X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_plcmd.c;h=1205e10b845288f25dbd240b3b612ae79d2f8b8a;hb=75c9ea6ab922bb9245879b578c67497c338b5c86;hp=d34cd267a9edd363a32605c083fefe5ba512e8bb;hpb=4967051aa734ece6ce44f0fb891a936e849c2fe2;p=samtools.git diff --git a/bam_plcmd.c b/bam_plcmd.c index d34cd26..1205e10 100644 --- a/bam_plcmd.c +++ b/bam_plcmd.c @@ -341,17 +341,19 @@ int bam_pileup(int argc, char *argv[]) d->max_depth = 0; d->tid = -1; d->mask = BAM_DEF_MASK; d->c = bam_maqcns_init(); + d->c->errmod = BAM_ERRMOD_MAQ2; // change the default model d->ido = bam_maqindel_opt_init(); - while ((c = getopt(argc, argv, "st:f:cT:N:r:l:d:im:gI:G:vM:S2aR:P")) >= 0) { + while ((c = getopt(argc, argv, "st:f:cT:N:r:l:d:im:gI:G:vM:S2aR:PA")) >= 0) { switch (c) { - case 'a': d->c->is_soap = 1; break; + case 'a': d->c->errmod = BAM_ERRMOD_SOAP; break; + case 'A': d->c->errmod = BAM_ERRMOD_MAQ; break; case 's': d->format |= BAM_PLF_SIMPLE; break; case 't': fn_list = strdup(optarg); break; case 'l': fn_pos = strdup(optarg); break; case 'f': fn_fa = strdup(optarg); break; case 'T': d->c->theta = atof(optarg); break; case 'N': d->c->n_hap = atoi(optarg); break; - case 'r': d->c->het_rate = atof(optarg); break; + case 'r': d->c->het_rate = atof(optarg); d->ido->r_snp = d->c->het_rate; break; case 'M': d->c->cap_mapQ = atoi(optarg); break; case 'd': d->max_depth = atoi(optarg); break; case 'c': d->format |= BAM_PLF_CNS; break; @@ -373,13 +375,15 @@ int bam_pileup(int argc, char *argv[]) default: fprintf(stderr, "Unrecognizd option '-%c'.\n", c); return 1; } } + if (d->c->errmod != BAM_ERRMOD_MAQ2) d->c->theta += 0.02; + if (d->c->theta > 1.0) d->c->theta = 1.0; if (fn_list) is_SAM = 1; if (optind == argc) { fprintf(stderr, "\n"); fprintf(stderr, "Usage: samtools pileup [options] |\n\n"); fprintf(stderr, "Option: -s simple (yet incomplete) pileup format\n"); fprintf(stderr, " -S the input is in SAM\n"); - fprintf(stderr, " -a use the SOAPsnp model for SNP calling\n"); + fprintf(stderr, " -A use the MAQ model for SNP calling\n"); fprintf(stderr, " -2 output the 2nd best call and quality\n"); fprintf(stderr, " -i only show lines/consensus with indels\n"); fprintf(stderr, " -m INT filtering reads with bits in INT [%d]\n", d->mask); @@ -388,7 +392,7 @@ int bam_pileup(int argc, char *argv[]) fprintf(stderr, " -t FILE list of reference sequences (force -S)\n"); fprintf(stderr, " -l FILE list of sites at which pileup is output\n"); fprintf(stderr, " -f FILE reference sequence in the FASTA format\n\n"); - fprintf(stderr, " -c output the maq consensus sequence\n"); + fprintf(stderr, " -c output the SOAPsnp consensus sequence\n"); fprintf(stderr, " -v print variants only (for -c)\n"); fprintf(stderr, " -g output in the GLFv3 format (suppressing -c/-i/-s)\n"); fprintf(stderr, " -T FLOAT theta in maq consensus calling model (for -c/-g) [%f]\n", d->c->theta); @@ -446,21 +450,74 @@ int bam_pileup(int argc, char *argv[]) * mpileup * ***********/ +#include +#include "bam2bcf.h" +#include "sample.h" + +#define MPLP_GLF 0x10 +#define MPLP_NO_COMP 0x20 +#define MPLP_NO_ORPHAN 0x40 +#define MPLP_REALN 0x80 +#define MPLP_NO_HALFTRIM 0x100 + typedef struct { - char *reg; + int max_mq, min_mq, flag, min_baseQ; + double theta; + char *reg, *fn_pos; faidx_t *fai; + kh_64_t *hash; } mplp_conf_t; typedef struct { bamFile fp; - bam_iterf_t iter; + bam_iter_t iter; + int min_mq, flag; + char *ref; } mplp_aux_t; +typedef struct { + int n; + int *n_plp, *m_plp; + bam_pileup1_t **plp; +} mplp_pileup_t; + static int mplp_func(void *data, bam1_t *b) { + extern int bam_realn(bam1_t *b, const char *ref); mplp_aux_t *ma = (mplp_aux_t*)data; - if (ma->iter) return bam_iterf_read(ma->fp, ma->iter, b); - return bam_read1(ma->fp, b); + int ret, cond = 0; + do { + cond = 0; + ret = ma->iter? bam_iter_read(ma->fp, ma->iter, b) : bam_read1(ma->fp, b); + if (ret < 0) break; + if (b->core.qual < ma->min_mq) cond = 1; + else if ((ma->flag&MPLP_NO_ORPHAN) && (b->core.flag&1) && !(b->core.flag&2)) cond = 1; + if (ma->ref && !cond && (ma->flag&MPLP_REALN)) bam_realn(b, ma->ref); + } while (cond); + return ret; +} + +static void group_smpl(mplp_pileup_t *m, bam_sample_t *sm, kstring_t *buf, + int n, char *const*fn, int *n_plp, const bam_pileup1_t **plp) +{ + int i, j; + memset(m->n_plp, 0, m->n * sizeof(int)); + for (i = 0; i < n; ++i) { + for (j = 0; j < n_plp[i]; ++j) { + const bam_pileup1_t *p = plp[i] + j; + uint8_t *q; + int id = -1; + q = bam_aux_get(p->b, "RG"); + if (q) id = bam_smpl_rg2smid(sm, fn[i], (char*)q+1, buf); + if (id < 0) id = bam_smpl_rg2smid(sm, fn[i], 0, buf); + assert(id >= 0 && id < m->n); + if (m->n_plp[id] == m->m_plp[id]) { + m->m_plp[id] = m->m_plp[id]? m->m_plp[id]<<1 : 8; + m->plp[id] = realloc(m->plp[id], sizeof(bam_pileup1_t) * m->m_plp[id]); + } + m->plp[id][m->n_plp[id]++] = *p; + } + } } static int mpileup(mplp_conf_t *conf, int n, char **fn) @@ -471,16 +528,35 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn) bam_mplp_t iter; bam_header_t *h = 0; char *ref; - // allocate + khash_t(64) *hash = 0; + + bcf_callaux_t *bca = 0; + bcf_callret1_t *bcr = 0; + bcf_call_t bc; + bcf_t *bp = 0; + bcf_hdr_t *bh = 0; + + bam_sample_t *sm = 0; + kstring_t buf; + mplp_pileup_t gplp; + + memset(&gplp, 0, sizeof(mplp_pileup_t)); + memset(&buf, 0, sizeof(kstring_t)); + memset(&bc, 0, sizeof(bcf_call_t)); data = calloc(n, sizeof(void*)); plp = calloc(n, sizeof(void*)); n_plp = calloc(n, sizeof(int*)); + sm = bam_smpl_init(); + // read the header and initialize data for (i = 0; i < n; ++i) { bam_header_t *h_tmp; data[i] = calloc(1, sizeof(mplp_aux_t)); + data[i]->min_mq = conf->min_mq; + data[i]->flag = conf->flag; data[i]->fp = bam_open(fn[i], "r"); h_tmp = bam_header_read(data[i]->fp); + bam_smpl_add(sm, fn[i], h_tmp->text); if (conf->reg) { int beg, end; bam_index_t *idx; @@ -494,7 +570,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn) exit(1); } if (i == 0) beg0 = beg, end0 = end; - data[i]->iter = bam_iterf_query(idx, tid, beg, end); + data[i]->iter = bam_iter_query(idx, tid, beg, end); bam_index_destroy(idx); } if (i == 0) h = h_tmp; @@ -503,40 +579,105 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn) bam_header_destroy(h_tmp); } } - // mpileup + gplp.n = sm->n; + gplp.n_plp = calloc(sm->n, sizeof(int)); + gplp.m_plp = calloc(sm->n, sizeof(int)); + gplp.plp = calloc(sm->n, sizeof(void*)); + + fprintf(stderr, "[%s] %d samples in %d input files\n", __func__, sm->n, n); + if (conf->fn_pos) hash = load_pos(conf->fn_pos, h); + // write the VCF header + if (conf->flag & MPLP_GLF) { + kstring_t s; + bh = calloc(1, sizeof(bcf_hdr_t)); + s.l = s.m = 0; s.s = 0; + bp = bcf_open("-", (conf->flag&MPLP_NO_COMP)? "wu" : "w"); + for (i = 0; i < h->n_targets; ++i) { + kputs(h->target_name[i], &s); + kputc('\0', &s); + } + bh->l_nm = s.l; + bh->name = malloc(s.l); + memcpy(bh->name, s.s, s.l); + s.l = 0; + for (i = 0; i < sm->n; ++i) { + kputs(sm->smpl[i], &s); kputc('\0', &s); + } + bh->l_smpl = s.l; + bh->sname = malloc(s.l); + memcpy(bh->sname, s.s, s.l); + bh->l_txt = 0; + free(s.s); + bcf_hdr_sync(bh); + bcf_hdr_write(bp, bh); + bca = bcf_call_init(-1., conf->min_baseQ); + bcr = calloc(sm->n, sizeof(bcf_callret1_t)); + } ref_tid = -1; ref = 0; iter = bam_mplp_init(n, mplp_func, (void**)data); while (bam_mplp_auto(iter, &tid, &pos, n_plp, plp) > 0) { if (conf->reg && (pos < beg0 || pos >= end0)) continue; // out of the region requested + if (hash) { + khint_t k; + k = kh_get(64, hash, (uint64_t)tid<<32 | pos); + if (k == kh_end(hash)) continue; + } if (tid != ref_tid) { free(ref); if (conf->fai) ref = fai_fetch(conf->fai, h->target_name[tid], &ref_len); + for (i = 0; i < n; ++i) data[i]->ref = ref; ref_tid = tid; } - printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N'); - for (i = 0; i < n; ++i) { - int j; - printf("\t%d\t", n_plp[i]); - if (n_plp[i] == 0) printf("*\t*"); - else { - for (j = 0; j < n_plp[i]; ++j) - pileup_seq(plp[i] + j, pos, ref_len, ref); - putchar('\t'); - for (j = 0; j < n_plp[i]; ++j) { - const bam_pileup1_t *p = plp[i] + j; - int c = bam1_qual(p->b)[p->qpos] + 33; - if (c > 126) c = 126; - putchar(c); + if (conf->flag & MPLP_GLF) { + int _ref0, ref16; + bcf1_t *b = calloc(1, sizeof(bcf1_t)); + group_smpl(&gplp, sm, &buf, n, fn, n_plp, plp); + _ref0 = (ref && pos < ref_len)? ref[pos] : 'N'; + ref16 = bam_nt16_table[_ref0]; + for (i = 0; i < gplp.n; ++i) + bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], ref16, bca, bcr + i); + bcf_call_combine(gplp.n, bcr, ref16, &bc); + bcf_call2bcf(tid, pos, &bc, b); + bcf_write(bp, bh, b); + bcf_destroy(b); + } else { + printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N'); + for (i = 0; i < n; ++i) { + int j; + printf("\t%d\t", n_plp[i]); + if (n_plp[i] == 0) printf("*\t*"); + else { + for (j = 0; j < n_plp[i]; ++j) + pileup_seq(plp[i] + j, pos, ref_len, ref); + putchar('\t'); + for (j = 0; j < n_plp[i]; ++j) { + const bam_pileup1_t *p = plp[i] + j; + int c = bam1_qual(p->b)[p->qpos] + 33; + if (c > 126) c = 126; + putchar(c); + } } } + putchar('\n'); } - putchar('\n'); } + + bcf_close(bp); + bam_smpl_destroy(sm); free(buf.s); + for (i = 0; i < gplp.n; ++i) free(gplp.plp[i]); + free(gplp.plp); free(gplp.n_plp); free(gplp.m_plp); + if (hash) { // free the hash table + khint_t k; + for (k = kh_begin(hash); k < kh_end(hash); ++k) + if (kh_exist(hash, k)) free(kh_val(hash, k)); + kh_destroy(64, hash); + } + bcf_hdr_destroy(bh); bcf_call_destroy(bca); free(bc.PL); free(bcr); bam_mplp_destroy(iter); bam_header_destroy(h); for (i = 0; i < n; ++i) { bam_close(data[i]->fp); - if (data[i]->iter) bam_iterf_destroy(data[i]->iter); + if (data[i]->iter) bam_iter_destroy(data[i]->iter); free(data[i]); } free(data); free(plp); free(ref); free(n_plp); @@ -548,17 +689,43 @@ int bam_mpileup(int argc, char *argv[]) int c; mplp_conf_t mplp; memset(&mplp, 0, sizeof(mplp_conf_t)); - while ((c = getopt(argc, argv, "f:r:")) >= 0) { + mplp.max_mq = 60; + mplp.theta = 1e-3; + mplp.min_baseQ = 13; + while ((c = getopt(argc, argv, "gf:r:l:M:q:t:Q:uaORH")) >= 0) { switch (c) { + case 't': mplp.theta = atof(optarg); break; case 'f': mplp.fai = fai_load(optarg); if (mplp.fai == 0) return 1; break; - case 'r': mplp.reg = strdup(optarg); + case 'r': mplp.reg = strdup(optarg); break; + case 'l': mplp.fn_pos = strdup(optarg); break; + case 'g': mplp.flag |= MPLP_GLF; break; + case 'u': mplp.flag |= MPLP_NO_COMP; break; + case 'a': mplp.flag |= MPLP_NO_ORPHAN | MPLP_REALN | MPLP_NO_HALFTRIM; break; + case 'O': mplp.flag |= MPLP_NO_ORPHAN; break; + case 'H': mplp.flag |= MPLP_NO_HALFTRIM; break; + case 'R': mplp.flag |= MPLP_REALN; break; + case 'M': mplp.max_mq = atoi(optarg); break; + case 'q': mplp.min_mq = atoi(optarg); break; + case 'Q': mplp.min_baseQ = atoi(optarg); break; } } if (argc == 1) { - fprintf(stderr, "Usage: samtools mpileup [-r reg] [-f in.fa] in1.bam [in2.bam [...]]\n"); + fprintf(stderr, "\n"); + fprintf(stderr, "Usage: samtools mpileup [options] in1.bam [in2.bam [...]]\n\n"); + fprintf(stderr, "Options: -f FILE reference sequence file [null]\n"); + fprintf(stderr, " -r STR region in which pileup is generated [null]\n"); + fprintf(stderr, " -l FILE list of positions (format: chr pos) [null]\n"); + fprintf(stderr, " -M INT cap mapping quality at INT [%d]\n", mplp.max_mq); + fprintf(stderr, " -Q INT min base quality [%d]\n", mplp.min_baseQ); + fprintf(stderr, " -q INT filter out alignment with MQ smaller than INT [%d]\n", mplp.min_mq); + fprintf(stderr, " -t FLOAT scaled mutation rate [%lg]\n", mplp.theta); + fprintf(stderr, " -g generate BCF output\n"); + fprintf(stderr, " -u do not compress BCF output\n"); + fprintf(stderr, "\n"); + fprintf(stderr, "Notes: Assuming diploid individuals.\n\n"); return 1; } mpileup(&mplp, argc - optind, argv + optind);