6 #include "bam_maqcns.h"
9 KHASH_SET_INIT_INT64(64)
11 #define BAM_PLF_SIMPLE 0x01
12 #define BAM_PLF_CNS 0x02
13 #define BAM_PLF_INDEL_ONLY 0x04
14 #define BAM_PLF_GLF 0x08
19 bam_maqindel_opt_t *ido;
23 int tid, len, last_pos;
26 glfFile fp; // for glf output only
29 char **bam_load_pos(const char *fn, int *_n);
30 void bam_init_header_hash(bam_header_t *header);
31 int32_t bam_get_tid(const bam_header_t *header, const char *seq_name);
33 static khash_t(64) *load_pos(const char *fn, bam_header_t *h)
39 bam_init_header_hash(h);
40 list = bam_load_pos(fn, &n);
42 for (i = 0; i < n; ++i) {
43 x = (uint64_t)bam_get_tid(h, list[i]) << 32;
46 x |= *((uint32_t*)s) - 1;
47 kh_put(64, hash, x, &tmp);
54 // an analogy to pileup_func() below
55 static int glt_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
57 pu_data_t *d = (pu_data_t*)data;
58 bam_maqindel_ret_t *r = 0;
63 fprintf(stderr, "[glt_func] reference sequence is required for generating GLT. Abort!\n");
66 if (d->hash && kh_get(64, d->hash, (uint64_t)tid<<32|pos) == kh_end(d->hash)) return 0;
67 if (d->fai && (int)tid != d->tid) {
68 if (d->ref) { // then write the end mark
69 memset(&g2, 0, sizeof(glf3_t));
70 g2.type = GLF_TYPE_END;
71 bgzf_write(d->fp, &g2, sizeof(glf3_t));
73 glf_ref_write(d->fp, d->h->target_name[tid]); // write reference
75 d->ref = fai_fetch(d->fai, d->h->target_name[tid], &d->len);
78 rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
79 g = bam_maqcns_glfgen(n, pu, bam_nt16_table[rb], d->c);
80 memcpy(&g2, g, sizeof(glf1_t));
81 g2.type = GLF_TYPE_NORMAL;
82 g2.offset = pos - d->last_pos;
84 bgzf_write(d->fp, &g2, sizeof(glf3_t));
85 r = bam_maqindel(n, pos, d->ido, pu, d->ref);
86 if (r) { // then write indel line
89 if (min > r->gl[0]) min = r->gl[0];
90 if (min > r->gl[1]) min = r->gl[1];
92 g2.type = GLF_TYPE_INDEL;
94 g2.lk[0] = r->gl[0] - min < 255? r->gl[0] - min : 255;
95 g2.lk[1] = r->gl[1] - min < 255? r->gl[1] - min : 255;
96 g2.lk[2] = g3 - min < 255? g3 - min : 255;
97 *(int16_t*)(g2.lk + 3) = (int16_t)r->indel1;
98 *(int16_t*)(g2.lk + 5) = (int16_t)r->indel2;
99 g2.min_lk = min < 255? min : 255;
100 bgzf_write(d->fp, &g2, sizeof(glf3_t));
101 if (r->indel1) bgzf_write(d->fp, r->s[0]+1, r->indel1>0? r->indel1 : -r->indel1);
102 if (r->indel2) bgzf_write(d->fp, r->s[1]+1, r->indel2>0? r->indel2 : -r->indel2);
103 bam_maqindel_ret_destroy(r);
109 static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
111 pu_data_t *d = (pu_data_t*)data;
112 bam_maqindel_ret_t *r = 0;
113 int i, j, rb, max_mapq = 0;
115 if (d->hash && kh_get(64, d->hash, (uint64_t)tid<<32|pos) == kh_end(d->hash)) return 0;
116 if (d->format & BAM_PLF_GLF) return glt_func(tid, pos, n, pu, data);
117 if (d->fai && (int)tid != d->tid) {
119 d->ref = fai_fetch(d->fai, d->h->target_name[tid], &d->len);
122 rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
123 if (d->format & BAM_PLF_INDEL_ONLY) {
124 for (i = 0; i < n; ++i)
125 if (pu[i].indel != 0) break;
126 if (i == n) return 0;
128 printf("%s\t%d\t%c\t", d->h->target_name[tid], pos + 1, rb);
129 if (d->format & BAM_PLF_CNS) { // consensus
130 int ref_q, rb4 = bam_nt16_table[rb];
131 x = bam_maqcns_call(n, pu, d->c);
133 if (rb4 != 15 && x>>28 != 15 && x>>28 != rb4) { // a SNP
134 ref_q = ((x>>24&0xf) == rb4)? x>>8&0xff : (x>>8&0xff) + (x&0xff);
135 if (ref_q > 255) ref_q = 255;
137 printf("%c\t%d\t%d\t%d\t", bam_nt16_rev_table[x>>28], x>>8&0xff, ref_q, x>>16&0xff);
139 if (d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) {
140 if (d->ref) // indel calling
141 r = bam_maqindel(n, pos, d->ido, pu, d->ref);
145 for (i = 0; i < n; ++i) {
146 const bam_pileup1_t *p = pu + i;
147 if (max_mapq < p->b->core.qual) max_mapq = p->b->core.qual;
148 if (p->is_head) printf("^%c", p->b->core.qual > 93? 126 : p->b->core.qual + 33);
150 int c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
151 if (toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
152 else c = bam1_strand(p->b)? tolower(c) : toupper(c);
155 printf("+%d", p->indel);
156 for (j = 1; j <= p->indel; ++j) {
157 c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + j)];
158 putchar(bam1_strand(p->b)? tolower(c) : toupper(c));
160 } else if (p->indel < 0) {
161 printf("%d", p->indel);
162 for (j = 1; j <= -p->indel; ++j) {
163 c = (d->ref && (int)pos+j < d->len)? d->ref[pos+j] : 'N';
164 putchar(bam1_strand(p->b)? tolower(c) : toupper(c));
168 if (p->is_tail) putchar('$');
171 for (i = 0; i < n; ++i) {
172 const bam_pileup1_t *p = pu + i;
173 int c = bam1_qual(p->b)[p->qpos] + 33;
174 if (c > 126) c = 126;
177 if (d->format & BAM_PLF_SIMPLE) {
179 for (i = 0; i < n; ++i) {
180 int c = pu[i].b->core.qual + 33;
181 if (c > 126) c = 126;
186 if (r) { // then print indel line
187 printf("%s\t%d\t*\t", d->h->target_name[tid], pos + 1);
188 if (r->gt < 2) printf("%s/%s\t", r->s[r->gt], r->s[r->gt]);
189 else printf("%s/%s\t", r->s[0], r->s[1]);
190 printf("%d\t%d\t", r->q_cns, r->q_ref);
191 printf("%d\t%d\t", max_mapq, n);
192 printf("%s\t%s\t", r->s[0], r->s[1]);
193 //printf("%d\t%d\t", r->gl[0], r->gl[1]);
194 printf("%d\t%d\t%d\t%d\n", r->cnt1, r->cnt2, r->cnt_ambi, r->cnt_anti);
195 bam_maqindel_ret_destroy(r);
200 int bam_pileup(int argc, char *argv[])
203 char *fn_list = 0, *fn_fa = 0, *fn_pos = 0;
204 pu_data_t *d = (pu_data_t*)calloc(1, sizeof(pu_data_t));
205 d->tid = -1; d->mask = BAM_DEF_MASK;
206 d->c = bam_maqcns_init();
207 d->ido = bam_maqindel_opt_init();
208 while ((c = getopt(argc, argv, "st:f:cT:N:r:l:im:gI:G:")) >= 0) {
210 case 's': d->format |= BAM_PLF_SIMPLE; break;
211 case 't': fn_list = strdup(optarg); break;
212 case 'l': fn_pos = strdup(optarg); break;
213 case 'f': fn_fa = strdup(optarg); break;
214 case 'T': d->c->theta = atof(optarg); break;
215 case 'N': d->c->n_hap = atoi(optarg); break;
216 case 'r': d->c->het_rate = atoi(optarg); break;
217 case 'c': d->format |= BAM_PLF_CNS; break;
218 case 'i': d->format |= BAM_PLF_INDEL_ONLY; break;
219 case 'm': d->mask = atoi(optarg); break;
220 case 'g': d->format |= BAM_PLF_GLF; break;
221 case 'I': d->ido->q_indel = atoi(optarg); break;
222 case 'G': d->ido->r_indel = atof(optarg); break;
223 default: fprintf(stderr, "Unrecognizd option '-%c'.\n", c); return 1;
226 if (optind == argc) {
227 fprintf(stderr, "\n");
228 fprintf(stderr, "Usage: bamtk pileup [options] <in.bam>|<in.sam>\n\n");
229 fprintf(stderr, "Option: -s simple (yet incomplete) pileup format\n");
230 fprintf(stderr, " -i only show lines/consensus with indels\n");
231 fprintf(stderr, " -m INT filtering reads with bits in INT [%d]\n", d->mask);
232 fprintf(stderr, " -t FILE list of reference sequences (assume the input is in SAM)\n");
233 fprintf(stderr, " -l FILE list of sites at which pileup is output\n");
234 fprintf(stderr, " -f FILE reference sequence in the FASTA format\n\n");
235 fprintf(stderr, " -c output the maq consensus sequence\n");
236 fprintf(stderr, " -g output in the extended GLT format (suppressing -c/-i/-s)\n");
237 fprintf(stderr, " -T FLOAT theta in maq consensus calling model (for -c/-g) [%f]\n", d->c->theta);
238 fprintf(stderr, " -N INT number of haplotypes in the sample (for -c/-g) [%d]\n", d->c->n_hap);
239 fprintf(stderr, " -r FLOAT prior of a difference between two haplotypes (for -c/-g) [%f]\n", d->c->het_rate);
240 fprintf(stderr, " -G FLOAT prior of an indel between two haplotypes (for -c/-g) [%f]\n", d->ido->r_indel);
241 fprintf(stderr, " -I INT phred prob. of an indel in sequencing/prep. (for -c/-g) [%d]\n", d->ido->q_indel);
242 fprintf(stderr, "\n");
243 free(fn_list); free(fn_fa); free(d);
246 if (fn_fa) d->fai = fai_load(fn_fa);
248 bam_maqcns_prepare(d->c);
249 if (d->format & BAM_PLF_GLF) {
251 h = glf_header_init();
252 d->fp = bgzf_fdopen(fileno(stdout), "w");
253 glf_header_write(d->fp, h);
254 glf_header_destroy(h);
260 bam_plbuf_t *buf = bam_plbuf_init(pileup_func, d);
261 bam_plbuf_set_mask(buf, d->mask);
262 d->h = sam_header_read2(fn_list);
263 if (fn_pos) d->hash = load_pos(fn_pos, d->h);
264 fp = sam_open(argv[optind]);
265 b = (bam1_t*)calloc(1, sizeof(bam1_t));
266 while ((ret = sam_read1(fp, d->h, b)) >= 0)
267 bam_plbuf_push(b, buf);
268 bam_plbuf_push(0, buf);
269 bam_plbuf_destroy(buf);
274 fp = (strcmp(argv[optind], "-") == 0)? bam_dopen(fileno(stdin), "r") : bam_open(argv[optind], "r");
275 d->h = bam_header_read(fp);
276 if (fn_pos) d->hash = load_pos(fn_pos, d->h);
277 bam_pileup_file(fp, d->mask, pileup_func, d);
280 if (d->format & BAM_PLF_GLF) bgzf_close(d->fp);
281 free(fn_pos); free(fn_list);
282 kh_destroy(64, d->hash);
283 bam_header_destroy(d->h);
284 if (d->fai) fai_destroy(d->fai);
285 bam_maqcns_destroy(d->c);
286 free(d->ido); free(d->ref); free(d);