annotate pyPRADA_1.2/tools/samtools-0.1.16/bcftools/call1.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 <unistd.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
2 #include <stdlib.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
3 #include <math.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
4 #include <zlib.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
5 #include <errno.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
6 #include "bcf.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
7 #include "prob1.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
8 #include "kstring.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
9 #include "time.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
10
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
11 #include "kseq.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
12 KSTREAM_INIT(gzFile, gzread, 16384)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
13
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
14 #define VC_NO_GENO 2
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
15 #define VC_BCFOUT 4
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
16 #define VC_CALL 8
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
17 #define VC_VARONLY 16
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
18 #define VC_VCFIN 32
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
19 #define VC_UNCOMP 64
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
20 #define VC_KEEPALT 256
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
21 #define VC_ACGT_ONLY 512
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
22 #define VC_QCALL 1024
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
23 #define VC_CALL_GT 2048
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
24 #define VC_ADJLD 4096
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
25 #define VC_NO_INDEL 8192
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
26 #define VC_ANNO_MAX 16384
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
27 #define VC_FIX_PL 32768
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
28 #define VC_EM 0x10000
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
29
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
30 typedef struct {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
31 int flag, prior_type, n1, n_sub, *sublist, n_perm;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
32 char *prior_file, **subsam, *fn_dict;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
33 uint8_t *ploidy;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
34 double theta, pref, indel_frac, min_perm_p, min_smpl_frac, min_lrt;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
35 void *bed;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
36 } viewconf_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
37
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
38 void *bed_read(const char *fn);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
39 void bed_destroy(void *_h);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
40 int bed_overlap(const void *_h, const char *chr, int beg, int end);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
41
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
42 typedef struct {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
43 double p[4];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
44 int mq, depth, is_tested, d[4];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
45 } anno16_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
46
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
47 static double ttest(int n1, int n2, int a[4])
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
48 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
49 extern double kf_betai(double a, double b, double x);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
50 double t, v, u1, u2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
51 if (n1 == 0 || n2 == 0 || n1 + n2 < 3) return 1.0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
52 u1 = (double)a[0] / n1; u2 = (double)a[2] / n2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
53 if (u1 <= u2) return 1.;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
54 t = (u1 - u2) / sqrt(((a[1] - n1 * u1 * u1) + (a[3] - n2 * u2 * u2)) / (n1 + n2 - 2) * (1./n1 + 1./n2));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
55 v = n1 + n2 - 2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
56 // printf("%d,%d,%d,%d,%lf,%lf,%lf\n", a[0], a[1], a[2], a[3], t, u1, u2);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
57 return t < 0.? 1. : .5 * kf_betai(.5*v, .5, v/(v+t*t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
58 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
59
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
60 static int test16_core(int anno[16], anno16_t *a)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
61 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
62 extern double kt_fisher_exact(int n11, int n12, int n21, int n22, double *_left, double *_right, double *two);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
63 double left, right;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
64 int i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
65 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
66 memcpy(a->d, anno, 4 * sizeof(int));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
67 a->depth = anno[0] + anno[1] + anno[2] + anno[3];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
68 a->is_tested = (anno[0] + anno[1] > 0 && anno[2] + anno[3] > 0);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
69 if (a->depth == 0) return -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
70 a->mq = (int)(sqrt((anno[9] + anno[11]) / a->depth) + .499);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
71 kt_fisher_exact(anno[0], anno[1], anno[2], anno[3], &left, &right, &a->p[0]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
72 for (i = 1; i < 4; ++i)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
73 a->p[i] = ttest(anno[0] + anno[1], anno[2] + anno[3], anno+4*i);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
74 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
75 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
76
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
77 static int test16(bcf1_t *b, anno16_t *a)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
78 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
79 char *p;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
80 int i, anno[16];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
81 a->p[0] = a->p[1] = a->p[2] = a->p[3] = 1.;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
82 a->d[0] = a->d[1] = a->d[2] = a->d[3] = 0.;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
83 a->mq = a->depth = a->is_tested = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
84 if ((p = strstr(b->info, "I16=")) == 0) return -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
85 p += 4;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
86 for (i = 0; i < 16; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
87 errno = 0; anno[i] = strtol(p, &p, 10);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
88 if (anno[i] == 0 && (errno == EINVAL || errno == ERANGE)) return -2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
89 ++p;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
90 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
91 return test16_core(anno, a);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
92 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
93
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
94 static void rm_info(bcf1_t *b, const char *key)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
95 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
96 char *p, *q;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
97 if ((p = strstr(b->info, key)) == 0) return;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
98 for (q = p; *q && *q != ';'; ++q);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
99 if (p > b->info && *(p-1) == ';') --p;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
100 memmove(p, q, b->l_str - (q - b->str));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
101 b->l_str -= q - p;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
102 bcf_sync(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
103 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
104
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
105 static int update_bcf1(bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p1rst_t *pr, double pref, int flag, double em[9])
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
106 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
107 kstring_t s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
108 int has_I16, is_var;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
109 double fq, r;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
110 anno16_t a;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
111
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
112 has_I16 = test16(b, &a) >= 0? 1 : 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
113 rm_info(b, "I16="); // FIXME: probably this function has a bug. If I move it below, I16 will not be removed!
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
114
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
115 memset(&s, 0, sizeof(kstring_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
116 kputc('\0', &s); kputs(b->ref, &s); kputc('\0', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
117 kputs(b->alt, &s); kputc('\0', &s); kputc('\0', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
118 kputs(b->info, &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
119 if (b->info[0]) kputc(';', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
120 { // print EM
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
121 if (em[0] >= 0) ksprintf(&s, "AF1=%.4g", 1 - em[0]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
122 if (em[4] >= 0 && em[4] <= 0.05) ksprintf(&s, ";G3=%.4g,%.4g,%.4g;HWE=%.3g", em[3], em[2], em[1], em[4]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
123 if (em[5] >= 0 && em[6] >= 0) ksprintf(&s, ";AF2=%.4g,%.4g", 1 - em[5], 1 - em[6]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
124 if (em[7] >= 0) ksprintf(&s, ";LRT=%.3g", em[7]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
125 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
126 if (pr == 0) { // if pr is unset, return
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
127 kputc('\0', &s); kputs(b->fmt, &s); kputc('\0', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
128 free(b->str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
129 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
130 bcf_sync(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
131 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
132 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
133
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
134 is_var = (pr->p_ref < pref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
135 r = is_var? pr->p_ref : pr->p_var;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
136
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
137 ksprintf(&s, ";CI95=%.4g,%.4g", pr->cil, pr->cih); // FIXME: when EM is not used, ";" should be omitted!
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
138 if (has_I16) ksprintf(&s, ";DP4=%d,%d,%d,%d;MQ=%d", a.d[0], a.d[1], a.d[2], a.d[3], a.mq);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
139 fq = pr->p_ref_folded < 0.5? -4.343 * log(pr->p_ref_folded) : 4.343 * log(pr->p_var_folded);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
140 if (fq < -999) fq = -999;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
141 if (fq > 999) fq = 999;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
142 ksprintf(&s, ";FQ=%.3g", fq);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
143 if (pr->cmp[0] >= 0.) { // two sample groups
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
144 int i, q[3];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
145 for (i = 1; i < 3; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
146 double x = pr->cmp[i] + pr->cmp[0]/2.;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
147 q[i] = x == 0? 255 : (int)(-4.343 * log(x) + .499);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
148 if (q[i] > 255) q[i] = 255;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
149 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
150 if (pr->perm_rank >= 0) ksprintf(&s, ";PR=%d", pr->perm_rank);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
151 ksprintf(&s, ";PCHI2=%.3g;PC2=%d,%d", q[1], q[2], pr->p_chi2);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
152 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
153 if (has_I16 && a.is_tested) ksprintf(&s, ";PV4=%.2g,%.2g,%.2g,%.2g", a.p[0], a.p[1], a.p[2], a.p[3]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
154 kputc('\0', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
155 kputs(b->fmt, &s); kputc('\0', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
156 free(b->str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
157 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
158 b->qual = r < 1e-100? 999 : -4.343 * log(r);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
159 if (b->qual > 999) b->qual = 999;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
160 bcf_sync(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
161 if (!is_var) bcf_shrink_alt(b, 1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
162 else if (!(flag&VC_KEEPALT))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
163 bcf_shrink_alt(b, pr->rank0 < 2? 2 : pr->rank0+1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
164 if (is_var && (flag&VC_CALL_GT)) { // call individual genotype
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
165 int i, x, old_n_gi = b->n_gi;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
166 s.m = b->m_str; s.l = b->l_str - 1; s.s = b->str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
167 kputs(":GT:GQ", &s); kputc('\0', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
168 b->m_str = s.m; b->l_str = s.l; b->str = s.s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
169 bcf_sync(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
170 for (i = 0; i < b->n_smpl; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
171 x = bcf_p1_call_gt(pa, pr->f_exp, i);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
172 ((uint8_t*)b->gi[old_n_gi].data)[i] = (x&3) == 0? 1<<3|1 : (x&3) == 1? 1 : 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
173 ((uint8_t*)b->gi[old_n_gi+1].data)[i] = x>>2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
174 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
175 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
176 return is_var;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
177 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
178
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
179 static char **read_samples(const char *fn, int *_n)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
180 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
181 gzFile fp;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
182 kstream_t *ks;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
183 kstring_t s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
184 int dret, n = 0, max = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
185 char **sam = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
186 *_n = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
187 s.l = s.m = 0; s.s = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
188 fp = gzopen(fn, "r");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
189 if (fp == 0) return 0; // fail to open file
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
190 ks = ks_init(fp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
191 while (ks_getuntil(ks, 0, &s, &dret) >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
192 int l;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
193 if (max == n) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
194 max = max? max<<1 : 4;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
195 sam = realloc(sam, sizeof(void*)*max);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
196 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
197 l = s.l;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
198 sam[n] = malloc(s.l + 2);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
199 strcpy(sam[n], s.s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
200 sam[n][l+1] = 2; // by default, diploid
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
201 if (dret != '\n') {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
202 if (ks_getuntil(ks, 0, &s, &dret) >= 0) { // read ploidy, 1 or 2
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
203 int x = (int)s.s[0] - '0';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
204 if (x == 1 || x == 2) sam[n][l+1] = x;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
205 else fprintf(stderr, "(%s) ploidy can only be 1 or 2; assume diploid\n", __func__);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
206 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
207 if (dret != '\n') ks_getuntil(ks, '\n', &s, &dret);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
208 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
209 ++n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
210 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
211 ks_destroy(ks);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
212 gzclose(fp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
213 free(s.s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
214 *_n = n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
215 return sam;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
216 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
217
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
218 static void write_header(bcf_hdr_t *h)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
219 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
220 kstring_t str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
221 str.l = h->l_txt? h->l_txt - 1 : 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
222 str.m = str.l + 1; str.s = h->txt;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
223 if (!strstr(str.s, "##INFO=<ID=DP,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
224 kputs("##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Raw read depth\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
225 if (!strstr(str.s, "##INFO=<ID=DP4,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
226 kputs("##INFO=<ID=DP4,Number=4,Type=Integer,Description=\"# high-quality ref-forward bases, ref-reverse, alt-forward and alt-reverse bases\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
227 if (!strstr(str.s, "##INFO=<ID=MQ,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
228 kputs("##INFO=<ID=MQ,Number=1,Type=Integer,Description=\"Root-mean-square mapping quality of covering reads\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
229 if (!strstr(str.s, "##INFO=<ID=FQ,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
230 kputs("##INFO=<ID=FQ,Number=1,Type=Float,Description=\"Phred probability of all samples being the same\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
231 if (!strstr(str.s, "##INFO=<ID=AF1,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
232 kputs("##INFO=<ID=AF1,Number=1,Type=Float,Description=\"Max-likelihood estimate of the site allele frequency of the first ALT allele\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
233 if (!strstr(str.s, "##INFO=<ID=G3,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
234 kputs("##INFO=<ID=G3,Number=3,Type=Float,Description=\"ML estimate of genotype frequencies\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
235 if (!strstr(str.s, "##INFO=<ID=HWE,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
236 kputs("##INFO=<ID=HWE,Number=1,Type=Float,Description=\"Chi^2 based HWE test P-value based on G3\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
237 if (!strstr(str.s, "##INFO=<ID=CI95,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
238 kputs("##INFO=<ID=CI95,Number=2,Type=Float,Description=\"Equal-tail Bayesian credible interval of the site allele frequency at the 95% level\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
239 if (!strstr(str.s, "##INFO=<ID=PV4,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
240 kputs("##INFO=<ID=PV4,Number=4,Type=Float,Description=\"P-values for strand bias, baseQ bias, mapQ bias and tail distance bias\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
241 if (!strstr(str.s, "##INFO=<ID=INDEL,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
242 kputs("##INFO=<ID=INDEL,Number=0,Type=Flag,Description=\"Indicates that the variant is an INDEL.\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
243 if (!strstr(str.s, "##INFO=<ID=PC2,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
244 kputs("##INFO=<ID=PC2,Number=2,Type=Integer,Description=\"Phred probability of the nonRef allele frequency in group1 samples being larger (,smaller) than in group2.\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
245 if (!strstr(str.s, "##INFO=<ID=PCHI2,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
246 kputs("##INFO=<ID=PCHI2,Number=1,Type=Float,Description=\"Posterior weighted chi^2 P-value for testing the association between group1 and group2 samples.\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
247 if (!strstr(str.s, "##INFO=<ID=QCHI2,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
248 kputs("##INFO=<ID=QCHI2,Number=1,Type=Integer,Description=\"Phred scaled PCHI2.\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
249 if (!strstr(str.s, "##INFO=<ID=RP,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
250 kputs("##INFO=<ID=PR,Number=1,Type=Integer,Description=\"# permutations yielding a smaller PCHI2.\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
251 if (!strstr(str.s, "##FORMAT=<ID=GT,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
252 kputs("##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
253 if (!strstr(str.s, "##FORMAT=<ID=GQ,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
254 kputs("##FORMAT=<ID=GQ,Number=1,Type=Integer,Description=\"Genotype Quality\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
255 if (!strstr(str.s, "##FORMAT=<ID=GL,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
256 kputs("##FORMAT=<ID=GL,Number=3,Type=Float,Description=\"Likelihoods for RR,RA,AA genotypes (R=ref,A=alt)\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
257 if (!strstr(str.s, "##FORMAT=<ID=DP,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
258 kputs("##FORMAT=<ID=DP,Number=1,Type=Integer,Description=\"# high-quality bases\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
259 if (!strstr(str.s, "##FORMAT=<ID=SP,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
260 kputs("##FORMAT=<ID=SP,Number=1,Type=Integer,Description=\"Phred-scaled strand bias P-value\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
261 if (!strstr(str.s, "##FORMAT=<ID=PL,"))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
262 kputs("##FORMAT=<ID=PL,Number=-1,Type=Integer,Description=\"List of Phred-scaled genotype likelihoods, number of values is (#ALT+1)*(#ALT+2)/2\">\n", &str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
263 h->l_txt = str.l + 1; h->txt = str.s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
264 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
265
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
266 double bcf_pair_freq(const bcf1_t *b0, const bcf1_t *b1, double f[4]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
267
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
268 int bcfview(int argc, char *argv[])
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
269 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
270 extern int bcf_2qcall(bcf_hdr_t *h, bcf1_t *b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
271 extern void bcf_p1_indel_prior(bcf_p1aux_t *ma, double x);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
272 extern int bcf_fix_gt(bcf1_t *b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
273 extern int bcf_anno_max(bcf1_t *b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
274 extern int bcf_shuffle(bcf1_t *b, int seed);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
275 bcf_t *bp, *bout = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
276 bcf1_t *b, *blast;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
277 int c, *seeds = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
278 uint64_t n_processed = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
279 viewconf_t vc;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
280 bcf_p1aux_t *p1 = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
281 bcf_hdr_t *hin, *hout;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
282 int tid, begin, end;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
283 char moder[4], modew[4];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
284
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
285 tid = begin = end = -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
286 memset(&vc, 0, sizeof(viewconf_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
287 vc.prior_type = vc.n1 = -1; vc.theta = 1e-3; vc.pref = 0.5; vc.indel_frac = -1.; vc.n_perm = 0; vc.min_perm_p = 0.01; vc.min_smpl_frac = 0; vc.min_lrt = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
288 while ((c = getopt(argc, argv, "FN1:l:cC:eHAGvbSuP:t:p:QgLi:IMs:D:U:X:d:")) >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
289 switch (c) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
290 case '1': vc.n1 = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
291 case 'l': vc.bed = bed_read(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
292 case 'D': vc.fn_dict = strdup(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
293 case 'F': vc.flag |= VC_FIX_PL; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
294 case 'N': vc.flag |= VC_ACGT_ONLY; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
295 case 'G': vc.flag |= VC_NO_GENO; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
296 case 'A': vc.flag |= VC_KEEPALT; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
297 case 'b': vc.flag |= VC_BCFOUT; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
298 case 'S': vc.flag |= VC_VCFIN; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
299 case 'c': vc.flag |= VC_CALL; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
300 case 'e': vc.flag |= VC_EM; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
301 case 'v': vc.flag |= VC_VARONLY | VC_CALL; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
302 case 'u': vc.flag |= VC_UNCOMP | VC_BCFOUT; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
303 case 'g': vc.flag |= VC_CALL_GT | VC_CALL; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
304 case 'I': vc.flag |= VC_NO_INDEL; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
305 case 'M': vc.flag |= VC_ANNO_MAX; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
306 case 't': vc.theta = atof(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
307 case 'p': vc.pref = atof(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
308 case 'i': vc.indel_frac = atof(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
309 case 'Q': vc.flag |= VC_QCALL; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
310 case 'L': vc.flag |= VC_ADJLD; break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
311 case 'U': vc.n_perm = atoi(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
312 case 'C': vc.min_lrt = atof(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
313 case 'X': vc.min_perm_p = atof(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
314 case 'd': vc.min_smpl_frac = atof(optarg); break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
315 case 's': vc.subsam = read_samples(optarg, &vc.n_sub);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
316 vc.ploidy = calloc(vc.n_sub + 1, 1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
317 for (tid = 0; tid < vc.n_sub; ++tid) vc.ploidy[tid] = vc.subsam[tid][strlen(vc.subsam[tid]) + 1];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
318 tid = -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
319 break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
320 case 'P':
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
321 if (strcmp(optarg, "full") == 0) vc.prior_type = MC_PTYPE_FULL;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
322 else if (strcmp(optarg, "cond2") == 0) vc.prior_type = MC_PTYPE_COND2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
323 else if (strcmp(optarg, "flat") == 0) vc.prior_type = MC_PTYPE_FLAT;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
324 else vc.prior_file = strdup(optarg);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
325 break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
326 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
327 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
328 if (argc == optind) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
329 fprintf(stderr, "\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
330 fprintf(stderr, "Usage: bcftools view [options] <in.bcf> [reg]\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
331 fprintf(stderr, "Input/output options:\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
332 fprintf(stderr, " -A keep all possible alternate alleles at variant sites\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
333 fprintf(stderr, " -b output BCF instead of VCF\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
334 fprintf(stderr, " -D FILE sequence dictionary for VCF->BCF conversion [null]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
335 fprintf(stderr, " -F PL generated by r921 or before (which generate old ordering)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
336 fprintf(stderr, " -G suppress all individual genotype information\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
337 fprintf(stderr, " -l FILE list of sites (chr pos) or regions (BED) to output [all sites]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
338 fprintf(stderr, " -L calculate LD for adjacent sites\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
339 fprintf(stderr, " -N skip sites where REF is not A/C/G/T\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
340 fprintf(stderr, " -Q output the QCALL likelihood format\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
341 fprintf(stderr, " -s FILE list of samples to use [all samples]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
342 fprintf(stderr, " -S input is VCF\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
343 fprintf(stderr, " -u uncompressed BCF output (force -b)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
344 fprintf(stderr, "\nConsensus/variant calling options:\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
345 fprintf(stderr, " -c SNP calling (force -e)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
346 fprintf(stderr, " -d FLOAT skip loci where less than FLOAT fraction of samples covered [0]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
347 fprintf(stderr, " -e likelihood based analyses\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
348 fprintf(stderr, " -g call genotypes at variant sites (force -c)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
349 fprintf(stderr, " -i FLOAT indel-to-substitution ratio [%.4g]\n", vc.indel_frac);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
350 fprintf(stderr, " -I skip indels\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
351 fprintf(stderr, " -p FLOAT variant if P(ref|D)<FLOAT [%.3g]\n", vc.pref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
352 fprintf(stderr, " -P STR type of prior: full, cond2, flat [full]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
353 fprintf(stderr, " -t FLOAT scaled substitution mutation rate [%.4g]\n", vc.theta);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
354 fprintf(stderr, " -v output potential variant sites only (force -c)\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
355 fprintf(stderr, "\nContrast calling and association test options:\n\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
356 fprintf(stderr, " -1 INT number of group-1 samples [0]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
357 fprintf(stderr, " -C FLOAT posterior constrast for LRT<FLOAT and P(ref|D)<0.5 [%g]\n", vc.min_lrt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
358 fprintf(stderr, " -U INT number of permutations for association testing (effective with -1) [0]\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
359 fprintf(stderr, " -X FLOAT only perform permutations for P(chi^2)<FLOAT [%g]\n", vc.min_perm_p);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
360 fprintf(stderr, "\n");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
361 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
362 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
363
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
364 if (vc.flag & VC_CALL) vc.flag |= VC_EM;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
365 if ((vc.flag & VC_VCFIN) && (vc.flag & VC_BCFOUT) && vc.fn_dict == 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
366 fprintf(stderr, "[%s] For VCF->BCF conversion please specify the sequence dictionary with -D\n", __func__);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
367 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
368 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
369 if (vc.n1 <= 0) vc.n_perm = 0; // TODO: give a warning here!
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
370 if (vc.n_perm > 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
371 seeds = malloc(vc.n_perm * sizeof(int));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
372 srand48(time(0));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
373 for (c = 0; c < vc.n_perm; ++c) seeds[c] = lrand48();
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
374 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
375 b = calloc(1, sizeof(bcf1_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
376 blast = calloc(1, sizeof(bcf1_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
377 strcpy(moder, "r");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
378 if (!(vc.flag & VC_VCFIN)) strcat(moder, "b");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
379 strcpy(modew, "w");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
380 if (vc.flag & VC_BCFOUT) strcat(modew, "b");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
381 if (vc.flag & VC_UNCOMP) strcat(modew, "u");
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
382 bp = vcf_open(argv[optind], moder);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
383 hin = hout = vcf_hdr_read(bp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
384 if (vc.fn_dict && (vc.flag & VC_VCFIN))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
385 vcf_dictread(bp, hin, vc.fn_dict);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
386 bout = vcf_open("-", modew);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
387 if (!(vc.flag & VC_QCALL)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
388 if (vc.n_sub) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
389 vc.sublist = calloc(vc.n_sub, sizeof(int));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
390 hout = bcf_hdr_subsam(hin, vc.n_sub, vc.subsam, vc.sublist);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
391 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
392 if (vc.flag & VC_CALL) write_header(hout);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
393 vcf_hdr_write(bout, hout);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
394 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
395 if (vc.flag & VC_CALL) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
396 p1 = bcf_p1_init(hout->n_smpl, vc.ploidy);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
397 if (vc.prior_file) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
398 if (bcf_p1_read_prior(p1, vc.prior_file) < 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
399 fprintf(stderr, "[%s] fail to read the prior AFS.\n", __func__);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
400 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
401 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
402 } else bcf_p1_init_prior(p1, vc.prior_type, vc.theta);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
403 if (vc.n1 > 0 && vc.min_lrt > 0.) { // set n1
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
404 bcf_p1_set_n1(p1, vc.n1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
405 bcf_p1_init_subprior(p1, vc.prior_type, vc.theta);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
406 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
407 if (vc.indel_frac > 0.) bcf_p1_indel_prior(p1, vc.indel_frac); // otherwise use the default indel_frac
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
408 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
409 if (optind + 1 < argc && !(vc.flag&VC_VCFIN)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
410 void *str2id = bcf_build_refhash(hout);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
411 if (bcf_parse_region(str2id, argv[optind+1], &tid, &begin, &end) >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
412 bcf_idx_t *idx;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
413 idx = bcf_idx_load(argv[optind]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
414 if (idx) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
415 uint64_t off;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
416 off = bcf_idx_query(idx, tid, begin);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
417 if (off == 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
418 fprintf(stderr, "[%s] no records in the query region.\n", __func__);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
419 return 1; // FIXME: a lot of memory leaks...
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
420 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
421 bgzf_seek(bp->fp, off, SEEK_SET);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
422 bcf_idx_destroy(idx);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
423 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
424 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
425 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
426 while (vcf_read(bp, hin, b) > 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
427 int is_indel;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
428 double em[9];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
429 if ((vc.flag & VC_VARONLY) && strcmp(b->alt, "X") == 0) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
430 if ((vc.flag & VC_VARONLY) && vc.min_smpl_frac > 0.) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
431 extern int bcf_smpl_covered(const bcf1_t *b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
432 int n = bcf_smpl_covered(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
433 if ((double)n / b->n_smpl < vc.min_smpl_frac) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
434 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
435 if (vc.n_sub) bcf_subsam(vc.n_sub, vc.sublist, b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
436 if (vc.flag & VC_FIX_PL) bcf_fix_pl(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
437 is_indel = bcf_is_indel(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
438 if ((vc.flag & VC_NO_INDEL) && is_indel) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
439 if ((vc.flag & VC_ACGT_ONLY) && !is_indel) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
440 int x;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
441 if (b->ref[0] == 0 || b->ref[1] != 0) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
442 x = toupper(b->ref[0]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
443 if (x != 'A' && x != 'C' && x != 'G' && x != 'T') continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
444 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
445 if (vc.bed && !bed_overlap(vc.bed, hin->ns[b->tid], b->pos, b->pos + strlen(b->ref))) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
446 if (tid >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
447 int l = strlen(b->ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
448 l = b->pos + (l > 0? l : 1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
449 if (b->tid != tid || b->pos >= end) break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
450 if (!(l > begin && end > b->pos)) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
451 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
452 ++n_processed;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
453 if (vc.flag & VC_QCALL) { // output QCALL format; STOP here
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
454 bcf_2qcall(hout, b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
455 continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
456 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
457 if (vc.flag & VC_EM) bcf_em1(b, vc.n1, 0xff, em);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
458 else {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
459 int i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
460 for (i = 0; i < 9; ++i) em[i] = -1.;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
461 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
462 if (vc.flag & (VC_CALL|VC_ADJLD)) bcf_gl2pl(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
463 if (vc.flag & VC_CALL) { // call variants
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
464 bcf_p1rst_t pr;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
465 int calret = bcf_p1_cal(b, (em[7] >= 0 && em[7] < vc.min_lrt), p1, &pr);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
466 if (n_processed % 100000 == 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
467 fprintf(stderr, "[%s] %ld sites processed.\n", __func__, (long)n_processed);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
468 bcf_p1_dump_afs(p1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
469 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
470 if (pr.p_ref >= vc.pref && (vc.flag & VC_VARONLY)) continue;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
471 if (vc.n_perm && vc.n1 > 0 && pr.p_chi2 < vc.min_perm_p) { // permutation test
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
472 bcf_p1rst_t r;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
473 int i, n = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
474 for (i = 0; i < vc.n_perm; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
475 #ifdef BCF_PERM_LRT // LRT based permutation is much faster but less robust to artifacts
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
476 double x[9];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
477 bcf_shuffle(b, seeds[i]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
478 bcf_em1(b, vc.n1, 1<<7, x);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
479 if (x[7] < em[7]) ++n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
480 #else
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
481 bcf_shuffle(b, seeds[i]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
482 bcf_p1_cal(b, 1, p1, &r);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
483 if (pr.p_chi2 >= r.p_chi2) ++n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
484 #endif
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
485 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
486 pr.perm_rank = n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
487 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
488 if (calret >= 0) update_bcf1(b, p1, &pr, vc.pref, vc.flag, em);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
489 } else if (vc.flag & VC_EM) update_bcf1(b, 0, 0, 0, vc.flag, em);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
490 if (vc.flag & VC_ADJLD) { // compute LD
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
491 double f[4], r2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
492 if ((r2 = bcf_pair_freq(blast, b, f)) >= 0) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
493 kstring_t s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
494 s.m = s.l = 0; s.s = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
495 if (*b->info) kputc(';', &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
496 ksprintf(&s, "NEIR=%.3f;NEIF4=%.3f,%.3f,%.3f,%.3f", r2, f[0], f[1], f[2], f[3]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
497 bcf_append_info(b, s.s, s.l);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
498 free(s.s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
499 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
500 bcf_cpy(blast, b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
501 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
502 if (vc.flag & VC_ANNO_MAX) bcf_anno_max(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
503 if (vc.flag & VC_NO_GENO) { // do not output GENO fields
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
504 b->n_gi = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
505 b->fmt[0] = '\0';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
506 b->l_str = b->fmt - b->str + 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
507 } else bcf_fix_gt(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
508 vcf_write(bout, hout, b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
509 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
510 if (vc.prior_file) free(vc.prior_file);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
511 if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
512 if (hin != hout) bcf_hdr_destroy(hout);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
513 bcf_hdr_destroy(hin);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
514 bcf_destroy(b); bcf_destroy(blast);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
515 vcf_close(bp); vcf_close(bout);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
516 if (vc.fn_dict) free(vc.fn_dict);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
517 if (vc.ploidy) free(vc.ploidy);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
518 if (vc.n_sub) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
519 int i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
520 for (i = 0; i < vc.n_sub; ++i) free(vc.subsam[i]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
521 free(vc.subsam); free(vc.sublist);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
522 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
523 if (vc.bed) bed_destroy(vc.bed);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
524 if (seeds) free(seeds);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
525 if (p1) bcf_p1_destroy(p1);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
526 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
527 }