bwa_seq_t *load_reads(const char *fa_fn, uint32_t *n_seqs) {
bwa_seq_t *seqs, *part_seqs;
bwa_seqio_t *ks;
int n_part_seqs = 0, n_seqs_full = 0, n_seqs_loaded = 0;
clock_t t = clock();
ks = bwa_open_reads(BWA_MODE, fa_fn);
n_seqs_full = N_CHUNK_SEQS;
show_msg(__func__, "Loading reads from library %s...\n", fa_fn);
seqs = (bwa_seq_t*) calloc (N_DF_MAX_SEQS, sizeof(bwa_seq_t));
while ((part_seqs = bwa_read_seq(ks, N_CHUNK_SEQS, &n_part_seqs, BWA_MODE, 0))
!= 0) {
show_msg(__func__, "%d sequences loaded: %.2f sec... \n",
n_seqs_loaded + n_part_seqs, fa_fn, (float) (clock() - t) / CLOCKS_PER_SEC);
pe_reverse_seqs(part_seqs, n_part_seqs);
if ((n_seqs_loaded + n_part_seqs) > n_seqs_full) {
n_seqs_full += n_part_seqs + 2;
kroundup32(n_seqs_full);
seqs = (bwa_seq_t*) realloc(seqs, sizeof(bwa_seq_t) * n_seqs_full);
}
memmove(&seqs[n_seqs_loaded], part_seqs, sizeof(bwa_seq_t) * n_part_seqs);
free(part_seqs);
n_seqs_loaded += n_part_seqs;
}
bwa_seq_close(ks);
if (n_seqs_loaded < 1) {
err_fatal(__func__,
"No sequence in file %s, make sure the format is correct! \n",
fa_fn);
}
*n_seqs = n_seqs_loaded;
return seqs;
}
void bwa_aln_core(const char *prefix, const char *fn_fa, const gap_opt_t *opt)
{
int i, n_seqs, tot_seqs = 0;
bwa_seq_t *seqs;
bwa_seqio_t *ks;
clock_t t;
bwt_t *bwt;
// initialization
ks = bwa_open_reads(opt->mode, fn_fa);
{ // load BWT
char *str = (char*)calloc(strlen(prefix) + 10, 1);
strcpy(str, prefix); strcat(str, ".bwt"); bwt = bwt_restore_bwt(str);
free(str);
}
// core loop
err_fwrite(opt, sizeof(gap_opt_t), 1, stdout);
while ((seqs = bwa_read_seq(ks, 0x40000, &n_seqs, opt->mode, opt->trim_qual)) != 0) {
tot_seqs += n_seqs;
t = clock();
fprintf(stderr, "[bwa_aln_core] calculate SA coordinate... ");
#ifdef HAVE_PTHREAD
if (opt->n_threads <= 1) { // no multi-threading at all
bwa_cal_sa_reg_gap(0, bwt, n_seqs, seqs, opt);
} else {
pthread_t *tid;
pthread_attr_t attr;
thread_aux_t *data;
int j;
pthread_attr_init(&attr);
pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_JOINABLE);
data = (thread_aux_t*)calloc(opt->n_threads, sizeof(thread_aux_t));
tid = (pthread_t*)calloc(opt->n_threads, sizeof(pthread_t));
for (j = 0; j < opt->n_threads; ++j) {
data[j].tid = j; data[j].bwt = bwt;
data[j].n_seqs = n_seqs; data[j].seqs = seqs; data[j].opt = opt;
pthread_create(&tid[j], &attr, worker, data + j);
}
for (j = 0; j < opt->n_threads; ++j) pthread_join(tid[j], 0);
free(data); free(tid);
}
#else
bwa_cal_sa_reg_gap(0, bwt, n_seqs, seqs, opt);
#endif
fprintf(stderr, "%.2f sec\n", (float)(clock() - t) / CLOCKS_PER_SEC); t = clock();
t = clock();
fprintf(stderr, "[bwa_aln_core] write to the disk... ");
for (i = 0; i < n_seqs; ++i) {
bwa_seq_t *p = seqs + i;
err_fwrite(&p->n_aln, 4, 1, stdout);
if (p->n_aln) err_fwrite(p->aln, sizeof(bwt_aln1_t), p->n_aln, stdout);
}
fprintf(stderr, "%.2f sec\n", (float)(clock() - t) / CLOCKS_PER_SEC); t = clock();
bwa_free_read_seq(n_seqs, seqs);
fprintf(stderr, "[bwa_aln_core] %d sequences have been processed.\n", tot_seqs);
}
// destroy
bwt_destroy(bwt);
bwa_seq_close(ks);
}
void bwa_aln_core(const char *prefix, const char *fn_fa, const gap_opt_t *opt)
{
int i, n_seqs, tot_seqs = 0;
bwa_seq_t *seqs;
bwa_seqio_t *ks;
clock_t t;
bwt_t *bwt;
// initialization
ks = bwa_open_reads(opt->mode, fn_fa);
{ // load BWT
char *str = (char*)calloc(strlen(prefix) + 10, 1);
strcpy(str, prefix); strcat(str, ".bwt"); bwt = bwt_restore_bwt(str);
free(str);
}
// core loop
err_fwrite(opt, sizeof(gap_opt_t), 1, stdout);
while ((seqs = bwa_read_seq(ks, 0x40000, &n_seqs, opt->mode, opt->trim_qual)) != 0) {
tot_seqs += n_seqs;
t = clock();
fprintf(stderr, "[bwa_aln_core] calculate SA coordinate... ");
#ifdef THREAD
if (opt->n_threads <= 1) { // no multi-threading at all
bwa_cal_sa_reg_gap(0, bwt, n_seqs, seqs, opt);
} else {
DWORD ThreadID;
HANDLE *tid;
thread_aux_t *data;
int j;
data = (thread_aux_t*)calloc(opt->n_threads, sizeof(thread_aux_t));
tid = (HANDLE*)calloc(opt->n_threads, sizeof(HANDLE));
for (j = 0; j < opt->n_threads; ++j) {
data[j].tid = j; data[j].bwt = bwt;
data[j].n_seqs = n_seqs; data[j].seqs = seqs; data[j].opt = opt;
//create threads
tid[j] = CreateThread(NULL, 0, (LPTHREAD_START_ROUTINE) worker, data + j, 0, &ThreadID);
if (tid[j] == NULL)
{
printf("CreateThread error: %d\n", GetLastError());
return;
}
}
// Wait for all threads to terminate
WaitForMultipleObjects(opt->n_threads, &tid[0], TRUE, INFINITE);
free(data); free(tid);
}
#else
bwa_cal_sa_reg_gap(0, bwt, n_seqs, seqs, opt);
#endif
fprintf(stderr, "%.2f sec\n", (float)(clock() - t) / CLOCKS_PER_SEC); t = clock();
t = clock();
fprintf(stderr, "[bwa_aln_core] write to the disk... ");
for (i = 0; i < n_seqs; ++i) {
bwa_seq_t *p = seqs + i;
err_fwrite(&p->n_aln, 4, 1, stdout);
if (p->n_aln) err_fwrite(p->aln, sizeof(bwt_aln1_t), p->n_aln, stdout);
}
fprintf(stderr, "%.2f sec\n", (float)(clock() - t) / CLOCKS_PER_SEC); t = clock();
bwa_free_read_seq(n_seqs, seqs);
fprintf(stderr, "[bwa_aln_core] %d sequences have been processed.\n", tot_seqs);
}
// destroy
bwt_destroy(bwt);
bwa_seq_close(ks);
}