From 6cda28b3a4c6afcf8c739a478b6bcf1e514f96e9 Mon Sep 17 00:00:00 2001 From: Heng Li Date: Thu, 16 Sep 2010 17:45:49 +0000 Subject: [PATCH] * samtools-0.1.8-14 (r727) * allow to change the capQ parameter at the command line --- bam_md.c | 12 ++++++------ bam_plcmd.c | 17 +++++++++-------- bamtk.c | 2 +- 3 files changed, 16 insertions(+), 15 deletions(-) diff --git a/bam_md.c b/bam_md.c index eee767e..d2c84dc 100644 --- a/bam_md.c +++ b/bam_md.c @@ -110,14 +110,14 @@ void bam_fillmd1(bam1_t *b, char *ref, int is_equal) bam_fillmd1_core(b, ref, is_equal, 0); } -int bam_cap_mapQ(bam1_t *b, char *ref) +int bam_cap_mapQ(bam1_t *b, char *ref, int thres) { uint8_t *seq = bam1_seq(b), *qual = bam1_qual(b); uint32_t *cigar = bam1_cigar(b); bam1_core_t *c = &b->core; int i, x, y, mm, q, len, clip_l, clip_q; double t; - + if (thres < 0) thres = 40; // set the default mm = q = len = clip_l = clip_q = 0; for (i = y = 0, x = c->pos; i < c->n_cigar; ++i) { int j, l = cigar[i]>>4, op = cigar[i]&0xf; @@ -154,10 +154,10 @@ int bam_cap_mapQ(bam1_t *b, char *ref) for (i = 0, t = 1; i < mm; ++i) t *= (double)len / (i+1); t = q - 4.343 * log(t) + clip_q / 5.; - if (t > 40) return -1; + if (t > thres) return -1; if (t < 0) t = 0; - t = sqrt((40 - t) / 40) * 40; - fprintf(stderr, "%s %lf %d\n", bam1_qname(b), t, q); + t = sqrt((thres - t) / thres) * thres; +// fprintf(stderr, "%s %lf %d\n", bam1_qname(b), t, q); return (int)(t + .499); } @@ -320,7 +320,7 @@ int bam_fillmd(int argc, char *argv[]) } if (is_realn) bam_realn(b, ref); if (is_capQ) { - int q = bam_cap_mapQ(b, ref); + int q = bam_cap_mapQ(b, ref, 40); if (b->core.qual > q) b->core.qual = q; } if (ref) bam_fillmd1_core(b, ref, is_equal, max_nm); diff --git a/bam_plcmd.c b/bam_plcmd.c index e9b3507..20c5632 100644 --- a/bam_plcmd.c +++ b/bam_plcmd.c @@ -458,10 +458,9 @@ int bam_pileup(int argc, char *argv[]) #define MPLP_NO_COMP 0x20 #define MPLP_NO_ORPHAN 0x40 #define MPLP_REALN 0x80 -#define MPLP_CAPQ 0x100 typedef struct { - int max_mq, min_mq, flag, min_baseQ; + int max_mq, min_mq, flag, min_baseQ, capQ_thres; double theta; char *reg, *fn_pos; faidx_t *fai; @@ -471,7 +470,7 @@ typedef struct { typedef struct { bamFile fp; bam_iter_t iter; - int min_mq, flag, ref_id; + int min_mq, flag, ref_id, capQ_thres; char *ref; } mplp_aux_t; @@ -484,7 +483,7 @@ typedef struct { static int mplp_func(void *data, bam1_t *b) { extern int bam_realn(bam1_t *b, const char *ref); - extern int bam_cap_mapQ(bam1_t *b, char *ref); + extern int bam_cap_mapQ(bam1_t *b, char *ref, int thres); mplp_aux_t *ma = (mplp_aux_t*)data; int ret, skip = 0; do { @@ -493,8 +492,8 @@ static int mplp_func(void *data, bam1_t *b) if (ret < 0) break; skip = 0; if (has_ref && (ma->flag&MPLP_REALN)) bam_realn(b, ma->ref); - if ((ma->flag&MPLP_CAPQ) && has_ref) { - int q = bam_cap_mapQ(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; @@ -561,6 +560,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn) data[i] = calloc(1, sizeof(mplp_aux_t)); data[i]->min_mq = conf->min_mq; data[i]->flag = conf->flag; + data[i]->capQ_thres = conf->capQ_thres; 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); @@ -699,7 +699,8 @@ int bam_mpileup(int argc, char *argv[]) mplp.max_mq = 60; mplp.theta = 1e-3; mplp.min_baseQ = 13; - while ((c = getopt(argc, argv, "gf:r:l:M:q:t:Q:uaORC")) >= 0) { + mplp.capQ_thres = 0; + while ((c = getopt(argc, argv, "gf:r:l:M:q:t:Q:uaORC:")) >= 0) { switch (c) { case 't': mplp.theta = atof(optarg); break; case 'f': @@ -712,8 +713,8 @@ int bam_mpileup(int argc, char *argv[]) case 'u': mplp.flag |= MPLP_NO_COMP; break; case 'a': mplp.flag |= MPLP_NO_ORPHAN | MPLP_REALN; break; case 'O': mplp.flag |= MPLP_NO_ORPHAN; break; - case 'C': mplp.flag |= MPLP_CAPQ; 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; diff --git a/bamtk.c b/bamtk.c index ce4ebd5..0ee62bc 100644 --- a/bamtk.c +++ b/bamtk.c @@ -9,7 +9,7 @@ #endif #ifndef PACKAGE_VERSION -#define PACKAGE_VERSION "0.1.8-13 (r715)" +#define PACKAGE_VERSION "0.1.8-14 (r727)" #endif int bam_taf2baf(int argc, char *argv[]); -- 2.39.2