annotate pyPRADA_1.2/tools/samtools-0.1.16/bam_plcmd.c @ 0:acc2ca1a3ba4

Uploaded
author siyuan
date Thu, 20 Feb 2014 00:44:58 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
1 #include <math.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
2 #include <stdio.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
3 #include <unistd.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
4 #include <ctype.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
5 #include <string.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
6 #include <errno.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
7 #include "sam.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
8 #include "faidx.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
9 #include "bam_maqcns.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
10 #include "khash.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
11 #include "glf.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
12 #include "kstring.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
13
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
14 typedef int *indel_list_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
15 KHASH_MAP_INIT_INT64(64, indel_list_t)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
16
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
17 #define BAM_PLF_SIMPLE 0x01
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
18 #define BAM_PLF_CNS 0x02
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
19 #define BAM_PLF_INDEL_ONLY 0x04
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
20 #define BAM_PLF_GLF 0x08
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
21 #define BAM_PLF_VAR_ONLY 0x10
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
22 #define BAM_PLF_2ND 0x20
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
23 #define BAM_PLF_RANBASE 0x40
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
24 #define BAM_PLF_1STBASE 0x80
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
25 #define BAM_PLF_ALLBASE 0x100
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
26 #define BAM_PLF_READPOS 0x200
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
27 #define BAM_PLF_NOBAQ 0x400
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
28
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
29 typedef struct {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
30 bam_header_t *h;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
31 bam_maqcns_t *c;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
32 bam_maqindel_opt_t *ido;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
33 faidx_t *fai;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
34 khash_t(64) *hash;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
35 uint32_t format;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
36 int tid, len, last_pos;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
37 int mask;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
38 int capQ_thres, min_baseQ;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
39 int max_depth; // for indel calling, ignore reads with the depth too high. 0 for unlimited
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
40 char *ref;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
41 glfFile fp_glf; // for glf output only
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
42 } pu_data_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
43
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
44 char **__bam_get_lines(const char *fn, int *_n);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
45 void bam_init_header_hash(bam_header_t *header);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
46 int32_t bam_get_tid(const bam_header_t *header, const char *seq_name);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
47
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
48 static khash_t(64) *load_pos(const char *fn, bam_header_t *h)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
49 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
50 char **list;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
51 int i, j, n, *fields, max_fields;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
52 khash_t(64) *hash;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
53 bam_init_header_hash(h);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
54 list = __bam_get_lines(fn, &n);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
55 hash = kh_init(64);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
56 max_fields = 0; fields = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
57 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
58 char *str = list[i];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
59 int chr, n_fields, ret;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
60 khint_t k;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
61 uint64_t x;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
62 n_fields = ksplit_core(str, 0, &max_fields, &fields);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
63 if (n_fields < 2) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
64 chr = bam_get_tid(h, str + fields[0]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
65 if (chr < 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
66 fprintf(stderr, "[load_pos] unknown reference sequence name: %s\n", str + fields[0]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
67 continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
68 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
69 x = (uint64_t)chr << 32 | (atoi(str + fields[1]) - 1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
70 k = kh_put(64, hash, x, &ret);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
71 if (ret == 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
72 fprintf(stderr, "[load_pos] position %s:%s has been loaded.\n", str+fields[0], str+fields[1]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
73 continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
74 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
75 kh_val(hash, k) = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
76 if (n_fields > 2) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
77 // count
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
78 for (j = 2; j < n_fields; ++j) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
79 char *s = str + fields[j];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
80 if ((*s != '+' && *s != '-') || !isdigit(s[1])) break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
81 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
82 if (j > 2) { // update kh_val()
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
83 int *q, y, z;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
84 q = kh_val(hash, k) = (int*)calloc(j - 1, sizeof(int));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
85 q[0] = j - 2; z = j; y = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
86 for (j = 2; j < z; ++j)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
87 q[y++] = atoi(str + fields[j]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
88 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
89 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
90 free(str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
91 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
92 free(list); free(fields);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
93 return hash;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
94 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
95
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
96 static inline int printw(int c, FILE *fp)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
97 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
98 char buf[16];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
99 int l, x;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
100 if (c == 0) return fputc('0', fp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
101 for (l = 0, x = c < 0? -c : c; x > 0; x /= 10) buf[l++] = x%10 + '0';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
102 if (c < 0) buf[l++] = '-';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
103 buf[l] = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
104 for (x = 0; x < l/2; ++x) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
105 int y = buf[x]; buf[x] = buf[l-1-x]; buf[l-1-x] = y;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
106 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
107 fputs(buf, fp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
108 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
109 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
110
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
111 // an analogy to pileup_func() below
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
112 static int glt3_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
113 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
114 pu_data_t *d = (pu_data_t*)data;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
115 bam_maqindel_ret_t *r = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
116 int rb, *proposed_indels = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
117 glf1_t *g;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
118 glf3_t *g3;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
119
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
120 if (d->fai == 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
121 fprintf(stderr, "[glt3_func] reference sequence is required for generating GLT. Abort!\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
122 exit(1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
123 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
124 if (d->hash) { // only output a list of sites
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
125 khint_t k = kh_get(64, d->hash, (uint64_t)tid<<32|pos);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
126 if (k == kh_end(d->hash)) return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
127 proposed_indels = kh_val(d->hash, k);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
128 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
129 g3 = glf3_init1();
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
130 if (d->fai && (int)tid != d->tid) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
131 if (d->ref) { // then write the end mark
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
132 g3->rtype = GLF3_RTYPE_END;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
133 glf3_write1(d->fp_glf, g3);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
134 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
135 glf3_ref_write(d->fp_glf, d->h->target_name[tid], d->h->target_len[tid]); // write reference
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
136 free(d->ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
137 d->ref = fai_fetch(d->fai, d->h->target_name[tid], &d->len);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
138 d->tid = tid;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
139 d->last_pos = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
140 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
141 rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
142 g = bam_maqcns_glfgen(n, pu, bam_nt16_table[rb], d->c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
143 memcpy(g3, g, sizeof(glf1_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
144 g3->rtype = GLF3_RTYPE_SUB;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
145 g3->offset = pos - d->last_pos;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
146 d->last_pos = pos;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
147 glf3_write1(d->fp_glf, g3);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
148 if (pos < d->len) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
149 int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
150 if (proposed_indels)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
151 r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
152 else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
153 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
154 if (r) { // then write indel line
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
155 int het = 3 * n, min;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
156 min = het;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
157 if (min > r->gl[0]) min = r->gl[0];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
158 if (min > r->gl[1]) min = r->gl[1];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
159 g3->ref_base = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
160 g3->rtype = GLF3_RTYPE_INDEL;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
161 memset(g3->lk, 0, 10);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
162 g3->lk[0] = r->gl[0] - min < 255? r->gl[0] - min : 255;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
163 g3->lk[1] = r->gl[1] - min < 255? r->gl[1] - min : 255;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
164 g3->lk[2] = het - min < 255? het - min : 255;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
165 g3->offset = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
166 g3->indel_len[0] = r->indel1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
167 g3->indel_len[1] = r->indel2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
168 g3->min_lk = min < 255? min : 255;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
169 g3->max_len = (abs(r->indel1) > abs(r->indel2)? abs(r->indel1) : abs(r->indel2)) + 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
170 g3->indel_seq[0] = strdup(r->s[0]+1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
171 g3->indel_seq[1] = strdup(r->s[1]+1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
172 glf3_write1(d->fp_glf, g3);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
173 bam_maqindel_ret_destroy(r);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
174 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
175 free(g);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
176 glf3_destroy1(g3);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
177 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
178 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
179
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
180 static inline void pileup_seq(const bam_pileup1_t *p, int pos, int ref_len, const char *ref)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
181 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
182 int j;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
183 if (p->is_head) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
184 putchar('^');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
185 putchar(p->b->core.qual > 93? 126 : p->b->core.qual + 33);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
186 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
187 if (!p->is_del) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
188 int c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos)];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
189 if (ref) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
190 int rb = pos < ref_len? ref[pos] : 'N';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
191 if (c == '=' || bam_nt16_table[c] == bam_nt16_table[rb]) c = bam1_strand(p->b)? ',' : '.';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
192 else c = bam1_strand(p->b)? tolower(c) : toupper(c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
193 } else {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
194 if (c == '=') c = bam1_strand(p->b)? ',' : '.';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
195 else c = bam1_strand(p->b)? tolower(c) : toupper(c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
196 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
197 putchar(c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
198 } else putchar(p->is_refskip? (bam1_strand(p->b)? '<' : '>') : '*');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
199 if (p->indel > 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
200 putchar('+'); printw(p->indel, stdout);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
201 for (j = 1; j <= p->indel; ++j) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
202 int c = bam_nt16_rev_table[bam1_seqi(bam1_seq(p->b), p->qpos + j)];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
203 putchar(bam1_strand(p->b)? tolower(c) : toupper(c));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
204 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
205 } else if (p->indel < 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
206 printw(p->indel, stdout);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
207 for (j = 1; j <= -p->indel; ++j) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
208 int c = (ref && (int)pos+j < ref_len)? ref[pos+j] : 'N';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
209 putchar(bam1_strand(p->b)? tolower(c) : toupper(c));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
210 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
211 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
212 if (p->is_tail) putchar('$');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
213 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
214
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
215 static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu, void *data)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
216 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
217 pu_data_t *d = (pu_data_t*)data;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
218 bam_maqindel_ret_t *r = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
219 int i, rb, rms_mapq = -1, *proposed_indels = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
220 uint64_t rms_aux;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
221 uint32_t cns = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
222
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
223 // if GLF is required, suppress -c completely
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
224 if (d->format & BAM_PLF_GLF) return glt3_func(tid, pos, n, pu, data);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
225 // if d->hash is initialized, only output the sites in the hash table
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
226 if (d->hash) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
227 khint_t k = kh_get(64, d->hash, (uint64_t)tid<<32|pos);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
228 if (k == kh_end(d->hash)) return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
229 proposed_indels = kh_val(d->hash, k);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
230 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
231 // update d->ref if necessary
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
232 if (d->fai && (int)tid != d->tid) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
233 free(d->ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
234 d->ref = faidx_fetch_seq(d->fai, d->h->target_name[tid], 0, 0x7fffffff, &d->len);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
235 d->tid = tid;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
236 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
237 rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
238 // when the indel-only mode is asked for, return if no reads mapped with indels
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
239 if (d->format & BAM_PLF_INDEL_ONLY) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
240 for (i = 0; i < n; ++i)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
241 if (pu[i].indel != 0) break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
242 if (i == n) return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
243 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
244 // call the consensus and indel
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
245 if (d->format & BAM_PLF_CNS) { // call consensus
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
246 if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE)) { // use a random base or the 1st base as the consensus call
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
247 const bam_pileup1_t *p = (d->format & BAM_PLF_1STBASE)? pu : pu + (int)(drand48() * n);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
248 int q = bam1_qual(p->b)[p->qpos];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
249 int mapQ = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
250 uint32_t b = bam1_seqi(bam1_seq(p->b), p->qpos);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
251 cns = b<<28 | 0xf<<24 | mapQ<<16 | q<<8;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
252 } else if (d->format & BAM_PLF_ALLBASE) { // collapse all bases
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
253 uint64_t rmsQ = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
254 uint32_t b = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
255 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
256 const bam_pileup1_t *p = pu + i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
257 int q = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
258 b |= bam1_seqi(bam1_seq(p->b), p->qpos);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
259 rmsQ += q * q;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
260 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
261 rmsQ = (uint64_t)(sqrt((double)rmsQ / n) + .499);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
262 cns = b<<28 | 0xf<<24 | rmsQ<<16 | 60<<8;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
263 } else {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
264 glf1_t *g = bam_maqcns_glfgen(n, pu, bam_nt16_table[rb], d->c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
265 cns = g->depth == 0? (0xfu<<28 | 0xf<<24) : glf2cns(g, (int)(d->c->q_r + .499));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
266 free(g);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
267 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
268 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
269 if ((d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) && d->ref && pos < d->len) { // call indels
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
270 int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
271 if (proposed_indels) // the first element gives the size of the array
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
272 r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
273 else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
274 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
275 // when only variant sites are asked for, test if the site is a variant
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
276 if ((d->format & BAM_PLF_CNS) && (d->format & BAM_PLF_VAR_ONLY)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
277 if (!(bam_nt16_table[rb] != 15 && cns>>28 != 15 && cns>>28 != bam_nt16_table[rb])) { // not a SNP
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
278 if (!(r && (r->gt == 2 || strcmp(r->s[r->gt], "*")))) { // not an indel
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
279 if (r) bam_maqindel_ret_destroy(r);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
280 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
281 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
282 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
283 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
284 // print the first 3 columns
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
285 fputs(d->h->target_name[tid], stdout); putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
286 printw(pos+1, stdout); putchar('\t'); putchar(rb); putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
287 // print consensus information if required
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
288 if (d->format & BAM_PLF_CNS) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
289 putchar(bam_nt16_rev_table[cns>>28]); putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
290 printw(cns>>8&0xff, stdout); putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
291 printw(cns&0xff, stdout); putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
292 printw(cns>>16&0xff, stdout); putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
293 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
294 // print pileup sequences
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
295 printw(n, stdout); putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
296 for (i = 0; i < n; ++i)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
297 pileup_seq(pu + i, pos, d->len, d->ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
298 // finalize rms_mapq
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
299 if (d->format & BAM_PLF_CNS) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
300 for (i = rms_aux = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
301 const bam_pileup1_t *p = pu + i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
302 int tmp = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
303 rms_aux += tmp * tmp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
304 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
305 rms_aux = (uint64_t)(sqrt((double)rms_aux / n) + .499);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
306 if (rms_mapq < 0) rms_mapq = rms_aux;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
307 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
308 putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
309 // print quality
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
310 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
311 const bam_pileup1_t *p = pu + i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
312 int c = bam1_qual(p->b)[p->qpos] + 33;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
313 if (c > 126) c = 126;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
314 putchar(c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
315 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
316 if (d->format & BAM_PLF_2ND) { // print 2nd calls and qualities
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
317 const unsigned char *q;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
318 putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
319 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
320 const bam_pileup1_t *p = pu + i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
321 q = bam_aux_get(p->b, "E2");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
322 putchar(q? q[p->qpos + 1] : 'N');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
323 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
324 putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
325 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
326 const bam_pileup1_t *p = pu + i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
327 q = bam_aux_get(p->b, "U2");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
328 putchar(q? q[p->qpos + 1] : '!');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
329 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
330 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
331 // print mapping quality if -s is flagged on the command line
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
332 if (d->format & BAM_PLF_SIMPLE) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
333 putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
334 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
335 int c = pu[i].b->core.qual + 33;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
336 if (c > 126) c = 126;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
337 putchar(c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
338 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
339 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
340 // print read position
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
341 if (d->format & BAM_PLF_READPOS) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
342 putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
343 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
344 int x = pu[i].qpos;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
345 int l = pu[i].b->core.l_qseq;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
346 printw(x < l/2? x+1 : -((l-1)-x+1), stdout); putchar(',');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
347 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
348 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
349 putchar('\n');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
350 // print the indel line if r has been calculated. This only happens if:
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
351 // a) -c or -i are flagged, AND b) the reference sequence is available
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
352 if (r) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
353 printf("%s\t%d\t*\t", d->h->target_name[tid], pos + 1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
354 if (r->gt < 2) printf("%s/%s\t", r->s[r->gt], r->s[r->gt]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
355 else printf("%s/%s\t", r->s[0], r->s[1]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
356 printf("%d\t%d\t", r->q_cns, r->q_ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
357 printf("%d\t%d\t", rms_mapq, n);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
358 printf("%s\t%s\t", r->s[0], r->s[1]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
359 //printf("%d\t%d\t", r->gl[0], r->gl[1]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
360 printf("%d\t%d\t%d\t", r->cnt1, r->cnt2, r->cnt_anti);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
361 printf("%d\t%d\n", r->cnt_ref, r->cnt_ambi);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
362 bam_maqindel_ret_destroy(r);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
363 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
364 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
365 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
366
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
367 int bam_pileup(int argc, char *argv[])
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
368 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
369 int c, is_SAM = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
370 char *fn_list = 0, *fn_fa = 0, *fn_pos = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
371 pu_data_t *d = (pu_data_t*)calloc(1, sizeof(pu_data_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
372 d->max_depth = 1024; d->tid = -1; d->mask = BAM_DEF_MASK; d->min_baseQ = 13;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
373 d->c = bam_maqcns_init();
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
374 d->c->errmod = BAM_ERRMOD_MAQ2; // change the default model
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
375 d->ido = bam_maqindel_opt_init();
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
376 while ((c = getopt(argc, argv, "st:f:cT:N:r:l:d:im:gI:G:vM:S2aR:PAQ:C:B")) >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
377 switch (c) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
378 case 'Q': d->c->min_baseQ = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
379 case 'C': d->capQ_thres = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
380 case 'B': d->format |= BAM_PLF_NOBAQ; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
381 case 'a': d->c->errmod = BAM_ERRMOD_SOAP; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
382 case 'A': d->c->errmod = BAM_ERRMOD_MAQ; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
383 case 's': d->format |= BAM_PLF_SIMPLE; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
384 case 't': fn_list = strdup(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
385 case 'l': fn_pos = strdup(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
386 case 'f': fn_fa = strdup(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
387 case 'T': d->c->theta = atof(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
388 case 'N': d->c->n_hap = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
389 case 'r': d->c->het_rate = atof(optarg); d->ido->r_snp = d->c->het_rate; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
390 case 'M': d->c->cap_mapQ = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
391 case 'd': d->max_depth = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
392 case 'c': d->format |= BAM_PLF_CNS; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
393 case 'i': d->format |= BAM_PLF_INDEL_ONLY; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
394 case 'v': d->format |= BAM_PLF_VAR_ONLY; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
395 case 'm': d->mask = strtol(optarg, 0, 0); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
396 case 'g': d->format |= BAM_PLF_GLF; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
397 case '2': d->format |= BAM_PLF_2ND; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
398 case 'P': d->format |= BAM_PLF_READPOS; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
399 case 'I': d->ido->q_indel = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
400 case 'G': d->ido->r_indel = atof(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
401 case 'S': is_SAM = 1; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
402 case 'R':
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
403 if (strcmp(optarg, "random") == 0) d->format |= BAM_PLF_RANBASE;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
404 else if (strcmp(optarg, "first") == 0) d->format |= BAM_PLF_1STBASE;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
405 else if (strcmp(optarg, "all") == 0) d->format |= BAM_PLF_ALLBASE;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
406 else fprintf(stderr, "[bam_pileup] unrecognized -R\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
407 break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
408 default: fprintf(stderr, "Unrecognizd option '-%c'.\n", c); return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
409 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
410 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
411 if (d->c->errmod != BAM_ERRMOD_MAQ2) d->c->theta += 0.02;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
412 if (d->c->theta > 1.0) d->c->theta = 1.0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
413 if (fn_list) is_SAM = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
414 if (optind == argc) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
415 fprintf(stderr, "\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
416 fprintf(stderr, "Usage: samtools pileup [options] <in.bam>|<in.sam>\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
417 fprintf(stderr, "Option: -s simple (yet incomplete) pileup format\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
418 fprintf(stderr, " -S the input is in SAM\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
419 fprintf(stderr, " -B disable BAQ computation\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
420 fprintf(stderr, " -A use the original MAQ model for SNP calling (DEPRECATED)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
421 fprintf(stderr, " -2 output the 2nd best call and quality\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
422 fprintf(stderr, " -i only show lines/consensus with indels\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
423 fprintf(stderr, " -Q INT min base quality (possibly capped by BAQ) [%d]\n", d->c->min_baseQ);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
424 fprintf(stderr, " -C INT coefficient for adjusting mapQ of poor mappings [%d]\n", d->capQ_thres);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
425 fprintf(stderr, " -m INT filtering reads with bits in INT [0x%x]\n", d->mask);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
426 fprintf(stderr, " -M INT cap mapping quality at INT [%d]\n", d->c->cap_mapQ);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
427 fprintf(stderr, " -d INT limit maximum depth for indels [%d]\n", d->max_depth);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
428 fprintf(stderr, " -t FILE list of reference sequences (force -S)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
429 fprintf(stderr, " -l FILE list of sites at which pileup is output\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
430 fprintf(stderr, " -f FILE reference sequence in the FASTA format\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
431 fprintf(stderr, " -c compute the consensus sequence\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
432 fprintf(stderr, " -v print variants only (for -c)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
433 fprintf(stderr, " -g output in the GLFv3 format (DEPRECATED)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
434 fprintf(stderr, " -T FLOAT theta in maq consensus calling model (for -c) [%.4g]\n", d->c->theta);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
435 fprintf(stderr, " -N INT number of haplotypes in the sample (for -c) [%d]\n", d->c->n_hap);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
436 fprintf(stderr, " -r FLOAT prior of a difference between two haplotypes (for -c) [%.4g]\n", d->c->het_rate);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
437 fprintf(stderr, " -G FLOAT prior of an indel between two haplotypes (for -c) [%.4g]\n", d->ido->r_indel);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
438 fprintf(stderr, " -I INT phred prob. of an indel in sequencing/prep. (for -c) [%d]\n", d->ido->q_indel);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
439 fprintf(stderr, "\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
440 fprintf(stderr, "Warning: Please use the `mpileup' command instead `pileup'. `Pileup' is deprecated!\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
441 free(fn_list); free(fn_fa); free(d);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
442 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
443 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
444 if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE|BAM_PLF_ALLBASE)) d->format |= BAM_PLF_CNS;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
445 if (fn_fa) d->fai = fai_load(fn_fa);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
446 if (d->format & (BAM_PLF_CNS|BAM_PLF_GLF)) bam_maqcns_prepare(d->c); // consensus calling
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
447 if (d->format & BAM_PLF_GLF) { // for glf output
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
448 glf3_header_t *h;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
449 h = glf3_header_init();
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
450 d->fp_glf = bgzf_fdopen(fileno(stdout), "w");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
451 glf3_header_write(d->fp_glf, h);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
452 glf3_header_destroy(h);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
453 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
454 if (d->fai == 0 && (d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
455 fprintf(stderr, "[bam_pileup] indels will not be called when -f is absent.\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
456 if (fn_fa && is_SAM && fn_list == 0) fn_list = samfaipath(fn_fa);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
457
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
458 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
459 samfile_t *fp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
460 fp = is_SAM? samopen(argv[optind], "r", fn_list) : samopen(argv[optind], "rb", 0);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
461 if (fp == 0 || fp->header == 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
462 fprintf(stderr, "[bam_pileup] fail to read the header: non-exisiting file or wrong format.\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
463 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
464 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
465 d->h = fp->header;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
466 if (fn_pos) d->hash = load_pos(fn_pos, d->h);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
467 { // run pileup
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
468 extern int bam_prob_realn(bam1_t *b, const char *ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
469 extern int bam_cap_mapQ(bam1_t *b, char *ref, int thres);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
470 bam1_t *b;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
471 int ret, tid, pos, n_plp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
472 bam_plp_t iter;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
473 const bam_pileup1_t *plp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
474 b = bam_init1();
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
475 iter = bam_plp_init(0, 0);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
476 bam_plp_set_mask(iter, d->mask);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
477 while ((ret = samread(fp, b)) >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
478 int skip = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
479 if ((int)b->core.tid < 0) break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
480 // update d->ref if necessary
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
481 if (d->fai && (int)b->core.tid != d->tid) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
482 free(d->ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
483 d->ref = faidx_fetch_seq(d->fai, d->h->target_name[b->core.tid], 0, 0x7fffffff, &d->len);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
484 d->tid = b->core.tid;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
485 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
486 if (d->ref && (d->format&BAM_PLF_CNS) && !(d->format&BAM_PLF_NOBAQ)) bam_prob_realn(b, d->ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
487 if (d->ref && (d->format&BAM_PLF_CNS) && d->capQ_thres > 10) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
488 int q = bam_cap_mapQ(b, d->ref, d->capQ_thres);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
489 if (q < 0) skip = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
490 else if (b->core.qual > q) b->core.qual = q;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
491 } else if (b->core.flag&BAM_FUNMAP) skip = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
492 else if ((d->format&BAM_PLF_CNS) && (b->core.flag&1) && !(b->core.flag&2)) skip = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
493 if (skip) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
494 bam_plp_push(iter, b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
495 while ((plp = bam_plp_next(iter, &tid, &pos, &n_plp)) != 0)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
496 pileup_func(tid, pos, n_plp, plp, d);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
497 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
498 bam_plp_push(iter, 0);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
499 while ((plp = bam_plp_next(iter, &tid, &pos, &n_plp)) != 0)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
500 pileup_func(tid, pos, n_plp, plp, d);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
501 bam_plp_destroy(iter);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
502 bam_destroy1(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
503 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
504 samclose(fp); // d->h will be destroyed here
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
505 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
506
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
507 // free
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
508 if (d->format & BAM_PLF_GLF) bgzf_close(d->fp_glf);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
509 if (fn_pos) { // free the hash table
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
510 khint_t k;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
511 for (k = kh_begin(d->hash); k < kh_end(d->hash); ++k)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
512 if (kh_exist(d->hash, k)) free(kh_val(d->hash, k));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
513 kh_destroy(64, d->hash);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
514 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
515 free(fn_pos); free(fn_list); free(fn_fa);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
516 if (d->fai) fai_destroy(d->fai);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
517 bam_maqcns_destroy(d->c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
518 free(d->ido); free(d->ref); free(d);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
519 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
520 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
521
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
522 /***********
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
523 * mpileup *
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
524 ***********/
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
525
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
526 #include <assert.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
527 #include "bam2bcf.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
528 #include "sample.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
529
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
530 #define MPLP_GLF 0x10
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
531 #define MPLP_NO_COMP 0x20
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
532 #define MPLP_NO_ORPHAN 0x40
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
533 #define MPLP_REALN 0x80
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
534 #define MPLP_FMT_DP 0x100
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
535 #define MPLP_FMT_SP 0x200
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
536 #define MPLP_NO_INDEL 0x400
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
537 #define MPLP_EXT_BAQ 0x800
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
538 #define MPLP_ILLUMINA13 0x1000
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
539
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
540 void *bed_read(const char *fn);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
541 void bed_destroy(void *_h);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
542 int bed_overlap(const void *_h, const char *chr, int beg, int end);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
543
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
544 typedef struct {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
545 int max_mq, min_mq, flag, min_baseQ, capQ_thres, max_depth, max_indel_depth;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
546 int openQ, extQ, tandemQ, min_support; // for indels
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
547 double min_frac; // for indels
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
548 char *reg, *pl_list;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
549 faidx_t *fai;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
550 void *bed, *rghash;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
551 } mplp_conf_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
552
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
553 typedef struct {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
554 bamFile fp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
555 bam_iter_t iter;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
556 bam_header_t *h;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
557 int ref_id;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
558 char *ref;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
559 const mplp_conf_t *conf;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
560 } mplp_aux_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
561
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
562 typedef struct {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
563 int n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
564 int *n_plp, *m_plp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
565 bam_pileup1_t **plp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
566 } mplp_pileup_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
567
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
568 static int mplp_func(void *data, bam1_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
569 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
570 extern int bam_realn(bam1_t *b, const char *ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
571 extern int bam_prob_realn_core(bam1_t *b, const char *ref, int);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
572 extern int bam_cap_mapQ(bam1_t *b, char *ref, int thres);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
573 mplp_aux_t *ma = (mplp_aux_t*)data;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
574 int ret, skip = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
575 do {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
576 int has_ref;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
577 ret = ma->iter? bam_iter_read(ma->fp, ma->iter, b) : bam_read1(ma->fp, b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
578 if (ret < 0) break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
579 if (b->core.tid < 0 || (b->core.flag&BAM_FUNMAP)) { // exclude unmapped reads
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
580 skip = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
581 continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
582 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
583 if (ma->conf->bed) { // test overlap
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
584 skip = !bed_overlap(ma->conf->bed, ma->h->target_name[b->core.tid], b->core.pos, bam_calend(&b->core, bam1_cigar(b)));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
585 if (skip) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
586 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
587 if (ma->conf->rghash) { // exclude read groups
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
588 uint8_t *rg = bam_aux_get(b, "RG");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
589 skip = (rg && bcf_str2id(ma->conf->rghash, (const char*)(rg+1)) >= 0);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
590 if (skip) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
591 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
592 if (ma->conf->flag & MPLP_ILLUMINA13) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
593 int i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
594 uint8_t *qual = bam1_qual(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
595 for (i = 0; i < b->core.l_qseq; ++i)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
596 qual[i] = qual[i] > 31? qual[i] - 31 : 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
597 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
598 has_ref = (ma->ref && ma->ref_id == b->core.tid)? 1 : 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
599 skip = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
600 if (has_ref && (ma->conf->flag&MPLP_REALN)) bam_prob_realn_core(b, ma->ref, (ma->conf->flag & MPLP_EXT_BAQ)? 3 : 1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
601 if (has_ref && ma->conf->capQ_thres > 10) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
602 int q = bam_cap_mapQ(b, ma->ref, ma->conf->capQ_thres);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
603 if (q < 0) skip = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
604 else if (b->core.qual > q) b->core.qual = q;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
605 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
606 else if (b->core.qual < ma->conf->min_mq) skip = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
607 else if ((ma->conf->flag&MPLP_NO_ORPHAN) && (b->core.flag&1) && !(b->core.flag&2)) skip = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
608 } while (skip);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
609 return ret;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
610 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
611
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
612 static void group_smpl(mplp_pileup_t *m, bam_sample_t *sm, kstring_t *buf,
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
613 int n, char *const*fn, int *n_plp, const bam_pileup1_t **plp)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
614 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
615 int i, j;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
616 memset(m->n_plp, 0, m->n * sizeof(int));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
617 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
618 for (j = 0; j < n_plp[i]; ++j) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
619 const bam_pileup1_t *p = plp[i] + j;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
620 uint8_t *q;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
621 int id = -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
622 q = bam_aux_get(p->b, "RG");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
623 if (q) id = bam_smpl_rg2smid(sm, fn[i], (char*)q+1, buf);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
624 if (id < 0) id = bam_smpl_rg2smid(sm, fn[i], 0, buf);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
625 if (id < 0 || id >= m->n) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
626 assert(q); // otherwise a bug
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
627 fprintf(stderr, "[%s] Read group %s used in file %s but absent from the header or an alignment missing read group.\n", __func__, (char*)q+1, fn[i]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
628 exit(1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
629 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
630 if (m->n_plp[id] == m->m_plp[id]) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
631 m->m_plp[id] = m->m_plp[id]? m->m_plp[id]<<1 : 8;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
632 m->plp[id] = realloc(m->plp[id], sizeof(bam_pileup1_t) * m->m_plp[id]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
633 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
634 m->plp[id][m->n_plp[id]++] = *p;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
635 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
636 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
637 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
638
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
639 static int mpileup(mplp_conf_t *conf, int n, char **fn)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
640 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
641 extern void *bcf_call_add_rg(void *rghash, const char *hdtext, const char *list);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
642 extern void bcf_call_del_rghash(void *rghash);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
643 mplp_aux_t **data;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
644 int i, tid, pos, *n_plp, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid, max_depth, max_indel_depth;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
645 const bam_pileup1_t **plp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
646 bam_mplp_t iter;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
647 bam_header_t *h = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
648 char *ref;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
649 void *rghash = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
650
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
651 bcf_callaux_t *bca = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
652 bcf_callret1_t *bcr = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
653 bcf_call_t bc;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
654 bcf_t *bp = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
655 bcf_hdr_t *bh = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
656
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
657 bam_sample_t *sm = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
658 kstring_t buf;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
659 mplp_pileup_t gplp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
660
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
661 memset(&gplp, 0, sizeof(mplp_pileup_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
662 memset(&buf, 0, sizeof(kstring_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
663 memset(&bc, 0, sizeof(bcf_call_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
664 data = calloc(n, sizeof(void*));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
665 plp = calloc(n, sizeof(void*));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
666 n_plp = calloc(n, sizeof(int*));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
667 sm = bam_smpl_init();
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
668
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
669 // read the header and initialize data
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
670 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
671 bam_header_t *h_tmp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
672 data[i] = calloc(1, sizeof(mplp_aux_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
673 data[i]->fp = strcmp(fn[i], "-") == 0? bam_dopen(fileno(stdin), "r") : bam_open(fn[i], "r");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
674 data[i]->conf = conf;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
675 h_tmp = bam_header_read(data[i]->fp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
676 data[i]->h = i? h : h_tmp; // for i==0, "h" has not been set yet
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
677 bam_smpl_add(sm, fn[i], h_tmp->text);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
678 rghash = bcf_call_add_rg(rghash, h_tmp->text, conf->pl_list);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
679 if (conf->reg) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
680 int beg, end;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
681 bam_index_t *idx;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
682 idx = bam_index_load(fn[i]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
683 if (idx == 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
684 fprintf(stderr, "[%s] fail to load index for %d-th input.\n", __func__, i+1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
685 exit(1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
686 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
687 if (bam_parse_region(h_tmp, conf->reg, &tid, &beg, &end) < 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
688 fprintf(stderr, "[%s] malformatted region or wrong seqname for %d-th input.\n", __func__, i+1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
689 exit(1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
690 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
691 if (i == 0) beg0 = beg, end0 = end;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
692 data[i]->iter = bam_iter_query(idx, tid, beg, end);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
693 bam_index_destroy(idx);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
694 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
695 if (i == 0) h = h_tmp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
696 else {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
697 // FIXME: to check consistency
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
698 bam_header_destroy(h_tmp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
699 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
700 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
701 gplp.n = sm->n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
702 gplp.n_plp = calloc(sm->n, sizeof(int));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
703 gplp.m_plp = calloc(sm->n, sizeof(int));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
704 gplp.plp = calloc(sm->n, sizeof(void*));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
705
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
706 fprintf(stderr, "[%s] %d samples in %d input files\n", __func__, sm->n, n);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
707 // write the VCF header
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
708 if (conf->flag & MPLP_GLF) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
709 kstring_t s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
710 bh = calloc(1, sizeof(bcf_hdr_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
711 s.l = s.m = 0; s.s = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
712 bp = bcf_open("-", (conf->flag&MPLP_NO_COMP)? "wu" : "w");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
713 for (i = 0; i < h->n_targets; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
714 kputs(h->target_name[i], &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
715 kputc('\0', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
716 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
717 bh->l_nm = s.l;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
718 bh->name = malloc(s.l);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
719 memcpy(bh->name, s.s, s.l);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
720 s.l = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
721 for (i = 0; i < sm->n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
722 kputs(sm->smpl[i], &s); kputc('\0', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
723 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
724 bh->l_smpl = s.l;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
725 bh->sname = malloc(s.l);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
726 memcpy(bh->sname, s.s, s.l);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
727 bh->txt = malloc(strlen(BAM_VERSION) + 64);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
728 bh->l_txt = 1 + sprintf(bh->txt, "##samtoolsVersion=%s\n", BAM_VERSION);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
729 free(s.s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
730 bcf_hdr_sync(bh);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
731 bcf_hdr_write(bp, bh);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
732 bca = bcf_call_init(-1., conf->min_baseQ);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
733 bcr = calloc(sm->n, sizeof(bcf_callret1_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
734 bca->rghash = rghash;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
735 bca->openQ = conf->openQ, bca->extQ = conf->extQ, bca->tandemQ = conf->tandemQ;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
736 bca->min_frac = conf->min_frac;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
737 bca->min_support = conf->min_support;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
738 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
739 ref_tid = -1; ref = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
740 iter = bam_mplp_init(n, mplp_func, (void**)data);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
741 max_depth = conf->max_depth;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
742 if (max_depth * sm->n > 1<<20)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
743 fprintf(stderr, "(%s) Max depth is above 1M. Potential memory hog!\n", __func__);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
744 if (max_depth * sm->n < 8000) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
745 max_depth = 8000 / sm->n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
746 fprintf(stderr, "<%s> Set max per-file depth to %d\n", __func__, max_depth);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
747 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
748 max_indel_depth = conf->max_indel_depth * sm->n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
749 bam_mplp_set_maxcnt(iter, max_depth);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
750 while (bam_mplp_auto(iter, &tid, &pos, n_plp, plp) > 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
751 if (conf->reg && (pos < beg0 || pos >= end0)) continue; // out of the region requested
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
752 if (conf->bed && tid >= 0 && !bed_overlap(conf->bed, h->target_name[tid], pos, pos+1)) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
753 if (tid != ref_tid) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
754 free(ref); ref = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
755 if (conf->fai) ref = fai_fetch(conf->fai, h->target_name[tid], &ref_len);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
756 for (i = 0; i < n; ++i) data[i]->ref = ref, data[i]->ref_id = tid;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
757 ref_tid = tid;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
758 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
759 if (conf->flag & MPLP_GLF) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
760 int total_depth, _ref0, ref16;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
761 bcf1_t *b = calloc(1, sizeof(bcf1_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
762 for (i = total_depth = 0; i < n; ++i) total_depth += n_plp[i];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
763 group_smpl(&gplp, sm, &buf, n, fn, n_plp, plp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
764 _ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
765 ref16 = bam_nt16_table[_ref0];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
766 for (i = 0; i < gplp.n; ++i)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
767 bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], ref16, bca, bcr + i);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
768 bcf_call_combine(gplp.n, bcr, ref16, &bc);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
769 bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
770 (conf->flag&MPLP_FMT_SP), 0, 0);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
771 bcf_write(bp, bh, b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
772 bcf_destroy(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
773 // call indels
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
774 if (!(conf->flag&MPLP_NO_INDEL) && total_depth < max_indel_depth && bcf_call_gap_prep(gplp.n, gplp.n_plp, gplp.plp, pos, bca, ref, rghash) >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
775 for (i = 0; i < gplp.n; ++i)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
776 bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], -1, bca, bcr + i);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
777 if (bcf_call_combine(gplp.n, bcr, -1, &bc) >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
778 b = calloc(1, sizeof(bcf1_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
779 bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
780 (conf->flag&MPLP_FMT_SP), bca, ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
781 bcf_write(bp, bh, b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
782 bcf_destroy(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
783 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
784 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
785 } else {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
786 printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
787 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
788 int j;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
789 printf("\t%d\t", n_plp[i]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
790 if (n_plp[i] == 0) printf("*\t*");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
791 else {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
792 for (j = 0; j < n_plp[i]; ++j)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
793 pileup_seq(plp[i] + j, pos, ref_len, ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
794 putchar('\t');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
795 for (j = 0; j < n_plp[i]; ++j) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
796 const bam_pileup1_t *p = plp[i] + j;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
797 int c = bam1_qual(p->b)[p->qpos] + 33;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
798 if (c > 126) c = 126;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
799 putchar(c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
800 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
801 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
802 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
803 putchar('\n');
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
804 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
805 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
806
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
807 bcf_close(bp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
808 bam_smpl_destroy(sm); free(buf.s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
809 for (i = 0; i < gplp.n; ++i) free(gplp.plp[i]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
810 free(gplp.plp); free(gplp.n_plp); free(gplp.m_plp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
811 bcf_call_del_rghash(rghash);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
812 bcf_hdr_destroy(bh); bcf_call_destroy(bca); free(bc.PL); free(bcr);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
813 bam_mplp_destroy(iter);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
814 bam_header_destroy(h);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
815 for (i = 0; i < n; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
816 bam_close(data[i]->fp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
817 if (data[i]->iter) bam_iter_destroy(data[i]->iter);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
818 free(data[i]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
819 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
820 free(data); free(plp); free(ref); free(n_plp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
821 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
822 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
823
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
824 #define MAX_PATH_LEN 1024
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
825 static int read_file_list(const char *file_list,int *n,char **argv[])
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
826 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
827 char buf[MAX_PATH_LEN];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
828 int len, nfiles;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
829 char **files;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
830
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
831 FILE *fh = fopen(file_list,"r");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
832 if ( !fh )
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
833 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
834 fprintf(stderr,"%s: %s\n", file_list,strerror(errno));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
835 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
836 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
837
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
838 // Speed is not an issue here, determine the number of files by reading the file twice
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
839 nfiles = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
840 while ( fgets(buf,MAX_PATH_LEN,fh) ) nfiles++;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
841
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
842 if ( fseek(fh, 0L, SEEK_SET) )
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
843 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
844 fprintf(stderr,"%s: %s\n", file_list,strerror(errno));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
845 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
846 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
847
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
848 files = calloc(nfiles,sizeof(char*));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
849 nfiles = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
850 while ( fgets(buf,MAX_PATH_LEN,fh) )
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
851 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
852 len = strlen(buf);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
853 while ( len>0 && isspace(buf[len-1]) ) len--;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
854 if ( !len ) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
855
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
856 files[nfiles] = malloc(sizeof(char)*(len+1));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
857 strncpy(files[nfiles],buf,len);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
858 files[nfiles][len] = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
859 nfiles++;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
860 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
861 fclose(fh);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
862 if ( !nfiles )
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
863 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
864 fprintf(stderr,"No files read from %s\n", file_list);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
865 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
866 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
867 *argv = files;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
868 *n = nfiles;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
869 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
870 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
871 #undef MAX_PATH_LEN
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
872
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
873 int bam_mpileup(int argc, char *argv[])
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
874 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
875 int c;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
876 const char *file_list = NULL;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
877 char **fn = NULL;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
878 int nfiles = 0, use_orphan = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
879 mplp_conf_t mplp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
880 memset(&mplp, 0, sizeof(mplp_conf_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
881 mplp.max_mq = 60;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
882 mplp.min_baseQ = 13;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
883 mplp.capQ_thres = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
884 mplp.max_depth = 250; mplp.max_indel_depth = 250;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
885 mplp.openQ = 40; mplp.extQ = 20; mplp.tandemQ = 100;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
886 mplp.min_frac = 0.002; mplp.min_support = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
887 mplp.flag = MPLP_NO_ORPHAN | MPLP_REALN;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
888 while ((c = getopt(argc, argv, "Agf:r:l:M:q:Q:uaRC:BDSd:L:b:P:o:e:h:Im:F:EG:6")) >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
889 switch (c) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
890 case 'f':
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
891 mplp.fai = fai_load(optarg);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
892 if (mplp.fai == 0) return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
893 break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
894 case 'd': mplp.max_depth = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
895 case 'r': mplp.reg = strdup(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
896 case 'l': mplp.bed = bed_read(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
897 case 'P': mplp.pl_list = strdup(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
898 case 'g': mplp.flag |= MPLP_GLF; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
899 case 'u': mplp.flag |= MPLP_NO_COMP | MPLP_GLF; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
900 case 'a': mplp.flag |= MPLP_NO_ORPHAN | MPLP_REALN; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
901 case 'B': mplp.flag &= ~MPLP_REALN; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
902 case 'R': mplp.flag |= MPLP_REALN; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
903 case 'D': mplp.flag |= MPLP_FMT_DP; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
904 case 'S': mplp.flag |= MPLP_FMT_SP; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
905 case 'I': mplp.flag |= MPLP_NO_INDEL; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
906 case 'E': mplp.flag |= MPLP_EXT_BAQ; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
907 case '6': mplp.flag |= MPLP_ILLUMINA13; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
908 case 'C': mplp.capQ_thres = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
909 case 'M': mplp.max_mq = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
910 case 'q': mplp.min_mq = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
911 case 'Q': mplp.min_baseQ = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
912 case 'b': file_list = optarg; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
913 case 'o': mplp.openQ = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
914 case 'e': mplp.extQ = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
915 case 'h': mplp.tandemQ = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
916 case 'A': use_orphan = 1; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
917 case 'F': mplp.min_frac = atof(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
918 case 'm': mplp.min_support = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
919 case 'L': mplp.max_indel_depth = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
920 case 'G': {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
921 FILE *fp_rg;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
922 char buf[1024];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
923 mplp.rghash = bcf_str2id_init();
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
924 if ((fp_rg = fopen(optarg, "r")) == 0)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
925 fprintf(stderr, "(%s) Fail to open file %s. Continue anyway.\n", __func__, optarg);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
926 while (!feof(fp_rg) && fscanf(fp_rg, "%s", buf) > 0) // this is not a good style, but forgive me...
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
927 bcf_str2id_add(mplp.rghash, strdup(buf));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
928 fclose(fp_rg);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
929 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
930 break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
931 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
932 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
933 if (use_orphan) mplp.flag &= ~MPLP_NO_ORPHAN;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
934 if (argc == 1) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
935 fprintf(stderr, "\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
936 fprintf(stderr, "Usage: samtools mpileup [options] in1.bam [in2.bam [...]]\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
937 fprintf(stderr, "Input options:\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
938 fprintf(stderr, " -6 assume the quality is in the Illumina-1.3+ encoding\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
939 fprintf(stderr, " -A count anomalous read pairs\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
940 fprintf(stderr, " -B disable BAQ computation\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
941 fprintf(stderr, " -b FILE list of input BAM files [null]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
942 fprintf(stderr, " -d INT max per-BAM depth to avoid excessive memory usage [%d]\n", mplp.max_depth);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
943 fprintf(stderr, " -E extended BAQ for higher sensitivity but lower specificity\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
944 fprintf(stderr, " -f FILE faidx indexed reference sequence file [null]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
945 fprintf(stderr, " -G FILE exclude read groups listed in FILE [null]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
946 fprintf(stderr, " -l FILE list of positions (chr pos) or regions (BED) [null]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
947 fprintf(stderr, " -M INT cap mapping quality at INT [%d]\n", mplp.max_mq);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
948 fprintf(stderr, " -r STR region in which pileup is generated [null]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
949 fprintf(stderr, " -q INT skip alignments with mapQ smaller than INT [%d]\n", mplp.min_mq);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
950 fprintf(stderr, " -Q INT skip bases with baseQ/BAQ smaller than INT [%d]\n", mplp.min_baseQ);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
951 fprintf(stderr, "\nOutput options:\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
952 fprintf(stderr, " -D output per-sample DP in BCF (require -g/-u)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
953 fprintf(stderr, " -g generate BCF output (genotype likelihoods)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
954 fprintf(stderr, " -S output per-sample strand bias P-value in BCF (require -g/-u)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
955 fprintf(stderr, " -u generate uncompress BCF output\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
956 fprintf(stderr, "\nSNP/INDEL genotype likelihoods options (effective with `-g' or `-u'):\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
957 fprintf(stderr, " -e INT Phred-scaled gap extension seq error probability [%d]\n", mplp.extQ);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
958 fprintf(stderr, " -F FLOAT minimum fraction of gapped reads for candidates [%g]\n", mplp.min_frac);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
959 fprintf(stderr, " -h INT coefficient for homopolymer errors [%d]\n", mplp.tandemQ);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
960 fprintf(stderr, " -I do not perform indel calling\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
961 fprintf(stderr, " -L INT max per-sample depth for INDEL calling [%d]\n", mplp.max_indel_depth);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
962 fprintf(stderr, " -m INT minimum gapped reads for indel candidates [%d]\n", mplp.min_support);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
963 fprintf(stderr, " -o INT Phred-scaled gap open sequencing error probability [%d]\n", mplp.openQ);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
964 fprintf(stderr, " -P STR comma separated list of platforms for indels [all]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
965 fprintf(stderr, "\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
966 fprintf(stderr, "Notes: Assuming diploid individuals.\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
967 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
968 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
969 if (file_list) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
970 if ( read_file_list(file_list,&nfiles,&fn) ) return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
971 mpileup(&mplp,nfiles,fn);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
972 for (c=0; c<nfiles; c++) free(fn[c]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
973 free(fn);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
974 } else mpileup(&mplp, argc - optind, argv + optind);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
975 if (mplp.rghash) bcf_str2id_thorough_destroy(mplp.rghash);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
976 free(mplp.reg); free(mplp.pl_list);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
977 if (mplp.fai) fai_destroy(mplp.fai);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
978 if (mplp.bed) bed_destroy(mplp.bed);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
979 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
980 }