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         int ret;
  37         s = (bam_flagstat_t*)calloc(1, sizeof(bam_flagstat_t));
  38         b = bam_init1();
  39         c = &b->core;
  40         while ((ret = bam_read1(fp, b)) >= 0)
  41                 flagstat_loop(s, c);
  42         bam_destroy1(b);
  43         if (ret != -1)
  44                 fprintf(stderr, "[bam_flagstat_core] Truncated file? Continue anyway.\n");
  45         return s;
  46 }
  47 int bam_flagstat(int argc, char *argv[])
  48 {
  49         bamFile fp;
  50         bam_header_t *header;
  51         bam_flagstat_t *s;
  52         if (argc == optind) {
  53                 fprintf(stderr, "Usage: samtools flagstat <in.bam>\n");
  54                 return 1;
  55         }
  56         fp = strcmp(argv[optind], "-")? bam_open(argv[optind], "r") : bam_dopen(fileno(stdin), "r");
  57         assert(fp);
  58         header = bam_header_read(fp);
  59         s = bam_flagstat_core(fp);
  60         printf("%lld in total\n", s->n_reads);
  61         printf("%lld QC failure\n", s->n_qcfail);
  62         printf("%lld duplicates\n", s->n_dup);
  63         printf("%lld mapped (%.2f%%)\n", s->n_mapped, (float)s->n_mapped / s->n_reads * 100.0);
  64         printf("%lld paired in sequencing\n", s->n_pair_all);
  65         printf("%lld read1\n", s->n_read1);
  66         printf("%lld read2\n", s->n_read2);
  67         printf("%lld properly paired (%.2f%%)\n", s->n_pair_good, (float)s->n_pair_good / s->n_pair_all * 100.0);
  68         printf("%lld with itself and mate mapped\n", s->n_pair_map);
  69         printf("%lld singletons (%.2f%%)\n", s->n_sgltn, (float)s->n_sgltn / s->n_pair_all * 100.0);
  70         printf("%lld with mate mapped to a different chr\n", s->n_diffchr);
  71         printf("%lld with mate mapped to a different chr (mapQ>=5)\n", s->n_diffhigh);
  72         free(s);
  73         bam_header_destroy(header);
  74         bam_close(fp);
  75         return 0;
  76 }