8 #include "sam_header.h"
13 KHASH_SET_INIT_STR(rg)
15 // When counting records instead of printing them,
16 // data passed to the bam_fetch callback is encapsulated in this struct.
22 typedef khash_t(rg) *rghash_t;
24 // FIXME: we'd better use no global variables...
25 static rghash_t g_rghash = 0;
26 static int g_min_mapQ = 0, g_flag_on = 0, g_flag_off = 0;
27 static float g_subsam = -1;
28 static char *g_library, *g_rg;
31 void *bed_read(const char *fn);
32 void bed_destroy(void *_h);
33 int bed_overlap(const void *_h, const char *chr, int beg, int end);
35 static inline int __g_skip_aln(const bam_header_t *h, const bam1_t *b)
37 if (b->core.qual < g_min_mapQ || ((b->core.flag & g_flag_on) != g_flag_on) || (b->core.flag & g_flag_off))
39 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))))
42 int x = (int)(g_subsam + .499);
43 uint32_t k = __ac_X31_hash_string(bam1_qname(b)) + x;
44 if (k%1024 / 1024.0 >= g_subsam - x) return 1;
46 if (g_rg || g_rghash) {
47 uint8_t *s = bam_aux_get(b, "RG");
49 if (g_rg) return (strcmp(g_rg, (char*)(s + 1)) == 0)? 0 : 1;
51 khint_t k = kh_get(rg, g_rghash, (char*)(s + 1));
52 return (k != kh_end(g_rghash))? 0 : 1;
57 const char *p = bam_get_library((bam_header_t*)h, b);
58 return (p && strcmp(p, g_library) == 0)? 0 : 1;
63 static char *drop_rg(char *hdtxt, rghash_t h, int *len)
65 char *p = hdtxt, *q, *r, *s;
67 memset(&str, 0, sizeof(kstring_t));
71 if (q == 0) q = p + strlen(p);
72 if (q - p < 3) break; // the line is too short; then stop
73 if (strncmp(p, "@RG\t", 4) == 0) {
76 if ((r = strstr(p, "\tID:")) != 0) {
78 for (s = r; *s != '\0' && *s != '\n' && *s != '\t'; ++s);
82 if (k != kh_end(h)) toprint = 1;
86 kputsn(p, q - p, &str); kputc('\n', &str);
94 // callback function for bam_fetch() that prints nonskipped records
95 static int view_func(const bam1_t *b, void *data)
97 if (!__g_skip_aln(((samfile_t*)data)->header, b))
98 samwrite((samfile_t*)data, b);
102 // callback function for bam_fetch() that counts nonskipped records
103 static int count_func(const bam1_t *b, void *data)
105 if (!__g_skip_aln(((count_func_data_t*)data)->header, b)) {
106 (*((count_func_data_t*)data)->count)++;
111 static int usage(int is_long_help);
113 int main_samview(int argc, char *argv[])
115 int c, is_header = 0, is_header_only = 0, is_bamin = 1, ret = 0, compress_level = -1, is_bamout = 0, is_count = 0;
116 int of_type = BAM_OFDEC, is_long_help = 0;
118 samfile_t *in = 0, *out = 0;
119 char in_mode[5], out_mode[5], *fn_out = 0, *fn_list = 0, *fn_ref = 0, *fn_rg = 0;
121 /* parse command-line options */
122 strcpy(in_mode, "r"); strcpy(out_mode, "w");
123 while ((c = getopt(argc, argv, "Sbct:h1Ho:q:f:F:ul:r:xX?T:R:L:s:")) >= 0) {
125 case 's': g_subsam = atof(optarg); break;
126 case 'c': is_count = 1; break;
127 case 'S': is_bamin = 0; break;
128 case 'b': is_bamout = 1; break;
129 case 't': fn_list = strdup(optarg); is_bamin = 0; break;
130 case 'h': is_header = 1; break;
131 case 'H': is_header_only = 1; break;
132 case 'o': fn_out = strdup(optarg); break;
133 case 'f': g_flag_on = strtol(optarg, 0, 0); break;
134 case 'F': g_flag_off = strtol(optarg, 0, 0); break;
135 case 'q': g_min_mapQ = atoi(optarg); break;
136 case 'u': compress_level = 0; break;
137 case '1': compress_level = 1; break;
138 case 'l': g_library = strdup(optarg); break;
139 case 'L': g_bed = bed_read(optarg); break;
140 case 'r': g_rg = strdup(optarg); break;
141 case 'R': fn_rg = strdup(optarg); break;
142 case 'x': of_type = BAM_OFHEX; break;
143 case 'X': of_type = BAM_OFSTR; break;
144 case '?': is_long_help = 1; break;
145 case 'T': fn_ref = strdup(optarg); is_bamin = 0; break;
146 default: return usage(is_long_help);
149 if (compress_level >= 0) is_bamout = 1;
150 if (is_header_only) is_header = 1;
151 if (is_bamout) strcat(out_mode, "b");
153 if (of_type == BAM_OFHEX) strcat(out_mode, "x");
154 else if (of_type == BAM_OFSTR) strcat(out_mode, "X");
156 if (is_bamin) strcat(in_mode, "b");
157 if (is_header) strcat(out_mode, "h");
158 if (compress_level >= 0) {
160 tmp[0] = compress_level + '0'; tmp[1] = '\0';
161 strcat(out_mode, tmp);
163 if (argc == optind) return usage(is_long_help); // potential memory leak...
165 // read the list of read groups
170 g_rghash = kh_init(rg);
171 fp_rg = fopen(fn_rg, "r");
172 while (!feof(fp_rg) && fscanf(fp_rg, "%s", buf) > 0) // this is not a good style, but bear me...
173 kh_put(rg, g_rghash, strdup(buf), &ret); // we'd better check duplicates...
177 // generate the fn_list if necessary
178 if (fn_list == 0 && fn_ref) fn_list = samfaipath(fn_ref);
179 // open file handlers
180 if ((in = samopen(argv[optind], in_mode, fn_list)) == 0) {
181 fprintf(pysamerr, "[main_samview] fail to open \"%s\" for reading.\n", argv[optind]);
185 if (in->header == 0) {
186 fprintf(pysamerr, "[main_samview] fail to read the header from \"%s\".\n", argv[optind]);
190 if (g_rghash) { // FIXME: I do not know what "bam_header_t::n_text" is for...
193 tmp = drop_rg(in->header->text, g_rghash, &l);
194 free(in->header->text);
195 in->header->text = tmp;
196 in->header->l_text = l;
198 if (!is_count && (out = samopen(fn_out? fn_out : "-", out_mode, in->header)) == 0) {
199 fprintf(pysamerr, "[main_samview] fail to open \"%s\" for writing.\n", fn_out? fn_out : "standard output");
203 if (is_header_only) goto view_end; // no need to print alignments
205 if (argc == optind + 1) { // convert/print the entire file
206 bam1_t *b = bam_init1();
208 while ((r = samread(in, b)) >= 0) { // read one alignment from `in'
209 if (!__g_skip_aln(in->header, b)) {
210 if (!is_count) samwrite(out, b); // write the alignment to `out'
215 fprintf(pysamerr, "[main_samview] truncated file.\n");
219 } else { // retrieve alignments in specified regions
221 bam_index_t *idx = 0;
222 if (is_bamin) idx = bam_index_load(argv[optind]); // load BAM index
223 if (idx == 0) { // index is unavailable
224 fprintf(pysamerr, "[main_samview] random alignment retrieval only works for indexed BAM files.\n");
228 for (i = optind + 1; i < argc; ++i) {
229 int tid, beg, end, result;
230 bam_parse_region(in->header, argv[i], &tid, &beg, &end); // parse a region in the format like `chr2:100-200'
231 if (tid < 0) { // reference name is not found
232 fprintf(pysamerr, "[main_samview] region \"%s\" specifies an unknown reference name. Continue anyway.\n", argv[i]);
237 count_func_data_t count_data = { in->header, &count };
238 result = bam_fetch(in->x.bam, idx, tid, beg, end, &count_data, count_func);
240 result = bam_fetch(in->x.bam, idx, tid, beg, end, out, view_func);
242 fprintf(pysamerr, "[main_samview] retrieval of region \"%s\" failed due to truncated file or corrupt BAM index file\n", argv[i]);
247 bam_index_destroy(idx); // destroy the BAM index
251 if (is_count && ret == 0) {
252 printf("%d\n", count);
254 // close files, free and return
255 free(fn_list); free(fn_ref); free(fn_out); free(g_library); free(g_rg); free(fn_rg);
256 if (g_bed) bed_destroy(g_bed);
259 for (k = 0; k < kh_end(g_rghash); ++k)
260 if (kh_exist(g_rghash, k)) free((char*)kh_key(g_rghash, k));
261 kh_destroy(rg, g_rghash);
269 static int usage(int is_long_help)
271 fprintf(pysamerr, "\n");
272 fprintf(pysamerr, "Usage: samtools view [options] <in.bam>|<in.sam> [region1 [...]]\n\n");
273 fprintf(pysamerr, "Options: -b output BAM\n");
274 fprintf(pysamerr, " -h print header for the SAM output\n");
275 fprintf(pysamerr, " -H print header only (no alignments)\n");
276 fprintf(pysamerr, " -S input is SAM\n");
277 fprintf(pysamerr, " -u uncompressed BAM output (force -b)\n");
278 fprintf(pysamerr, " -1 fast compression (force -b)\n");
279 fprintf(pysamerr, " -x output FLAG in HEX (samtools-C specific)\n");
280 fprintf(pysamerr, " -X output FLAG in string (samtools-C specific)\n");
281 fprintf(pysamerr, " -c print only the count of matching records\n");
282 fprintf(pysamerr, " -L FILE output alignments overlapping the input BED FILE [null]\n");
283 fprintf(pysamerr, " -t FILE list of reference names and lengths (force -S) [null]\n");
284 fprintf(pysamerr, " -T FILE reference sequence file (force -S) [null]\n");
285 fprintf(pysamerr, " -o FILE output file name [stdout]\n");
286 fprintf(pysamerr, " -R FILE list of read groups to be outputted [null]\n");
287 fprintf(pysamerr, " -f INT required flag, 0 for unset [0]\n");
288 fprintf(pysamerr, " -F INT filtering flag, 0 for unset [0]\n");
289 fprintf(pysamerr, " -q INT minimum mapping quality [0]\n");
290 fprintf(pysamerr, " -l STR only output reads in library STR [null]\n");
291 fprintf(pysamerr, " -r STR only output reads in read group STR [null]\n");
292 fprintf(pysamerr, " -s FLOAT fraction of templates to subsample; integer part as seed [-1]\n");
293 fprintf(pysamerr, " -? longer help\n");
294 fprintf(pysamerr, "\n");
296 fprintf(pysamerr, "Notes:\n\
298 1. By default, this command assumes the file on the command line is in\n\
299 the BAM format and it prints the alignments in SAM. If `-t' is\n\
300 applied, the input file is assumed to be in the SAM format. The\n\
301 file supplied with `-t' is SPACE/TAB delimited with the first two\n\
302 fields of each line consisting of the reference name and the\n\
303 corresponding sequence length. The `.fai' file generated by `faidx'\n\
304 can be used here. This file may be empty if reads are unaligned.\n\
306 2. SAM->BAM conversion: `samtools view -bT ref.fa in.sam.gz'.\n\
308 3. BAM->SAM conversion: `samtools view in.bam'.\n\
310 4. A region should be presented in one of the following formats:\n\
311 `chr1', `chr2:1,000' and `chr3:1000-2,000'. When a region is\n\
312 specified, the input alignment file must be an indexed BAM file.\n\
314 5. Option `-u' is preferred over `-b' when the output is piped to\n\
315 another samtools command.\n\
317 6. In a string FLAG, each character represents one bit with\n\
318 p=0x1 (paired), P=0x2 (properly paired), u=0x4 (unmapped),\n\
319 U=0x8 (mate unmapped), r=0x10 (reverse), R=0x20 (mate reverse)\n\
320 1=0x40 (first), 2=0x80 (second), s=0x100 (not primary), \n\
321 f=0x200 (failure) and d=0x400 (duplicate). Note that `-x' and\n\
322 `-X' are samtools-C specific. Picard and older samtools do not\n\
323 support HEX or string flags.\n\
328 int main_import(int argc, char *argv[])
333 fprintf(pysamerr, "Usage: bamtk import <in.ref_list> <in.sam> <out.bam>\n");
337 argv2 = calloc(6, sizeof(char*));
338 argv2[0] = "import", argv2[1] = "-o", argv2[2] = argv[3], argv2[3] = "-bt", argv2[4] = argv[1], argv2[5] = argv[2];
339 ret = main_samview(argc2, argv2);
344 int8_t seq_comp_table[16] = { 0, 8, 4, 12, 2, 10, 9, 14, 1, 6, 5, 13, 3, 11, 7, 15 };
346 int main_bam2fq(int argc, char *argv[])
354 fprintf(pysamerr, "Usage: samtools bam2fq <in.bam>\n");
357 fp = strcmp(argv[1], "-")? bam_open(argv[1], "r") : bam_dopen(fileno(stdin), "r");
358 if (fp == 0) return 1;
359 h = bam_header_read(fp);
363 while (bam_read1(fp, b) >= 0) {
364 int i, qlen = b->core.l_qseq;
366 putchar('@'); fputs(bam1_qname(b), stdout);
367 if ((b->core.flag & 0x40) && !(b->core.flag & 0x80)) puts("/1");
368 else if ((b->core.flag & 0x80) && !(b->core.flag & 0x40)) puts("/2");
370 if (max_buf < qlen + 1) {
373 buf = realloc(buf, max_buf);
377 for (i = 0; i < qlen; ++i)
378 buf[i] = bam1_seqi(seq, i);
379 if (b->core.flag & 16) { // reverse complement
380 for (i = 0; i < qlen>>1; ++i) {
381 int8_t t = seq_comp_table[buf[qlen - 1 - i]];
382 buf[qlen - 1 - i] = seq_comp_table[buf[i]];
385 if (qlen&1) buf[i] = seq_comp_table[buf[i]];
387 for (i = 0; i < qlen; ++i)
388 buf[i] = bam_nt16_rev_table[buf[i]];
392 for (i = 0; i < qlen; ++i)
393 buf[i] = 33 + seq[i];
394 if (b->core.flag & 16) { // reverse
395 for (i = 0; i < qlen>>1; ++i) {
396 int8_t t = buf[qlen - 1 - i];
397 buf[qlen - 1 - i] = buf[i];
405 bam_header_destroy(h);