|
0
|
1 #include <stdlib.h>
|
|
|
2 #include <string.h>
|
|
|
3 #include <unistd.h>
|
|
|
4 #include "kstring.h"
|
|
|
5 #include "bam.h"
|
|
|
6
|
|
|
7 void bam_template_cigar(bam1_t *b1, bam1_t *b2, kstring_t *str)
|
|
|
8 {
|
|
|
9 bam1_t *swap;
|
|
|
10 int i, end;
|
|
|
11 uint32_t *cigar;
|
|
|
12 str->l = 0;
|
|
|
13 if (b1->core.tid != b2->core.tid || b1->core.tid < 0) return; // coordinateless or not on the same chr; skip
|
|
|
14 if (b1->core.pos > b2->core.pos) swap = b1, b1 = b2, b2 = swap; // make sure b1 has a smaller coordinate
|
|
|
15 kputc((b1->core.flag & BAM_FREAD1)? '1' : '2', str); // segment index
|
|
|
16 kputc((b1->core.flag & BAM_FREVERSE)? 'R' : 'F', str); // strand
|
|
|
17 for (i = 0, cigar = bam1_cigar(b1); i < b1->core.n_cigar; ++i) {
|
|
|
18 kputw(bam_cigar_oplen(cigar[i]), str);
|
|
|
19 kputc(bam_cigar_opchr(cigar[i]), str);
|
|
|
20 }
|
|
|
21 end = bam_calend(&b1->core, cigar);
|
|
|
22 kputw(b2->core.pos - end, str);
|
|
|
23 kputc('T', str);
|
|
|
24 kputc((b2->core.flag & BAM_FREAD1)? '1' : '2', str); // segment index
|
|
|
25 kputc((b2->core.flag & BAM_FREVERSE)? 'R' : 'F', str); // strand
|
|
|
26 for (i = 0, cigar = bam1_cigar(b2); i < b2->core.n_cigar; ++i) {
|
|
|
27 kputw(bam_cigar_oplen(cigar[i]), str);
|
|
|
28 kputc(bam_cigar_opchr(cigar[i]), str);
|
|
|
29 }
|
|
|
30 bam_aux_append(b1, "CT", 'Z', str->l+1, (uint8_t*)str->s);
|
|
|
31 }
|
|
|
32
|
|
|
33 // currently, this function ONLY works if each read has one hit
|
|
|
34 void bam_mating_core(bamFile in, bamFile out, int remove_reads)
|
|
|
35 {
|
|
|
36 bam_header_t *header;
|
|
|
37 bam1_t *b[2];
|
|
|
38 int curr, has_prev, pre_end = 0, cur_end;
|
|
|
39 kstring_t str;
|
|
|
40
|
|
|
41 str.l = str.m = 0; str.s = 0;
|
|
|
42 header = bam_header_read(in);
|
|
|
43 bam_header_write(out, header);
|
|
|
44
|
|
|
45 b[0] = bam_init1();
|
|
|
46 b[1] = bam_init1();
|
|
|
47 curr = 0; has_prev = 0;
|
|
|
48 while (bam_read1(in, b[curr]) >= 0) {
|
|
|
49 bam1_t *cur = b[curr], *pre = b[1-curr];
|
|
|
50 if (cur->core.tid < 0)
|
|
|
51 {
|
|
|
52 if ( !remove_reads ) bam_write1(out, cur);
|
|
|
53 continue;
|
|
|
54 }
|
|
|
55 cur_end = bam_calend(&cur->core, bam1_cigar(cur));
|
|
|
56 if (cur_end > (int)header->target_len[cur->core.tid]) cur->core.flag |= BAM_FUNMAP;
|
|
|
57 if (cur->core.flag & BAM_FSECONDARY)
|
|
|
58 {
|
|
|
59 if ( !remove_reads ) bam_write1(out, cur);
|
|
|
60 continue; // skip secondary alignments
|
|
|
61 }
|
|
|
62 if (has_prev) {
|
|
|
63 if (strcmp(bam1_qname(cur), bam1_qname(pre)) == 0) { // identical pair name
|
|
|
64 cur->core.mtid = pre->core.tid; cur->core.mpos = pre->core.pos;
|
|
|
65 pre->core.mtid = cur->core.tid; pre->core.mpos = cur->core.pos;
|
|
|
66 if (pre->core.tid == cur->core.tid && !(cur->core.flag&(BAM_FUNMAP|BAM_FMUNMAP))
|
|
|
67 && !(pre->core.flag&(BAM_FUNMAP|BAM_FMUNMAP))) // set TLEN/ISIZE
|
|
|
68 {
|
|
|
69 uint32_t cur5, pre5;
|
|
|
70 cur5 = (cur->core.flag&BAM_FREVERSE)? cur_end : cur->core.pos;
|
|
|
71 pre5 = (pre->core.flag&BAM_FREVERSE)? pre_end : pre->core.pos;
|
|
|
72 cur->core.isize = pre5 - cur5; pre->core.isize = cur5 - pre5;
|
|
|
73 } else cur->core.isize = pre->core.isize = 0;
|
|
|
74 if (pre->core.flag&BAM_FREVERSE) cur->core.flag |= BAM_FMREVERSE;
|
|
|
75 else cur->core.flag &= ~BAM_FMREVERSE;
|
|
|
76 if (cur->core.flag&BAM_FREVERSE) pre->core.flag |= BAM_FMREVERSE;
|
|
|
77 else pre->core.flag &= ~BAM_FMREVERSE;
|
|
|
78 if (cur->core.flag & BAM_FUNMAP) { pre->core.flag |= BAM_FMUNMAP; pre->core.flag &= ~BAM_FPROPER_PAIR; }
|
|
|
79 if (pre->core.flag & BAM_FUNMAP) { cur->core.flag |= BAM_FMUNMAP; cur->core.flag &= ~BAM_FPROPER_PAIR; }
|
|
|
80 bam_template_cigar(pre, cur, &str);
|
|
|
81 bam_write1(out, pre);
|
|
|
82 bam_write1(out, cur);
|
|
|
83 has_prev = 0;
|
|
|
84 } else { // unpaired or singleton
|
|
|
85 pre->core.mtid = -1; pre->core.mpos = -1; pre->core.isize = 0;
|
|
|
86 if (pre->core.flag & BAM_FPAIRED) {
|
|
|
87 pre->core.flag |= BAM_FMUNMAP;
|
|
|
88 pre->core.flag &= ~BAM_FMREVERSE & ~BAM_FPROPER_PAIR;
|
|
|
89 }
|
|
|
90 bam_write1(out, pre);
|
|
|
91 }
|
|
|
92 } else has_prev = 1;
|
|
|
93 curr = 1 - curr;
|
|
|
94 pre_end = cur_end;
|
|
|
95 }
|
|
|
96 if (has_prev) bam_write1(out, b[1-curr]);
|
|
|
97 bam_header_destroy(header);
|
|
|
98 bam_destroy1(b[0]);
|
|
|
99 bam_destroy1(b[1]);
|
|
|
100 free(str.s);
|
|
|
101 }
|
|
|
102
|
|
|
103 void usage()
|
|
|
104 {
|
|
|
105 fprintf(stderr,"Usage: samtools fixmate <in.nameSrt.bam> <out.nameSrt.bam>\n");
|
|
|
106 fprintf(stderr,"Options:\n");
|
|
|
107 fprintf(stderr," -r remove unmapped reads and secondary alignments\n");
|
|
|
108 exit(1);
|
|
|
109 }
|
|
|
110
|
|
|
111 int bam_mating(int argc, char *argv[])
|
|
|
112 {
|
|
|
113 bamFile in, out;
|
|
|
114 int c, remove_reads=0;
|
|
|
115 while ((c = getopt(argc, argv, "r")) >= 0) {
|
|
|
116 switch (c) {
|
|
|
117 case 'r': remove_reads=1; break;
|
|
|
118 }
|
|
|
119 }
|
|
|
120 if (optind+1 >= argc) usage();
|
|
|
121 in = (strcmp(argv[optind], "-") == 0)? bam_dopen(fileno(stdin), "r") : bam_open(argv[optind], "r");
|
|
|
122 out = (strcmp(argv[optind+1], "-") == 0)? bam_dopen(fileno(stdout), "w") : bam_open(argv[optind+1], "w");
|
|
|
123 bam_mating_core(in, out, remove_reads);
|
|
|
124 bam_close(in); bam_close(out);
|
|
|
125 return 0;
|
|
|
126 }
|
|
|
127
|
|
|
128
|
