#include "sam_header.h"
#include "sam.h"
#include "faidx.h"
+#include "kstring.h"
#include "khash.h"
KHASH_SET_INIT_STR(rg)
// data passed to the bam_fetch callback is encapsulated in this struct.
typedef struct {
bam_header_t *header;
- int *count;
+ int64_t *count; // int does overflow for very big BAMs
} count_func_data_t;
typedef khash_t(rg) *rghash_t;
-rghash_t g_rghash = 0;
-static int g_min_mapQ = 0, g_flag_on = 0, g_flag_off = 0;
+// FIXME: we'd better use no global variables...
+static rghash_t g_rghash = 0;
+static int g_min_mapQ = 0, g_flag_on = 0, g_flag_off = 0, g_qual_scale = 0, g_min_qlen = 0;
+static uint32_t g_subsam_seed = 0;
+static double g_subsam_frac = -1.;
static char *g_library, *g_rg;
-static int g_sol2sanger_tbl[128];
+static void *g_bed;
-static void sol2sanger(bam1_t *b)
+void *bed_read(const char *fn);
+void bed_destroy(void *_h);
+int bed_overlap(const void *_h, const char *chr, int beg, int end);
+
+static int process_aln(const bam_header_t *h, bam1_t *b)
{
- int l;
- uint8_t *qual = bam1_qual(b);
- if (g_sol2sanger_tbl[30] == 0) {
- for (l = 0; l != 128; ++l) {
- g_sol2sanger_tbl[l] = (int)(10.0 * log(1.0 + pow(10.0, (l - 64 + 33) / 10.0)) / log(10.0) + .499);
- if (g_sol2sanger_tbl[l] >= 93) g_sol2sanger_tbl[l] = 93;
+ if (g_qual_scale > 1) {
+ int i;
+ uint8_t *qual = bam1_qual(b);
+ for (i = 0; i < b->core.l_qseq; ++i) {
+ int c = qual[i] * g_qual_scale;
+ qual[i] = c < 93? c : 93;
}
}
- for (l = 0; l < b->core.l_qseq; ++l) {
- int q = qual[l];
- if (q > 127) q = 127;
- qual[l] = g_sol2sanger_tbl[q];
+ if (g_min_qlen > 0) {
+ int k, qlen = 0;
+ uint32_t *cigar = bam1_cigar(b);
+ for (k = 0; k < b->core.n_cigar; ++k)
+ if ((bam_cigar_type(bam_cigar_op(cigar[k]))&1) || bam_cigar_op(cigar[k]) == BAM_CHARD_CLIP)
+ qlen += bam_cigar_oplen(cigar[k]);
+ if (qlen < g_min_qlen) return 1;
}
-}
-
-static inline int __g_skip_aln(const bam_header_t *h, const bam1_t *b)
-{
if (b->core.qual < g_min_mapQ || ((b->core.flag & g_flag_on) != g_flag_on) || (b->core.flag & g_flag_off))
return 1;
+ if (g_bed && b->core.tid >= 0 && !bed_overlap(g_bed, h->target_name[b->core.tid], b->core.pos, bam_calend(&b->core, bam1_cigar(b))))
+ return 1;
+ if (g_subsam_frac > 0.) {
+ uint32_t k = __ac_X31_hash_string(bam1_qname(b)) + g_subsam_seed;
+ if ((double)(k&0xffffff) / 0x1000000 >= g_subsam_frac) return 1;
+ }
if (g_rg || g_rghash) {
uint8_t *s = bam_aux_get(b, "RG");
if (s) {
return 0;
}
+static char *drop_rg(char *hdtxt, rghash_t h, int *len)
+{
+ char *p = hdtxt, *q, *r, *s;
+ kstring_t str;
+ memset(&str, 0, sizeof(kstring_t));
+ while (1) {
+ int toprint = 0;
+ q = strchr(p, '\n');
+ if (q == 0) q = p + strlen(p);
+ if (q - p < 3) break; // the line is too short; then stop
+ if (strncmp(p, "@RG\t", 4) == 0) {
+ int c;
+ khint_t k;
+ if ((r = strstr(p, "\tID:")) != 0) {
+ r += 4;
+ for (s = r; *s != '\0' && *s != '\n' && *s != '\t'; ++s);
+ c = *s; *s = '\0';
+ k = kh_get(rg, h, r);
+ *s = c;
+ if (k != kh_end(h)) toprint = 1;
+ }
+ } else toprint = 1;
+ if (toprint) {
+ kputsn(p, q - p, &str); kputc('\n', &str);
+ }
+ p = q + 1;
+ }
+ *len = str.l;
+ return str.s;
+}
+
// callback function for bam_fetch() that prints nonskipped records
static int view_func(const bam1_t *b, void *data)
{
- if (!__g_skip_aln(((samfile_t*)data)->header, b))
+ if (!process_aln(((samfile_t*)data)->header, (bam1_t*)b))
samwrite((samfile_t*)data, b);
return 0;
}
// callback function for bam_fetch() that counts nonskipped records
static int count_func(const bam1_t *b, void *data)
{
- if (!__g_skip_aln(((count_func_data_t*)data)->header, b)) {
+ if (!process_aln(((count_func_data_t*)data)->header, (bam1_t*)b)) {
(*((count_func_data_t*)data)->count)++;
}
return 0;
int main_samview(int argc, char *argv[])
{
- int c, is_header = 0, is_header_only = 0, is_bamin = 1, ret = 0, compress_level = -1, is_bamout = 0, slx2sngr = 0, is_count = 0;
- int of_type = BAM_OFDEC, is_long_help = 0;
- int count = 0;
+ int c, is_header = 0, is_header_only = 0, is_bamin = 1, ret = 0, compress_level = -1, is_bamout = 0, is_count = 0;
+ int of_type = BAM_OFDEC, is_long_help = 0, n_threads = 0;
+ int64_t count = 0;
samfile_t *in = 0, *out = 0;
- char in_mode[5], out_mode[5], *fn_out = 0, *fn_list = 0, *fn_ref = 0, *fn_rg = 0;
+ char in_mode[5], out_mode[5], *fn_out = 0, *fn_list = 0, *fn_ref = 0, *fn_rg = 0, *q;
/* parse command-line options */
strcpy(in_mode, "r"); strcpy(out_mode, "w");
- while ((c = getopt(argc, argv, "Sbct:h1Ho:q:f:F:ul:r:xX?T:CR:")) >= 0) {
+ while ((c = getopt(argc, argv, "SbBct:h1Ho:q:f:F:ul:r:xX?T:R:L:s:Q:@:m:")) >= 0) {
switch (c) {
+ case 's':
+ if ((g_subsam_seed = strtol(optarg, &q, 10)) != 0) {
+ srand(g_subsam_seed);
+ g_subsam_seed = rand();
+ }
+ g_subsam_frac = strtod(q, &q);
+ break;
+ case 'm': g_min_qlen = atoi(optarg); break;
case 'c': is_count = 1; break;
- case 'C': slx2sngr = 1; break;
case 'S': is_bamin = 0; break;
case 'b': is_bamout = 1; break;
case 't': fn_list = strdup(optarg); is_bamin = 0; break;
case 'u': compress_level = 0; break;
case '1': compress_level = 1; break;
case 'l': g_library = strdup(optarg); break;
+ case 'L': g_bed = bed_read(optarg); break;
case 'r': g_rg = strdup(optarg); break;
case 'R': fn_rg = strdup(optarg); break;
case 'x': of_type = BAM_OFHEX; break;
case 'X': of_type = BAM_OFSTR; break;
case '?': is_long_help = 1; break;
case 'T': fn_ref = strdup(optarg); is_bamin = 0; break;
+ case 'B': bam_no_B = 1; break;
+ case 'Q': g_qual_scale = atoi(optarg); break;
+ case '@': n_threads = strtol(optarg, 0, 0); break;
default: return usage(is_long_help);
}
}
ret = 1;
goto view_end;
}
+ if (g_rghash) { // FIXME: I do not know what "bam_header_t::n_text" is for...
+ char *tmp;
+ int l;
+ tmp = drop_rg(in->header->text, g_rghash, &l);
+ free(in->header->text);
+ in->header->text = tmp;
+ in->header->l_text = l;
+ }
if (!is_count && (out = samopen(fn_out? fn_out : "-", out_mode, in->header)) == 0) {
fprintf(stderr, "[main_samview] fail to open \"%s\" for writing.\n", fn_out? fn_out : "standard output");
ret = 1;
goto view_end;
}
+ if (n_threads > 1) samthreads(out, n_threads, 256);
if (is_header_only) goto view_end; // no need to print alignments
if (argc == optind + 1) { // convert/print the entire file
bam1_t *b = bam_init1();
int r;
while ((r = samread(in, b)) >= 0) { // read one alignment from `in'
- if (!__g_skip_aln(in->header, b)) {
- if (slx2sngr) sol2sanger(b);
+ if (!process_aln(in->header, b)) {
if (!is_count) samwrite(out, b); // write the alignment to `out'
count++;
}
view_end:
if (is_count && ret == 0) {
- printf("%d\n", count);
+ printf("%ld\n", count); // compilers on some platforms may complain about printing int64_t with %ld
}
// close files, free and return
free(fn_list); free(fn_ref); free(fn_out); free(g_library); free(g_rg); free(fn_rg);
+ if (g_bed) bed_destroy(g_bed);
if (g_rghash) {
khint_t k;
for (k = 0; k < kh_end(g_rghash); ++k)
fprintf(stderr, " -x output FLAG in HEX (samtools-C specific)\n");
fprintf(stderr, " -X output FLAG in string (samtools-C specific)\n");
fprintf(stderr, " -c print only the count of matching records\n");
+ fprintf(stderr, " -B collapse the backward CIGAR operation\n");
+ fprintf(stderr, " -@ INT number of BAM compression threads [0]\n");
+ fprintf(stderr, " -L FILE output alignments overlapping the input BED FILE [null]\n");
fprintf(stderr, " -t FILE list of reference names and lengths (force -S) [null]\n");
fprintf(stderr, " -T FILE reference sequence file (force -S) [null]\n");
fprintf(stderr, " -o FILE output file name [stdout]\n");
fprintf(stderr, " -q INT minimum mapping quality [0]\n");
fprintf(stderr, " -l STR only output reads in library STR [null]\n");
fprintf(stderr, " -r STR only output reads in read group STR [null]\n");
+ fprintf(stderr, " -s FLOAT fraction of templates to subsample; integer part as seed [-1]\n");
fprintf(stderr, " -? longer help\n");
fprintf(stderr, "\n");
if (is_long_help)
free(argv2);
return ret;
}
+
+int8_t seq_comp_table[16] = { 0, 8, 4, 12, 2, 10, 9, 14, 1, 6, 5, 13, 3, 11, 7, 15 };
+
+int main_bam2fq(int argc, char *argv[])
+{
+ bamFile fp;
+ bam_header_t *h;
+ bam1_t *b;
+ int8_t *buf;
+ int max_buf, c, no12 = 0;
+ while ((c = getopt(argc, argv, "n")) > 0)
+ if (c == 'n') no12 = 1;
+ if (argc == 1) {
+ fprintf(stderr, "Usage: samtools bam2fq <in.bam>\n");
+ return 1;
+ }
+ fp = strcmp(argv[optind], "-")? bam_open(argv[optind], "r") : bam_dopen(fileno(stdin), "r");
+ if (fp == 0) return 1;
+ h = bam_header_read(fp);
+ b = bam_init1();
+ buf = 0;
+ max_buf = 0;
+ while (bam_read1(fp, b) >= 0) {
+ int i, qlen = b->core.l_qseq;
+ uint8_t *seq;
+ putchar('@'); fputs(bam1_qname(b), stdout);
+ if (no12) putchar('\n');
+ else {
+ if ((b->core.flag & 0x40) && !(b->core.flag & 0x80)) puts("/1");
+ else if ((b->core.flag & 0x80) && !(b->core.flag & 0x40)) puts("/2");
+ else putchar('\n');
+ }
+ if (max_buf < qlen + 1) {
+ max_buf = qlen + 1;
+ kroundup32(max_buf);
+ buf = realloc(buf, max_buf);
+ }
+ buf[qlen] = 0;
+ seq = bam1_seq(b);
+ for (i = 0; i < qlen; ++i)
+ buf[i] = bam1_seqi(seq, i);
+ if (b->core.flag & 16) { // reverse complement
+ for (i = 0; i < qlen>>1; ++i) {
+ int8_t t = seq_comp_table[buf[qlen - 1 - i]];
+ buf[qlen - 1 - i] = seq_comp_table[buf[i]];
+ buf[i] = t;
+ }
+ if (qlen&1) buf[i] = seq_comp_table[buf[i]];
+ }
+ for (i = 0; i < qlen; ++i)
+ buf[i] = bam_nt16_rev_table[buf[i]];
+ puts((char*)buf);
+ puts("+");
+ seq = bam1_qual(b);
+ for (i = 0; i < qlen; ++i)
+ buf[i] = 33 + seq[i];
+ if (b->core.flag & 16) { // reverse
+ for (i = 0; i < qlen>>1; ++i) {
+ int8_t t = buf[qlen - 1 - i];
+ buf[qlen - 1 - i] = buf[i];
+ buf[i] = t;
+ }
+ }
+ puts((char*)buf);
+ }
+ free(buf);
+ bam_destroy1(b);
+ bam_header_destroy(h);
+ bam_close(fp);
+ return 0;
+}
projects / samtools.git / blobdiff