X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_plcmd.c;h=53972de86f62bb717872d1cf6f4b651aac13ba4b;hb=998b354209b48256a4ffa6fa2be118ec7e021081;hp=708e63e7e8b600baf166e23979990711d9bf5829;hpb=7795f54fa8aed1e0b1edff61890e9fe25fdb7fe9;p=samtools.git diff --git a/bam_plcmd.c b/bam_plcmd.c index 708e63e..53972de 100644 --- a/bam_plcmd.c +++ b/bam_plcmd.c @@ -5,7 +5,6 @@ #include "sam.h" #include "faidx.h" #include "bam_maqcns.h" -#include "bam_mcns.h" #include "khash.h" #include "glf.h" #include "kstring.h" @@ -342,12 +341,12 @@ 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->is_soap = 1; // change the default model + 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:PA")) >= 0) { switch (c) { - case 'a': d->c->is_soap = 1; break; - case 'A': d->c->is_soap = 0; 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; @@ -376,6 +375,8 @@ 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"); @@ -449,8 +450,16 @@ int bam_pileup(int argc, char *argv[]) * mpileup * ***********/ +#include +#include "bam2bcf.h" +#include "sample.h" + +#define MPLP_GLF 0x10 +#define MPLP_NO_COMP 0x20 + typedef struct { - int vcf, max_mq, min_mq; + int max_mq, min_mq, flag, min_baseQ; + double theta; char *reg, *fn_pos; faidx_t *fai; kh_64_t *hash; @@ -462,6 +471,12 @@ typedef struct { int min_mq; } 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) { mplp_aux_t *ma = (mplp_aux_t*)data; @@ -472,20 +487,57 @@ static int mplp_func(void *data, bam1_t *b) 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) { mplp_aux_t **data; - mc_aux_t *ma = 0; int i, tid, pos, *n_plp, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid; const bam_pileup1_t **plp; bam_mplp_t iter; bam_header_t *h = 0; char *ref; khash_t(64) *hash = 0; - // allocate + + 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; @@ -493,6 +545,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn) data[i]->min_mq = conf->min_mq; 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; @@ -515,24 +568,40 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn) bam_header_destroy(h_tmp); } } + 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->vcf) { + if (conf->flag & MPLP_GLF) { kstring_t s; + bh = calloc(1, sizeof(bcf_hdr_t)); s.l = s.m = 0; s.s = 0; - kputs("#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT", &s); - for (i = 0; i < n; ++i) { - const char *p; - kputc('\t', &s); - if ((p = strstr(fn[i], ".bam")) != 0) - kputsn(fn[i], p - fn[i], &s); - else kputs(fn[i], &s); + 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); } - puts(s.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)); } - // mpileup - if (conf->vcf) ma = mc_init(n); 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) { @@ -547,42 +616,18 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn) if (conf->fai) ref = fai_fetch(conf->fai, h->target_name[tid], &ref_len); ref_tid = tid; } - if (conf->vcf) { - double f0, f; // the reference allele frequency - int j, _ref, _alt, _ref0, depth, rms_q, _ref0b; - uint64_t sqr_sum; - _ref0 = _ref0b = (ref && pos < ref_len)? ref[pos] : 'N'; - _ref0 = bam_nt16_nt4_table[bam_nt16_table[_ref0]]; - f = f0 = mc_freq0(_ref0, n_plp, plp, ma, &_ref, &_alt); - if (f >= 0.0) { - double flast = f; - for (j = 0; j < 10; ++j) { - f = mc_freq_iter(flast, ma); - if (fabs(f - flast) < 1e-3) break; - flast = f; - } - } - printf("%s\t%d\t.\t%c\t", h->target_name[tid], pos + 1, _ref0b); - if (_ref0 == _ref) putchar("ACGTN"[_alt]); - else printf("%c,%c", "ACGTN"[_ref], "ACGTN"[_alt]); - printf("\t0\t"); // FIXME: currently these not available - if (f0 < 0.) printf("Q13\t"); - else printf(".\t"); - for (i = depth = 0, sqr_sum = 0; i < n; ++i) { - depth += n_plp[i]; - for (j = 0; j < n_plp[i]; ++j) { - int q = plp[i][j].b->core.qual; - if (q > conf->max_mq) q = conf->max_mq; - sqr_sum += q * q; - } - } - rms_q = (int)(sqrt((double)sqr_sum / depth) + .499); - printf("DP=%d;MQ=%d;AF=%.3lg", depth, rms_q, f0<0.?0.:1.-f); - printf("\tDP"); - // output genotype information; FIXME: to be implmented... - for (i = 0; i < n; ++i) - printf("\t%d", n_plp[i]); - putchar('\n'); + 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) { @@ -604,13 +649,18 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn) 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); } - mc_destroy(ma); + 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) { @@ -628,21 +678,37 @@ int bam_mpileup(int argc, char *argv[]) mplp_conf_t mplp; memset(&mplp, 0, sizeof(mplp_conf_t)); mplp.max_mq = 60; - while ((c = getopt(argc, argv, "f:r:l:VM:q:")) >= 0) { + mplp.theta = 1e-3; + mplp.min_baseQ = 13; + while ((c = getopt(argc, argv, "gf:r:l:M:q:t:Q:u")) >= 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); break; case 'l': mplp.fn_pos = strdup(optarg); break; - case 'V': mplp.vcf = 1; break; + case 'g': mplp.flag |= MPLP_GLF; break; + case 'u': mplp.flag |= MPLP_NO_COMP; 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] [-l pos] 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 GLF output\n"); + fprintf(stderr, "\n"); + fprintf(stderr, "Notes: Assuming diploid individuals.\n\n"); return 1; } mpileup(&mplp, argc - optind, argv + optind);