* samtools-0.1.12-2 (r877)

 * allow to fine control the selection of indel candidates. The current
   setting is okay for lowCov and highCov with ~100 samples, but it
   skips too many indels for highCov with >250 samples.

diff --git a/bam2bcf.c b/bam2bcf.c
index 088635c0b04a549b762d1a3b28efad82f4bb5c80..08120fc46c2a3902e01582b45de0659d4b333eb6 100644 (file)
--- a/bam2bcf.c
+++ b/bam2bcf.c
@@ -23,6 +23,8 @@ bcf_callaux_t *bcf_call_init(double theta, int min_baseQ)
        bca->openQ = 40; bca->extQ = 20; bca->tandemQ = 100;
        bca->min_baseQ = min_baseQ;
        bca->e = errmod_init(1. - theta);
+       bca->min_frac = 0.002;
+       bca->min_support = 1;
        return bca;
 }
 

diff --git a/bam2bcf.h b/bam2bcf.h
index 26b022c08e89e3597bd6e400690d42bd7ecabe38..958567241936a4c54feae501c725ad77a6e3f1be 100644 (file)
--- a/bam2bcf.h
+++ b/bam2bcf.h
@@ -9,7 +9,9 @@
 
 typedef struct __bcf_callaux_t {
        int capQ, min_baseQ;
-       int openQ, extQ, tandemQ;
+       int openQ, extQ, tandemQ; // for indels
+       int min_support; // for collecting indel candidates
+       double min_frac; // for collecting indel candidates
        // for internal uses
        int max_bases;
        int indel_types[4];

diff --git a/bam2bcf_indel.c b/bam2bcf_indel.c
index 16241d0c0399fb7e727951f2d8f64d86798e25d1..239fb8d731040b4c2eea951040c2eaf8334c5c7e 100644 (file)
--- a/bam2bcf_indel.c
+++ b/bam2bcf_indel.c
@@ -11,7 +11,6 @@ KHASH_SET_INIT_STR(rg)
 
 #define MINUS_CONST 0x10000000
 #define INDEL_WINDOW_SIZE 50
-#define MIN_SUPPORT_COEF 500
 
 void *bcf_call_add_rg(void *_hash, const char *hdtext, const char *list)
 {
@@ -165,7 +164,7 @@ int bcf_call_gap_prep(int n, int *n_plp, bam_pileup1_t **plp, int pos, bcf_calla
                // squeeze out identical types
                for (i = 1, n_types = 1; i < m; ++i)
                        if (aux[i] != aux[i-1]) ++n_types;
-               if (n_types == 1 || n_alt * MIN_SUPPORT_COEF < n_tot) { // no indels or too few supporting reads
+               if (n_types == 1 || (double)n_alt / n_tot < bca->min_frac || n_alt < bca->min_support) { // then skip
                        free(aux); return -1;
                }
                types = (int*)calloc(n_types, sizeof(int));

diff --git a/bam_plcmd.c b/bam_plcmd.c
index b562b94c7a2f150bd26b3b19b6343c1d8b09dbbc..1aa93614f35db8891e7bbe1fd955e1100efdbed0 100644 (file)
--- a/bam_plcmd.c
+++ b/bam_plcmd.c
@@ -536,7 +536,8 @@ int bam_pileup(int argc, char *argv[])
 
 typedef struct {
        int max_mq, min_mq, flag, min_baseQ, capQ_thres, max_depth;
-       int openQ, extQ, tandemQ;
+       int openQ, extQ, tandemQ, min_support; // for indels
+       double min_frac; // for indels
        char *reg, *fn_pos, *pl_list;
        faidx_t *fai;
        kh_64_t *hash;
@@ -702,6 +703,8 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
                bcr = calloc(sm->n, sizeof(bcf_callret1_t));
                bca->rghash = rghash;
                bca->openQ = conf->openQ, bca->extQ = conf->extQ, bca->tandemQ = conf->tandemQ;
+               bca->min_frac = conf->min_frac;
+               bca->min_support = conf->min_support;
        }
        ref_tid = -1; ref = 0;
        iter = bam_mplp_init(n, mplp_func, (void**)data);
@@ -858,8 +861,9 @@ int bam_mpileup(int argc, char *argv[])
        mplp.capQ_thres = 0;
        mplp.max_depth = 250;
        mplp.openQ = 40; mplp.extQ = 20; mplp.tandemQ = 100;
+       mplp.min_frac = 0.002; mplp.min_support = 1;
        mplp.flag = MPLP_NO_ORPHAN | MPLP_REALN;
-       while ((c = getopt(argc, argv, "Agf:r:l:M:q:Q:uaRC:BDSd:b:P:o:e:h:I")) >= 0) {
+       while ((c = getopt(argc, argv, "Agf:r:l:M:q:Q:uaRC:BDSd:b:P:o:e:h:Im:F:")) >= 0) {
                switch (c) {
                case 'f':
                        mplp.fai = fai_load(optarg);
@@ -886,6 +890,8 @@ int bam_mpileup(int argc, char *argv[])
                case 'e': mplp.extQ = atoi(optarg); break;
                case 'h': mplp.tandemQ = atoi(optarg); break;
                case 'A': use_orphan = 1; break;
+               case 'F': mplp.min_frac = atof(optarg); break;
+               case 'm': mplp.min_support = atoi(optarg); break;
                }
        }
        if (use_orphan) mplp.flag &= ~MPLP_NO_ORPHAN;
@@ -904,6 +910,8 @@ int bam_mpileup(int argc, char *argv[])
                fprintf(stderr, "         -o INT      Phred-scaled gap open sequencing error probability [%d]\n", mplp.openQ);
                fprintf(stderr, "         -e INT      Phred-scaled gap extension seq error probability [%d]\n", mplp.extQ);
                fprintf(stderr, "         -h INT      coefficient for homopolyer errors [%d]\n", mplp.tandemQ);
+               fprintf(stderr, "         -m INT      minimum gapped reads for indel candidates [%d]\n", mplp.min_support);
+               fprintf(stderr, "         -F FLOAT    minimum fraction of gapped reads for candidates [%g]\n", mplp.min_frac);
                fprintf(stderr, "         -A          use anomalous read pairs in SNP/INDEL calling\n");
                fprintf(stderr, "         -g          generate BCF output\n");
                fprintf(stderr, "         -u          do not compress BCF output\n");

diff --git a/bamtk.c b/bamtk.c
index a5a0da385224cdb79c20072dd2da3e5328f91af3..9dbdfdeb5c19d50bb78e10501929ebaa2ffd4aea 100644 (file)
--- a/bamtk.c
+++ b/bamtk.c
@@ -9,7 +9,7 @@
 #endif
 
 #ifndef PACKAGE_VERSION
-#define PACKAGE_VERSION "0.1.12-1 (r873)"
+#define PACKAGE_VERSION "0.1.12-2 (r877)"
 #endif
 
 int bam_taf2baf(int argc, char *argv[]);