bam_mate.c

   1 #include <stdlib.h>
   2 #include <string.h>
   3 #include "bam.h"
   4
   5 // currently, this function ONLY works if each read has one hit
   6 void bam_mating_core(bamFile in, bamFile out)
   7 {
   8         bam_header_t *header;
   9         bam1_t *b[2];
  10         int curr, has_prev;
  11
  12         header = bam_header_read(in);
  13         bam_header_write(out, header);
  14
  15         b[0] = bam_init1();
  16         b[1] = bam_init1();
  17         curr = 0; has_prev = 0;
  18         while (bam_read1(in, b[curr]) >= 0) {
  19                 bam1_t *cur = b[curr], *pre = b[1-curr];
  20                 if (has_prev) {
  21                         if (strcmp(bam1_qname(cur), bam1_qname(pre)) == 0) { // identical pair name
  22                                 cur->core.mtid = pre->core.tid; cur->core.mpos = pre->core.pos;
  23                                 pre->core.mtid = cur->core.tid; pre->core.mpos = cur->core.pos;
  24                                 if (pre->core.tid == cur->core.tid && !(cur->core.flag&(BAM_FUNMAP|BAM_FMUNMAP))
  25                                         && !(pre->core.flag&(BAM_FUNMAP|BAM_FMUNMAP)))
  26                                 {
  27                                         uint32_t cur5, pre5;
  28                                         cur5 = (cur->core.flag&BAM_FREVERSE)? bam_calend(&cur->core, bam1_cigar(cur)) : cur->core.pos;
  29                                         pre5 = (pre->core.flag&BAM_FREVERSE)? bam_calend(&pre->core, bam1_cigar(pre)) : pre->core.pos;
  30                                         cur->core.isize = pre5 - cur5; pre->core.isize = cur5 - pre5;
  31                                 } else cur->core.isize = pre->core.isize = 0;
  32                                 if (pre->core.flag&BAM_FREVERSE) cur->core.flag |= BAM_FMREVERSE;
  33                                 else cur->core.flag &= ~BAM_FMREVERSE;
  34                                 if (cur->core.flag&BAM_FREVERSE) pre->core.flag |= BAM_FMREVERSE;
  35                                 else pre->core.flag &= ~BAM_FMREVERSE;
  36                                 if (cur->core.flag & BAM_FUNMAP) { pre->core.flag |= BAM_FMUNMAP; pre->core.flag &= ~BAM_FPROPER_PAIR; }
  37                                 if (pre->core.flag & BAM_FUNMAP) { cur->core.flag |= BAM_FMUNMAP; cur->core.flag &= ~BAM_FPROPER_PAIR; }
  38                                 bam_write1(out, pre);
  39                                 bam_write1(out, cur);
  40                                 has_prev = 0;
  41                         } else { // unpaired or singleton
  42                                 pre->core.mtid = -1; pre->core.mpos = -1; pre->core.isize = 0;
  43                                 if (pre->core.flag & BAM_FPAIRED) {
  44                                         pre->core.flag |= BAM_FMUNMAP;
  45                                         pre->core.flag &= ~BAM_FMREVERSE & ~BAM_FPROPER_PAIR;
  46                                 }
  47                                 bam_write1(out, pre);
  48                         }
  49                 } else has_prev = 1;
  50                 curr = 1 - curr;
  51         }
  52         if (has_prev) bam_write1(out, b[1-curr]);
  53         bam_header_destroy(header);
  54         bam_destroy1(b[0]);
  55         bam_destroy1(b[1]);
  56 }
  57
  58 int bam_mating(int argc, char *argv[])
  59 {
  60         bamFile in, out;
  61         if (argc < 3) {
  62                 fprintf(stderr, "samtools fixmate <in.nameSrt.bam> <out.nameSrt.bam>\n");
  63                 return 1;
  64         }
  65         in = (strcmp(argv[1], "-") == 0)? bam_dopen(fileno(stdin), "r") : bam_open(argv[1], "r");
  66     out = (strcmp(argv[2], "-") == 0)? bam_dopen(fileno(stdout), "w") : bam_open(argv[2], "w");
  67         bam_mating_core(in, out);
  68         bam_close(in); bam_close(out);
  69         return 0;
  70 }