int bam_merge(int argc, char *argv[])
{
int c, is_by_qname = 0, flag = 0, ret = 0;
char *fn_headers = NULL, *reg = 0;
while ((c = getopt(argc, argv, "h:nruR:")) >= 0) {
switch (c) {
case 'r': flag |= MERGE_RG; break;
case 'h': fn_headers = strdup(optarg); break;
case 'n': is_by_qname = 1; break;
case 'u': flag |= MERGE_UNCOMP; break;
case 'R': reg = strdup(optarg); break;
}
}
if (optind + 2 >= argc) {
fprintf(stderr, "\n");
fprintf(stderr, "Usage: samtools merge [-nr] [-h inh.sam] <out.bam> <in1.bam> <in2.bam> [...]\n\n");
fprintf(stderr, "Options: -n sort by read names\n");
fprintf(stderr, " -r attach RG tag (inferred from file names)\n");
fprintf(stderr, " -u uncompressed BAM output\n");
fprintf(stderr, " -R STR merge file in the specified region STR [all]\n");
fprintf(stderr, " -h FILE copy the header in FILE to <out.bam> [in1.bam]\n\n");
fprintf(stderr, "Note: Samtools' merge does not reconstruct the @RG dictionary in the header. Users\n");
fprintf(stderr, " must provide the correct header with -h, or uses Picard which properly maintains\n");
fprintf(stderr, " the header dictionary in merging.\n\n");
return 1;
}
if (bam_merge_core(is_by_qname, argv[optind], fn_headers, argc - optind - 1, argv + optind + 1, flag, reg) < 0) ret = 1;
free(reg);
free(fn_headers);
return ret;
}
int bam_merge(int argc, char *argv[])
{
int c, is_by_qname = 0, flag = 0, ret = 0;
char *fn_headers = NULL, *reg = 0;
while (0) {
//while ((c = getopt(argc, argv, "h:nru1R:f")) >= 0) {
switch (c) {
case 'r': flag |= MERGE_RG; break;
case 'f': flag |= MERGE_FORCE; break;
case 'h': fn_headers = strdup(optarg); break;
case 'n': is_by_qname = 1; break;
case '1': flag |= MERGE_LEVEL1; break;
case 'u': flag |= MERGE_UNCOMP; break;
case 'R': reg = strdup(optarg); break;
}
}
if (optind + 2 >= argc) {
fprintf(stderr, "\n");
fprintf(stderr, "Usage: samtools merge [-nr] [-h inh.sam] <out.bam> <in1.bam> <in2.bam> [...]\n\n");
fprintf(stderr, "Options: -n sort by read names\n");
fprintf(stderr, " -r attach RG tag (inferred from file names)\n");
fprintf(stderr, " -u uncompressed BAM output\n");
fprintf(stderr, " -f overwrite the output BAM if exist\n");
fprintf(stderr, " -1 compress level 1\n");
fprintf(stderr, " -R STR merge file in the specified region STR [all]\n");
fprintf(stderr, " -h FILE copy the header in FILE to <out.bam> [in1.bam]\n\n");
fprintf(stderr, "Note: Samtools' merge does not reconstruct the @RG dictionary in the header. Users\n");
fprintf(stderr, " must provide the correct header with -h, or uses Picard which properly maintains\n");
fprintf(stderr, " the header dictionary in merging.\n\n");
return 1;
}
if (!(flag & MERGE_FORCE) && strcmp(argv[optind], "-")) {
FILE *fp = fopen(argv[optind], "rb");
if (fp != NULL) {
fclose(fp);
fprintf(stderr, "[%s] File '%s' exists. Please apply '-f' to overwrite. Abort.\n", __func__, argv[optind]);
return 1;
}
}
if (bam_merge_core(is_by_qname, argv[optind], fn_headers, argc - optind - 1, argv + optind + 1, flag, reg) < 0) ret = 1;
free(reg);
free(fn_headers);
return ret;
}
/*!
@abstract Sort an unsorted BAM file based on the chromosome order
and the leftmost position of an alignment
@param is_by_qname whether to sort by query name
@param fn name of the file to be sorted
@param prefix prefix of the output and the temporary files; upon
sucessess, prefix.bam will be written.
@param max_mem approxiate maximum memory (very inaccurate)
@discussion It may create multiple temporary subalignment files
and then merge them by calling bam_merge_core(). This function is
NOT thread safe.
*/
void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size_t max_mem, int is_stdout)
{
int n, ret, k, i;
size_t mem;
bam_header_t *header;
bamFile fp;
bam1_t *b, **buf;
g_is_by_qname = is_by_qname;
n = k = 0; mem = 0;
fp = strcmp(fn, "-")? bam_open(fn, "r") : bam_dopen(fileno(stdin), "r");
if (fp == 0) {
fprintf(stderr, "[bam_sort_core] fail to open file %s\n", fn);
return;
}
header = bam_header_read(fp);
if (is_by_qname) change_SO(header, "queryname");
else change_SO(header, "coordinate");
buf = (bam1_t**)calloc(max_mem / BAM_CORE_SIZE, sizeof(bam1_t*));
// write sub files
for (;;) {
if (buf[k] == 0) buf[k] = (bam1_t*)calloc(1, sizeof(bam1_t));
b = buf[k];
if ((ret = bam_read1(fp, b)) < 0) break;
mem += ret;
++k;
if (mem >= max_mem) {
sort_blocks(n++, k, buf, prefix, header, 0);
mem = 0; k = 0;
}
}
if (ret != -1)
fprintf(stderr, "[bam_sort_core] truncated file. Continue anyway.\n");
if (n == 0) sort_blocks(-1, k, buf, prefix, header, is_stdout);
else { // then merge
char **fns, *fnout;
fprintf(stderr, "[bam_sort_core] merging from %d files...\n", n+1);
sort_blocks(n++, k, buf, prefix, header, 0);
fnout = (char*)calloc(strlen(prefix) + 20, 1);
if (is_stdout) sprintf(fnout, "-");
else sprintf(fnout, "%s.bam", prefix);
fns = (char**)calloc(n, sizeof(char*));
for (i = 0; i < n; ++i) {
fns[i] = (char*)calloc(strlen(prefix) + 20, 1);
sprintf(fns[i], "%s.%.4d.bam", prefix, i);
}
bam_merge_core(is_by_qname, fnout, 0, n, fns, 0, 0);
free(fnout);
for (i = 0; i < n; ++i) {
unlink(fns[i]);
free(fns[i]);
}
free(fns);
}
for (k = 0; (size_t)k < max_mem / BAM_CORE_SIZE; ++k) {
if (buf[k]) {
free(buf[k]->data);
free(buf[k]);
}
}
free(buf);
bam_header_destroy(header);
bam_close(fp);
}
