int
tmap_refseq_refinfo_main(int argc, char *argv[])
{
int c, help=0;
tmap_refseq_t *refseq = NULL;
tmap_file_t *fp_anno = NULL;
char *fn_anno = NULL;
char *fn_fasta = NULL;
while((c = getopt(argc, argv, "vh")) >= 0) {
switch(c) {
case 'v': tmap_progress_set_verbosity(1); break;
case 'h': help = 1; break;
default: return 1;
}
}
if(1 != argc - optind || 1 == help) {
tmap_file_fprintf(tmap_file_stderr, "Usage: %s %s [-vh] <in.fasta>\n", PACKAGE, argv[0]);
return 1;
}
fn_fasta = argv[optind];
// Note: 'tmap_file_stdout' should not have been previously modified
tmap_file_stdout = tmap_file_fdopen(fileno(stdout), "wb", TMAP_FILE_NO_COMPRESSION);
// allocate some memory
refseq = tmap_calloc(1, sizeof(tmap_refseq_t), "refseq");
refseq->is_rev = 0;
refseq->is_shm = 0;
// read the annotation file
fn_anno = tmap_get_file_name(fn_fasta, TMAP_ANNO_FILE);
fp_anno = tmap_file_fopen(fn_anno, "rb", TMAP_ANNO_COMPRESSION);
tmap_refseq_read_anno(fp_anno, refseq);
tmap_file_fclose(fp_anno);
free(fn_anno);
// no need to read in the pac
refseq->seq = NULL;
// print the header
tmap_refseq_print_header(tmap_file_stdout, refseq);
// destroy
tmap_refseq_destroy(refseq);
// close the output
tmap_file_fclose(tmap_file_stdout);
return 0;
}
int
tmap_vswbm_main(int argc, char *argv[])
{
int32_t seq_len = 150;
int32_t tlen = 256;
int32_t n_iter = 1000;
int32_t n_sub_iter = 1;
int32_t vsw_type = 0;
int c;
while((c = getopt(argc, argv, "q:t:n:N:H:h")) >= 0) {
switch(c) {
case 'q':
seq_len = atoi(optarg); break;
case 't':
tlen = atoi(optarg); break;
case 'n':
n_iter = atoi(optarg); break;
case 'N':
n_sub_iter = atoi(optarg); break;
case 'H':
vsw_type = atoi(optarg); break;
case 'h':
default:
return usage(seq_len, tlen, n_iter, n_sub_iter, vsw_type);
}
}
if(argc != optind || seq_len > tlen) {
return usage(seq_len, tlen, n_iter, n_sub_iter, vsw_type);
}
tmap_progress_set_verbosity(1);
tmap_progress_print2("starting benchmark");
tmap_vsw_bm_core(seq_len, tlen, n_iter, n_sub_iter, vsw_type);
tmap_progress_print2("ending benchmark");
return 0;
}
int
tmap_refseq_fasta2pac_main(int argc, char *argv[])
{
int c, help=0;
while((c = getopt(argc, argv, "vh")) >= 0) {
switch(c) {
case 'v': tmap_progress_set_verbosity(1); break;
case 'h': help = 1; break;
default: return 1;
}
}
if(1 != argc - optind || 1 == help) {
tmap_file_fprintf(tmap_file_stderr, "Usage: %s %s [-vh] <in.fasta>\n", PACKAGE, argv[0]);
return 1;
}
tmap_refseq_fasta2pac(argv[optind], TMAP_FILE_NO_COMPRESSION);
return 0;
}
int
tmap_index(int argc, char *argv[])
{
int c;
tmap_index_opt_t opt;
opt.fn_fasta = NULL;
opt.occ_interval = TMAP_BWT_OCC_INTERVAL;
opt.hash_width = INT32_MAX;
opt.sa_interval = TMAP_SA_INTERVAL;
opt.is_large = -1;
opt.check_hash = 1;
if(2 == argc && 0 == strcmp("--version", argv[1])) {
tmap_file_stdout = tmap_file_fdopen(fileno(stdout), "wb", TMAP_FILE_NO_COMPRESSION);
tmap_file_fprintf(tmap_file_stdout, "%s\n", tmap_refseq_get_version_format(PACKAGE_VERSION));
tmap_file_fclose(tmap_file_stdout);
return 0;
}
while((c = getopt(argc, argv, "f:o:i:w:a:hvH")) >= 0) {
switch(c) {
case 'f':
opt.fn_fasta = tmap_strdup(optarg); break;
case 'o':
opt.occ_interval = atoi(optarg); break;
case 'i':
opt.sa_interval = atoi(optarg); break;
case 'w':
opt.hash_width = atoi(optarg); break;
case 'a':
if(0 == strcmp("is", optarg)) opt.is_large = 0;
else if(0 == strcmp("bwtsw", optarg)) opt.is_large = 1;
else tmap_error("Option -a value not correct", Exit, CommandLineArgument);
break;
case 'v':
tmap_progress_set_verbosity(1); break;
case 'h':
return usage(&opt);
case 'H':
opt.check_hash = 0; break;
default:
return usage(&opt);
}
}
if(argc != optind || 1 == argc) {
return usage(&opt);
}
if(NULL == opt.fn_fasta) {
tmap_error("required option -f", Exit, CommandLineArgument);
}
if(opt.occ_interval < TMAP_BWT_OCC_MOD || 0 != (opt.occ_interval % 2) || 0 != (opt.occ_interval % TMAP_BWT_OCC_MOD)) {
tmap_error("option -o out of range", Exit, CommandLineArgument);
}
if(opt.hash_width < 0) {
tmap_error("option -w out of range", Exit, CommandLineArgument);
}
if(opt.sa_interval <= 0 || (1 < opt.sa_interval && 0 != (opt.sa_interval % 2))) {
tmap_error("option -i out of range", Exit, CommandLineArgument);
}
tmap_index_core(&opt);
free(opt.fn_fasta);
tmap_progress_print2("terminating successfully");
return 0;
}
int
tmap_seqs_io_sff2sam_main(int argc, char *argv[])
{
int c, help = 0;
tmap_seqs_io_t *io_in = NULL;
tmap_seqs_t *seqs = NULL;
char **sam_rg = NULL;
int32_t sam_rg_num = 0;
int bidirectional = 0, sam_flowspace_tags = 0;
int out_type = 0;
tmap_sam_io_t *io_out = NULL;
bam_header_t *header = NULL; // BAM Header
int32_t i;
/*
uint8_t *key_seq = NULL;
int key_seq_len = 0;
*/
while((c = getopt(argc, argv, "DGR:Yvh")) >= 0) {
switch(c) {
case 'D': bidirectional = 1; break;
case 'G': break;
case 'R':
sam_rg = tmap_realloc(sam_rg, (1+sam_rg_num) * sizeof(char*), "sam_rg");
sam_rg[sam_rg_num] = tmap_strdup(optarg);
sam_rg_num++;
break;
case 'Y': sam_flowspace_tags = 1; break;
case 'v': tmap_progress_set_verbosity(1); break;
case 'h': help = 1; break;
default: return 1;
}
}
if(1 != argc - optind || 1 == help) {
tmap_file_fprintf(tmap_file_stderr, "Usage: %s %s [-R -Y -v -h] <in.sff>\n", PACKAGE, argv[0]);
return 1;
}
// input
io_in = tmap_seqs_io_init(&argv[optind], 1, TMAP_SEQ_TYPE_SFF, TMAP_FILE_NO_COMPRESSION, 0l, 0l);
// BAM Header
header = tmap_seqs_io_to_bam_header(NULL, io_in, sam_rg, sam_rg_num, argc, argv);
// open the output file
switch(out_type) {
case 0: // SAM
io_out = tmap_sam_io_init2("-", "wh", header);
break;
case 1:
io_out = tmap_sam_io_init2("-", "wb", header);
break;
case 2:
io_out = tmap_sam_io_init2("-", "wbu", header);
break;
default:
tmap_bug();
}
// destroy the BAM Header
bam_header_destroy(header);
header = NULL;
seqs = tmap_seqs_init(TMAP_SEQ_TYPE_SFF);
while(0 < tmap_seqs_io_read(io_in, seqs, io_out->fp->header->header)) {
bam1_t *b = NULL;
tmap_seq_t *seq = seqs->seqs[0];
b = tmap_sam_convert_unmapped(seq, sam_flowspace_tags, bidirectional, NULL,
0, 0, 0,
0, 0, 0,
"\tlq:i:%d\trq:i:%d\tla:i:%d\trq:i:%d",
seq->data.sff->rheader->clip_qual_left,
seq->data.sff->rheader->clip_qual_right,
seq->data.sff->rheader->clip_adapter_left,
seq->data.sff->rheader->clip_adapter_right);
if(samwrite(io_out->fp, b) <= 0) {
tmap_error("Error writing the SAM file", Exit, WriteFileError);
}
bam_destroy1(b);
tmap_seqs_destroy(seqs);
seqs = tmap_seqs_init(TMAP_SEQ_TYPE_SFF);
}
tmap_seqs_destroy(seqs);
// free memory
tmap_seqs_io_destroy(io_in);
tmap_sam_io_destroy(io_out);
for(i=0;i<sam_rg_num;i++) {
free(sam_rg[i]);
}
free(sam_rg);
return 0;
}
int
tmap_refseq_pac2fasta_main(int argc, char *argv[])
{
int c, help=0, amb=0;
uint32_t i, j, k;
char *fn_fasta = NULL;
tmap_refseq_t *refseq = NULL;
while((c = getopt(argc, argv, "avh")) >= 0) {
switch(c) {
case 'a': amb = 1; break;
case 'v': tmap_progress_set_verbosity(1); break;
case 'h': help = 1; break;
default: return 1;
}
}
if(1 != argc - optind || 1 == help) {
tmap_file_fprintf(tmap_file_stderr, "Usage: %s %s [-avh] <in.fasta>\n", PACKAGE, argv[0]);
return 1;
}
fn_fasta = argv[optind];
// Note: 'tmap_file_stdout' should not have been previously modified
tmap_file_stdout = tmap_file_fdopen(fileno(stdout), "wb", TMAP_FILE_NO_COMPRESSION);
// read in the reference sequence
refseq = tmap_refseq_read(fn_fasta, 0);
for(i=0;i<refseq->num_annos;i++) {
tmap_file_fprintf(tmap_file_stdout, ">%s", refseq->annos[i].name->s); // new line handled later
for(j=k=0;j<refseq->annos[i].len;j++) {
if(0 == (j % TMAP_REFSEQ_FASTA_LINE_LENGTH)) {
tmap_file_fprintf(tmap_file_stdout, "\n");
}
if(1 == amb && 0 < refseq->annos[i].num_amb) {
// move the next ambiguous region
while(k < refseq->annos[i].num_amb && refseq->annos[i].amb_positions_end[k] < j+1) {
k++;
}
// check for the ambiguous region
if(k < refseq->annos[i].num_amb
&& 0 == tmap_interval_overlap(j+1, j+1, refseq->annos[i].amb_positions_start[k], refseq->annos[i].amb_positions_end[k])) {
tmap_file_fprintf(tmap_file_stdout, "%c", tmap_iupac_int_to_char[refseq->annos[i].amb_bases[k]]);
}
else {
tmap_file_fprintf(tmap_file_stdout, "%c", "ACGTN"[(int)tmap_refseq_seq_i(refseq, j + refseq->annos[i].offset)]);
}
}
else {
tmap_file_fprintf(tmap_file_stdout, "%c", "ACGTN"[(int)tmap_refseq_seq_i(refseq, j + refseq->annos[i].offset)]);
}
}
tmap_file_fprintf(tmap_file_stdout, "\n");
}
// destroy
tmap_refseq_destroy(refseq);
// close the output
tmap_file_fclose(tmap_file_stdout);
return 0;
}