fasta-file "
"[ --blast_args BLASTN_ARGUMENTS ]\n", argv[0]);
opt_config_print_usage(conf);
exit(1);
}
system("rm CaBLAST_results.xml");
if (!search_flags.hide_progress)
fprintf(stderr, "Loading database data\n\n");
db = cb_database_r_init(args->args[0],
(search_flags.load_coarse_db ||
search_flags.load_coarse_residues),
(search_flags.load_coarse_db ||
search_flags.load_coarse_links),
search_flags.load_compressed_db,
search_flags.link_block_size);
dbsize = read_int_from_file(8, db->coarse_db->db->file_params);
if (!search_flags.hide_progress)
fprintf(stderr, "Running coarse BLAST\n\n");
blast_coarse(args, dbsize);
if (NULL == (query_file = fopen(args->args[1], "r"))) {
fprintf(stderr, "fopen: 'fopen %s' failed: %s\n",
args->args[1], strerror(errno));
exit(1);
}
queries = ds_vector_create();
query = fasta_read_next(query_file, "");
while (query) {
ds_vector_append(queries, (void *)query);
query = fasta_read_next(query_file, "");
}
fclose(query_file);
if (!search_flags.hide_progress)
fprintf(stderr, "Processing coarse BLAST hits for fine BLAST\n\n");
if (search_flags.show_hit_info)
if (NULL == (test_hits_file = fopen("CaBLAST_hits.txt", "w"))) {
fprintf(stderr, "fopen: 'fopen %s' failed: %s\n",
"CaBLAST_hits.txt", strerror(errno));
exit(1);
}
//Parse the XML file generated from coarse BLAST and get its iterations.
doc = xmlReadFile("CaBLAST_temp_blast_results.xml", NULL, 0);
if (doc == NULL) {
fprintf(stderr, "Could not parse CaBLAST_temp_blast_results.xml\n");
return 0;
}
root = xmlDocGetRootElement(doc);
iterations = get_blast_iterations(root);
for (i = 0; i < iterations->size; i++) {
if (!search_flags.hide_progress) {
int32_t digits_full = floor(log10((double)iterations->size)),
digits_i = floor(log10((double)i)), spaces;
char *bar = progress_bar(i, iterations->size);
spaces = digits_full - digits_i;
fprintf(stderr, "\r");
fprintf(stderr, "iteration: %d/%d", i+1, iterations->size);
for (j = 0; j < spaces; j++)
putc(' ', stderr);
fprintf(stderr, " %s ", bar);
free(bar);
}
/*Expand any BLAST hits we got from the current query sequence during
coarse BLAST.*/
expanded_hits = expand_blast_hits(iterations, i, db);
for (j = 0; j < expanded_hits->size; j++)
ds_vector_append(oseqs, ds_vector_get(expanded_hits, j));
ds_vector_free_no_data(expanded_hits);
}
write_fine_fasta(oseqs);
for (i = 0; i < oseqs->size; i++)
cb_hit_expansion_free(
(struct cb_hit_expansion *)ds_vector_get(oseqs, i));
ds_vector_free_no_data(oseqs);
blast_fine(args, dbsize);
if (!search_flags.hide_progress)
fprintf(stderr, "\n"); //Make a newline after the progress bar
for (i = 0; i < queries->size; i++)
fasta_free_seq((struct fasta_seq *)ds_vector_get(queries, i));
ds_vector_free_no_data(queries);
if (search_flags.show_hit_info)
fclose(test_hits_file);
//Free the XML data and expanded hits
for (i = 0; i < iterations->size; i++) {
struct DSVector *iteration =
(struct DSVector *)ds_vector_get(iterations, i);
for (j = 0; j < iteration->size; j++) {
struct hit *h = (struct hit *)ds_vector_get(iteration, j);
ds_vector_free(h->hsps);
free(h);
}
}
ds_vector_free_no_data(iterations);
cb_database_r_free(db);
xmlFreeDoc(doc);
/*Free the coarse BLAST results file if the --no-cleanup flag is not being
used.*/
if (!search_flags.no_cleanup)
system("rm CaBLAST_temp_blast_results.xml");
opt_args_free(args);
opt_config_free(conf);
return 0;
}
