3 /* This program demonstrates how to generate pileup from multiple BAMs
4 * simutaneously, to achieve random access and to use the BED interface.
5 * To compile this program separately, you may:
7 * gcc -g -O2 -Wall -o bam2depth -D_MAIN_BAM2DEPTH bam2depth.c -L. -lbam -lz
15 typedef struct { // auxiliary data structure
16 bamFile fp; // the file handler
17 bam_iter_t iter; // NULL if a region not specified
18 int min_mapQ; // mapQ filter
21 void *bed_read(const char *fn); // read a BED or position list file
22 void bed_destroy(void *_h); // destroy the BED data structure
23 int bed_overlap(const void *_h, const char *chr, int beg, int end); // test if chr:beg-end overlaps
25 // This function reads a BAM alignment from one BAM file.
26 static int read_bam(void *data, bam1_t *b) // read level filters better go here to avoid pileup
28 aux_t *aux = (aux_t*)data; // data in fact is a pointer to an auxiliary structure
29 int ret = aux->iter? bam_iter_read(aux->fp, aux->iter, b) : bam_read1(aux->fp, b);
30 if ((int)b->core.qual < aux->min_mapQ) b->core.flag |= BAM_FUNMAP;
34 #ifdef _MAIN_BAM2DEPTH
35 int main(int argc, char *argv[])
37 int main_depth(int argc, char *argv[])
40 int i, n, tid, beg, end, pos, *n_plp, baseQ = 0, mapQ = 0;
41 const bam_pileup1_t **plp;
42 char *reg = 0; // specified region
43 void *bed = 0; // BED data structure
44 bam_header_t *h = 0; // BAM header of the 1st input
48 // parse the command line
49 while ((n = getopt(argc, argv, "r:b:q:Q:")) >= 0) {
51 case 'r': reg = strdup(optarg); break; // parsing a region requires a BAM header
52 case 'b': bed = bed_read(optarg); break; // BED or position list file can be parsed now
53 case 'q': baseQ = atoi(optarg); break; // base quality threshold
54 case 'Q': mapQ = atoi(optarg); break; // mapping quality threshold
58 fprintf(pysamerr, "Usage: bam2depth [-r reg] [-q baseQthres] [-Q mapQthres] [-b in.bed] <in1.bam> [...]\n");
62 // initialize the auxiliary data structures
63 n = argc - optind; // the number of BAMs on the command line
64 data = calloc(n, sizeof(void*)); // data[i] for the i-th input
65 beg = 0; end = 1<<30; tid = -1; // set the default region
66 for (i = 0; i < n; ++i) {
68 data[i] = calloc(1, sizeof(aux_t));
69 data[i]->fp = bam_open(argv[optind+i], "r"); // open BAM
70 data[i]->min_mapQ = mapQ; // set the mapQ filter
71 htmp = bam_header_read(data[i]->fp); // read the BAM header
73 h = htmp; // keep the header of the 1st BAM
74 if (reg) bam_parse_region(h, reg, &tid, &beg, &end); // also parse the region
75 } else bam_header_destroy(htmp); // if not the 1st BAM, trash the header
76 if (tid >= 0) { // if a region is specified and parsed successfully
77 bam_index_t *idx = bam_index_load(argv[optind+i]); // load the index
78 data[i]->iter = bam_iter_query(idx, tid, beg, end); // set the iterator
79 bam_index_destroy(idx); // the index is not needed any more; phase out of the memory
83 // the core multi-pileup loop
84 mplp = bam_mplp_init(n, read_bam, (void**)data); // initialization
85 n_plp = calloc(n, sizeof(int)); // n_plp[i] is the number of covering reads from the i-th BAM
86 plp = calloc(n, sizeof(void*)); // plp[i] points to the array of covering reads (internal in mplp)
87 while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0) { // come to the next covered position
88 if (pos < beg || pos >= end) continue; // out of range; skip
89 if (bed && bed_overlap(bed, h->target_name[tid], pos, pos + 1) == 0) continue; // not in BED; skip
90 fputs(h->target_name[tid], stdout); printf("\t%d", pos+1); // a customized printf() would be faster
91 for (i = 0; i < n; ++i) { // base level filters have to go here
93 for (j = 0; j < n_plp[i]; ++j) {
94 const bam_pileup1_t *p = plp[i] + j; // DON'T modfity plp[][] unless you really know
95 if (p->is_del || p->is_refskip) ++m; // having dels or refskips at tid:pos
96 else if (bam1_qual(p->b)[p->qpos] < baseQ) ++m; // low base quality
98 printf("\t%d", n_plp[i] - m); // this the depth to output
102 free(n_plp); free(plp);
103 bam_mplp_destroy(mplp);
105 bam_header_destroy(h);
106 for (i = 0; i < n; ++i) {
107 bam_close(data[i]->fp);
108 if (data[i]->iter) bam_iter_destroy(data[i]->iter);
111 free(data); free(reg);
112 if (bed) bed_destroy(bed);