#include "sam.h"
#include "faidx.h"
#include "bam_maqcns.h"
+#include "bam2bcf.h"
#include "khash.h"
#include "glf.h"
#include "kstring.h"
#define BAM_PLF_INDEL_ONLY 0x04
#define BAM_PLF_GLF 0x08
#define BAM_PLF_VAR_ONLY 0x10
+#define BAM_PLF_2ND 0x20
+#define BAM_PLF_RANBASE 0x40
+#define BAM_PLF_1STBASE 0x80
+#define BAM_PLF_ALLBASE 0x100
+#define BAM_PLF_READPOS 0x200
typedef struct {
bam_header_t *h;
uint32_t format;
int tid, len, last_pos;
int mask;
+ int max_depth; // for indel calling, ignore reads with the depth too high. 0 for unlimited
char *ref;
glfFile fp_glf; // for glf output only
} pu_data_t;
g3->offset = pos - d->last_pos;
d->last_pos = pos;
glf3_write1(d->fp_glf, g3);
- if (proposed_indels)
- r = bam_maqindel(n, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
- else r = bam_maqindel(n, pos, d->ido, pu, d->ref, 0, 0);
+ if (pos < d->len) {
+ int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
+ if (proposed_indels)
+ r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+ else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
+ }
if (r) { // then write indel line
int het = 3 * n, min;
min = het;
return 0;
}
+static void pileup_seq(const bam_pileup1_t *p, int pos, int ref_len, const char *ref)
+{
+ if (p->is_head) printf("^%c", p->b->core.qual > 93? 126 : p->b->core.qual + 33);
+ if (!p->is_del) {
+ int j, rb, c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
+ rb = (ref && pos < ref_len)? ref[pos] : 'N';
+ if (c == '=' || toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
+ else c = bam1_strand(p->b)? tolower(c) : toupper(c);
+ putchar(c);
+ if (p->indel > 0) {
+ printf("+%d", p->indel);
+ for (j = 1; j <= p->indel; ++j) {
+ c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + j)];
+ putchar(bam1_strand(p->b)? tolower(c) : toupper(c));
+ }
+ } else if (p->indel < 0) {
+ printf("%d", p->indel);
+ for (j = 1; j <= -p->indel; ++j) {
+ c = (ref && (int)pos+j < ref_len)? ref[pos+j] : 'N';
+ putchar(bam1_strand(p->b)? tolower(c) : toupper(c));
+ }
+ }
+ } else putchar('*');
+ if (p->is_tail) putchar('$');
+}
+
static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
{
pu_data_t *d = (pu_data_t*)data;
bam_maqindel_ret_t *r = 0;
- int i, j, rb, rms_mapq = -1, *proposed_indels = 0;
+ int i, rb, rms_mapq = -1, *proposed_indels = 0;
uint64_t rms_aux;
uint32_t cns = 0;
// update d->ref if necessary
if (d->fai && (int)tid != d->tid) {
free(d->ref);
- d->ref = fai_fetch(d->fai, d->h->target_name[tid], &d->len);
+ d->ref = faidx_fetch_seq(d->fai, d->h->target_name[tid], 0, 0x7fffffff, &d->len);
d->tid = tid;
}
rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
if (i == n) return 0;
}
// call the consensus and indel
- if (d->format & BAM_PLF_CNS) // call consensus
- cns = bam_maqcns_call(n, pu, d->c);
- if ((d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) && d->ref) { // call indels
- if (proposed_indels) // the first element gives the size of the array
- r = bam_maqindel(n, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
- else r = bam_maqindel(n, pos, d->ido, pu, d->ref, 0, 0);
+ if (d->format & BAM_PLF_CNS) { // call consensus
+ if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE)) { // use a random base or the 1st base as the consensus call
+ const bam_pileup1_t *p = (d->format & BAM_PLF_1STBASE)? pu : pu + (int)(drand48() * n);
+ int q = bam1_qual(p->b)[p->qpos];
+ int mapQ = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
+ uint32_t b = bam1_seqi(bam1_seq(p->b), p->qpos);
+ cns = b<<28 | 0xf<<24 | mapQ<<16 | q<<8;
+ } else if (d->format & BAM_PLF_ALLBASE) { // collapse all bases
+ uint64_t rmsQ = 0;
+ uint32_t b = 0;
+ for (i = 0; i < n; ++i) {
+ const bam_pileup1_t *p = pu + i;
+ int q = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
+ b |= bam1_seqi(bam1_seq(p->b), p->qpos);
+ rmsQ += q * q;
+ }
+ rmsQ = (uint64_t)(sqrt((double)rmsQ / n) + .499);
+ cns = b<<28 | 0xf<<24 | rmsQ<<16 | 60<<8;
+ } else cns = bam_maqcns_call(n, pu, d->c);
+ }
+ if ((d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) && d->ref && pos < d->len) { // call indels
+ int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
+ if (proposed_indels) // the first element gives the size of the array
+ r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+ else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
}
// when only variant sites are asked for, test if the site is a variant
if ((d->format & BAM_PLF_CNS) && (d->format & BAM_PLF_VAR_ONLY)) {
const bam_pileup1_t *p = pu + i;
int tmp = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
rms_aux += tmp * tmp;
- if (p->is_head) printf("^%c", p->b->core.qual > 93? 126 : p->b->core.qual + 33);
- if (!p->is_del) {
- int c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
- if (c == '=' || toupper(c) == toupper(rb)) c = bam1_strand(p->b)? ',' : '.';
- else c = bam1_strand(p->b)? tolower(c) : toupper(c);
- putchar(c);
- if (p->indel > 0) {
- printf("+%d", p->indel);
- for (j = 1; j <= p->indel; ++j) {
- c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + j)];
- putchar(bam1_strand(p->b)? tolower(c) : toupper(c));
- }
- } else if (p->indel < 0) {
- printf("%d", p->indel);
- for (j = 1; j <= -p->indel; ++j) {
- c = (d->ref && (int)pos+j < d->len)? d->ref[pos+j] : 'N';
- putchar(bam1_strand(p->b)? tolower(c) : toupper(c));
- }
- }
- } else putchar('*');
- if (p->is_tail) putchar('$');
+ pileup_seq(p, pos, d->len, d->ref);
}
// finalize rms_mapq
rms_aux = (uint64_t)(sqrt((double)rms_aux / n) + .499);
if (c > 126) c = 126;
putchar(c);
}
+ if (d->format & BAM_PLF_2ND) { // print 2nd calls and qualities
+ const unsigned char *q;
+ putchar('\t');
+ for (i = 0; i < n; ++i) {
+ const bam_pileup1_t *p = pu + i;
+ q = bam_aux_get(p->b, "E2");
+ putchar(q? q[p->qpos + 1] : 'N');
+ }
+ putchar('\t');
+ for (i = 0; i < n; ++i) {
+ const bam_pileup1_t *p = pu + i;
+ q = bam_aux_get(p->b, "U2");
+ putchar(q? q[p->qpos + 1] : '!');
+ }
+ }
// print mapping quality if -s is flagged on the command line
if (d->format & BAM_PLF_SIMPLE) {
putchar('\t');
putchar(c);
}
}
+ // print read position
+ if (d->format & BAM_PLF_READPOS) {
+ putchar('\t');
+ for (i = 0; i < n; ++i) {
+ int x = pu[i].qpos;
+ int l = pu[i].b->core.l_qseq;
+ printf("%d,", x < l/2? x+1 : -((l-1)-x+1));
+ }
+ }
putchar('\n');
// print the indel line if r has been calculated. This only happens if:
// a) -c or -i are flagged, AND b) the reference sequence is available
printf("%d\t%d\t", rms_mapq, n);
printf("%s\t%s\t", r->s[0], r->s[1]);
//printf("%d\t%d\t", r->gl[0], r->gl[1]);
- printf("%d\t%d\t%d\n", r->cnt1, r->cnt2, r->cnt_anti);
+ printf("%d\t%d\t%d\t", r->cnt1, r->cnt2, r->cnt_anti);
+ printf("%d\t%d\n", r->cnt_ref, r->cnt_ambi);
bam_maqindel_ret_destroy(r);
}
return 0;
int c, is_SAM = 0;
char *fn_list = 0, *fn_fa = 0, *fn_pos = 0;
pu_data_t *d = (pu_data_t*)calloc(1, sizeof(pu_data_t));
+ 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:im:gI:G:vM:S")) >= 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->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;
case 'i': d->format |= BAM_PLF_INDEL_ONLY; break;
case 'v': d->format |= BAM_PLF_VAR_ONLY; break;
case 'm': d->mask = strtol(optarg, 0, 0); break;
case 'g': d->format |= BAM_PLF_GLF; break;
+ case '2': d->format |= BAM_PLF_2ND; break;
+ case 'P': d->format |= BAM_PLF_READPOS; break;
case 'I': d->ido->q_indel = atoi(optarg); break;
case 'G': d->ido->r_indel = atof(optarg); break;
case 'S': is_SAM = 1; break;
+ case 'R':
+ if (strcmp(optarg, "random") == 0) d->format |= BAM_PLF_RANBASE;
+ else if (strcmp(optarg, "first") == 0) d->format |= BAM_PLF_1STBASE;
+ else if (strcmp(optarg, "all") == 0) d->format |= BAM_PLF_ALLBASE;
+ else fprintf(stderr, "[bam_pileup] unrecognized -R\n");
+ break;
default: fprintf(stderr, "Unrecognizd option '-%c'.\n", c); return 1;
}
}
fprintf(stderr, "Usage: samtools pileup [options] <in.bam>|<in.sam>\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 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);
fprintf(stderr, " -M INT cap mapping quality at INT [%d]\n", d->c->cap_mapQ);
- fprintf(stderr, " -t FILE list of reference sequences (assume the input is in SAM)\n");
+ fprintf(stderr, " -d INT limit maximum depth for indels [unlimited]\n");
+ 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);
free(fn_list); free(fn_fa); free(d);
return 1;
}
+ if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE|BAM_PLF_ALLBASE)) d->format |= BAM_PLF_CNS;
if (fn_fa) d->fai = fai_load(fn_fa);
if (d->format & (BAM_PLF_CNS|BAM_PLF_GLF)) bam_maqcns_prepare(d->c); // consensus calling
if (d->format & BAM_PLF_GLF) { // for glf output
}
if (d->fai == 0 && (d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)))
fprintf(stderr, "[bam_pileup] indels will not be called when -f is absent.\n");
+ if (fn_fa && is_SAM && fn_list == 0) fn_list = samfaipath(fn_fa);
+
{
samfile_t *fp;
fp = is_SAM? samopen(argv[optind], "r", fn_list) : samopen(argv[optind], "rb", 0);
free(d->ido); free(d->ref); free(d);
return 0;
}
+
+/***********
+ * mpileup *
+ ***********/
+
+#define MPLP_GLF 0x10
+#define MPLP_NO_COMP 0x20
+
+typedef struct {
+ 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_iter_t iter;
+ int min_mq;
+} mplp_aux_t;
+
+static int mplp_func(void *data, bam1_t *b)
+{
+ mplp_aux_t *ma = (mplp_aux_t*)data;
+ int ret;
+ do {
+ ret = ma->iter? bam_iter_read(ma->fp, ma->iter, b) : bam_read1(ma->fp, b);
+ } while (b->core.qual < ma->min_mq && ret >= 0);
+ return ret;
+}
+
+static int mpileup(mplp_conf_t *conf, int n, char **fn)
+{
+ mplp_aux_t **data;
+ 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;
+
+ bcf_callaux_t *bca = 0;
+ bcf_callret1_t *bcr = 0;
+ bcf_call_t bc;
+ bcf_t *bp = 0;
+ bcf_hdr_t *bh = 0;
+
+ memset(&bc, 0, sizeof(bcf_call_t));
+ data = calloc(n, sizeof(void*));
+ plp = calloc(n, sizeof(void*));
+ n_plp = calloc(n, sizeof(int*));
+
+ // 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]->fp = bam_open(fn[i], "r");
+ h_tmp = bam_header_read(data[i]->fp);
+ if (conf->reg) {
+ int beg, end;
+ bam_index_t *idx;
+ idx = bam_index_load(fn[i]);
+ if (idx == 0) {
+ fprintf(stderr, "[%s] fail to load index for %d-th input.\n", __func__, i+1);
+ exit(1);
+ }
+ if (bam_parse_region(h_tmp, conf->reg, &tid, &beg, &end) < 0) {
+ fprintf(stderr, "[%s] malformatted region or wrong seqname for %d-th input.\n", __func__, i+1);
+ exit(1);
+ }
+ if (i == 0) beg0 = beg, end0 = end;
+ data[i]->iter = bam_iter_query(idx, tid, beg, end);
+ bam_index_destroy(idx);
+ }
+ if (i == 0) h = h_tmp;
+ else {
+ // FIXME: to check consistency
+ bam_header_destroy(h_tmp);
+ }
+ }
+ 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 < n; ++i) {
+ const char *p;
+ if ((p = strstr(fn[i], ".bam")) != 0)
+ kputsn(fn[i], p - fn[i], &s);
+ else kputs(fn[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);
+ }
+ // mpileup
+ if (conf->flag & MPLP_GLF) {
+ bca = bcf_call_init(-1., conf->min_baseQ);
+ bcr = calloc(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);
+ ref_tid = tid;
+ }
+ if (conf->flag & MPLP_GLF) {
+ int _ref0, ref16;
+ bcf1_t *b = calloc(1, sizeof(bcf1_t));
+ _ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
+ ref16 = bam_nt16_table[_ref0];
+ for (i = 0; i < n; ++i)
+ bcf_call_glfgen(n_plp[i], plp[i], ref16, bca, bcr + i);
+ bcf_call_combine(n, bcr, ref16, &bc);
+ bcf_call2bcf(tid, pos, &bc, b);
+ bcf_write(bp, bh, b);
+ //fprintf(stderr, "%d,%d,%d\n", b->tid, b->pos, b->l_str);
+ 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');
+ }
+ }
+ bcf_close(bp);
+ 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_iter_destroy(data[i]->iter);
+ free(data[i]);
+ }
+ free(data); free(plp); free(ref); free(n_plp);
+ return 0;
+}
+
+int bam_mpileup(int argc, char *argv[])
+{
+ int c;
+ mplp_conf_t mplp;
+ memset(&mplp, 0, sizeof(mplp_conf_t));
+ mplp.max_mq = 60;
+ 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 '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, "\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);
+ free(mplp.reg);
+ if (mplp.fai) fai_destroy(mplp.fai);
+ return 0;
+}
projects / samtools.git / blobdiff