#define BAM_PLF_1STBASE 0x80
#define BAM_PLF_ALLBASE 0x100
#define BAM_PLF_READPOS 0x200
+#define BAM_PLF_NOBAQ 0x400
typedef struct {
bam_header_t *h;
uint32_t format;
int tid, len, last_pos;
int mask;
+ int capQ_thres, min_baseQ;
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
return hash;
}
+static inline int printw(int c, FILE *fp)
+{
+ char buf[16];
+ int l, x;
+ if (c == 0) return fputc('0', fp);
+ for (l = 0, x = c < 0? -c : c; x > 0; x /= 10) buf[l++] = x%10 + '0';
+ if (c < 0) buf[l++] = '-';
+ buf[l] = 0;
+ for (x = 0; x < l/2; ++x) {
+ int y = buf[x]; buf[x] = buf[l-1-x]; buf[l-1-x] = y;
+ }
+ fputs(buf, fp);
+ return 0;
+}
+
// an analogy to pileup_func() below
static int glt3_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
{
return 0;
}
-static void pileup_seq(const bam_pileup1_t *p, int pos, int ref_len, const char *ref)
+static inline 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);
+ int j;
+ if (p->is_head) {
+ putchar('^');
+ putchar(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);
+ int c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
+ if (ref) {
+ int rb = pos < ref_len? ref[pos] : 'N';
+ if (c == '=' || bam_nt16_table[c] == bam_nt16_table[rb]) c = bam1_strand(p->b)? ',' : '.';
+ else c = bam1_strand(p->b)? tolower(c) : toupper(c);
+ } else {
+ if (c == '=') 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(p->is_refskip? (bam1_strand(p->b)? '<' : '>') : '*');
+ if (p->indel > 0) {
+ putchar('+'); printw(p->indel, stdout);
+ for (j = 1; j <= p->indel; ++j) {
+ int 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) {
+ printw(p->indel, stdout);
+ for (j = 1; j <= -p->indel; ++j) {
+ int 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('$');
}
}
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);
+ } else {
+ glf1_t *g = bam_maqcns_glfgen(n, pu, bam_nt16_table[rb], d->c);
+ cns = g->depth == 0? (0xfu<<28 | 0xf<<24) : glf2cns(g, (int)(d->c->q_r + .499));
+ free(g);
+ }
}
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;
}
// 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)) {
- if (!(bam_nt16_table[rb] != 15 && cns>>28 != bam_nt16_table[rb])) { // not a SNP
+ if (!(bam_nt16_table[rb] != 15 && cns>>28 != 15 && cns>>28 != bam_nt16_table[rb])) { // not a SNP
if (!(r && (r->gt == 2 || strcmp(r->s[r->gt], "*")))) { // not an indel
if (r) bam_maqindel_ret_destroy(r);
return 0;
}
}
// print the first 3 columns
- printf("%s\t%d\t%c\t", d->h->target_name[tid], pos + 1, rb);
+ fputs(d->h->target_name[tid], stdout); putchar('\t');
+ printw(pos+1, stdout); putchar('\t'); putchar(rb); putchar('\t');
// print consensus information if required
if (d->format & BAM_PLF_CNS) {
- int ref_q, rb4 = bam_nt16_table[rb];
- ref_q = 0;
- if (rb4 != 15 && cns>>28 != 15 && cns>>28 != rb4) { // a SNP
- ref_q = ((cns>>24&0xf) == rb4)? cns>>8&0xff : (cns>>8&0xff) + (cns&0xff);
- if (ref_q > 255) ref_q = 255;
- }
- rms_mapq = cns>>16&0xff;
- printf("%c\t%d\t%d\t%d\t", bam_nt16_rev_table[cns>>28], cns>>8&0xff, ref_q, rms_mapq);
+ putchar(bam_nt16_rev_table[cns>>28]); putchar('\t');
+ printw(cns>>8&0xff, stdout); putchar('\t');
+ printw(cns&0xff, stdout); putchar('\t');
+ printw(cns>>16&0xff, stdout); putchar('\t');
}
// print pileup sequences
- printf("%d\t", n);
- rms_aux = 0; // we need to recalculate rms_mapq when -c is not flagged on the command line
- for (i = 0; i < n; ++i) {
- 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;
- pileup_seq(p, pos, d->len, d->ref);
- }
+ printw(n, stdout); putchar('\t');
+ for (i = 0; i < n; ++i)
+ pileup_seq(pu + i, pos, d->len, d->ref);
// finalize rms_mapq
- rms_aux = (uint64_t)(sqrt((double)rms_aux / n) + .499);
- if (rms_mapq < 0) rms_mapq = rms_aux;
+ if (d->format & BAM_PLF_CNS) {
+ for (i = rms_aux = 0; i < n; ++i) {
+ 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;
+ }
+ rms_aux = (uint64_t)(sqrt((double)rms_aux / n) + .499);
+ if (rms_mapq < 0) rms_mapq = rms_aux;
+ }
putchar('\t');
// print quality
for (i = 0; i < n; ++i) {
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));
+ printw(x < l/2? x+1 : -((l-1)-x+1), stdout); putchar(',');
}
}
putchar('\n');
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->max_depth = 1024; d->tid = -1; d->mask = BAM_DEF_MASK; d->min_baseQ = 13;
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:PA")) >= 0) {
+ while ((c = getopt(argc, argv, "st:f:cT:N:r:l:d:im:gI:G:vM:S2aR:PAQ:C:B")) >= 0) {
switch (c) {
+ case 'Q': d->c->min_baseQ = atoi(optarg); break;
+ case 'C': d->capQ_thres = atoi(optarg); break;
+ case 'B': d->format |= BAM_PLF_NOBAQ; 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;
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, " -B disable BAQ computation\n");
+ fprintf(stderr, " -A use the original MAQ model for SNP calling (DEPRECATED)\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, " -Q INT min base quality (possibly capped by BAQ) [%d]\n", d->c->min_baseQ);
+ fprintf(stderr, " -C INT coefficient for adjusting mapQ of poor mappings [%d]\n", d->capQ_thres);
+ fprintf(stderr, " -m INT filtering reads with bits in INT [0x%x]\n", d->mask);
fprintf(stderr, " -M INT cap mapping quality at INT [%d]\n", d->c->cap_mapQ);
- fprintf(stderr, " -d INT limit maximum depth for indels [unlimited]\n");
+ fprintf(stderr, " -d INT limit maximum depth for indels [%d]\n", d->max_depth);
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 SOAPsnp consensus sequence\n");
+ fprintf(stderr, " -c compute the 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);
- fprintf(stderr, " -N INT number of haplotypes in the sample (for -c/-g) [%d]\n", d->c->n_hap);
- fprintf(stderr, " -r FLOAT prior of a difference between two haplotypes (for -c/-g) [%f]\n", d->c->het_rate);
- fprintf(stderr, " -G FLOAT prior of an indel between two haplotypes (for -c/-g) [%f]\n", d->ido->r_indel);
- fprintf(stderr, " -I INT phred prob. of an indel in sequencing/prep. (for -c/-g) [%d]\n", d->ido->q_indel);
+ fprintf(stderr, " -g output in the GLFv3 format (DEPRECATED)\n");
+ fprintf(stderr, " -T FLOAT theta in maq consensus calling model (for -c) [%.4g]\n", d->c->theta);
+ fprintf(stderr, " -N INT number of haplotypes in the sample (for -c) [%d]\n", d->c->n_hap);
+ fprintf(stderr, " -r FLOAT prior of a difference between two haplotypes (for -c) [%.4g]\n", d->c->het_rate);
+ fprintf(stderr, " -G FLOAT prior of an indel between two haplotypes (for -c) [%.4g]\n", d->ido->r_indel);
+ fprintf(stderr, " -I INT phred prob. of an indel in sequencing/prep. (for -c) [%d]\n", d->ido->q_indel);
fprintf(stderr, "\n");
free(fn_list); free(fn_fa); free(d);
return 1;
}
d->h = fp->header;
if (fn_pos) d->hash = load_pos(fn_pos, d->h);
- sampileup(fp, d->mask, pileup_func, d);
+ { // run pileup
+ extern int bam_prob_realn(bam1_t *b, const char *ref);
+ extern int bam_cap_mapQ(bam1_t *b, char *ref, int thres);
+ bam1_t *b;
+ int ret, tid, pos, n_plp;
+ bam_plp_t iter;
+ const bam_pileup1_t *plp;
+ b = bam_init1();
+ iter = bam_plp_init(0, 0);
+ bam_plp_set_mask(iter, d->mask);
+ while ((ret = samread(fp, b)) >= 0) {
+ int skip = 0;
+ // update d->ref if necessary
+ if (d->fai && (int)b->core.tid != d->tid) {
+ free(d->ref);
+ d->ref = faidx_fetch_seq(d->fai, d->h->target_name[b->core.tid], 0, 0x7fffffff, &d->len);
+ d->tid = b->core.tid;
+ }
+ if (d->ref && (d->format&BAM_PLF_CNS) && !(d->format&BAM_PLF_NOBAQ)) bam_prob_realn(b, d->ref);
+ if (d->ref && (d->format&BAM_PLF_CNS) && d->capQ_thres > 10) {
+ int q = bam_cap_mapQ(b, d->ref, d->capQ_thres);
+ if (q < 0) skip = 1;
+ else if (b->core.qual > q) b->core.qual = q;
+ } else if (b->core.flag&BAM_FUNMAP) skip = 1;
+ else if ((d->format&BAM_PLF_CNS) && (b->core.flag&1) && !(b->core.flag&2)) skip = 1;
+ if (skip) continue;
+ bam_plp_push(iter, b);
+ while ((plp = bam_plp_next(iter, &tid, &pos, &n_plp)) != 0)
+ pileup_func(tid, pos, n_plp, plp, d);
+ }
+ bam_plp_push(iter, 0);
+ while ((plp = bam_plp_next(iter, &tid, &pos, &n_plp)) != 0)
+ pileup_func(tid, pos, n_plp, plp, d);
+ bam_plp_destroy(iter);
+ bam_destroy1(b);
+ }
samclose(fp); // d->h will be destroyed here
}
#define MPLP_GLF 0x10
#define MPLP_NO_COMP 0x20
+#define MPLP_NO_ORPHAN 0x40
+#define MPLP_REALN 0x80
typedef struct {
- int max_mq, min_mq, flag, min_baseQ;
- double theta;
+ int max_mq, min_mq, flag, min_baseQ, capQ_thres;
char *reg, *fn_pos;
faidx_t *fai;
kh_64_t *hash;
typedef struct {
bamFile fp;
bam_iter_t iter;
- int min_mq;
+ int min_mq, flag, ref_id, capQ_thres;
+ char *ref;
} mplp_aux_t;
typedef struct {
static int mplp_func(void *data, bam1_t *b)
{
+ extern int bam_realn(bam1_t *b, const char *ref);
+ extern int bam_prob_realn(bam1_t *b, const char *ref);
+ extern int bam_cap_mapQ(bam1_t *b, char *ref, int thres);
mplp_aux_t *ma = (mplp_aux_t*)data;
- int ret;
+ int ret, skip = 0;
do {
+ int has_ref = (ma->ref && ma->ref_id == b->core.tid)? 1 : 0;
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);
+ if (ret < 0) break;
+ skip = 0;
+ if (has_ref && (ma->flag&MPLP_REALN)) bam_prob_realn(b, ma->ref);
+ if (has_ref && ma->capQ_thres > 10) {
+ int q = bam_cap_mapQ(b, ma->ref, ma->capQ_thres);
+ if (q < 0) skip = 1;
+ else if (b->core.qual > q) b->core.qual = q;
+ } else if (b->core.flag&BAM_FUNMAP) skip = 1;
+ else if (b->core.qual < ma->min_mq) skip = 1;
+ else if ((ma->flag&MPLP_NO_ORPHAN) && (b->core.flag&1) && !(b->core.flag&2)) skip = 1;
+ } while (skip);
return ret;
}
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");
+ data[i]->flag = conf->flag;
+ data[i]->capQ_thres = conf->capQ_thres;
+ data[i]->fp = strcmp(fn[i], "-") == 0? bam_dopen(fileno(stdin), "r") : 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) {
if (k == kh_end(hash)) continue;
}
if (tid != ref_tid) {
- free(ref);
+ free(ref); ref = 0;
if (conf->fai) ref = fai_fetch(conf->fai, h->target_name[tid], &ref_len);
+ for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid;
ref_tid = tid;
}
if (conf->flag & MPLP_GLF) {
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) {
+ mplp.capQ_thres = 0;
+ mplp.flag = MPLP_NO_ORPHAN | MPLP_REALN;
+ while ((c = getopt(argc, argv, "gf:r:l:M:q:Q:uaORC:B")) >= 0) {
switch (c) {
- case 't': mplp.theta = atof(optarg); break;
case 'f':
mplp.fai = fai_load(optarg);
if (mplp.fai == 0) return 1;
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 'u': mplp.flag |= MPLP_NO_COMP | MPLP_GLF; break;
+ case 'a': mplp.flag |= MPLP_NO_ORPHAN | MPLP_REALN; break;
+ case 'B': mplp.flag &= ~MPLP_REALN & ~MPLP_NO_ORPHAN; break;
+ case 'O': mplp.flag |= MPLP_NO_ORPHAN; break;
+ case 'R': mplp.flag |= MPLP_REALN; break;
+ case 'C': mplp.capQ_thres = atoi(optarg); 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;
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, " -g generate BCF output\n");
+ fprintf(stderr, " -u do not compress BCF output\n");
+ fprintf(stderr, " -B disable BAQ computation\n");
fprintf(stderr, "\n");
fprintf(stderr, "Notes: Assuming diploid individuals.\n\n");
return 1;