bam_stat.c

   1 #include "bam.h"
   2
   3 typedef struct {
   4         long long n_reads, n_mapped, n_pair_all, n_pair_map, n_pair_good;
   5         long long n_sgltn, n_read1, n_read2;
   6         long long n_qcfail, n_dup;
   7         long long n_diffchr, n_diffhigh;
   8 } bam_flagstat_t;
   9
  10 #define flagstat_loop(s, c) do {                                                                                \
  11                 ++(s)->n_reads;                                                                                                 \
  12                 if ((c)->flag & BAM_FPAIRED) {                                                                  \
  13                         ++(s)->n_pair_all;                                                                                      \
  14                         if ((c)->flag & BAM_FPROPER_PAIR) ++(s)->n_pair_good;           \
  15                         if ((c)->flag & BAM_FREAD1) ++(s)->n_read1;                                     \
  16                         if ((c)->flag & BAM_FREAD2) ++(s)->n_read2;                                     \
  17                         if ((c)->flag & BAM_FMUNMAP) ++(s)->n_sgltn;                            \
  18                         if (!((c)->flag & BAM_FUNMAP) && !((c)->flag & BAM_FMUNMAP)) { \
  19                                 ++(s)->n_pair_map;                                                                              \
  20                                 if ((c)->mtid != (c)->tid) {                                                    \
  21                                         ++(s)->n_diffchr;                                                                       \
  22                                         if ((c)->qual >= 5) ++(s)->n_diffhigh;                          \
  23                                 }                                                                                                               \
  24                         }                                                                                                                       \
  25                 }                                                                                                                               \
  26                 if (!((c)->flag & BAM_FUNMAP)) ++(s)->n_mapped;                                 \
  27                 if ((c)->flag & BAM_FQCFAIL) ++(s)->n_qcfail;                                   \
  28                 if ((c)->flag & BAM_FDUP) ++(s)->n_dup;                                                 \
  29         } while (0)
  30
  31 bam_flagstat_t *bam_flagstat_core(bamFile fp)
  32 {
  33         bam_flagstat_t *s;
  34         bam1_t *b;
  35         bam1_core_t *c;
  36         s = (bam_flagstat_t*)calloc(1, sizeof(bam_flagstat_t));
  37         b = bam_init1();
  38         c = &b->core;
  39         while (bam_read1(fp, b) >= 0) {
  40                 flagstat_loop(s, c);
  41         }
  42         bam_destroy1(b);
  43         return s;
  44 }
  45 int bam_flagstat(int argc, char *argv[])
  46 {
  47         bamFile fp;
  48         bam_header_t *header;
  49         bam_flagstat_t *s;
  50         if (argc == optind) {
  51                 fprintf(stderr, "Usage: samtools flagstat <in.bam>\n");
  52                 return 1;
  53         }
  54         fp = strcmp(argv[optind], "-")? bam_open(argv[optind], "r") : bam_dopen(fileno(stdin), "r");
  55         assert(fp);
  56         header = bam_header_read(fp);
  57         s = bam_flagstat_core(fp);
  58         printf("%lld in total\n", s->n_reads);
  59         printf("%lld QC failure\n", s->n_qcfail);
  60         printf("%lld duplicates\n", s->n_dup);
  61         printf("%lld mapped\n", s->n_mapped);
  62         printf("%lld paired in sequencing\n", s->n_pair_all);
  63         printf("%lld read1\n", s->n_read1);
  64         printf("%lld read2\n", s->n_read2);
  65         printf("%lld properly paired\n", s->n_pair_good);
  66         printf("%lld with both itself and its mate mapped\n", s->n_pair_map);
  67         printf("%lld singletons\n", s->n_sgltn);
  68         printf("%lld with its mate mapped to a different chr\n", s->n_diffchr);
  69         free(s);
  70         bam_header_destroy(header);
  71         bam_close(fp);
  72         return 0;
  73 }