bam2depth: Enabled reading the list of BAMs from a file list

[samtools.git] / bam2depth.c

diff --git a/bam2depth.c b/bam2depth.c
index ca36b89b6ec2a7ff0339c67dd9d4f2ba6e828b0f..02311efc35db9155a5fc4d3174ad5e38506bea82 100644 (file)
--- a/bam2depth.c
+++ b/bam2depth.c
@@ -13,7 +13,7 @@
 typedef struct {     // auxiliary data structure
        bamFile fp;      // the file handler
        bam_iter_t iter; // NULL if a region not specified
-       int min_mapQ;    // mapQ filter
+       int min_mapQ, min_len; // mapQ filter; length filter
 } aux_t;
 
 void *bed_read(const char *fn); // read a BED or position list file
@@ -25,40 +25,65 @@ static int read_bam(void *data, bam1_t *b) // read level filters better go here
 {
        aux_t *aux = (aux_t*)data; // data in fact is a pointer to an auxiliary structure
        int ret = aux->iter? bam_iter_read(aux->fp, aux->iter, b) : bam_read1(aux->fp, b);
-       if ((int)b->core.qual < aux->min_mapQ) b->core.flag |= BAM_FUNMAP;
+       if (!(b->core.flag&BAM_FUNMAP)) {
+               if ((int)b->core.qual < aux->min_mapQ) b->core.flag |= BAM_FUNMAP;
+               else if (aux->min_len && bam_cigar2qlen(&b->core, bam1_cigar(b)) < aux->min_len) b->core.flag |= BAM_FUNMAP;
+       }
        return ret;
 }
 
+int read_file_list(const char *file_list,int *n,char **argv[]);
+
 #ifdef _MAIN_BAM2DEPTH
 int main(int argc, char *argv[])
 #else
 int main_depth(int argc, char *argv[])
 #endif
 {
-       int i, n, tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0;
+       int i, n, tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0, min_len = 0, nfiles;
        const bam_pileup1_t **plp;
        char *reg = 0; // specified region
        void *bed = 0; // BED data structure
+    char *file_list = NULL, **fn = NULL;
        bam_header_t *h = 0; // BAM header of the 1st input
        aux_t **data;
        bam_mplp_t mplp;
 
        // parse the command line
-       while ((n = getopt(argc, argv, "r:b:q:Q:")) >= 0) {
+       while ((n = getopt(argc, argv, "r:b:q:Q:l:f:")) >= 0) {
                switch (n) {
+                       case 'l': min_len = atoi(optarg); break; // minimum query length
                        case 'r': reg = strdup(optarg); break;   // parsing a region requires a BAM header
                        case 'b': bed = bed_read(optarg); break; // BED or position list file can be parsed now
                        case 'q': baseQ = atoi(optarg); break;   // base quality threshold
                        case 'Q': mapQ = atoi(optarg); break;    // mapping quality threshold
+                       case 'f': file_list = optarg; break;
                }
        }
-       if (optind == argc) {
-               fprintf(stderr, "Usage: bam2depth [-r reg] [-q baseQthres] [-Q mapQthres] [-b in.bed] <in1.bam> [...]\n");
+       if (optind == argc && !file_list) {
+        fprintf(stderr, "\n");
+        fprintf(stderr, "Usage: samtools depth [options] in1.bam [in2.bam [...]]\n");
+        fprintf(stderr, "Options:\n");
+        fprintf(stderr, "   -b <bed>            list of positions or regions\n");
+        fprintf(stderr, "   -f <list>           list of input BAM filenames, one per line [null]\n");
+        fprintf(stderr, "   -l <int>            minQLen\n");
+        fprintf(stderr, "   -q <int>            base quality threshold\n");
+        fprintf(stderr, "   -Q <int>            mapping quality threshold\n");
+        fprintf(stderr, "   -r <chr:from-to>    region\n");
+        fprintf(stderr, "\n");
                return 1;
        }
 
        // initialize the auxiliary data structures
-       n = argc - optind; // the number of BAMs on the command line
+    if (file_list) 
+    {
+        if ( read_file_list(file_list,&nfiles,&fn) ) return 1;
+        n = nfiles;
+        argv = fn;
+        optind = 0;
+    }
+    else
+        n = argc - optind; // the number of BAMs on the command line
        data = calloc(n, sizeof(void*)); // data[i] for the i-th input
        beg = 0; end = 1<<30; tid = -1;  // set the default region
        for (i = 0; i < n; ++i) {
@@ -66,6 +91,7 @@ int main_depth(int argc, char *argv[])
                data[i] = calloc(1, sizeof(aux_t));
                data[i]->fp = bam_open(argv[optind+i], "r"); // open BAM
                data[i]->min_mapQ = mapQ;                    // set the mapQ filter
+               data[i]->min_len  = min_len;                 // set the qlen filter
                htmp = bam_header_read(data[i]->fp);         // read the BAM header
                if (i == 0) {
                        h = htmp; // keep the header of the 1st BAM
@@ -108,5 +134,10 @@ int main_depth(int argc, char *argv[])
        }
        free(data); free(reg);
        if (bed) bed_destroy(bed);
+    if ( file_list )
+    {
+        for (i=0; i<n; i++) free(fn[i]);
+        free(fn);
+    }
        return 0;
 }