* allow to set min base q

 * fixed a bug in mpileup -u

diff --git a/bam2bcf.c b/bam2bcf.c
index d69c771da80ef14ed925e549548f988a766ae680..e64e3357c04d5cd61bc7ecdc3e6af6d0fd859182 100644 (file)
--- a/bam2bcf.c
+++ b/bam2bcf.c
@@ -12,18 +12,19 @@ extern      void ks_introsort_uint32_t(size_t n, uint32_t a[]);
 #define CALL_DEFTHETA 0.85f
 
 struct __bcf_callaux_t {
-       int max_info, capQ;
+       int max_info, capQ, min_baseQ;
        double *fk;
        uint32_t *info;
 };
 
-bcf_callaux_t *bcf_call_init(double theta)
+bcf_callaux_t *bcf_call_init(double theta, int min_baseQ)
 {
        bcf_callaux_t *bca;
        int n;
        if (theta <= 0.) theta = CALL_DEFTHETA;
        bca = calloc(1, sizeof(bcf_callaux_t));
        bca->capQ = 60;
+       bca->min_baseQ = min_baseQ;
        bca->fk = calloc(CALL_MAX, sizeof(double));
        bca->fk[0] = 1.;
        for (n = 1; n < CALL_MAX; ++n)
@@ -89,6 +90,7 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base /*4-bit*/, bcf
                uint32_t q, x = 0, qq;
                if (p->is_del || (p->b->core.flag&BAM_FUNMAP)) continue; // skip unmapped reads and deleted bases
                q = (uint32_t)bam1_qual(p->b)[p->qpos]; // base quality
+               if (q < bca->min_baseQ) continue;
                x |= (uint32_t)bam1_strand(p->b) << 18 | q << 8 | p->b->core.qual;
                if (p->b->core.qual < q) q = p->b->core.qual; // cap the overall quality at mapping quality
                x |= q << 24;
@@ -137,17 +139,9 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base /*4-bit*/, bcf
                        p[j*5+k] = p[k*5+j] = 3.01 * (aux.c[j] + aux.c[k]) + tmp;
                }
        }
-       return 0;
+       return r->depth;
 }
 
-/*
-  1) Find the top 2 bases (from esum[4]).
-
-  2) If the reference base is among the top 2, consider the locus is
-     potentially biallelic and set call->a[2] as -1; otherwise, the
-     locus is potentially triallelic. If the reference is ambiguous,
-     take the weakest call as the pseudo-reference.
- */
 int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/, bcf_call_t *call)
 {
        int ref4, i, j;

diff --git a/bam2bcf.h b/bam2bcf.h
index 896108ac8a1f26bf4c67bb7694fd4f71da4658cd..655174f585f663b54c0e10327df45eddb8942616 100644 (file)
--- a/bam2bcf.h
+++ b/bam2bcf.h
@@ -24,7 +24,7 @@ typedef struct {
 extern "C" {
 #endif
 
-       bcf_callaux_t *bcf_call_init(double theta);
+       bcf_callaux_t *bcf_call_init(double theta, int min_baseQ);
        void bcf_call_destroy(bcf_callaux_t *bca);
        int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base /*4-bit*/, bcf_callaux_t *bca, bcf_callret1_t *r);
        int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/, bcf_call_t *call);

diff --git a/bam_plcmd.c b/bam_plcmd.c
index 8affabfd078f0a18452e65c29d60a39b0071e0ba..2c3178c23874c27750bbaa3c3296d2bace171b4a 100644 (file)
--- a/bam_plcmd.c
+++ b/bam_plcmd.c
@@ -454,11 +454,12 @@ int bam_pileup(int argc, char *argv[])
 #define MPLP_VAR   0x2
 #define MPLP_AFALL 0x8
 #define MPLP_GLF   0x10
+#define MPLP_NO_COMP 0x20
 
 #define MPLP_AFS_BLOCK 0x10000
 
 typedef struct {
-       int max_mq, min_mq, prior_type, flag;
+       int max_mq, min_mq, prior_type, flag, min_baseQ;
        double theta;
        char *reg, *fn_pos;
        faidx_t *fai;
@@ -538,7 +539,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
        if (conf->flag & MPLP_GLF) {
                kstring_t s;
                s.l = s.m = 0; s.s = 0;
-               bp = bcf_open("-", "w");
+               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);
@@ -583,7 +584,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
        }
        // mpileup
        if (conf->flag & MPLP_GLF) {
-               bca = bcf_call_init(-1.);
+               bca = bcf_call_init(-1., conf->min_baseQ);
                bcr = calloc(n, sizeof(bcf_callret1_t));
        } else if (conf->flag & MPLP_VCF) {
                ma = mc_init(n);
@@ -716,7 +717,8 @@ int bam_mpileup(int argc, char *argv[])
        mplp.max_mq = 60;
        mplp.prior_type = MC_PTYPE_FULL;
        mplp.theta = 1e-3;
-       while ((c = getopt(argc, argv, "gvVcFSP:f:r:l:VM:q:t:")) >= 0) {
+       mplp.min_baseQ = 13;
+       while ((c = getopt(argc, argv, "gvVcFSP:f:r:l:VM:q:t:Q:u")) >= 0) {
                switch (c) {
                case 't': mplp.theta = atof(optarg); break;
                case 'P':
@@ -739,8 +741,10 @@ int bam_mpileup(int argc, char *argv[])
                case 'c': mplp.flag |= MPLP_VCF; break;
                case 'F': mplp.flag |= MPLP_AFALL; break;
                case 'v': mplp.flag |= MPLP_VAR; 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 (mplp.flag&MPLP_GLF) mplp.flag &= ~MPLP_VCF;
@@ -754,6 +758,7 @@ int bam_mpileup(int argc, char *argv[])
                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, "         -P STR      prior: full, flat, cond2 [full]\n");
+               fprintf(stderr, "         -Q INT      min base quality [%d]\n", mplp.min_baseQ);
                fprintf(stderr, "         -c          generate VCF output (consensus calling)\n");
                fprintf(stderr, "         -g          generate GLF output\n");
                fprintf(stderr, "         -v          show variant sites only\n");

diff --git a/bcftools/bcf.c b/bcftools/bcf.c
index f8b55c2cf02161828e4faa35665ca12d986978cc..e40631975af4351436efdc938c01dca054567c85 100644 (file)
--- a/bcftools/bcf.c
+++ b/bcftools/bcf.c
@@ -17,9 +17,9 @@ bcf_t *bcf_open(const char *fn, const char *mode)
        bcf_t *b;
        b = calloc(1, sizeof(bcf_t));
        if (strchr(mode, 'w')) {
-               b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdout), "w");
+               b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdout), mode);
        } else {
-               b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), "r");
+               b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), mode);
        }
        b->fp->owned_file = 1;
        if (strchr(mode, 'r')) bcf_hdr_read(b);