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 +1,617 @@
|
|
1
|
+
#include <assert.h>
|
2
|
+
#include <string.h>
|
3
|
+
#include <stdio.h>
|
4
|
+
#include "gfa-priv.h"
|
5
|
+
#include "kalloc.h"
|
6
|
+
#include "ksort.h"
|
7
|
+
#include "khashl.h" // make it compatible with kalloc
|
8
|
+
#include "kdq.h"
|
9
|
+
#include "kvec-km.h"
|
10
|
+
|
11
|
+
int gfa_ed_dbg = 0;
|
12
|
+
|
13
|
+
/***************
|
14
|
+
* Preparation *
|
15
|
+
***************/
|
16
|
+
|
17
|
+
void gfa_edopt_init(gfa_edopt_t *opt)
|
18
|
+
{
|
19
|
+
memset(opt, 0, sizeof(gfa_edopt_t));
|
20
|
+
opt->bw_dyn = opt->max_lag = opt->s_term = -1;
|
21
|
+
opt->max_chk = 1000;
|
22
|
+
}
|
23
|
+
|
24
|
+
gfa_edseq_t *gfa_edseq_init(const gfa_t *g)
|
25
|
+
{
|
26
|
+
uint32_t i, n_vtx = gfa_n_vtx(g);
|
27
|
+
gfa_edseq_t *es;
|
28
|
+
GFA_MALLOC(es, n_vtx);
|
29
|
+
for (i = 0; i < g->n_seg; ++i) {
|
30
|
+
const gfa_seg_t *s = &g->seg[i];
|
31
|
+
char *t;
|
32
|
+
int32_t j;
|
33
|
+
GFA_MALLOC(t, s->len + 1);
|
34
|
+
for (j = 0; j < s->len; ++j)
|
35
|
+
t[s->len - j - 1] = gfa_comp_table[(uint8_t)s->seq[j]];
|
36
|
+
t[s->len] = 0;
|
37
|
+
es[i<<1].seq = (char*)s->seq;
|
38
|
+
es[i<<1|1].seq = t;
|
39
|
+
es[i<<1].len = es[i<<1|1].len = s->len;
|
40
|
+
}
|
41
|
+
return es;
|
42
|
+
}
|
43
|
+
|
44
|
+
void gfa_edseq_destroy(int32_t n_seg, gfa_edseq_t *es)
|
45
|
+
{
|
46
|
+
int32_t i;
|
47
|
+
for (i = 0; i < n_seg; ++i)
|
48
|
+
free((char*)es[i<<1|1].seq);
|
49
|
+
free(es);
|
50
|
+
}
|
51
|
+
|
52
|
+
/*****************
|
53
|
+
* Edit distance *
|
54
|
+
*****************/
|
55
|
+
|
56
|
+
#define GWF_DIAG_SHIFT 0x40000000
|
57
|
+
|
58
|
+
static inline uint64_t gwf_gen_vd(uint32_t v, int32_t d)
|
59
|
+
{
|
60
|
+
return (uint64_t)v<<32 | (GWF_DIAG_SHIFT + d);
|
61
|
+
}
|
62
|
+
|
63
|
+
/*
|
64
|
+
* Diagonal interval
|
65
|
+
*/
|
66
|
+
typedef struct {
|
67
|
+
uint64_t vd0, vd1;
|
68
|
+
} gwf_intv_t;
|
69
|
+
|
70
|
+
typedef kvec_t(gwf_intv_t) gwf_intv_v;
|
71
|
+
|
72
|
+
#define intvd_key(x) ((x).vd0)
|
73
|
+
KRADIX_SORT_INIT(gwf_intv, gwf_intv_t, intvd_key, 8)
|
74
|
+
|
75
|
+
static int gwf_intv_is_sorted(int32_t n_a, const gwf_intv_t *a)
|
76
|
+
{
|
77
|
+
int32_t i;
|
78
|
+
for (i = 1; i < n_a; ++i)
|
79
|
+
if (a[i-1].vd0 > a[i].vd0) break;
|
80
|
+
return (i == n_a);
|
81
|
+
}
|
82
|
+
|
83
|
+
// merge overlapping intervals; input must be sorted
|
84
|
+
static size_t gwf_intv_merge_adj(size_t n, gwf_intv_t *a)
|
85
|
+
{
|
86
|
+
size_t i, k;
|
87
|
+
uint64_t st, en;
|
88
|
+
if (n == 0) return 0;
|
89
|
+
st = a[0].vd0, en = a[0].vd1;
|
90
|
+
for (i = 1, k = 0; i < n; ++i) {
|
91
|
+
if (a[i].vd0 > en) {
|
92
|
+
a[k].vd0 = st, a[k++].vd1 = en;
|
93
|
+
st = a[i].vd0, en = a[i].vd1;
|
94
|
+
} else en = en > a[i].vd1? en : a[i].vd1;
|
95
|
+
}
|
96
|
+
a[k].vd0 = st, a[k++].vd1 = en;
|
97
|
+
return k;
|
98
|
+
}
|
99
|
+
|
100
|
+
// merge two sorted interval lists
|
101
|
+
static size_t gwf_intv_merge2(gwf_intv_t *a, size_t n_b, const gwf_intv_t *b, size_t n_c, const gwf_intv_t *c)
|
102
|
+
{
|
103
|
+
size_t i = 0, j = 0, k = 0;
|
104
|
+
while (i < n_b && j < n_c) {
|
105
|
+
if (b[i].vd0 <= c[j].vd0)
|
106
|
+
a[k++] = b[i++];
|
107
|
+
else a[k++] = c[j++];
|
108
|
+
}
|
109
|
+
while (i < n_b) a[k++] = b[i++];
|
110
|
+
while (j < n_c) a[k++] = c[j++];
|
111
|
+
return gwf_intv_merge_adj(k, a);
|
112
|
+
}
|
113
|
+
|
114
|
+
/*
|
115
|
+
* Diagonal
|
116
|
+
*/
|
117
|
+
typedef struct { // a diagonal
|
118
|
+
uint64_t vd; // higher 32 bits: vertex ID; lower 32 bits: diagonal+0x4000000
|
119
|
+
int32_t k;
|
120
|
+
int32_t len;
|
121
|
+
uint32_t xo; // higher 31 bits: anti diagonal; lower 1 bit: out-of-order or not
|
122
|
+
int32_t t;
|
123
|
+
} gwf_diag_t;
|
124
|
+
|
125
|
+
typedef kvec_t(gwf_diag_t) gwf_diag_v;
|
126
|
+
|
127
|
+
#define ed_key(x) ((x).vd)
|
128
|
+
KRADIX_SORT_INIT(gwf_ed, gwf_diag_t, ed_key, 8)
|
129
|
+
|
130
|
+
KDQ_INIT(gwf_diag_t)
|
131
|
+
|
132
|
+
// push (v,d,k) to the end of the queue
|
133
|
+
static inline void gwf_diag_push(void *km, gwf_diag_v *a, uint32_t v, int32_t d, int32_t k, uint32_t x, uint32_t ooo, int32_t t)
|
134
|
+
{
|
135
|
+
gwf_diag_t *p;
|
136
|
+
kv_pushp(gwf_diag_t, km, *a, &p);
|
137
|
+
p->vd = gwf_gen_vd(v, d), p->k = k, p->xo = x<<1|ooo, p->t = t;
|
138
|
+
}
|
139
|
+
|
140
|
+
// determine the wavefront on diagonal (v,d)
|
141
|
+
static inline int32_t gwf_diag_update(gwf_diag_t *p, uint32_t v, int32_t d, int32_t k, uint32_t x, uint32_t ooo, int32_t t)
|
142
|
+
{
|
143
|
+
uint64_t vd = gwf_gen_vd(v, d);
|
144
|
+
if (p->vd == vd) {
|
145
|
+
p->xo = p->k > k? p->xo : x<<1|ooo;
|
146
|
+
p->t = p->k > k? p->t : t;
|
147
|
+
p->k = p->k > k? p->k : k;
|
148
|
+
return 0;
|
149
|
+
}
|
150
|
+
return 1;
|
151
|
+
}
|
152
|
+
|
153
|
+
static int gwf_diag_is_sorted(int32_t n_a, const gwf_diag_t *a)
|
154
|
+
{
|
155
|
+
int32_t i;
|
156
|
+
for (i = 1; i < n_a; ++i)
|
157
|
+
if (a[i-1].vd > a[i].vd) break;
|
158
|
+
return (i == n_a);
|
159
|
+
}
|
160
|
+
|
161
|
+
// sort a[]. This uses the gwf_diag_t::ooo field to speed up sorting.
|
162
|
+
static void gwf_diag_sort(int32_t n_a, gwf_diag_t *a, void *km, gwf_diag_v *ooo)
|
163
|
+
{
|
164
|
+
int32_t i, j, k, n_b, n_c;
|
165
|
+
gwf_diag_t *b, *c;
|
166
|
+
|
167
|
+
kv_resize(gwf_diag_t, km, *ooo, n_a);
|
168
|
+
for (i = 0, n_c = 0; i < n_a; ++i)
|
169
|
+
if (a[i].xo&1) ++n_c;
|
170
|
+
n_b = n_a - n_c;
|
171
|
+
b = ooo->a, c = b + n_b;
|
172
|
+
for (i = j = k = 0; i < n_a; ++i) {
|
173
|
+
if (a[i].xo&1) c[k++] = a[i];
|
174
|
+
else b[j++] = a[i];
|
175
|
+
}
|
176
|
+
radix_sort_gwf_ed(c, c + n_c);
|
177
|
+
for (k = 0; k < n_c; ++k) c[k].xo &= 0xfffffffeU;
|
178
|
+
|
179
|
+
i = j = k = 0;
|
180
|
+
while (i < n_b && j < n_c) {
|
181
|
+
if (b[i].vd <= c[j].vd)
|
182
|
+
a[k++] = b[i++];
|
183
|
+
else a[k++] = c[j++];
|
184
|
+
}
|
185
|
+
while (i < n_b) a[k++] = b[i++];
|
186
|
+
while (j < n_c) a[k++] = c[j++];
|
187
|
+
}
|
188
|
+
|
189
|
+
// remove diagonals not on the wavefront
|
190
|
+
static int32_t gwf_diag_dedup(int32_t n_a, gwf_diag_t *a, void *km, gwf_diag_v *ooo)
|
191
|
+
{
|
192
|
+
int32_t i, n, st;
|
193
|
+
if (!gwf_diag_is_sorted(n_a, a))
|
194
|
+
gwf_diag_sort(n_a, a, km, ooo);
|
195
|
+
for (i = 1, st = 0, n = 0; i <= n_a; ++i) {
|
196
|
+
if (i == n_a || a[i].vd != a[st].vd) {
|
197
|
+
int32_t j, max_j = st;
|
198
|
+
if (st + 1 < i)
|
199
|
+
for (j = st + 1; j < i; ++j) // choose the far end (i.e. the wavefront)
|
200
|
+
if (a[max_j].k < a[j].k) max_j = j;
|
201
|
+
a[n++] = a[max_j];
|
202
|
+
st = i;
|
203
|
+
}
|
204
|
+
}
|
205
|
+
return n;
|
206
|
+
}
|
207
|
+
|
208
|
+
// use forbidden bands to remove diagonals not on the wavefront
|
209
|
+
static int32_t gwf_mixed_dedup(int32_t n_a, gwf_diag_t *a, int32_t n_b, gwf_intv_t *b)
|
210
|
+
{
|
211
|
+
int32_t i = 0, j = 0, k = 0;
|
212
|
+
while (i < n_a && j < n_b) {
|
213
|
+
if (a[i].vd >= b[j].vd0 && a[i].vd < b[j].vd1) ++i;
|
214
|
+
else if (a[i].vd >= b[j].vd1) ++j;
|
215
|
+
else a[k++] = a[i++];
|
216
|
+
}
|
217
|
+
while (i < n_a) a[k++] = a[i++];
|
218
|
+
return k;
|
219
|
+
}
|
220
|
+
|
221
|
+
/*
|
222
|
+
* Traceback stack
|
223
|
+
*/
|
224
|
+
KHASHL_MAP_INIT(KH_LOCAL, gwf_map64_t, gwf_map64, uint64_t, int32_t, kh_hash_uint64, kh_eq_generic)
|
225
|
+
|
226
|
+
typedef struct {
|
227
|
+
int32_t v;
|
228
|
+
int32_t pre;
|
229
|
+
} gwf_trace_t;
|
230
|
+
|
231
|
+
typedef kvec_t(gwf_trace_t) gwf_trace_v;
|
232
|
+
|
233
|
+
static int32_t gwf_trace_push(void *km, gwf_trace_v *a, int32_t v, int32_t pre, gwf_map64_t *h)
|
234
|
+
{
|
235
|
+
uint64_t key = (uint64_t)v << 32 | (uint32_t)pre;
|
236
|
+
khint_t k;
|
237
|
+
int absent;
|
238
|
+
k = gwf_map64_put(h, key, &absent);
|
239
|
+
if (absent) {
|
240
|
+
gwf_trace_t *p;
|
241
|
+
kv_pushp(gwf_trace_t, km, *a, &p);
|
242
|
+
p->v = v, p->pre = pre;
|
243
|
+
kh_val(h, k) = a->n - 1;
|
244
|
+
return a->n - 1;
|
245
|
+
}
|
246
|
+
return kh_val(h, k);
|
247
|
+
}
|
248
|
+
|
249
|
+
/*
|
250
|
+
* Core GWFA routine
|
251
|
+
*/
|
252
|
+
KHASHL_INIT(KH_LOCAL, gwf_set64_t, gwf_set64, uint64_t, kh_hash_dummy, kh_eq_generic)
|
253
|
+
|
254
|
+
typedef struct {
|
255
|
+
void *km;
|
256
|
+
gwf_set64_t *ha; // hash table for adjacency
|
257
|
+
gwf_map64_t *ht; // hash table for traceback
|
258
|
+
gwf_intv_v intv;
|
259
|
+
gwf_intv_v tmp, swap;
|
260
|
+
gwf_diag_v ooo;
|
261
|
+
gwf_trace_v t;
|
262
|
+
} gwf_edbuf_t;
|
263
|
+
|
264
|
+
// remove diagonals not on the wavefront
|
265
|
+
static int32_t gwf_dedup(gwf_edbuf_t *buf, int32_t n_a, gwf_diag_t *a)
|
266
|
+
{
|
267
|
+
if (buf->intv.n + buf->tmp.n > 0) {
|
268
|
+
if (!gwf_intv_is_sorted(buf->tmp.n, buf->tmp.a))
|
269
|
+
radix_sort_gwf_intv(buf->tmp.a, buf->tmp.a + buf->tmp.n);
|
270
|
+
kv_copy(gwf_intv_t, buf->km, buf->swap, buf->intv);
|
271
|
+
kv_resize(gwf_intv_t, buf->km, buf->intv, buf->intv.n + buf->tmp.n);
|
272
|
+
buf->intv.n = gwf_intv_merge2(buf->intv.a, buf->swap.n, buf->swap.a, buf->tmp.n, buf->tmp.a);
|
273
|
+
}
|
274
|
+
n_a = gwf_diag_dedup(n_a, a, buf->km, &buf->ooo);
|
275
|
+
if (buf->intv.n > 0)
|
276
|
+
n_a = gwf_mixed_dedup(n_a, a, buf->intv.n, buf->intv.a);
|
277
|
+
return n_a;
|
278
|
+
}
|
279
|
+
|
280
|
+
// remove diagonals that lag far behind the furthest wavefront
|
281
|
+
static int32_t gwf_prune(int32_t n_a, gwf_diag_t *a, uint32_t max_lag, int32_t bw_dyn)
|
282
|
+
{
|
283
|
+
int32_t i, j, iq, dq, max_i = -1;
|
284
|
+
uint32_t max_x = 0;
|
285
|
+
gwf_diag_t *q;
|
286
|
+
for (i = 0; i < n_a; ++i)
|
287
|
+
if (a[i].xo>>1 > max_x)
|
288
|
+
max_x = a[i].xo>>1, max_i = i;
|
289
|
+
q = &a[max_i];
|
290
|
+
iq = (int32_t)q->vd - GWF_DIAG_SHIFT + q->k;
|
291
|
+
dq = (int32_t)(q->xo>>1) - iq - iq;
|
292
|
+
for (i = j = 0; i < n_a; ++i) {
|
293
|
+
gwf_diag_t *p = &a[i];
|
294
|
+
int32_t ip = (int32_t)p->vd - GWF_DIAG_SHIFT + p->k;
|
295
|
+
int32_t dp = (int32_t)(p->xo>>1) - ip - ip;
|
296
|
+
int32_t w = dp > dq? dp - dq : dq - dp;
|
297
|
+
if (bw_dyn >= 0 && w > bw_dyn) continue;
|
298
|
+
if ((p->xo>>1) + max_lag < max_x) continue;
|
299
|
+
a[j++] = *p;
|
300
|
+
}
|
301
|
+
return j;
|
302
|
+
}
|
303
|
+
|
304
|
+
// reach the wavefront
|
305
|
+
static inline int32_t gwf_extend1(int32_t d, int32_t k, int32_t vl, const char *ts, int32_t ql, const char *qs)
|
306
|
+
{
|
307
|
+
int32_t max_k = (ql - d < vl? ql - d : vl) - 1;
|
308
|
+
const char *ts_ = ts + 1, *qs_ = qs + d + 1;
|
309
|
+
#if 0
|
310
|
+
// int32_t i = k + d; while (k + 1 < vl && i + 1 < ql && ts[k+1] == q[i+1]) ++k, ++i;
|
311
|
+
while (k < max_k && *(ts_ + k) == *(qs_ + k))
|
312
|
+
++k;
|
313
|
+
#else
|
314
|
+
uint64_t cmp = 0;
|
315
|
+
while (k + 7 < max_k) {
|
316
|
+
uint64_t x = *(uint64_t*)(ts_ + k); // warning: unaligned memory access
|
317
|
+
uint64_t y = *(uint64_t*)(qs_ + k);
|
318
|
+
cmp = x ^ y;
|
319
|
+
if (cmp == 0) k += 8;
|
320
|
+
else break;
|
321
|
+
}
|
322
|
+
if (cmp)
|
323
|
+
k += __builtin_ctzl(cmp) >> 3; // on x86, this is done via the BSR instruction: https://www.felixcloutier.com/x86/bsr
|
324
|
+
else if (k + 7 >= max_k)
|
325
|
+
while (k < max_k && *(ts_ + k) == *(qs_ + k)) // use this for generic CPUs. It is slightly faster than the unoptimized version
|
326
|
+
++k;
|
327
|
+
#endif
|
328
|
+
return k;
|
329
|
+
}
|
330
|
+
|
331
|
+
// This is essentially Landau-Vishkin for linear sequences. The function speeds up alignment to long vertices. Not really necessary.
|
332
|
+
static void gwf_ed_extend_batch(void *km, const gfa_t *g, const gfa_edseq_t *es, int32_t ql, const char *q, int32_t n, gwf_diag_t *a, gwf_diag_v *B,
|
333
|
+
kdq_t(gwf_diag_t) *A, gwf_intv_v *tmp_intv, gfa_edrst_t *r)
|
334
|
+
{
|
335
|
+
int32_t j, m;
|
336
|
+
int32_t v = a->vd>>32;
|
337
|
+
int32_t vl = es[v].len;
|
338
|
+
const char *ts = es[v].seq;
|
339
|
+
gwf_diag_t *b;
|
340
|
+
|
341
|
+
// wfa_extend
|
342
|
+
for (j = 0; j < n; ++j) {
|
343
|
+
int32_t k;
|
344
|
+
k = gwf_extend1((int32_t)a[j].vd - GWF_DIAG_SHIFT, a[j].k, vl, ts, ql, q);
|
345
|
+
a[j].len = k - a[j].k;
|
346
|
+
a[j].xo += a[j].len << 2;
|
347
|
+
a[j].k = k;
|
348
|
+
}
|
349
|
+
|
350
|
+
// wfa_next
|
351
|
+
kv_resize(gwf_diag_t, km, *B, B->n + n + 2);
|
352
|
+
b = &B->a[B->n];
|
353
|
+
b[0].vd = a[0].vd - 1;
|
354
|
+
b[0].xo = a[0].xo + 2; // 2 == 1<<1
|
355
|
+
b[0].k = a[0].k + 1;
|
356
|
+
b[0].t = a[0].t;
|
357
|
+
b[1].vd = a[0].vd;
|
358
|
+
b[1].xo = n == 1 || a[0].k > a[1].k? a[0].xo + 4 : a[1].xo + 2;
|
359
|
+
b[1].t = n == 1 || a[0].k > a[1].k? a[0].t : a[1].t;
|
360
|
+
b[1].k = (n == 1 || a[0].k > a[1].k? a[0].k : a[1].k) + 1;
|
361
|
+
for (j = 1; j < n - 1; ++j) {
|
362
|
+
uint32_t x = a[j-1].xo + 2;
|
363
|
+
int32_t k = a[j-1].k, t = a[j-1].t;
|
364
|
+
x = k > a[j].k + 1? x : a[j].xo + 4;
|
365
|
+
t = k > a[j].k + 1? t : a[j].t;
|
366
|
+
k = k > a[j].k + 1? k : a[j].k + 1;
|
367
|
+
x = k > a[j+1].k + 1? x : a[j+1].xo + 2;
|
368
|
+
t = k > a[j+1].k + 1? t : a[j+1].t;
|
369
|
+
k = k > a[j+1].k + 1? k : a[j+1].k + 1;
|
370
|
+
b[j+1].vd = a[j].vd, b[j+1].k = k, b[j+1].xo = x, b[j+1].t = t;
|
371
|
+
}
|
372
|
+
if (n >= 2) {
|
373
|
+
b[n].vd = a[n-1].vd;
|
374
|
+
b[n].xo = a[n-2].k > a[n-1].k + 1? a[n-2].xo + 2 : a[n-1].xo + 4;
|
375
|
+
b[n].t = a[n-2].k > a[n-1].k + 1? a[n-2].t : a[n-1].t;
|
376
|
+
b[n].k = a[n-2].k > a[n-1].k + 1? a[n-2].k : a[n-1].k + 1;
|
377
|
+
}
|
378
|
+
b[n+1].vd = a[n-1].vd + 1;
|
379
|
+
b[n+1].xo = a[n-1].xo + 2;
|
380
|
+
b[n+1].t = a[n-1].t;
|
381
|
+
b[n+1].k = a[n-1].k;
|
382
|
+
|
383
|
+
// drop out-of-bound cells
|
384
|
+
//if (a[n-1].k == vl - 1) b[n+1].k = vl; // insertion to the end of a vertex is handled elsewhere. FIXME: this line leads to wrong result for MHC-57 and MHC-HG002.2
|
385
|
+
for (j = 0; j < n; ++j) {
|
386
|
+
gwf_diag_t *p = &a[j];
|
387
|
+
if (p->k == vl - 1 || (int32_t)p->vd - GWF_DIAG_SHIFT + p->k == ql - 1)
|
388
|
+
p->xo |= 1, *kdq_pushp(gwf_diag_t, A) = *p;
|
389
|
+
}
|
390
|
+
for (j = 0, m = 0; j < n + 2; ++j) {
|
391
|
+
gwf_diag_t *p = &b[j];
|
392
|
+
int32_t d = (int32_t)p->vd - GWF_DIAG_SHIFT;
|
393
|
+
if (d + p->k < ql && p->k < vl) {
|
394
|
+
b[m++] = *p;
|
395
|
+
} else if (p->k == vl) {
|
396
|
+
gwf_intv_t *q;
|
397
|
+
kv_pushp(gwf_intv_t, km, *tmp_intv, &q);
|
398
|
+
q->vd0 = gwf_gen_vd(v, d), q->vd1 = q->vd0 + 1;
|
399
|
+
}
|
400
|
+
}
|
401
|
+
B->n += m;
|
402
|
+
}
|
403
|
+
|
404
|
+
// wfa_extend and wfa_next combined
|
405
|
+
static gwf_diag_t *gwf_ed_extend(gwf_edbuf_t *buf, const gfa_edopt_t *opt, const gfa_t *g, const gfa_edseq_t *es, int32_t s, int32_t ql, const char *q,
|
406
|
+
uint32_t v1, int32_t off1, int32_t *end_tb, int32_t *n_a_, gwf_diag_t *a, gfa_edrst_t *r)
|
407
|
+
{
|
408
|
+
int32_t i, x, n = *n_a_, do_dedup = 1;
|
409
|
+
kdq_t(gwf_diag_t) *A;
|
410
|
+
gwf_diag_v B = {0,0,0};
|
411
|
+
gwf_diag_t *b;
|
412
|
+
|
413
|
+
r->end_v = (uint32_t)-1;
|
414
|
+
r->end_off = *end_tb = -1;
|
415
|
+
buf->tmp.n = 0;
|
416
|
+
gwf_set64_clear(buf->ha); // hash table $h to avoid visiting a vertex twice
|
417
|
+
for (i = 0, x = 1; i < 32; ++i, x <<= 1)
|
418
|
+
if (x >= n) break;
|
419
|
+
if (i < 4) i = 4;
|
420
|
+
A = kdq_init2(gwf_diag_t, buf->km, i); // $A is a queue
|
421
|
+
kv_resize(gwf_diag_t, buf->km, B, n * 2);
|
422
|
+
#if 0 // unoptimized version without calling gwf_ed_extend_batch() at all. The final result will be the same.
|
423
|
+
A->count = n;
|
424
|
+
memcpy(A->a, a, n * sizeof(*a));
|
425
|
+
#else // optimized for long vertices.
|
426
|
+
for (x = 0, i = 1; i <= n; ++i) {
|
427
|
+
if (i == n || a[i].vd != a[i-1].vd + 1) {
|
428
|
+
gwf_ed_extend_batch(buf->km, g, es, ql, q, i - x, &a[x], &B, A, &buf->tmp, r);
|
429
|
+
x = i;
|
430
|
+
}
|
431
|
+
}
|
432
|
+
if (kdq_size(A) == 0) do_dedup = 0;
|
433
|
+
#endif
|
434
|
+
kfree(buf->km, a); // $a is not used as it has been copied to $A
|
435
|
+
|
436
|
+
while (kdq_size(A)) {
|
437
|
+
gwf_diag_t t;
|
438
|
+
uint32_t v, x0;
|
439
|
+
int32_t ooo, d, k, i, vl;
|
440
|
+
|
441
|
+
t = *kdq_shift(gwf_diag_t, A);
|
442
|
+
ooo = t.xo&1, v = t.vd >> 32; // vertex
|
443
|
+
d = (int32_t)t.vd - GWF_DIAG_SHIFT; // diagonal
|
444
|
+
k = t.k; // wavefront position on the vertex
|
445
|
+
vl = es[v].len; // $vl is the vertex length
|
446
|
+
k = gwf_extend1(d, k, vl, es[v].seq, ql, q);
|
447
|
+
i = k + d; // query position
|
448
|
+
x0 = (t.xo >> 1) + ((k - t.k) << 1); // current anti diagonal
|
449
|
+
|
450
|
+
if (k + 1 < vl && i + 1 < ql) { // the most common case: the wavefront is in the middle
|
451
|
+
int32_t push1 = 1, push2 = 1;
|
452
|
+
if (B.n >= 2) push1 = gwf_diag_update(&B.a[B.n - 2], v, d-1, k+1, x0 + 1, ooo, t.t);
|
453
|
+
if (B.n >= 1) push2 = gwf_diag_update(&B.a[B.n - 1], v, d, k+1, x0 + 2, ooo, t.t);
|
454
|
+
if (push1) gwf_diag_push(buf->km, &B, v, d-1, k+1, x0 + 1, 1, t.t);
|
455
|
+
if (push2 || push1) gwf_diag_push(buf->km, &B, v, d, k+1, x0 + 2, 1, t.t);
|
456
|
+
gwf_diag_push(buf->km, &B, v, d+1, k, x0 + 1, ooo, t.t);
|
457
|
+
} else if (i + 1 < ql) { // k + 1 == g->len[v]; reaching the end of the vertex but not the end of query
|
458
|
+
int32_t nv = gfa_arc_n(g, v), j, n_ext = 0, tw = -1;
|
459
|
+
gfa_arc_t *av = gfa_arc_a(g, v);
|
460
|
+
gwf_intv_t *p;
|
461
|
+
kv_pushp(gwf_intv_t, buf->km, buf->tmp, &p);
|
462
|
+
p->vd0 = gwf_gen_vd(v, d), p->vd1 = p->vd0 + 1;
|
463
|
+
if (opt->traceback) tw = gwf_trace_push(buf->km, &buf->t, v, t.t, buf->ht);
|
464
|
+
for (j = 0; j < nv; ++j) { // traverse $v's neighbors
|
465
|
+
uint32_t w = av[j].w; // $w is next to $v
|
466
|
+
int32_t ol = av[j].ow;
|
467
|
+
int absent;
|
468
|
+
gwf_set64_put(buf->ha, (uint64_t)w<<32 | (i + 1), &absent); // test if ($w,$i) has been visited
|
469
|
+
if (q[i + 1] == es[w].seq[ol]) { // can be extended to the next vertex without a mismatch
|
470
|
+
++n_ext;
|
471
|
+
if (absent) {
|
472
|
+
gwf_diag_t *p;
|
473
|
+
p = kdq_pushp(gwf_diag_t, A);
|
474
|
+
p->vd = gwf_gen_vd(w, i + 1 - ol), p->k = ol, p->xo = (x0+2)<<1 | 1, p->t = tw;
|
475
|
+
}
|
476
|
+
} else if (absent) {
|
477
|
+
gwf_diag_push(buf->km, &B, w, i - ol, ol, x0 + 1, 1, tw);
|
478
|
+
gwf_diag_push(buf->km, &B, w, i + 1 - ol, ol, x0 + 2, 1, tw);
|
479
|
+
}
|
480
|
+
}
|
481
|
+
if (nv == 0 || n_ext != nv) // add an insertion to the target; this *might* cause a duplicate in corner cases
|
482
|
+
gwf_diag_push(buf->km, &B, v, d+1, k, x0 + 1, 1, t.t);
|
483
|
+
} else if (v1 == (uint32_t)-1 || (v == v1 && k == off1)) { // i + 1 == ql
|
484
|
+
r->end_v = v, r->end_off = k, r->wlen = x0 - i - 1, *end_tb = t.t, *n_a_ = 0;
|
485
|
+
kdq_destroy(gwf_diag_t, A);
|
486
|
+
kfree(buf->km, B.a);
|
487
|
+
return 0;
|
488
|
+
} else if (k + 1 < vl) { // i + 1 == ql; reaching the end of the query but not the end of the vertex
|
489
|
+
gwf_diag_push(buf->km, &B, v, d-1, k+1, x0 + 1, ooo, t.t); // add an deletion; this *might* case a duplicate in corner cases
|
490
|
+
} else if (v != v1) { // i + 1 == ql && k + 1 == g->len[v]; not reaching the last vertex $v1
|
491
|
+
int32_t nv = gfa_arc_n(g, v), j, tw = -1;
|
492
|
+
const gfa_arc_t *av = gfa_arc_a(g, v);
|
493
|
+
if (opt->traceback) tw = gwf_trace_push(buf->km, &buf->t, v, t.t, buf->ht);
|
494
|
+
for (j = 0; j < nv; ++j)
|
495
|
+
gwf_diag_push(buf->km, &B, av[j].w, i - av[j].ow, av[j].ow, x0 + 1, 1, tw); // deleting the first base on the next vertex
|
496
|
+
} else { // may come here when k>off1 (due to banding); do nothing in this case
|
497
|
+
}
|
498
|
+
}
|
499
|
+
|
500
|
+
kdq_destroy(gwf_diag_t, A);
|
501
|
+
*n_a_ = n = B.n, b = B.a;
|
502
|
+
|
503
|
+
if (do_dedup) *n_a_ = n = gwf_dedup(buf, n, b);
|
504
|
+
if (opt->max_lag > 0 && n > opt->max_chk && ((s+1)&0xf) == 0)
|
505
|
+
*n_a_ = n = gwf_prune(n, b, opt->max_lag, opt->bw_dyn);
|
506
|
+
return b;
|
507
|
+
}
|
508
|
+
|
509
|
+
static void gwf_traceback(gwf_edbuf_t *buf, int32_t end_v, int32_t end_tb, gfa_edrst_t *path)
|
510
|
+
{
|
511
|
+
int32_t i = end_tb, n = 1;
|
512
|
+
while (i >= 0 && buf->t.a[i].v >= 0)
|
513
|
+
++n, i = buf->t.a[i].pre;
|
514
|
+
KMALLOC(buf->km, path->v, n);
|
515
|
+
i = end_tb, n = 0;
|
516
|
+
path->v[n++] = end_v;
|
517
|
+
while (i >= 0 && buf->t.a[i].v >= 0)
|
518
|
+
path->v[n++] = buf->t.a[i].v, i = buf->t.a[i].pre;
|
519
|
+
path->nv = n;
|
520
|
+
for (i = 0; i < path->nv>>1; ++i)
|
521
|
+
n = path->v[i], path->v[i] = path->v[path->nv - 1 - i], path->v[path->nv - 1 - i] = n;
|
522
|
+
}
|
523
|
+
|
524
|
+
static void gwf_ed_print_diag(const gfa_t *g, size_t n, gwf_diag_t *a) // for debugging only
|
525
|
+
{
|
526
|
+
size_t i;
|
527
|
+
for (i = 0; i < n; ++i) {
|
528
|
+
int32_t d = (int32_t)a[i].vd - GWF_DIAG_SHIFT;
|
529
|
+
printf("Z\t%d\t%s\t%d\t%d\t%d\n", d + a[i].k, g->seg[(a[i].vd>>32)>>1].name, d, a[i].k, a[i].xo>>1);
|
530
|
+
}
|
531
|
+
}
|
532
|
+
|
533
|
+
static void gwf_ed_print_intv(size_t n, gwf_intv_t *a) // for debugging only
|
534
|
+
{
|
535
|
+
size_t i;
|
536
|
+
for (i = 0; i < n; ++i)
|
537
|
+
printf("Z\t%d\t%d\t%d\n", (int32_t)(a[i].vd0>>32), (int32_t)a[i].vd0 - GWF_DIAG_SHIFT, (int32_t)a[i].vd1 - GWF_DIAG_SHIFT);
|
538
|
+
}
|
539
|
+
|
540
|
+
typedef struct {
|
541
|
+
const gfa_t *g;
|
542
|
+
const gfa_edseq_t *es;
|
543
|
+
const gfa_edopt_t *opt;
|
544
|
+
int32_t ql;
|
545
|
+
const char *q;
|
546
|
+
gwf_edbuf_t buf;
|
547
|
+
int32_t s, n_a;
|
548
|
+
gwf_diag_t *a;
|
549
|
+
int32_t end_tb;
|
550
|
+
} gfa_edbuf_t;
|
551
|
+
|
552
|
+
void *gfa_ed_init(void *km, const gfa_edopt_t *opt, const gfa_t *g, const gfa_edseq_t *es, int32_t ql, const char *q, uint32_t v0, int32_t off0)
|
553
|
+
{
|
554
|
+
gfa_edbuf_t *z;
|
555
|
+
KCALLOC(km, z, 1);
|
556
|
+
z->buf.km = km;
|
557
|
+
z->opt = opt;
|
558
|
+
z->g = g, z->es = es;
|
559
|
+
z->ql = ql, z->q = q;
|
560
|
+
z->buf.ha = gwf_set64_init2(km);
|
561
|
+
z->buf.ht = gwf_map64_init2(km);
|
562
|
+
kv_resize(gwf_trace_t, km, z->buf.t, 16);
|
563
|
+
KCALLOC(km, z->a, 1);
|
564
|
+
z->a[0].vd = gwf_gen_vd(v0, -off0), z->a[0].k = off0 - 1, z->a[0].xo = 0;
|
565
|
+
if (z->opt->traceback) z->a[0].t = gwf_trace_push(km, &z->buf.t, -1, -1, z->buf.ht);
|
566
|
+
z->n_a = 1;
|
567
|
+
return z;
|
568
|
+
}
|
569
|
+
|
570
|
+
void gfa_ed_step(void *z_, uint32_t v1, int32_t off1, int32_t s_term, gfa_edrst_t *r)
|
571
|
+
{
|
572
|
+
gfa_edbuf_t *z = (gfa_edbuf_t*)z_;
|
573
|
+
const gfa_edopt_t *opt = z->opt;
|
574
|
+
if (s_term < 0 && z->opt->s_term >= 0) s_term = z->opt->s_term;
|
575
|
+
r->n_end = 0, r->n_iter = 0;
|
576
|
+
while (z->n_a > 0) {
|
577
|
+
z->a = gwf_ed_extend(&z->buf, opt, z->g, z->es, z->s, z->ql, z->q, v1, off1, &z->end_tb, &z->n_a, z->a, r);
|
578
|
+
r->n_iter += z->n_a; // + z->buf.intv.n;
|
579
|
+
if (r->end_off >= 0 || z->n_a == 0) break;
|
580
|
+
if (r->n_end > 0) break;
|
581
|
+
if (s_term >= 0 && z->s >= s_term) break;
|
582
|
+
if (z->opt->i_term > 0 && r->n_iter > z->opt->i_term) break;
|
583
|
+
++z->s;
|
584
|
+
if (gfa_ed_dbg >= 1) {
|
585
|
+
printf("[%s] dist=%d, n=%d, n_intv=%ld, n_tb=%ld\n", __func__, z->s, z->n_a, z->buf.intv.n, z->buf.t.n);
|
586
|
+
if (gfa_ed_dbg == 2) gwf_ed_print_diag(z->g, z->n_a, z->a);
|
587
|
+
if (gfa_ed_dbg == 3) gwf_ed_print_intv(z->buf.intv.n, z->buf.intv.a);
|
588
|
+
}
|
589
|
+
}
|
590
|
+
if (opt->traceback && r->end_off >= 0)
|
591
|
+
gwf_traceback(&z->buf, r->end_v, z->end_tb, r);
|
592
|
+
r->s = r->end_v != (uint32_t)-1? z->s : -1;
|
593
|
+
}
|
594
|
+
|
595
|
+
void gfa_ed_destroy(void *z_)
|
596
|
+
{
|
597
|
+
gfa_edbuf_t *z = (gfa_edbuf_t*)z_;
|
598
|
+
void *km = z->buf.km;
|
599
|
+
kfree(km, z->a);
|
600
|
+
gwf_set64_destroy(z->buf.ha);
|
601
|
+
gwf_map64_destroy(z->buf.ht);
|
602
|
+
kfree(km, z->buf.ooo.a);
|
603
|
+
kfree(km, z->buf.intv.a);
|
604
|
+
kfree(km, z->buf.tmp.a);
|
605
|
+
kfree(km, z->buf.swap.a);
|
606
|
+
kfree(km, z->buf.t.a);
|
607
|
+
kfree(km, z);
|
608
|
+
}
|
609
|
+
|
610
|
+
int32_t gfa_edit_dist(void *km, const gfa_edopt_t *opt, const gfa_t *g, const gfa_edseq_t *es, int32_t ql, const char *q, uint32_t v0, int32_t off0, gfa_edrst_t *rst)
|
611
|
+
{
|
612
|
+
void *z;
|
613
|
+
z = gfa_ed_init(km, opt, g, es, ql, q, v0, off0);
|
614
|
+
gfa_ed_step(z, (uint32_t)-1, -1, -1, rst);
|
615
|
+
gfa_ed_destroy(z);
|
616
|
+
return rst->s;
|
617
|
+
}
|