ruby-minigraph 0.0.20.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/LICENSE.txt +21 -0
- data/README.md +62 -0
- data/ext/Rakefile +56 -0
- data/ext/cmappy/cmappy.c +7 -0
- data/ext/cmappy/cmappy.h +8 -0
- data/ext/minigraph/LICENSE.txt +23 -0
- data/ext/minigraph/Makefile +66 -0
- data/ext/minigraph/NEWS.md +317 -0
- data/ext/minigraph/README.md +207 -0
- data/ext/minigraph/algo.c +194 -0
- data/ext/minigraph/algo.h +33 -0
- data/ext/minigraph/asm-call.c +147 -0
- data/ext/minigraph/bseq.c +133 -0
- data/ext/minigraph/bseq.h +76 -0
- data/ext/minigraph/cal_cov.c +139 -0
- data/ext/minigraph/doc/example1.png +0 -0
- data/ext/minigraph/doc/example2.png +0 -0
- data/ext/minigraph/doc/examples.graffle +0 -0
- data/ext/minigraph/format.c +241 -0
- data/ext/minigraph/galign.c +140 -0
- data/ext/minigraph/gchain1.c +532 -0
- data/ext/minigraph/gcmisc.c +223 -0
- data/ext/minigraph/gfa-aug.c +260 -0
- data/ext/minigraph/gfa-base.c +526 -0
- data/ext/minigraph/gfa-bbl.c +372 -0
- data/ext/minigraph/gfa-ed.c +617 -0
- data/ext/minigraph/gfa-io.c +395 -0
- data/ext/minigraph/gfa-priv.h +154 -0
- data/ext/minigraph/gfa.h +166 -0
- data/ext/minigraph/ggen.c +182 -0
- data/ext/minigraph/ggen.h +21 -0
- data/ext/minigraph/ggsimple.c +570 -0
- data/ext/minigraph/gmap.c +211 -0
- data/ext/minigraph/index.c +230 -0
- data/ext/minigraph/kalloc.c +224 -0
- data/ext/minigraph/kalloc.h +82 -0
- data/ext/minigraph/kavl.h +414 -0
- data/ext/minigraph/kdq.h +134 -0
- data/ext/minigraph/ketopt.h +116 -0
- data/ext/minigraph/khashl.h +348 -0
- data/ext/minigraph/krmq.h +474 -0
- data/ext/minigraph/kseq.h +256 -0
- data/ext/minigraph/ksort.h +164 -0
- data/ext/minigraph/kstring.h +165 -0
- data/ext/minigraph/kthread.c +159 -0
- data/ext/minigraph/kthread.h +15 -0
- data/ext/minigraph/kvec-km.h +105 -0
- data/ext/minigraph/kvec.h +110 -0
- data/ext/minigraph/lchain.c +441 -0
- data/ext/minigraph/main.c +301 -0
- data/ext/minigraph/map-algo.c +500 -0
- data/ext/minigraph/mgpriv.h +128 -0
- data/ext/minigraph/minigraph.1 +359 -0
- data/ext/minigraph/minigraph.h +176 -0
- data/ext/minigraph/miniwfa.c +834 -0
- data/ext/minigraph/miniwfa.h +95 -0
- data/ext/minigraph/misc/mgutils.js +1451 -0
- data/ext/minigraph/misc.c +12 -0
- data/ext/minigraph/options.c +134 -0
- data/ext/minigraph/shortk.c +251 -0
- data/ext/minigraph/sketch.c +109 -0
- data/ext/minigraph/sys.c +147 -0
- data/ext/minigraph/sys.h +20 -0
- data/ext/minigraph/test/MT-chimp.fa +277 -0
- data/ext/minigraph/test/MT-human.fa +239 -0
- data/ext/minigraph/test/MT-orangA.fa +276 -0
- data/ext/minigraph/test/MT.gfa +19 -0
- data/ext/minigraph/tex/Makefile +13 -0
- data/ext/minigraph/tex/minigraph.bib +676 -0
- data/ext/minigraph/tex/minigraph.tex +986 -0
- data/ext/minigraph/tex/plots/CHM13-f1-90.bb.anno.gp +42 -0
- data/ext/minigraph/tex/plots/CHM13-f1-90.bb.anno.tbl +13 -0
- data/ext/minigraph/tex/plots/CHM13-f1-90.bb.mini-inter-none.win.gp +269 -0
- data/ext/minigraph/tex/plots/CHM13-f1-90.bb.mini-inter-none.win.sh +7 -0
- data/ext/minigraph/tex/plots/CHM13v1.cen.bed +23 -0
- data/ext/minigraph/tex/plots/CHM13v1.size +23 -0
- data/ext/minigraph/tex/plots/anno2tbl.js +40 -0
- data/ext/minigraph/tex/plots/bedutils.js +367 -0
- data/ext/minigraph/tex/plots/chr-plot.js +130 -0
- data/ext/minigraph/tex/plots/gen-anno.mak +24 -0
- data/ext/minigraph.patch +21 -0
- data/lib/minigraph/ffi/constants.rb +230 -0
- data/lib/minigraph/ffi/functions.rb +70 -0
- data/lib/minigraph/ffi/mappy.rb +8 -0
- data/lib/minigraph/ffi.rb +27 -0
- data/lib/minigraph/version.rb +5 -0
- data/lib/minigraph.rb +72 -0
- metadata +159 -0
@@ -0,0 +1,834 @@
|
|
1
|
+
#include <string.h>
|
2
|
+
#include <assert.h>
|
3
|
+
#include <stdlib.h>
|
4
|
+
#include <stdio.h>
|
5
|
+
#include "miniwfa.h"
|
6
|
+
#include "kalloc.h"
|
7
|
+
|
8
|
+
/*
|
9
|
+
* Default setting
|
10
|
+
*/
|
11
|
+
void mwf_opt_init(mwf_opt_t *opt)
|
12
|
+
{
|
13
|
+
memset(opt, 0, sizeof(*opt));
|
14
|
+
opt->x = 4; // corresponding SW score: m=1, x=3, o1=4, e1=3/2, o2=15, e2=1/2
|
15
|
+
opt->o1 = 4, opt->e1 = 2;
|
16
|
+
opt->o2 = 15, opt->e2 = 1;
|
17
|
+
opt->kmer = 13, opt->max_occ = 2, opt->min_len = 30;
|
18
|
+
}
|
19
|
+
|
20
|
+
/*
|
21
|
+
* Structs and simple functions for traceback
|
22
|
+
*/
|
23
|
+
typedef struct {
|
24
|
+
int32_t lo, hi;
|
25
|
+
uint8_t *x;
|
26
|
+
} wf_tb1_t;
|
27
|
+
|
28
|
+
typedef struct {
|
29
|
+
int32_t m, n;
|
30
|
+
wf_tb1_t *a;
|
31
|
+
} wf_tb_t;
|
32
|
+
|
33
|
+
static wf_tb1_t *wf_tb_add(void *km, wf_tb_t *tb, int32_t lo, int32_t hi)
|
34
|
+
{
|
35
|
+
wf_tb1_t *p;
|
36
|
+
if (tb->n == tb->m) {
|
37
|
+
tb->m += (tb->m>>1) + 4;
|
38
|
+
tb->a = Krealloc(km, wf_tb1_t, tb->a, tb->m);
|
39
|
+
}
|
40
|
+
p = &tb->a[tb->n++];
|
41
|
+
p->lo = lo, p->hi = hi;
|
42
|
+
p->x = Kcalloc(km, uint8_t, hi - lo + 1);
|
43
|
+
return p;
|
44
|
+
}
|
45
|
+
|
46
|
+
typedef struct {
|
47
|
+
int32_t m, n;
|
48
|
+
uint32_t *cigar;
|
49
|
+
} wf_cigar_t;
|
50
|
+
|
51
|
+
static void wf_cigar_push1(void *km, wf_cigar_t *c, int32_t op, int32_t len)
|
52
|
+
{
|
53
|
+
if (c->n && op == (c->cigar[c->n-1]&0xf)) {
|
54
|
+
c->cigar[c->n-1] += len<<4;
|
55
|
+
} else {
|
56
|
+
if (c->n == c->m) {
|
57
|
+
c->m = c->m + (c->m>>1) + 4;
|
58
|
+
c->cigar = Krealloc(km, uint32_t, c->cigar, c->m);
|
59
|
+
}
|
60
|
+
c->cigar[c->n++] = len<<4 | op;
|
61
|
+
}
|
62
|
+
}
|
63
|
+
|
64
|
+
/*
|
65
|
+
* The stripe data structure
|
66
|
+
*/
|
67
|
+
#define WF_NEG_INF (-0x40000000)
|
68
|
+
|
69
|
+
typedef struct {
|
70
|
+
int32_t lo, hi;
|
71
|
+
int32_t *mem, *H, *E1, *E2, *F1, *F2;
|
72
|
+
} wf_slice_t;
|
73
|
+
|
74
|
+
typedef struct {
|
75
|
+
int32_t s, top, n, max_pen, lo, hi;
|
76
|
+
wf_slice_t *a;
|
77
|
+
} wf_stripe_t;
|
78
|
+
|
79
|
+
void wf_stripe_add(void *km, wf_stripe_t *wf, int32_t lo, int32_t hi)
|
80
|
+
{
|
81
|
+
int32_t i, n, m1 = wf->max_pen + 1, m2 = m1 * 2;
|
82
|
+
wf_slice_t *f;
|
83
|
+
++wf->s;
|
84
|
+
++wf->top;
|
85
|
+
if (wf->top == wf->n) wf->top = 0;
|
86
|
+
f = &wf->a[wf->top];
|
87
|
+
f->lo = lo, f->hi = hi;
|
88
|
+
n = hi - lo + 1;
|
89
|
+
kfree(km, f->mem);
|
90
|
+
f->mem = Kmalloc(km, int32_t, 5 * (n + m2));
|
91
|
+
f->H = f->mem + m1;
|
92
|
+
f->E1 = f->H + n + m2;
|
93
|
+
f->F1 = f->E1 + n + m2;
|
94
|
+
f->E2 = f->F1 + n + m2;
|
95
|
+
f->F2 = f->E2 + n + m2;
|
96
|
+
for (i = -m1; i < 0; ++i)
|
97
|
+
f->H[i] = f->E1[i] = f->E2[i] = f->F1[i] = f->F2[i] = WF_NEG_INF;
|
98
|
+
for (i = n; i < n + m1; ++i)
|
99
|
+
f->H[i] = f->E1[i] = f->E2[i] = f->F1[i] = f->F2[i] = WF_NEG_INF;
|
100
|
+
f->H -= lo, f->E1 -= lo, f->E2 -= lo, f->F1 -= lo, f->F2 -= lo; // such that f->H[lo] points to 0
|
101
|
+
}
|
102
|
+
|
103
|
+
static wf_stripe_t *wf_stripe_init(void *km, int32_t max_pen)
|
104
|
+
{
|
105
|
+
int32_t i;
|
106
|
+
wf_stripe_t *wf;
|
107
|
+
wf = Kcalloc(km, wf_stripe_t, 1);
|
108
|
+
wf->max_pen = max_pen;
|
109
|
+
wf->n = max_pen + 1;
|
110
|
+
wf->a = Kcalloc(km, wf_slice_t, wf->n);
|
111
|
+
wf->lo = wf->hi = 0;
|
112
|
+
for (i = 0; i < wf->n; ++i) {
|
113
|
+
wf_slice_t *f;
|
114
|
+
wf_stripe_add(km, wf, 0, 0);
|
115
|
+
f = &wf->a[wf->top];
|
116
|
+
f->H[0] = f->E1[0] = f->E2[0] = f->F1[0] = f->F2[0] = WF_NEG_INF;
|
117
|
+
}
|
118
|
+
wf->s = 0;
|
119
|
+
wf->a[wf->top].H[0] = -1;
|
120
|
+
return wf;
|
121
|
+
}
|
122
|
+
|
123
|
+
static void wf_stripe_destroy(void *km, wf_stripe_t *wf)
|
124
|
+
{
|
125
|
+
int32_t i;
|
126
|
+
for (i = 0; i < wf->n; ++i)
|
127
|
+
kfree(km, wf->a[i].mem);
|
128
|
+
kfree(km, wf->a);
|
129
|
+
kfree(km, wf);
|
130
|
+
}
|
131
|
+
|
132
|
+
static inline wf_slice_t *wf_stripe_get(const wf_stripe_t *wf, int32_t x)
|
133
|
+
{
|
134
|
+
int32_t y = wf->top - x;
|
135
|
+
if (y < 0) y += wf->n;
|
136
|
+
return &wf->a[y];
|
137
|
+
}
|
138
|
+
|
139
|
+
static inline int good_diag(int32_t d, int32_t k, int32_t tl, int32_t ql) // check if (d,k) falls within the DP matrix
|
140
|
+
{
|
141
|
+
return ((k >= -1 && k < tl) && (d + k >= -1 && d + k < ql));
|
142
|
+
}
|
143
|
+
|
144
|
+
static void wf_stripe_shrink(wf_stripe_t *wf, int32_t tl, int32_t ql)
|
145
|
+
{
|
146
|
+
int32_t j, d;
|
147
|
+
for (d = wf->lo; d <= wf->hi; ++d) {
|
148
|
+
for (j = 0; j < wf->n; ++j) {
|
149
|
+
wf_slice_t *p = &wf->a[(wf->top + 1 + j) % wf->n];
|
150
|
+
if (d < p->lo || d > p->hi) continue;
|
151
|
+
if (good_diag(d, p->H[d], tl, ql)) break;
|
152
|
+
if (good_diag(d, p->E1[d], tl, ql) || good_diag(d, p->F1[d], tl, ql)) break;
|
153
|
+
if (good_diag(d, p->E2[d], tl, ql) || good_diag(d, p->F2[d], tl, ql)) break;
|
154
|
+
}
|
155
|
+
if (j < wf->n) break; // stop when we see a "good diagonal" in the stripe
|
156
|
+
}
|
157
|
+
assert(d <= wf->hi); // should never happen
|
158
|
+
wf->lo = d;
|
159
|
+
for (d = wf->hi; d >= wf->lo; --d) {
|
160
|
+
for (j = 0; j < wf->n; ++j) {
|
161
|
+
wf_slice_t *p = &wf->a[(wf->top + 1 + j) % wf->n];
|
162
|
+
if (d < p->lo || d > p->hi) continue;
|
163
|
+
if (good_diag(d, p->H[d], tl, ql)) break;
|
164
|
+
if (good_diag(d, p->E1[d], tl, ql) || good_diag(d, p->F1[d], tl, ql)) break;
|
165
|
+
if (good_diag(d, p->E2[d], tl, ql) || good_diag(d, p->F2[d], tl, ql)) break;
|
166
|
+
}
|
167
|
+
if (j < wf->n) break;
|
168
|
+
}
|
169
|
+
assert(d >= wf->lo);
|
170
|
+
wf->hi = d;
|
171
|
+
}
|
172
|
+
|
173
|
+
typedef struct {
|
174
|
+
int32_t s, d;
|
175
|
+
} wf_chkpt_t;
|
176
|
+
|
177
|
+
/*
|
178
|
+
* Extend a diagonal along exact matches
|
179
|
+
*/
|
180
|
+
|
181
|
+
// pad strings with distinct characters
|
182
|
+
static void wf_pad_str(void *km, int32_t tl, const char *ts, int32_t ql, const char *qs, char **pts, char **pqs)
|
183
|
+
{
|
184
|
+
uint8_t t[256];
|
185
|
+
int32_t i, c1 = -1, c2 = -1;
|
186
|
+
char *s1, *s2;
|
187
|
+
*pts = *pqs = 0;
|
188
|
+
// collect all used characters
|
189
|
+
memset(t, 0, 256);
|
190
|
+
for (i = 0; i < tl; ++i)
|
191
|
+
if (t[(uint8_t)ts[i]] == 0)
|
192
|
+
t[(uint8_t)ts[i]] = 1;
|
193
|
+
for (i = 0; i < ql; ++i)
|
194
|
+
if (t[(uint8_t)qs[i]] == 0)
|
195
|
+
t[(uint8_t)qs[i]] = 1;
|
196
|
+
for (i = 0; i < 256; ++i)
|
197
|
+
if (t[i] == 0) {
|
198
|
+
if (c1 < 0) c1 = i;
|
199
|
+
else if (c2 < 0) c2 = i;
|
200
|
+
}
|
201
|
+
if (c1 < 0 || c2 < 0) return; // The two strings use >=255 characters. Unlikely for bio strings.
|
202
|
+
s1 = Kmalloc(km, char, tl + ql + 16); // the two strings are allocated together
|
203
|
+
s2 = s1 + tl + 8;
|
204
|
+
memcpy(s1, ts, tl);
|
205
|
+
for (i = tl; i < tl + 8; ++i) s1[i] = c1; // pad with c1
|
206
|
+
memcpy(s2, qs, ql);
|
207
|
+
for (i = ql; i < ql + 8; ++i) s2[i] = c2; // pad with c2
|
208
|
+
*pts = s1, *pqs = s2;
|
209
|
+
}
|
210
|
+
|
211
|
+
// Extend a diagonal along exact matches.
|
212
|
+
static inline int32_t wf_extend1_padded(const char *ts, const char *qs, int32_t k, int32_t d)
|
213
|
+
{
|
214
|
+
uint64_t cmp = 0;
|
215
|
+
const char *ts_ = ts + 1;
|
216
|
+
const char *qs_ = qs + d + 1;
|
217
|
+
while (1) {
|
218
|
+
uint64_t x = *(uint64_t*)(ts_ + k); // warning: unaligned memory access
|
219
|
+
uint64_t y = *(uint64_t*)(qs_ + k);
|
220
|
+
cmp = x ^ y;
|
221
|
+
if (cmp == 0) k += 8;
|
222
|
+
else break;
|
223
|
+
}
|
224
|
+
k += __builtin_ctzl(cmp) >> 3;
|
225
|
+
return k;
|
226
|
+
}
|
227
|
+
|
228
|
+
/*
|
229
|
+
* Core wf_next() routines
|
230
|
+
*/
|
231
|
+
|
232
|
+
// Force loop vectorization. Learned from WFA.
|
233
|
+
#if defined(__clang__)
|
234
|
+
#define PRAGMA_LOOP_VECTORIZE _Pragma("clang loop vectorize(enable)")
|
235
|
+
#elif defined(__GNUC__)
|
236
|
+
#define PRAGMA_LOOP_VECTORIZE _Pragma("GCC ivdep")
|
237
|
+
#else
|
238
|
+
#define PRAGMA_LOOP_VECTORIZE _Pragma("ivdep")
|
239
|
+
#endif
|
240
|
+
|
241
|
+
#define wf_max(a, b) ((a) >= (b)? (a) : (b))
|
242
|
+
|
243
|
+
static void wf_next_prep(void *km, const mwf_opt_t *opt, wf_stripe_t *wf, int32_t lo, int32_t hi,
|
244
|
+
int32_t **H, int32_t **E1, int32_t **F1, int32_t **E2, int32_t **F2,
|
245
|
+
const int32_t **pHx, const int32_t **pHo1, const int32_t **pHo2,
|
246
|
+
const int32_t **pE1, const int32_t **pF1, const int32_t **pE2, const int32_t **pF2)
|
247
|
+
{
|
248
|
+
const wf_slice_t *fx, *fo1, *fo2, *fe1, *fe2;
|
249
|
+
wf_slice_t *ft;
|
250
|
+
wf_stripe_add(km, wf, lo, hi);
|
251
|
+
ft = &wf->a[wf->top];
|
252
|
+
fx = wf_stripe_get(wf, opt->x);
|
253
|
+
fo1 = wf_stripe_get(wf, opt->o1 + opt->e1);
|
254
|
+
fo2 = wf_stripe_get(wf, opt->o2 + opt->e2);
|
255
|
+
fe1 = wf_stripe_get(wf, opt->e1);
|
256
|
+
fe2 = wf_stripe_get(wf, opt->e2);
|
257
|
+
*pHx = fx->H, *pHo1 = fo1->H, *pHo2 = fo2->H, *pE1 = fe1->E1, *pE2 = fe2->E2, *pF1 = fe1->F1, *pF2 = fe2->F2;
|
258
|
+
*H = ft->H, *E1 = ft->E1, *E2 = ft->E2, *F1 = ft->F1, *F2 = ft->F2;
|
259
|
+
}
|
260
|
+
|
261
|
+
static void wf_next_score(int32_t lo, int32_t hi, int32_t *H, int32_t *E1, int32_t *F1, int32_t *E2, int32_t *F2,
|
262
|
+
const int32_t *pHx, const int32_t *pHo1, const int32_t *pHo2,
|
263
|
+
const int32_t *pE1, const int32_t *pF1, const int32_t *pE2, const int32_t *pF2)
|
264
|
+
{
|
265
|
+
int32_t d;
|
266
|
+
PRAGMA_LOOP_VECTORIZE
|
267
|
+
for (d = lo; d <= hi; ++d) {
|
268
|
+
int32_t h, f, e;
|
269
|
+
E1[d] = wf_max(pHo1[d-1], pE1[d-1]);
|
270
|
+
E2[d] = wf_max(pHo2[d-1], pE2[d-1]);
|
271
|
+
e = wf_max(E1[d], E2[d]);
|
272
|
+
F1[d] = wf_max(pHo1[d+1], pF1[d+1]) + 1;
|
273
|
+
F2[d] = wf_max(pHo2[d+1], pF2[d+1]) + 1;
|
274
|
+
f = wf_max(F1[d], F2[d]);
|
275
|
+
h = wf_max(e, f);
|
276
|
+
H[d] = wf_max(pHx[d] + 1, h);
|
277
|
+
// if (H[d] >= -1) fprintf(stderr, "s=%d, d=%d, k=%d, (%d,%d)\n", wf->s, d, H[d], E1[d], F1[d]);
|
278
|
+
}
|
279
|
+
}
|
280
|
+
|
281
|
+
static void wf_next_tb(int32_t lo, int32_t hi, int32_t *H, int32_t *E1, int32_t *F1, int32_t *E2, int32_t *F2, uint8_t *ax,
|
282
|
+
const int32_t *pHx, const int32_t *pHo1, const int32_t *pHo2,
|
283
|
+
const int32_t *pE1, const int32_t *pF1, const int32_t *pE2, const int32_t *pF2)
|
284
|
+
{
|
285
|
+
int32_t d;
|
286
|
+
PRAGMA_LOOP_VECTORIZE
|
287
|
+
for (d = lo; d <= hi; ++d) {
|
288
|
+
int32_t h, f, e;
|
289
|
+
uint8_t x = 0, ze, zf, z;
|
290
|
+
x |= pHo1[d-1] >= pE1[d-1]? 0 : 0x08;
|
291
|
+
E1[d] = wf_max(pHo1[d-1], pE1[d-1]);
|
292
|
+
x |= pHo2[d-1] >= pE2[d-1]? 0 : 0x20;
|
293
|
+
E2[d] = wf_max(pHo2[d-1], pE2[d-1]);
|
294
|
+
ze = E1[d] >= E2[d]? 1 : 3;
|
295
|
+
e = wf_max(E1[d], E2[d]);
|
296
|
+
x |= pHo1[d+1] >= pF1[d+1]? 0 : 0x10;
|
297
|
+
F1[d] = wf_max(pHo1[d+1], pF1[d+1]) + 1;
|
298
|
+
x |= pHo2[d+1] >= pF2[d+1]? 0 : 0x40;
|
299
|
+
F2[d] = wf_max(pHo2[d+1], pF2[d+1]) + 1;
|
300
|
+
zf = F1[d] >= F2[d]? 2 : 4;
|
301
|
+
f = wf_max(F1[d], F2[d]);
|
302
|
+
z = e >= f? ze : zf;
|
303
|
+
h = wf_max(e, f);
|
304
|
+
z = pHx[d] + 1 >= h? 0 : z;
|
305
|
+
H[d] = wf_max(pHx[d] + 1, h);
|
306
|
+
ax[d] = x | z;
|
307
|
+
}
|
308
|
+
}
|
309
|
+
|
310
|
+
/*
|
311
|
+
* Core algorithm
|
312
|
+
*/
|
313
|
+
static void wf_next_basic(void *km, void *km_tb, const mwf_opt_t *opt, wf_stripe_t *wf, wf_tb_t *tb, int32_t lo, int32_t hi)
|
314
|
+
{
|
315
|
+
int32_t *H, *E1, *E2, *F1, *F2;
|
316
|
+
const int32_t *pHx, *pHo1, *pHo2, *pE1, *pE2, *pF1, *pF2;
|
317
|
+
wf_next_prep(km, opt, wf, lo, hi, &H, &E1, &F1, &E2, &F2, &pHx, &pHo1, &pHo2, &pE1, &pF1, &pE2, &pF2);
|
318
|
+
if (tb) {
|
319
|
+
uint8_t *ax;
|
320
|
+
ax = wf_tb_add(km_tb, tb, lo, hi)->x - lo;
|
321
|
+
wf_next_tb(lo, hi, H, E1, F1, E2, F2, ax, pHx, pHo1, pHo2, pE1, pF1, pE2, pF2);
|
322
|
+
} else {
|
323
|
+
wf_next_score(lo, hi, H, E1, F1, E2, F2, pHx, pHo1, pHo2, pE1, pF1, pE2, pF2);
|
324
|
+
}
|
325
|
+
if (H[lo] >= -1 || E1[lo] >= -1 || F1[lo] >= -1 || E2[lo] >= -1 || F2[lo] >= -1) wf->lo = lo;
|
326
|
+
if (H[hi] >= -1 || E1[hi] >= -1 || F1[hi] >= -1 || E2[hi] >= -1 || F2[hi] >= -1) wf->hi = hi;
|
327
|
+
}
|
328
|
+
|
329
|
+
static uint32_t *wf_traceback(void *km, const mwf_opt_t *opt, wf_tb_t *tb, int32_t t_end, const char *ts, int32_t q_end, const char *qs, int32_t last, int32_t *n_cigar)
|
330
|
+
{
|
331
|
+
wf_cigar_t cigar = {0,0,0};
|
332
|
+
int32_t i = q_end, k = t_end, s = tb->n - 1;
|
333
|
+
while (i >= 0 && k >= 0) {
|
334
|
+
int32_t k0 = k, j, x, state, ext;
|
335
|
+
if (last == 0) { // if the previous state is 0, check exact matches
|
336
|
+
while (i >= 0 && k >= 0 && qs[i] == ts[k])
|
337
|
+
--i, --k;
|
338
|
+
if (k0 - k > 0)
|
339
|
+
wf_cigar_push1(km, &cigar, 7, k0 - k);
|
340
|
+
if (i < 0 || k < 0) break;
|
341
|
+
}
|
342
|
+
assert(s >= 0);
|
343
|
+
j = i - k - tb->a[s].lo;
|
344
|
+
assert(j <= tb->a[s].hi - tb->a[s].lo);
|
345
|
+
x = tb->a[s].x[j];
|
346
|
+
state = last == 0? x&7 : last;
|
347
|
+
ext = state > 0? x>>(state+2)&1 : 0; // whether an extension
|
348
|
+
//fprintf(stderr, "s=%d, %d->%d, ext=%d%d%d%d, i=%d, k=%d\n", s, last, state, x>>3&1, x>>4&1, x>>5&1, x>>6&1, i, k);
|
349
|
+
if (state == 0) {
|
350
|
+
wf_cigar_push1(km, &cigar, 8, 1);
|
351
|
+
--i, --k, s -= opt->x;
|
352
|
+
} else if (state == 1) {
|
353
|
+
wf_cigar_push1(km, &cigar, 1, 1);
|
354
|
+
--i, s -= ext? opt->e1 : opt->o1 + opt->e1;
|
355
|
+
} else if (state == 3) {
|
356
|
+
wf_cigar_push1(km, &cigar, 1, 1);
|
357
|
+
--i, s -= ext? opt->e2 : opt->o2 + opt->e2;
|
358
|
+
} else if (state == 2) {
|
359
|
+
wf_cigar_push1(km, &cigar, 2, 1);
|
360
|
+
--k, s -= ext? opt->e1 : opt->o1 + opt->e1;
|
361
|
+
} else if (state == 4) {
|
362
|
+
wf_cigar_push1(km, &cigar, 2, 1);
|
363
|
+
--k, s -= ext? opt->e2 : opt->o2 + opt->e2;
|
364
|
+
} else abort();
|
365
|
+
last = state > 0 && ext? state : 0;
|
366
|
+
}
|
367
|
+
if (opt->flag&MWF_F_DEBUG) fprintf(stderr, "s0=%d, s=%d, i=%d, k=%d\n", tb->n-1, s, i, k);
|
368
|
+
if (i >= 0) wf_cigar_push1(km, &cigar, 1, i + 1);
|
369
|
+
else if (k >= 0) wf_cigar_push1(km, &cigar, 2, k + 1);
|
370
|
+
for (i = 0; i < cigar.n>>1; ++i) { // reverse to the input order
|
371
|
+
uint32_t t = cigar.cigar[i];
|
372
|
+
cigar.cigar[i] = cigar.cigar[cigar.n - i - 1];
|
373
|
+
cigar.cigar[cigar.n - i - 1] = t;
|
374
|
+
}
|
375
|
+
*n_cigar = cigar.n;
|
376
|
+
return cigar.cigar;
|
377
|
+
}
|
378
|
+
|
379
|
+
// pts and pqs MUST BE padded with wf_pad_str()
|
380
|
+
static void mwf_wfa_core(void *km, const mwf_opt_t *opt, int32_t tl, const char *pts, int32_t ql, const char *pqs, int32_t n_seg, wf_chkpt_t *seg, mwf_rst_t *r)
|
381
|
+
{
|
382
|
+
int32_t max_pen, sid, is_tb = !!(opt->flag&MWF_F_CIGAR), last_state = 0, stopped = 0;
|
383
|
+
wf_stripe_t *wf;
|
384
|
+
wf_tb_t tb = {0,0,0};
|
385
|
+
void *km_tb, *km_st;
|
386
|
+
|
387
|
+
memset(r, 0, sizeof(*r));
|
388
|
+
km_tb = is_tb && !(opt->flag&MWF_F_NO_KALLOC)? km_init2(km, 0) : 0;
|
389
|
+
km_st = !(opt->flag&MWF_F_NO_KALLOC)? km_init2(km, 0) : 0;
|
390
|
+
max_pen = opt->x;
|
391
|
+
max_pen = max_pen > opt->o1 + opt->e1? max_pen : opt->o1 + opt->e1;
|
392
|
+
max_pen = max_pen > opt->o2 + opt->e2? max_pen : opt->o2 + opt->e2;
|
393
|
+
wf = wf_stripe_init(km_st, max_pen);
|
394
|
+
assert(pts);
|
395
|
+
|
396
|
+
sid = 0;
|
397
|
+
while (1) {
|
398
|
+
wf_slice_t *p = &wf->a[wf->top];
|
399
|
+
int32_t d, lo, hi, *H = p->H;
|
400
|
+
for (d = p->lo; d <= p->hi; ++d) {
|
401
|
+
int32_t k = 0;
|
402
|
+
if (H[d] < -1 || d + H[d] < -1 || H[d] >= tl || d + H[d] >= ql) continue;
|
403
|
+
k = wf_extend1_padded(pts, pqs, H[d], d);
|
404
|
+
//fprintf(stderr, "[s=%d] [%d,%d]:%d %d->%d,%d,%d,%d,%d\n", wf->s, p->lo, p->hi, d, H[d], k, wf->a[wf->top].E1[d], wf->a[wf->top].F1[d], wf->a[wf->top].E2[d], wf->a[wf->top].F2[d]);
|
405
|
+
if (k == tl - 1 && d + k == ql - 1) {
|
406
|
+
if (k == H[d] && is_tb)
|
407
|
+
last_state = tb.a[tb.n-1].x[d - tb.a[tb.n-1].lo] & 7;
|
408
|
+
break;
|
409
|
+
}
|
410
|
+
H[d] = k;
|
411
|
+
}
|
412
|
+
if (d <= p->hi) break;
|
413
|
+
if (is_tb && seg && sid < n_seg && seg[sid].s == wf->s) {
|
414
|
+
assert(seg[sid].d >= wf->lo && seg[sid].d <= wf->hi);
|
415
|
+
wf->lo = wf->hi = seg[sid++].d;
|
416
|
+
}
|
417
|
+
lo = wf->lo > -tl? wf->lo - 1 : -tl;
|
418
|
+
hi = wf->hi < ql? wf->hi + 1 : ql;
|
419
|
+
wf_next_basic(km_st, km_tb, opt, wf, is_tb? &tb : 0, lo, hi);
|
420
|
+
if ((wf->s&0xff) == 0) wf_stripe_shrink(wf, tl, ql);
|
421
|
+
r->n_iter += hi - lo + 1;
|
422
|
+
if ((opt->max_iter > 0 && r->n_iter > opt->max_iter) || (opt->max_s > 0 && wf->s > opt->max_s)) {
|
423
|
+
stopped = 1;
|
424
|
+
break;
|
425
|
+
}
|
426
|
+
}
|
427
|
+
r->s = stopped? -1 : wf->s;
|
428
|
+
if (is_tb && !stopped)
|
429
|
+
r->cigar = wf_traceback(km, opt, &tb, tl-1, pts, ql-1, pqs, last_state, &r->n_cigar);
|
430
|
+
if (km_st == 0) wf_stripe_destroy(km_st, wf);
|
431
|
+
else km_destroy(km_st);
|
432
|
+
km_destroy(km_tb);
|
433
|
+
if (is_tb && !stopped)
|
434
|
+
r->cigar = (uint32_t*)krelocate(km, r->cigar, r->n_cigar * sizeof(*r->cigar));
|
435
|
+
}
|
436
|
+
|
437
|
+
/*
|
438
|
+
* Low-memory mode
|
439
|
+
*/
|
440
|
+
typedef struct {
|
441
|
+
int32_t n, n_intv, max_s;
|
442
|
+
int32_t *x;
|
443
|
+
uint64_t *intv;
|
444
|
+
} wf_ss_t; // snapshot
|
445
|
+
|
446
|
+
typedef struct {
|
447
|
+
int32_t n, m;
|
448
|
+
wf_ss_t *a;
|
449
|
+
} wf_sss_t;
|
450
|
+
|
451
|
+
static void wf_snapshot1(void *km, wf_stripe_t *sf, wf_ss_t *ss)
|
452
|
+
{
|
453
|
+
int32_t j, k, t;
|
454
|
+
ss->n = 0, ss->max_s = sf->s;
|
455
|
+
for (j = 0; j < sf->n; ++j)
|
456
|
+
ss->n += 5 * (sf->a[j].hi - sf->a[j].lo + 1);
|
457
|
+
ss->x = Kmalloc(km, int32_t, ss->n);
|
458
|
+
ss->n_intv = sf->n;
|
459
|
+
ss->intv = Kmalloc(km, uint64_t, ss->n_intv);
|
460
|
+
for (j = 0, t = 0; j < sf->n; ++j) {
|
461
|
+
wf_slice_t *p;
|
462
|
+
k = (sf->top + 1 + j) % sf->n;
|
463
|
+
p = &sf->a[k];
|
464
|
+
ss->intv[j] = (uint64_t)p->lo << 32 | (p->hi - p->lo + 1) * 5;
|
465
|
+
for (k = p->lo; k <= p->hi; ++k) {
|
466
|
+
ss->x[t] = p->H[k], p->H[k] = t++;
|
467
|
+
ss->x[t] = p->E1[k], p->E1[k] = t++;
|
468
|
+
ss->x[t] = p->F1[k], p->F1[k] = t++;
|
469
|
+
ss->x[t] = p->E2[k], p->E2[k] = t++;
|
470
|
+
ss->x[t] = p->F2[k], p->F2[k] = t++;
|
471
|
+
}
|
472
|
+
}
|
473
|
+
assert(t == ss->n);
|
474
|
+
}
|
475
|
+
|
476
|
+
static void wf_snapshot(void *km, wf_sss_t *sss, wf_stripe_t *sf)
|
477
|
+
{
|
478
|
+
if (sss->n == sss->m) {
|
479
|
+
sss->m += (sss->m>>1) + 8;
|
480
|
+
sss->a = Krealloc(km, wf_ss_t, sss->a, sss->m);
|
481
|
+
}
|
482
|
+
wf_snapshot1(km, sf, &sss->a[sss->n++]);
|
483
|
+
}
|
484
|
+
|
485
|
+
static void wf_snapshot_free(void *km, wf_sss_t *sss)
|
486
|
+
{
|
487
|
+
int32_t j;
|
488
|
+
for (j = 0; j < sss->n; ++j) {
|
489
|
+
kfree(km, sss->a[j].x);
|
490
|
+
kfree(km, sss->a[j].intv);
|
491
|
+
}
|
492
|
+
kfree(km, sss->a);
|
493
|
+
}
|
494
|
+
|
495
|
+
static void wf_next_seg(void *km, const mwf_opt_t *opt, uint8_t *xbuf, wf_stripe_t *wf, wf_stripe_t *sf, int32_t lo, int32_t hi)
|
496
|
+
{
|
497
|
+
int32_t d, *H, *E1, *E2, *F1, *F2;
|
498
|
+
const int32_t *pHx, *pHo1, *pHo2, *pE1, *pE2, *pF1, *pF2;
|
499
|
+
uint8_t *ax = xbuf - lo;
|
500
|
+
|
501
|
+
wf_next_prep(km, opt, wf, lo, hi, &H, &E1, &F1, &E2, &F2, &pHx, &pHo1, &pHo2, &pE1, &pF1, &pE2, &pF2);
|
502
|
+
wf_next_tb(lo, hi, H, E1, F1, E2, F2, ax, pHx, pHo1, pHo2, pE1, pF1, pE2, pF2);
|
503
|
+
wf_next_prep(km, opt, sf, lo, hi, &H, &E1, &F1, &E2, &F2, &pHx, &pHo1, &pHo2, &pE1, &pF1, &pE2, &pF2);
|
504
|
+
PRAGMA_LOOP_VECTORIZE
|
505
|
+
for (d = lo; d <= hi; ++d) { // FIXME: merge this loop into the loop in wf_next_tb(). I tried but couldn't make clang vectorize.
|
506
|
+
uint8_t x = ax[d];
|
507
|
+
int32_t a, b, e1, f1, e2, f2, h;
|
508
|
+
a = pHo1[d-1], b = pE1[d-1];
|
509
|
+
e1 = E1[d] = (x&0x08) == 0? a : b;
|
510
|
+
a = pHo1[d+1], b = pF1[d+1];
|
511
|
+
f1 = F1[d] = (x&0x10) == 0? a : b;
|
512
|
+
a = pHo2[d-1], b = pE2[d-1];
|
513
|
+
e2 = E2[d] = (x&0x20) == 0? a : b;
|
514
|
+
a = pHo2[d+1], b = pF2[d+1];
|
515
|
+
f2 = F2[d] = (x&0x40) == 0? a : b;
|
516
|
+
x &= 7;
|
517
|
+
h = pHx[d];
|
518
|
+
h = x == 1? e1 : h;
|
519
|
+
h = x == 2? f1 : h;
|
520
|
+
h = x == 3? e2 : h;
|
521
|
+
h = x == 4? f2 : h;
|
522
|
+
H[d] = h;
|
523
|
+
}
|
524
|
+
if (H[lo] >= -1 || E1[lo] >= -1 || F1[lo] >= -1 || E2[lo] >= -1 || F2[lo] >= -1) wf->lo = lo;
|
525
|
+
if (H[hi] >= -1 || E1[hi] >= -1 || F1[hi] >= -1 || E2[hi] >= -1 || F2[hi] >= -1) wf->hi = hi;
|
526
|
+
}
|
527
|
+
|
528
|
+
static wf_chkpt_t *wf_traceback_seg(void *km, wf_sss_t *sss, int32_t last, int32_t *n_seg)
|
529
|
+
{
|
530
|
+
int32_t j;
|
531
|
+
wf_chkpt_t *seg;
|
532
|
+
*n_seg = sss->n;
|
533
|
+
seg = Kmalloc(km, wf_chkpt_t, sss->n);
|
534
|
+
for (j = sss->n - 1; j >= 0; --j) {
|
535
|
+
int32_t k, m;
|
536
|
+
wf_ss_t *p = &sss->a[j];
|
537
|
+
for (k = 0, m = 0; k < p->n_intv; ++k) {
|
538
|
+
if (last >= m && last < m + (int32_t)p->intv[k])
|
539
|
+
break;
|
540
|
+
m += (int32_t)p->intv[k];
|
541
|
+
}
|
542
|
+
assert(k < p->n_intv);
|
543
|
+
seg[j].s = p->max_s - (p->n_intv - k - 1);
|
544
|
+
seg[j].d = (int32_t)(p->intv[k]>>32) + (last - m) / 5;
|
545
|
+
last = p->x[last];
|
546
|
+
}
|
547
|
+
assert(last == -1);
|
548
|
+
return seg;
|
549
|
+
}
|
550
|
+
|
551
|
+
wf_chkpt_t *mwf_wfa_seg(void *km, const mwf_opt_t *opt, int32_t tl, const char *pts, int32_t ql, const char *pqs, int32_t *n_seg_)
|
552
|
+
{
|
553
|
+
int32_t max_pen, last, n_seg;
|
554
|
+
wf_stripe_t *wf, *sf;
|
555
|
+
wf_sss_t sss = {0,0,0};
|
556
|
+
uint8_t *xbuf;
|
557
|
+
wf_chkpt_t *seg;
|
558
|
+
void *km_st;
|
559
|
+
|
560
|
+
km_st = !(opt->flag&MWF_F_NO_KALLOC)? km_init2(km, 0) : 0;
|
561
|
+
max_pen = opt->x;
|
562
|
+
max_pen = max_pen > opt->o1 + opt->e1? max_pen : opt->o1 + opt->e1;
|
563
|
+
max_pen = max_pen > opt->o2 + opt->e2? max_pen : opt->o2 + opt->e2;
|
564
|
+
xbuf = Kcalloc(km_st, uint8_t, tl + ql + 1);
|
565
|
+
wf = wf_stripe_init(km_st, max_pen);
|
566
|
+
sf = wf_stripe_init(km_st, max_pen);
|
567
|
+
assert(pts);
|
568
|
+
|
569
|
+
while (1) {
|
570
|
+
wf_slice_t *p = &wf->a[wf->top];
|
571
|
+
int32_t d, lo, hi, *H = p->H;
|
572
|
+
for (d = p->lo; d <= p->hi; ++d) {
|
573
|
+
int32_t k;
|
574
|
+
if (H[d] < -1 || d + H[d] < -1 || H[d] >= tl || d + H[d] >= ql) continue;
|
575
|
+
k = wf_extend1_padded(pts, pqs, H[d], d);
|
576
|
+
if (k == tl - 1 && d + k == ql - 1) {
|
577
|
+
last = sf->a[sf->top].H[d];
|
578
|
+
break;
|
579
|
+
}
|
580
|
+
H[d] = k;
|
581
|
+
}
|
582
|
+
if (d <= p->hi) break;
|
583
|
+
lo = wf->lo > -tl? wf->lo - 1 : -tl;
|
584
|
+
hi = wf->hi < ql? wf->hi + 1 : ql;
|
585
|
+
if ((wf->s + 1) % opt->step == 0)
|
586
|
+
wf_snapshot(km_st, &sss, sf);
|
587
|
+
wf_next_seg(km_st, opt, xbuf, wf, sf, lo, hi);
|
588
|
+
if ((wf->s&0xff) == 0) wf_stripe_shrink(wf, tl, ql);
|
589
|
+
}
|
590
|
+
seg = wf_traceback_seg(km, &sss, last, &n_seg);
|
591
|
+
if (km_st == 0) {
|
592
|
+
wf_snapshot_free(km_st, &sss);
|
593
|
+
wf_stripe_destroy(km_st, wf);
|
594
|
+
wf_stripe_destroy(km_st, sf);
|
595
|
+
kfree(km_st, xbuf);
|
596
|
+
} else km_destroy(km_st);
|
597
|
+
|
598
|
+
seg = (wf_chkpt_t*)krelocate(km, seg, n_seg * sizeof(*seg));
|
599
|
+
*n_seg_ = n_seg;
|
600
|
+
return seg;
|
601
|
+
}
|
602
|
+
|
603
|
+
void mwf_wfa_exact(void *km, const mwf_opt_t *opt, int32_t tl, const char *ts, int32_t ql, const char *qs, mwf_rst_t *r)
|
604
|
+
{
|
605
|
+
int32_t n_seg = 0;
|
606
|
+
wf_chkpt_t *seg = 0;
|
607
|
+
char *pts, *pqs;
|
608
|
+
|
609
|
+
wf_pad_str(km, tl, ts, ql, qs, &pts, &pqs);
|
610
|
+
if (opt->step > 0)
|
611
|
+
seg = mwf_wfa_seg(km, opt, tl, pts, ql, pqs, &n_seg);
|
612
|
+
mwf_wfa_core(km, opt, tl, pts, ql, pqs, n_seg, seg, r);
|
613
|
+
kfree(km, seg);
|
614
|
+
kfree(km, pts);
|
615
|
+
}
|
616
|
+
|
617
|
+
/*
|
618
|
+
* Heuristics
|
619
|
+
*/
|
620
|
+
static int32_t mg_lis_64(void *km, int32_t n, const uint64_t *a, int32_t *b)
|
621
|
+
{
|
622
|
+
int32_t i, k, L = 0, *M, *P = b;
|
623
|
+
KMALLOC(km, M, n+1);
|
624
|
+
for (i = 0; i < n; ++i) {
|
625
|
+
int32_t lo = 1, hi = L, newL;
|
626
|
+
while (lo <= hi) {
|
627
|
+
int32_t mid = (lo + hi + 1) >> 1;
|
628
|
+
if (a[M[mid]] < a[i]) lo = mid + 1;
|
629
|
+
else hi = mid - 1;
|
630
|
+
}
|
631
|
+
newL = lo, P[i] = M[newL - 1], M[newL] = i;
|
632
|
+
if (newL > L) L = newL;
|
633
|
+
}
|
634
|
+
k = M[L];
|
635
|
+
memcpy(M, P, n * sizeof(int32_t));
|
636
|
+
for (i = L - 1; i >= 0; --i) b[i] = k, k = M[k];
|
637
|
+
kfree(km, M);
|
638
|
+
return L;
|
639
|
+
}
|
640
|
+
|
641
|
+
extern void radix_sort_gfa64(uint64_t*, uint64_t*);
|
642
|
+
extern unsigned char seq_nt4_table[256];
|
643
|
+
|
644
|
+
static int32_t mg_fc_kmer(int32_t len, const char *seq, int32_t rid, int32_t k, uint64_t *a)
|
645
|
+
{
|
646
|
+
int32_t i, l, n;
|
647
|
+
uint64_t x, mask = (1ULL<<k*2) - 1;
|
648
|
+
for (i = l = 0, x = 0, n = 0; i < len; ++i) {
|
649
|
+
int32_t c = seq_nt4_table[(uint8_t)seq[i]];
|
650
|
+
if (c < 4) {
|
651
|
+
x = (x << 2 | c) & mask;
|
652
|
+
if (++l >= k) a[n++] = (x<<1|rid) << 32 | i;
|
653
|
+
} else l = 0, x = 0;
|
654
|
+
}
|
655
|
+
return n;
|
656
|
+
}
|
657
|
+
|
658
|
+
static uint64_t *mg_chain(void *km, int32_t l1, const char *s1, int32_t l2, const char *s2, int32_t k, int32_t max_occ, int32_t *n_lis_)
|
659
|
+
{
|
660
|
+
int32_t i, n_a, n_b, m_b, i0, n_lis, *lis;
|
661
|
+
uint64_t *a, *b;
|
662
|
+
|
663
|
+
*n_lis_ = 0;
|
664
|
+
if (l1 < k || l2 < k) return 0;
|
665
|
+
assert(k >= 2 && k <= 15);
|
666
|
+
|
667
|
+
// collect k-mers
|
668
|
+
KMALLOC(km, a, l1 + l2);
|
669
|
+
n_a = mg_fc_kmer(l1, s1, 0, k, a);
|
670
|
+
n_a += mg_fc_kmer(l2, s2, 1, k, &a[n_a]);
|
671
|
+
radix_sort_gfa64(a, a + n_a);
|
672
|
+
|
673
|
+
// collect k-mer matches
|
674
|
+
n_b = m_b = 0, b = 0;
|
675
|
+
for (i0 = 0, i = 1; i <= n_a; ++i) {
|
676
|
+
if (i == n_a || a[i0]>>33 != a[i]>>33) {
|
677
|
+
if (i - i0 >= 2) {
|
678
|
+
int32_t j, s, t;
|
679
|
+
for (j = i0; j < i && (a[j]>>32&1) == 0; ++j) {}
|
680
|
+
if (j > i0 && j < i && j - i0 <= max_occ && i - j <= max_occ) {
|
681
|
+
for (s = i0; s < j; ++s)
|
682
|
+
for (t = j; t < i; ++t) {
|
683
|
+
if (n_b == m_b) KEXPAND(km, b, m_b);
|
684
|
+
b[n_b++] = a[s]<<32 | (uint32_t)a[t];
|
685
|
+
}
|
686
|
+
}
|
687
|
+
}
|
688
|
+
i0 = i;
|
689
|
+
}
|
690
|
+
}
|
691
|
+
kfree(km, a);
|
692
|
+
|
693
|
+
// find co-linear chain with LIS
|
694
|
+
radix_sort_gfa64(b, b + n_b);
|
695
|
+
for (i = 0; i < n_b; ++i)
|
696
|
+
b[i] = b[i]>>32 | b[i]<<32;
|
697
|
+
KMALLOC(km, lis, n_b);
|
698
|
+
n_lis = mg_lis_64(km, n_b, b, lis);
|
699
|
+
a = Kmalloc(km, uint64_t, n_lis);
|
700
|
+
for (i = 0; i < n_lis; ++i) a[i] = b[lis[i]];
|
701
|
+
kfree(km, lis);
|
702
|
+
kfree(km, b);
|
703
|
+
b = Kmalloc(km, uint64_t, n_lis);
|
704
|
+
memcpy(b, a, sizeof(uint64_t) * n_lis);
|
705
|
+
kfree(km, a);
|
706
|
+
*n_lis_ = n_lis;
|
707
|
+
for (i = 0; i < n_lis; ++i) // switch back, such that seq1 on the high bits
|
708
|
+
b[i] = b[i]>>32 | b[i]<<32;
|
709
|
+
return b;
|
710
|
+
}
|
711
|
+
|
712
|
+
static double mwf_ksim(void *km, int32_t l1, const char *s1, int32_t l2, const char *s2, int32_t k)
|
713
|
+
{
|
714
|
+
int32_t i, i0, j, n_a, n1 = 0, n2 = 0, t1 = 0, t2 = 0;
|
715
|
+
double p1, p2;
|
716
|
+
uint64_t *a;
|
717
|
+
if (l1 < k || l2 < k) return 0;
|
718
|
+
assert(k >= 2 && k <= 15);
|
719
|
+
KMALLOC(km, a, l1 + l2);
|
720
|
+
n_a = mg_fc_kmer(l1, s1, 0, k, a);
|
721
|
+
n_a += mg_fc_kmer(l2, s2, 1, k, &a[n_a]);
|
722
|
+
radix_sort_gfa64(a, a + n_a);
|
723
|
+
for (i0 = 0, i = 1; i <= n_a; ++i) {
|
724
|
+
if (i == n_a || a[i0]>>33 != a[i]>>33) {
|
725
|
+
int32_t m1, m2, min;
|
726
|
+
for (j = i0; j < i && (a[j]>>32&1) == 0; ++j) {}
|
727
|
+
m1 = j - i0, m2 = i - j;
|
728
|
+
min = m1 < m2? m1 : m2;
|
729
|
+
n1 += m1, n2 += m2;
|
730
|
+
if (m1 > 0 && m2 > 0)
|
731
|
+
t1 += min, t2 += min;
|
732
|
+
i0 = i;
|
733
|
+
}
|
734
|
+
}
|
735
|
+
kfree(km, a);
|
736
|
+
p1 = (double)t1 / n1, p2 = (double)t2 / n2;
|
737
|
+
return p1 > p2? p1 : p2;
|
738
|
+
}
|
739
|
+
|
740
|
+
#define kroundup32(x) (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
|
741
|
+
|
742
|
+
static void wf_cigar_push(void *km, wf_cigar_t *c, int32_t n_cigar, const uint32_t *cigar)
|
743
|
+
{
|
744
|
+
if (n_cigar == 0) return;
|
745
|
+
wf_cigar_push1(km, c, cigar[0]&0xf, cigar[0]>>4);
|
746
|
+
if (c->n + n_cigar - 1 > c->m) {
|
747
|
+
c->m = c->n + n_cigar - 1;
|
748
|
+
kroundup32(c->m);
|
749
|
+
c->cigar = Krealloc(km, uint32_t, c->cigar, c->m);
|
750
|
+
}
|
751
|
+
memcpy(&c->cigar[c->n], &cigar[1], sizeof(*cigar) * (n_cigar - 1));
|
752
|
+
c->n += n_cigar - 1;
|
753
|
+
}
|
754
|
+
|
755
|
+
static int32_t wf_anchor_filter(int32_t n, uint64_t *a, int32_t tl, int32_t ql, int32_t k, int32_t min_l)
|
756
|
+
{
|
757
|
+
int32_t i, st, x0, y0, x1, y1, j, l, m;
|
758
|
+
for (i = 0, x0 = y0 = x1 = y1 = 0, st = -1, l = 0; i <= n; ++i) {
|
759
|
+
int32_t x, y;
|
760
|
+
if (i == n) x = tl, y = ql;
|
761
|
+
else x = (int32_t)(a[i]>>32) + 1, y = (int32_t)a[i] + 1;
|
762
|
+
if (x - x0 != y - y0) {
|
763
|
+
//fprintf(stderr, "X\t%d\t(%d,%d) -> (%d,%d)\n", l, x0, y0, x, y);
|
764
|
+
if (l < min_l)
|
765
|
+
for (j = st > 0? st : 0; j < i; ++j)
|
766
|
+
a[j] = 0;
|
767
|
+
x0 = x, y0 = y, st = i, l = k;
|
768
|
+
} else l += x - x1;
|
769
|
+
x1 = x, y1 = y;
|
770
|
+
}
|
771
|
+
for (i = 0, m = 0; i < n; ++i)
|
772
|
+
if (a[i] != 0) a[m++] = a[i];
|
773
|
+
return m;
|
774
|
+
}
|
775
|
+
|
776
|
+
void mwf_wfa_chain(void *km, const mwf_opt_t *opt, int32_t tl, const char *ts, int32_t ql, const char *qs, mwf_rst_t *r)
|
777
|
+
{
|
778
|
+
int32_t n_a, i, x0, y0;
|
779
|
+
uint64_t *a;
|
780
|
+
void *km_wfa;
|
781
|
+
wf_cigar_t c = {0,0,0};
|
782
|
+
|
783
|
+
km_wfa = !(opt->flag&MWF_F_NO_KALLOC)? km_init2(km, 0) : 0;
|
784
|
+
a = mg_chain(km_wfa, tl, ts, ql, qs, opt->kmer, opt->max_occ, &n_a);
|
785
|
+
n_a = wf_anchor_filter(n_a, a, tl, ql, opt->kmer, opt->min_len);
|
786
|
+
r->s = 0;
|
787
|
+
for (i = 0, x0 = y0 = 0; i <= n_a; ++i) {
|
788
|
+
int32_t x1, y1;
|
789
|
+
if (i == n_a) x1 = tl, y1 = ql;
|
790
|
+
else x1 = (int32_t)(a[i]>>32) + 1, y1 = (int32_t)a[i] + 1;
|
791
|
+
if (i < n_a && x1 - x0 == y1 - y0 && x1 - x0 <= opt->kmer) {
|
792
|
+
if (opt->flag&MWF_F_CIGAR)
|
793
|
+
wf_cigar_push1(km, &c, 7, x1 - x0);
|
794
|
+
} else if (x0 < x1 && y0 < y1) {
|
795
|
+
if (x1 - x0 >= 10000 && y1 - y0 >= 10000 && mwf_ksim(km, x1 - x0, &ts[x0], y1 - y0, &qs[y0], opt->kmer) < 0.02) {
|
796
|
+
if (opt->flag&MWF_F_CIGAR) {
|
797
|
+
wf_cigar_push1(km, &c, 2, x1 - x0);
|
798
|
+
wf_cigar_push1(km, &c, 1, y1 - y0);
|
799
|
+
}
|
800
|
+
r->s += opt->o2 * 2 + opt->e2 * ((x1 - x0) + (y1 - y0));
|
801
|
+
} else {
|
802
|
+
mwf_rst_t q;
|
803
|
+
mwf_wfa_exact(km_wfa, opt, x1 - x0, &ts[x0], y1 - y0, &qs[y0], &q);
|
804
|
+
if (opt->flag&MWF_F_CIGAR)
|
805
|
+
wf_cigar_push(km, &c, q.n_cigar, q.cigar);
|
806
|
+
r->s += q.s;
|
807
|
+
kfree(km_wfa, q.cigar);
|
808
|
+
}
|
809
|
+
} else if (x0 < x1) {
|
810
|
+
wf_cigar_push1(km, &c, 2, x1 - x0);
|
811
|
+
r->s += opt->o2 + (x1 - x0) * opt->e2 < opt->o1 + (x1 - x0) * opt->e1? opt->o2 + (x1 - x0) * opt->e2 : opt->o1 + (x1 - x0) * opt->e1;
|
812
|
+
} else if (y0 < y1) {
|
813
|
+
wf_cigar_push1(km, &c, 1, y1 - y0);
|
814
|
+
r->s += opt->o2 + (y1 - y0) * opt->e2 < opt->o1 + (y1 - y0) * opt->e1? opt->o2 + (y1 - y0) * opt->e2 : opt->o1 + (y1 - y0) * opt->e1;
|
815
|
+
}
|
816
|
+
x0 = x1, y0 = y1;
|
817
|
+
}
|
818
|
+
if (km_wfa == 0) kfree(km_wfa, a);
|
819
|
+
km_destroy(km_wfa);
|
820
|
+
r->n_cigar = c.n, r->cigar = c.cigar;
|
821
|
+
r->cigar = (uint32_t*)krelocate(km, r->cigar, r->n_cigar * sizeof(*r->cigar));
|
822
|
+
}
|
823
|
+
|
824
|
+
void mwf_wfa_auto(void *km, const mwf_opt_t *opt0, int32_t tl, const char *ts, int32_t ql, const char *qs, mwf_rst_t *r)
|
825
|
+
{
|
826
|
+
mwf_opt_t opt = *opt0;
|
827
|
+
opt.step = 0, opt.max_iter = 100000000;
|
828
|
+
mwf_wfa_exact(km, &opt, tl, ts, ql, qs, r);
|
829
|
+
if (r->s < 0) {
|
830
|
+
if (opt.flag & MWF_F_CIGAR) opt.step = 5000;
|
831
|
+
opt.max_iter = -1;
|
832
|
+
mwf_wfa_chain(km, &opt, tl, ts, ql, qs, r);
|
833
|
+
}
|
834
|
+
}
|