-
Notifications
You must be signed in to change notification settings - Fork 1
/
xtree.c
1644 lines (1488 loc) · 58.6 KB
/
xtree.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// Gabriel Al-Ghalith. Efficient characterization of orthogonal metagenomic
// taxonomy and pathway coverage with CrossTree. 2018.
#define VER "CrossTree v0.92i by Gabe"
#define VNO 1
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <omp.h>
#include <stdint.h>
#ifndef _OPENMP
#define omp_get_max_threads() 1
#define omp_get_thread_num() 0
#include <time.h>
#define omp_get_wtime() ((double)clock()/CLOCKS_PER_SEC)
#define omp_set_num_threads(x) x
#endif
#include <sys/mman.h>
#include <sys/stat.h>
#ifndef kmer_t
#define kmer_t uint32_t
#endif
#ifndef rix_t
#define rix_t uint32_t
#endif
#define AMBIG 4
#include <math.h>
#include <zlib.h>
void * huge_malloc(size_t n) {
void *ptr = 0;
posix_memalign(&ptr, 1 << 21, n);
madvise(ptr, n, MADV_HUGEPAGE);
return ptr;
}
void * huge_calloc(size_t n) {
void *ptr = huge_malloc(n);
memset(ptr,0,n);
return ptr;
}
const uint8_t CONV[32] = {4,0,4,1,4,4,4,2,4,4,4,4,4,4,4,4,4,4,4,4,3,3,4,4,4,4,4,4,4,4,4,4};
const uint8_t RCONV[32] = {4,3,4,2,4,4,4,1,4,4,4,4,4,4,4,4,4,4,4,4,0,0,4,4,4,4,4,4,4,4,4,4};
#pragma pack(1)
typedef struct {
kmer_t sfx;
rix_t rix;
} KPod;
//typedef struct {uint32_t h1, h2;} hpair_t;
int strcmp_ptr(const void* a, const void* b)
{ return strcmp(*(char**)a,*(char**)b); }
typedef union {uint32_t a[4]; __uint128_t n; uint8_t b[16];} MasterBin_t;
int binCmp(const void *a, const void *b) {
MasterBin_t *A = *(MasterBin_t **)a, *B = *(MasterBin_t **)b;
uint64_t valA = *(uint64_t *)(A->b+4), valB = *(uint64_t *)(B->b+4);
return valA < valB ? -1 : valB < valA;
}
static inline uint32_t prefix_to_num(char *s, int nl, int *err, uint32_t kshift) {
uint32_t nib = 0, k = kshift;
for (uint32_t i = 0; i < nl; ++i, k-=2) {
uint32_t c = CONV[31 & s[i]];
if (c == AMBIG) {*err = 1; return i;}
nib |= c << k;
}
return nib;
}
static inline uint32_t prefix_to_num_RC(char *s, int nl) {
uint32_t nib = 0, k = 0, c;
for (uint32_t i = 0; i < nl; ++i, k+=2)
c = RCONV[31 & s[i]],
nib |= c << k;
return nib;
}
static inline kmer_t dna_to_num(char *s, int nl, int *err, kmer_t kshift) {
kmer_t nib = 0, k = kshift;
for (uint32_t i = 0; i < nl; ++i, k-=2) {
kmer_t c = CONV[31 & s[i]];
if (c == AMBIG) {*err = 1; return i;}
nib |= c << k;
}
return nib;
}
static inline kmer_t dna_to_num_RC(char *s, int nl) {
kmer_t nib = 0, k = 0, c;
for (uint32_t i = 0; i < nl; ++i, k+=2)
c = RCONV[31 & s[i]],
nib |= c << k;
return nib;
}
static inline uint64_t binsearch_str(char **Strings, char *key, uint64_t N) {
uint64_t lo = 0, hi = N;
while (lo < hi) {
uint64_t mid = lo + ((hi-lo) >> 1);
int cmp = strcmp(key,Strings[mid]);
if (cmp > 0) lo = mid+1;
else if (cmp < 0) hi = mid;
else return mid;
}
return -1;
}
// Key string delimited by tab, null, newline...
static inline uint64_t binsearch_str_d(char **Strings, char *key, uint64_t N) {
uint64_t lo = 0, hi = N;
while (lo < hi) {
uint64_t mid = lo + ((hi-lo) >> 1);
char *a = key, *b = Strings[mid];
while (*b && *a==*b) ++a, ++b;
if (!*b && (!*a || *a == '\n' || *a=='\t')) return mid;
if (*a < *b) hi = mid;
else lo = mid+1;
}
return -1;
}
// returns first possible match as if null existed at position key_len, matched as far as possible
// A helper function will translate position via Map_h* minus ranks missing.
static inline uint64_t binsearch_str_L(char **Strings, char *key, uint64_t N, uint32_t key_len) {
uint64_t lo = 0, hi = N;
while (lo < hi) {
uint64_t mid = lo + ((hi - lo) >> 1);
int cmp = 0;
//strcmp(key,Strings[mid]);
uint32_t i = 0; char *ref = Strings[mid];
for (; i < key_len && ref[i]; ++i)
if (key[i] != ref[i]) break;
if (i == key_len /* && key[i] */ && ref[i]) cmp = -1;
else cmp = key[i] - ref[i];
//printf("Compared key:\n%s\nagainst ref:\n%s\nuntil pos %u: %d [lo=%lu, mid=%lu, hi=%lu]\n",
// key,ref,key_len,cmp,lo,mid,hi);
//if (!cmp) return mid;
if (cmp > 0) lo = mid + 1;
else if (cmp < 0) hi = mid;
else return mid;
}
return lo;
}
static inline int cmpfunc(const void *a, const void *b) {
return *(uint64_t*)a < *(uint64_t*)b ? -1 :
*(uint64_t*)b < *(uint64_t*)a;
}
static inline int u16cmp(const void *a, const void *b) {
return *(uint16_t*)a < *(uint16_t*)b ? -1 :
*(uint16_t*)b < *(uint16_t*)a;
}
static inline int kpackcmp(const void *a, const void *b) {
KPod *k1 = (KPod *)a, *k2 = (KPod *)b;
if (k1->sfx < k2->sfx) return -1;
if (k2->sfx < k1->sfx) return 1;
if (k1->rix < k2->rix) return -1;
if (k2->rix < k1->rix) return 1;
return 0;
}
/* static inline KPod * WBS_k(KPod *KP, uint64_t Lx, kmer_t k) {
KPod *p=KP;
while (Lx) {
size_t w = (Lx >> 1) + 1;
if (p[w].sfx < k) p+=w, Lx-=w;
else if (p[w].sfx == k) return p+w;
else Lx = w-1;
}
return p->sfx==k ? p : 0;
} */
static inline uint64_t LBS_k(KPod *KP, uint64_t N, kmer_t k) {
uint64_t L = 0, R = N;
while (L < R) {
uint64_t m = (L + R) >> 1;
if (KP[m].sfx < k) L = m+1;
else R = m;
}
return KP[L].sfx == k ? L : -1;
}
static inline uint64_t get_queries(gzFile in, uint8_t **QBucket, char **HBucket, uint8_t *head, uint8_t *line, int qChunk, uint64_t szmax) {
uint64_t nq = 0; uint8_t *QB_ptr = *QBucket; char *H_ptr = *HBucket;
uint8_t *eol; int len;
while (nq < qChunk && QB_ptr - *QBucket <= szmax) {
if (!gzgets(in,head,(1 << 20)-1)) break;
eol = strchr(head,'\n');
*eol = 0;
len = eol-head+1;
memcpy(H_ptr,head+1,len-1);
HBucket[nq] = H_ptr;
H_ptr += len-1;
if (!gzgets(in,line,(1 << 28)-1)) break;
eol = strchr(line,'\n');
*eol = 0;
len = eol-line+1;
memcpy(QB_ptr,line,len);
QBucket[nq] = QB_ptr;
QB_ptr += len;
if (*head == '@' && (!gzgets(in,line,1 << 28) ||
!gzgets(in,line,1 << 28))) break;
++nq;
}
return nq;
}
#define USAGE1 "USAGE: xtree {BUILD,ALIGN} [options]\n "
#define USAGE2 USAGE1 "Options for both BUILD and ALIGN, with args: {seqs,log-out,threads,db}\n"
#define USAGE3 USAGE2 "BUILD Options\n With args: {map,comp,k,db-out} <arg>\n"
#define USAGE4 USAGE3 "ALIGN Options\n With args: {confidence,perq-out,ref-out,tax-out,cov-out,orthog-out}\n"
#define USAGE USAGE4 " Without args: {redistribute,shallow-lca,copymem}"
int main(int argc, char *argv[]) {
puts(VER);
char *dbPath = 0, *seqPath = 0, *ixPath = 0, *covPath = 0;
char *perqPath = 0, *taxPath = 0, *orthogPath = 0, *refPath = 0, *logPath = 0;
int doBuild = 0, threads = omp_get_max_threads(), comp = 0, kchoice = 0;
int doFullLCA = 1, doRedist = 0, doFastRedist = 0, doCopyMem = 0;
double conf = 0.33; // Reasonable default for compression lv 2
uint32_t nUniqMatches = 0;
// Robust parser
for (int a = 1; a < argc; ++a) {
if (!strcmp(argv[a],"BUILD")) doBuild = 1;
else if (!strcmp(argv[a],"--map")) ixPath = argv[++a];
else if (!strcmp(argv[a],"--comp")) comp = atoi(argv[++a]); //I
else if (!strcmp(argv[a],"--k")) kchoice = atoi(argv[++a]); //I
else if (!strcmp(argv[a],"ALIGN")) doBuild = 0;
else if (!strcmp(argv[a],"--confidence")) {
double ctemp = atof(argv[++a]);
if (ctemp <= 1) conf = ctemp, printf("Setting confprop %f\n",conf);
else nUniqMatches = ctemp, printf("Setting min uniq ref matches to %u\n",nUniqMatches);
}
else if (!strcmp(argv[a],"--perq-out")) perqPath = argv[++a];
else if (!strcmp(argv[a],"--ref-out")) refPath = argv[++a];
else if (!strcmp(argv[a],"--tax-out")) taxPath = argv[++a];
else if (!strcmp(argv[a],"--cov-out")) covPath = argv[++a];
else if (!strcmp(argv[a],"--orthog-out")) orthogPath = argv[++a];
else if (!strcmp(argv[a],"--redistribute")) doRedist = 1; //NA
else if (!strcmp(argv[a],"--fast-redistribute")) doRedist = 1, doFastRedist = 1; //NA
else if (!strcmp(argv[a],"--shallow-lca")) doFullLCA = 0; //NA
else if (!strcmp(argv[a],"--copymem")) doCopyMem = 1; //NA
// Options for both BUILD and ALIGN
else if (!strcmp(argv[a],"--seqs")) seqPath = argv[++a];
else if (!strcmp(argv[a],"--log-out")) logPath = argv[++a];
else if (!strcmp(argv[a],"--threads")) threads = atoi(argv[++a]);
else if (!strcmp(argv[a],"--db") || !strcmp(argv[a],"--db-out"))
dbPath = argv[++a];
else {printf("Unrecognized option: %s\n",argv[a]); exit(1);}
}
threads = threads > 256 ? 256 : threads;
omp_set_num_threads(threads);
printf("Using %d thread(s)\n",threads);
if (argc < 4) {puts(USAGE); exit(1);}
if (doBuild) {
uint32_t PL = 13, SL = sizeof(kmer_t)*4;
uint64_t K = PL+SL;
if (comp) printf("Setting compression level to %d\n",comp);
if (kchoice) K = kchoice;
SL = K - PL;
if (K < PL || !SL || SL > sizeof(kmer_t)*4) {printf("Bad K! [%lu]\n",K); exit(1);}
printf("Building DB with K=%lu [PL %d, SL %d]\n",K,PL,SL);
uint32_t kpre_shf = PL*2-2;
kmer_t kpst_shf = SL*2-2;
uint32_t pre_bshf = 32-(PL*2), pre_bshf_2 = pre_bshf+2;
kmer_t pst_bshf = sizeof(kmer_t)*8-(SL*2), pst_bshf_2 = pst_bshf+2;
FILE *in = fopen(seqPath,"rb");
if (!in) {printf("ERROR: bad FASTA input: %s\n",seqPath); exit(2);}
struct stat sb; int fno = fileno(in); fstat(fno,&sb);
uint64_t fsz = sb.st_size;
char *Raw = mmap(0, fsz+16, PROT_READ, MAP_SHARED, fno, 0);
madvise(Raw,fsz,MADV_SEQUENTIAL);
// We need: num records, num valid k-mers (each!), header locs
if (Raw[0] != '>') {puts("Uh oh. Input FASTA looks fishy."); exit(2);}
// Set up the tank for input -- up to 1 billion refs allowed
uint64_t nbins = 1<<(2*PL),
*Offsets = huge_malloc(((uint64_t)1<<30)*sizeof(*Offsets)),
*Nibs = huge_calloc((nbins+1)*sizeof(*Nibs));
if (!Offsets || !Nibs) {puts("ERROR:OOM Offsets"); exit(3);}
uint32_t ns = 0;
double wtime = omp_get_wtime();
#pragma omp parallel for
for (uint64_t z = 0; z < fsz; ++z) {
if (Raw[z] > 64 && Raw[z-1] == '\n') {
uint32_t ix;
#pragma omp atomic capture
ix = ns++;
uint64_t x = z, y = z;
while (Raw[y] && Raw[y] != '\n') ++y;
uint32_t num, slideSafe = 0;
Offsets[ix] = x; // Offsets contain start of sequence
while (x + K <= y) {
int err=0, a = 0;
if (slideSafe) {
uint32_t c = CONV[31 & Raw[x+PL-1]];
if (c==AMBIG) {x+=PL; slideSafe=0; continue;}
num = (num << pre_bshf_2 >> pre_bshf) | c;
} else {
num = prefix_to_num(Raw+x,PL,&err,kpre_shf);
if (err) {x+= num+1; slideSafe = 0; continue;}
slideSafe = 1;
}
while (a < comp && !CONV[31 & Raw[x+a-comp]]) ++a; // compression
if (a == comp)
#pragma omp atomic
++Nibs[num];
++x;
}
}
}
printf("There were %u records here (%f s)\n",ns,omp_get_wtime()-wtime);
if (ns > 65535 && sizeof(rix_t) == 2)
{puts("ERROR: too many refs (>65K)"); exit(2);}
//Offsets = realloc(Offsets,ns*sizeof(*Offsets));
qsort(Offsets,ns,sizeof(*Offsets),cmpfunc);
// Create the data structures
uint64_t totWords = 0;
for (uint64_t i = 0; i < nbins; ++i)
totWords += Nibs[i];
printf("In total, we need a structure that is %lu large.\n",totWords);
printf("Now allocating %f GB of RAM...\n",(double)totWords*sizeof(KPod)/1073741824);
wtime = omp_get_wtime();
KPod *KGrid = huge_malloc(totWords*sizeof(*KGrid));
uint64_t *KIx = huge_calloc((nbins+1)*sizeof(*KIx));
if (!KIx) {puts("OOM:KIx"); exit(3);}
for (uint64_t i = 1; i < nbins; ++i)
KIx[i] = KIx[i-1] + Nibs[i-1];
for (uint64_t i = 0; i < nbins; ++i) Nibs[i] = KIx[i];
#ifdef WSL
for (uint64_t i = 0; i < totWords; i+=4096/sizeof(*KGrid))
KGrid[i].sfx=1;
#endif
uint32_t mask32 = ((uint64_t)1 << (2*PL)) - 1;
kmer_t maskK = ((kmer_t)1 << (2*SL)) - 1;
if (SL==sizeof(kmer_t)*4) maskK = -1;
#pragma omp parallel for schedule(dynamic)
for (uint32_t i = 0; i < ns; ++i) {
uint64_t x = Offsets[i];
uint64_t y = x; while (Raw[y] && Raw[y] != '\n') ++y;
int slideSafe = 0; uint32_t num; kmer_t kmer;
while (x + K <= y) {
int err=0, a = 0;
if (slideSafe) {
uint32_t c = CONV[31 & Raw[x+PL-1]];
if (c==AMBIG) {x+=PL; slideSafe=0; continue;}
//num = (num << pre_bshf_2 >> pre_bshf) | c;
num = ((num << 2) | c) & mask32;
kmer_t k = CONV[31 & Raw[x+K-1]];
if (k==AMBIG) {x+=K; slideSafe=0; continue;}
//kmer = (kmer << pst_bshf_2 >> pst_bshf) | k;
kmer = ((kmer << 2) | k) & maskK;
} else {
num = prefix_to_num(Raw+x,PL,&err,kpre_shf);
if (err) {x+= num+1; slideSafe=0; continue;}
kmer = dna_to_num(Raw+x+PL,SL,&err,kpst_shf);
if (err) {x+= kmer+1; slideSafe=0; continue;}
slideSafe = 1;
}
while (a < comp && !CONV[31 & Raw[x+a-comp]]) ++a; // compression
if (a==comp) {
uint64_t pod_ix;
#pragma omp atomic capture
pod_ix = KIx[num]++;
KGrid[pod_ix] = (KPod){kmer,i};
}
++x;
}
}
printf("Time: %f\n",omp_get_wtime()-wtime);
uint64_t numK = 0;
#pragma omp parallel for schedule(dynamic,1024) reduction(+:numK)
for (uint64_t i = 0; i < nbins; ++i) {
if (KIx[i] <= Nibs[i]) continue;
KPod *start = KGrid + Nibs[i];
size_t num = KIx[i]-Nibs[i];
qsort(start,num,sizeof(*start),kpackcmp);
numK += num;
}
printf("There were %lu k-mers.\n",numK);
printf("Some stats on distributions!\n");
uint64_t n_dupe = 0, n_multi = 0;
#pragma omp parallel for schedule(dynamic,1024) reduction(+:n_dupe,n_multi)
for (uint64_t i = 0; i < nbins; ++i) {
if (KIx[i] <= Nibs[i]) continue;
for (uint64_t j = Nibs[i]+1; j < KIx[i]; ++j)
if (KGrid[j].sfx==KGrid[j-1].sfx) {
++n_multi;
if (KGrid[j].rix==KGrid[j-1].rix) ++n_dupe;
}
}
printf("Exact duplicates: %lu; across refs: %lu\n",n_dupe,n_multi);
// Now write the results
/* File structure:
1. Version byte and rix_t size [1]
2. Size of prefix [1]
3. Size of suffix [1]
4. Size of kmer_t [1]
5. Num refs [4]
6. Num kmers [8]
7. All prefix indices [1 << (2 * #2) x 8]
8. All kmer data [(#2 + #4) * #6]
9. Size of string data [8]
10. All strings [#9]
11. Number of Ix1 in map [4] [0 means skip rest of file]
12. String size for Ix1 [8]
13. Ix1 strings dump [#12]
14. Number of Ix2 in map [4] [can be 0/skipped if no h2 map]
15. String size of Ix2 [8]
16. Ix2 strings dump [#15]
//17. hpair_t dump [num ref by 8]
17. HPairs[0] dump [num ref by 4]
18. HPairs[1] dump [num ref by 4]
*/
uint64_t fileSz = 0, stringSz = 0;
#pragma omp parallel for reduction(+:stringSz)
for (uint32_t i = 0; i < ns; ++i) {
uint64_t x = Offsets[i];
uint64_t y = x; while (Raw[y] != '>') --y;
stringSz += x-y - 1; // -1 for the '>' we're on, -1 '\n', but +1 '\0'
}
fileSz = 24 + sizeof(*Nibs)*(nbins+1) + sizeof(*KGrid)*numK + stringSz;
printf("Initial file size = %lu\n",fileSz+4);
FILE *db = fopen(dbPath,"wb");
if (!db) {puts("I/O error: invalid db output file"); exit(1);}
setvbuf(db, 0, _IOFBF, 1<<22);
wtime = omp_get_wtime();
fputc((VNO << 4) | sizeof(rix_t),db);
fputc(PL,db); fputc(SL,db); fputc(sizeof(kmer_t),db);
size_t wrote = 4;
wrote += fwrite(&ns,sizeof(ns),1,db);
wrote += fwrite(&numK,sizeof(numK),1,db);
uint64_t tally = 0;
for (uint64_t i = 0; i < nbins+1; ++i) {
wrote += fwrite(&tally,sizeof(*Nibs),1,db);
tally += KIx[i]-Nibs[i];
}
printf("First write: %f s.\n",omp_get_wtime()-wtime);
wtime = omp_get_wtime();
for (uint64_t i = 0; i < nbins; ++i) {
if (KIx[i] <= Nibs[i]) continue;
uint64_t num = KIx[i]-Nibs[i];
fwrite(KGrid + Nibs[i], sizeof(*KGrid), num,db);
}
printf("Second write: %f s.\n",omp_get_wtime()-wtime);
wtime = omp_get_wtime();
fwrite(&stringSz,sizeof(stringSz),1,db);
for (uint32_t i = 0; i < ns; ++i) {
uint64_t x = Offsets[i];
uint64_t y = x; while (Raw[y] != '>') --y;
fwrite(Raw+y+1,1,x-y-2,db);fputc(0,db);
}
printf("Third write: %f s.\n",omp_get_wtime()-wtime);
// TODO: also write statistics; for each genome, total K etc
FILE *out = fopen(logPath,"wb");
if (!out) printf("No log file specified; won't produce tally\n");
else {
uint32_t *TotK_m = huge_calloc((uint64_t)ns*sizeof(*TotK_m));
uint32_t *TotUniq_m = huge_calloc((uint64_t)ns*sizeof(*TotUniq_m));
#pragma omp parallel
{
int tid = omp_get_thread_num();
uint32_t *TotK = TotK_m, *TotUniq = TotUniq_m;
if (tid)
TotK = huge_calloc((uint64_t)ns*sizeof(*TotK)),
TotUniq = huge_calloc((uint64_t)ns*sizeof(*TotUniq));
#pragma omp for schedule(dynamic,1024)
for (uint64_t i = 0; i < nbins; ++i) {
if (KIx[i] <= Nibs[i]) continue; // empty
uint32_t ambig = 0;
uint64_t end = Nibs[i]+(KIx[i]-Nibs[i]), nd;
kmer_t thisK = KGrid[Nibs[i]].sfx+1;
for (uint64_t j = Nibs[i]; j < end; j += nd) {
rix_t rix = KGrid[j].rix;
// If new k-mer, check if ambig, store max value
if (KGrid[j].sfx != thisK) {
thisK = KGrid[j].sfx;
ambig = 0;
for (uint64_t k = j+1; k < end && KGrid[k].sfx == thisK; ++k)
ambig |= KGrid[k].rix ^ rix;
}
// Find number of in-ref copies
nd = 1;
for (uint64_t k = j+1; k < end &&
KGrid[k].sfx == thisK && KGrid[k].rix == rix; ++k) ++nd;
// Increment the appropriate variables
if (!ambig) TotUniq[rix] += nd;
TotK[rix] += nd;
}
}
#pragma omp critical
if (tid) for (uint32_t i = 0; i < ns; ++i)
TotK_m[i] += TotK[i], TotUniq_m[i] += TotUniq[i];
}
fprintf(out,"Reference\tTotalKmers\tUniqKmers\n");
for (uint64_t i = 0; i < ns; ++i) {
uint64_t x = Offsets[i];
uint64_t y = x; while (Raw[y] != '>') --y;
fwrite(Raw+y+1,1,x-y-2,out); // write the ref name
fprintf(out,"\t%u\t%u\n",TotK_m[i],TotUniq_m[i]);
}
}
// Handle the H1/H2 mappings
if (!ixPath) { // If no ixMap was provided, finish up.
uint32_t zeroRef = 0;
fwrite(&zeroRef,sizeof(zeroRef),1,db);
exit(0);
}
wtime = omp_get_wtime();
// Read the mapping file in. Up to 3 columns are used (first is index)
// proof of concept -- read whole file, sort all 3 (perhaps parallel) by ptrs
// Then perform classical deduplication and data structure creation
free(Nibs); free(KGrid); free(KIx);
// New observation: can determine a priori number and placement of bins
// for all possible interpolations, given sorted map. Also search/store!
//gzFile map = gzopen(ixPath,"rb");
//int mapFsz = gzread(map,wholeMap,(uint64_t)1 << 38);
FILE *map = fopen(ixPath,"rb");
if (!map) {fprintf(stderr,"Can't open map: %s\n",ixPath); exit(2);}
uint64_t sz = 0;
fseeko(map,0,SEEK_END); sz = ftello(map); rewind(map);
if (sz < 2) {fprintf(stderr,"ERR: map malformatted\n"); exit(2);}
char *wholeMap = huge_calloc(sz+16);
size_t mapFsz = fread(wholeMap,1,sz,map);
if (mapFsz != sz) {puts("BAD MAP!"); exit(10101);}
uint64_t nL = 0;
if (wholeMap[sz-1]!='\n') ++nL;
#pragma omp parallel for reduction(+:nL)
for (uint64_t i = 0; i < sz; ++i)
nL += wholeMap[i] == '\n';
printf("Map contained %lu lines.\n",nL);
//typedef struct {char *str, *h1, *h2;} map_str_t;
//map_str_t *MapStrs = calloc(nL,sizeof(*MapStrs));
char **RefStr = calloc(nL,sizeof(*RefStr)),
**H1Str = calloc(nL,sizeof(*H1Str)),
**H2Str = calloc(nL,sizeof(*H2Str));
char *map_ptr = wholeMap;
int ncol = 2; uint64_t numTimesH2showed = 0;
for (uint64_t i = 0; i < nL; ++i) {
RefStr[i] = map_ptr;
while (*map_ptr != '\t' && *map_ptr != '\n') ++map_ptr;
if (*map_ptr == '\n') {puts("Bad map! Need >1 columns!"); exit(2);}
*map_ptr++ = 0;
H1Str[i] = map_ptr;
while (*map_ptr != '\t' && *map_ptr != '\n') ++map_ptr;
if (*map_ptr == '\n') {*map_ptr++ = 0; ncol = 1; /* printf("1col line %lu\n",i); */ continue;}
*map_ptr++ = 0;
H2Str[i] = map_ptr;
++numTimesH2showed;
while (*map_ptr != '\n') ++map_ptr;
*map_ptr++ = 0;
}
wholeMap[sz-1] = 0; // in case it's a newline
printf("Detected %d columns (a second showed up %lu times). Parsed. [%f]\n",ncol,numTimesH2showed,omp_get_wtime()-wtime);
wtime = omp_get_wtime();
uint64_t nuniq_ref = 0, nuniq_h1 = 0, nuniq_h2 = 0;
#pragma omp parallel sections
{
#pragma omp section
{
double wtimeL = omp_get_wtime();
qsort(RefStr,nL,sizeof(*RefStr),strcmp_ptr);
for (uint64_t i = 1; i < nL; ++i) // dedupe!
if (strcmp(RefStr[i],RefStr[i-1]))
RefStr[nuniq_ref++] = RefStr[i-1];
RefStr[nuniq_ref++] = RefStr[nL-1];
printf("Name sort complete [%f]\n",omp_get_wtime()-wtimeL);
}
#pragma omp section
{
double wtimeL = omp_get_wtime();
qsort(H1Str,nL,sizeof(*H1Str),strcmp_ptr);
for (uint64_t i = 1; i < nL; ++i) // dedupe!
if (strcmp(H1Str[i],H1Str[i-1]))
H1Str[nuniq_h1++] = H1Str[i-1];
H1Str[nuniq_h1++] = H1Str[nL-1];
printf("H1 sort complete [%f]\n",omp_get_wtime()-wtimeL);
}
#pragma omp section
{
double wtimeL = omp_get_wtime();
if (ncol > 1) {
qsort(H2Str,nL,sizeof(*H2Str),strcmp_ptr);
for (uint64_t i = 1; i < nL; ++i) // dedupe!
if (strcmp(H2Str[i],H2Str[i-1]))
H2Str[nuniq_h2++] = H2Str[i-1];
H2Str[nuniq_h2++] = H2Str[nL-1];
}
printf("H2 sort complete [%f]\n",omp_get_wtime()-wtimeL);
}
}
printf("All sorting complete. [%f]\n",omp_get_wtime()-wtime);
printf("Unique: %lu refs, %lu H1's, %lu H2's.\n",
nuniq_ref, nuniq_h1, nuniq_h2);
wtime = omp_get_wtime();
// Go thru each ref and match it to the list.
// Then lookup the h1 and h2 strings by proxy (past null).
// Get the ids of the h1 and h2's from their respective lists.
// Then add the pair [h1,h2] to the internal-ref-length struct.
//hpair_t *HPairs = malloc(sizeof(*HPairs)*ns);
uint32_t *HPairs[2] = { huge_calloc((uint64_t)ns*sizeof(*HPairs[0])),
huge_calloc((uint64_t)ns*sizeof(*HPairs[1])) };
#pragma omp parallel for schedule(dynamic,16)
for (uint32_t i = 0; i < ns; ++i) {
char *ref = Raw+Offsets[i];
while (*ref != '>') --ref;
uint64_t refmatch = binsearch_str_d(RefStr, ++ref, nuniq_ref);
if (refmatch == (uint64_t)-1) {
fprintf(stderr,"ERR: Map missing '%.*s'\n",
(int)(Raw+Offsets[i] - ref - 1), ref); exit(2);
}
char *h = RefStr[refmatch];
uint64_t h1match = 0, h2match = 0;
while (*h) ++h;
h1match = binsearch_str(H1Str,++h,nuniq_h1);
if (h1match == (uint64_t)-1) {puts("INTERNAL ERROR H1"); exit(9);}
if (ncol > 1) {
while (*h) ++h;
h2match = binsearch_str(H2Str,++h,nuniq_h2);
if (h2match == (uint64_t)-1) {puts("INTERNAL ERROR H2"); exit(9);}
}
//HPairs[i] = (hpair_t){h1match,h2match};
HPairs[0][i] = h1match, HPairs[1][i] = h2match;
}
// Should we make the taxonomy sublevel-map now or later?
// (In theory it can be made during read-in but...)
// Update: now made during read-in!
// Need to dump:
// 1. String data (for both headers only; tracking length)
// 2. The hpair_t dump -- update: now the two separate H arrays
/*
11. Number of Ix1 in map [4] [0 means skip rest of file]
12. String size for Ix1 [8]
13. Ix1 strings dump [#12]
14. Number of Ix2 in map [4] [can be 0/skipped if no h2 map]
15. String size of Ix2 [8]
16. Ix2 strings dump [#15]
//17. hpair_t dump [num ref by 8]
17. HPairs[0] dump [num ref by 4]
18. HPairs[1] dump [num ref by 4]
*/
stringSz = 0;
#pragma omp parallel for reduction(+:stringSz)
for (uint64_t i = 0; i < nuniq_h1; ++i)
stringSz += strlen(H1Str[i]) + 1;
fileSz += 4 + 8 + stringSz;
fwrite(&nuniq_h1,4,1,db); // # 11
fwrite(&stringSz,sizeof(stringSz),1,db); // # 12
for (uint64_t i = 0; i < nuniq_h1; ++i)
fwrite(H1Str[i],1,strlen(H1Str[i])+1,db); // #13
fileSz += 4; // H2
fwrite(&nuniq_h2,4,1,db); // #14
stringSz = 0;
if (ncol > 1) {
#pragma omp parallel for reduction(+:stringSz)
for (uint64_t i = 0; i < nuniq_h2; ++i)
stringSz += strlen(H2Str[i]) + 1;
fwrite(&stringSz,sizeof(stringSz),1,db); // #15
for (uint64_t i = 0; i < nuniq_h2; ++i)
fwrite(H2Str[i],1,strlen(H2Str[i])+1,db); // #16
} else fwrite(&nuniq_h2,8,1,db); // 15 & 16 w/no col2
fileSz += 8 + stringSz;
fileSz += (uint64_t)ns*sizeof(*HPairs[0]);
fwrite(HPairs[0],sizeof(*HPairs[0]),ns,db); // #17
if (nuniq_h2) fileSz += (uint64_t)ns*sizeof(*HPairs[1]),
fwrite(HPairs[1],sizeof(*HPairs[1]),ns,db); // #18
printf("Final filesize: %lu\n",fileSz);
exit(0);
}
/// Now handle the parsing and searching.
// TODO: make parser size-aware, not "num sequences" fixed
if (ixPath) puts("WARNING: map file only applicable during DB BUILD");
// Read the database in.
FILE *db = fopen(dbPath,"rb");
if (!db) {puts("ERROR: bad input"); exit(2);}
struct stat sb; int fno = fileno(db); fstat(fno,&sb);
uint64_t fsz = sb.st_size, place = 0;
char *Raw = mmap(0, fsz+16, PROT_READ, MAP_SHARED |
MAP_POPULATE, fno, 0);
madvise(Raw,fsz,MADV_WILLNEED);
double wtime = omp_get_wtime();
if (doCopyMem) {
puts("Copying database into local memory...");
char *copied = huge_malloc(fsz+16);
if (!copied) {puts("Can't do copymem on this system. Exiting..."); exit(3);}
memcpy(copied,Raw,fsz+1);
munmap(Raw,fsz+16);
Raw = copied;
printf("Copied into local memory [%f]\n",omp_get_wtime()-wtime);
wtime = omp_get_wtime();
}
uint32_t ver = Raw[0] >> 4, rixSz = (uint8_t)Raw[0] & 15,
PL = Raw[1], SL = Raw[2], ktSz = Raw[3];
uint32_t numRef = *(uint32_t *)(Raw+4);
uint64_t numK = *(uint64_t *)(Raw+8);
printf("DBv: %d, rixSz = %d, PL = %d, SL = %d, ktSz = %d\n",
ver, rixSz, PL, SL, ktSz);
printf("Number of refs = %u, kmers = %lu\n",numRef, numK);
if (sizeof(kmer_t)!=ktSz || sizeof(rix_t) != rixSz)
{puts("ERROR: wrong K or R size(s) for this DB"); exit(2);}
place = 16; //#1-6
// Initialize stats
uint64_t K = PL+SL;
uint32_t kpre_shf = PL*2-2;
kmer_t kpst_shf = SL*2-2;
uint32_t pre_bshf = 32-(PL*2), pre_bshf_2 = pre_bshf+2;
kmer_t pst_bshf = sizeof(kmer_t)*8-(SL*2), pst_bshf_2 = pst_bshf+2;
// Read the prefix array.
uint64_t *Nibs = (uint64_t *)(Raw+place);
uint64_t nbins = (uint64_t)1 << (2*PL);
place += (nbins+1) * sizeof(*Nibs); //#7
//printf("DEBUG: nbins: %lu, Nibs[nbins] = %lu\n",nbins,Nibs[nbins]);
KPod *KGrid = (KPod *)(Raw + place);
printf("Size of kpod = %ld\n",sizeof(*KGrid));
place += numK*sizeof(*KGrid); //#8
// Read the ref names
uint64_t stringSz = *(uint64_t *)(Raw + place);
place += sizeof(stringSz); //#9
char *RefRaw = Raw + place;
place += stringSz; //#10
char **RefNames = malloc(sizeof(*RefNames)*numRef); // defer
//printf("String size = %lu\n",stringSz);
//printf("String 1 = %s\n",RefRaw);
// Read the h1 and h2 lists
uint32_t nuniq_h1 = 0, nuniq_h2 = 0;
char *H1Raw = 0, *H2Raw = 0, **HStr[2] = {0,0}; // **H1Str = 0, **H2Str = 0;
//hpair_t *HPairs = 0;
uint32_t *HPairs[2] = { 0,0 };
nuniq_h1 = *(uint32_t *)(Raw+place);
//printf("nuniq_h1 = %u\n",nuniq_h1);
place += sizeof(nuniq_h1); //#11
if (nuniq_h1) {
HStr[0] = malloc((uint64_t)nuniq_h1*sizeof(*HStr[0]));
stringSz = *(uint64_t *)(Raw + place);
place += sizeof(stringSz); //#12
H1Raw = Raw + place;
place += stringSz; //#13
nuniq_h2 = *(uint32_t *)(Raw+place);
place += sizeof(nuniq_h2); //#14
if (nuniq_h2) HStr[1] = malloc((uint64_t)nuniq_h2*sizeof(*HStr[1]));
stringSz = *(uint64_t *)(Raw + place);
place += sizeof(stringSz); //#15
H2Raw = Raw + place;
place += stringSz; //#16
HPairs[0] = (uint32_t *)(Raw + place);
place += (uint64_t)numRef * sizeof(*HPairs[0]); // #17
if (nuniq_h2) HPairs[1] = (uint32_t *)(Raw + place);
place += !nuniq_h2? 0 : (uint64_t)numRef * sizeof(*HPairs[1]); // #18
}
uint32_t NUniqH[2] = {nuniq_h1, nuniq_h2};
printf("Read file of size = %lu (h1: %u, h2: %u)\n",place,nuniq_h1,nuniq_h2);
// Make data structures to contain bin mappings
// (first think how you want to count and create the interpolated
// bins to be accessed -- ideally you'd specify an interpolation
// and get the unique (interpolated) index back.
uint32_t **LBins[2] = {calloc(4096,sizeof(*LBins[0])), calloc(4096,sizeof(*LBins[1]))};
//uint32_t **Tbins_h1 = calloc(4096,sizeof(*Tbins_h1));
//uint32_t **Tbins_h2 = calloc(4096,sizeof(*Tbins_h2));
// Parse the strings in parallel
if (!nuniq_h1)
printf("WARNING: No taxonomy was included during DB formation\n");
//FILE *debug = fopen("debug.txt","wb"); // DEBUG ONLY
#pragma omp parallel sections
{
#pragma omp section
{
for (uint32_t i = 0; i < numRef; ++i) // Ref names
RefNames[i] = RefRaw, RefRaw = strchr(RefRaw,0)+1;
/* FILE *log = fopen("Dbg.txt","wb");
for (uint32_t i = 0; i < numRef; ++i)
fprintf(log,"%u\t%s\n",i,RefNames[i]);
fflush(log); fclose(log); */
//exit(10101);
}
#pragma omp section
{
//char **H1Str = HStr[0]; //uint32_t **Tbins_h1 = LBins[0];
for (uint32_t i = 0; i < nuniq_h1; ++i) // H1 names
HStr[0][i] = H1Raw, H1Raw = strchr(H1Raw,0)+1;
for (uint32_t i = 0; i < nuniq_h1; ++i) {
char *ref = HStr[0][i], *ptr = ref-1;
int lv = 0;
while (ptr = strchr(ptr+1,';')) {
//fprintf(debug,"Searching: %s until %lu...\n --> %.*s\n",ref,ptr-ref,(int)(ptr-ref),ref);
int64_t find = binsearch_str_L(HStr[0],ref,nuniq_h1,ptr-ref);
//fprintf(debug,"%.*s\t%ld\n",(int)(ptr-ref),ref,find);
if (!LBins[0][lv]) {
LBins[0][lv] = calloc(nuniq_h1,sizeof(*LBins[0][lv]));
for (uint32_t j = 0; j < nuniq_h1; ++j) LBins[0][lv][j]=-1;
}
LBins[0][lv++][i] = find;
}
}
}
#pragma omp section
{
for (uint32_t i = 0; i < nuniq_h2; ++i) // H2 names
HStr[1][i] = H2Raw, H2Raw = strchr(H2Raw,0)+1;
for (uint32_t i = 0; i < nuniq_h2; ++i) {
char *ref = HStr[1][i], *ptr = ref-1;
int lv = 0;
while (ptr = strchr(ptr+1,';')) {
int64_t find = binsearch_str_L(HStr[1],ref,nuniq_h2,ptr-ref);
if (!LBins[1][lv]) {
LBins[1][lv] = calloc(nuniq_h2,sizeof(*LBins[1][lv]));
for (uint32_t j = 0; j < nuniq_h2; ++j) LBins[1][lv][j]=-1;
}
LBins[1][lv++][i] = find;
}
}
}
}
// debug: print everything out.
/* FILE *debug = fopen("debug.txt","wb");
for (uint32_t i = 0; i < numRef; ++i)
fprintf(debug,"%s\t%s\t%s\n",RefNames[i],H1Str[HPairs[i].h1],H2Str[HPairs[i].h2]);
fprintf(debug,"AND NOW THE H1 ACTION BOYS:\n");
for (uint32_t i = 0; i < nuniq_h1; ++i)
fprintf(debug,"%s\t%u\t%u\n",H1Str[i],Tbins_h1[i+1],Tbins_h1[i+1]-Tbins_h1[i]);
fprintf(debug,"AND NOW THE H2 ACTION BOYS:\n");
if (nuniq_h2) for (uint32_t i = 0; i < nuniq_h2; ++i)
fprintf(debug,"%s\t%u\t%u\n",H2Str[i],Tbins_h2[i+1],Tbins_h2[i+1]-Tbins_h2[i]);
// Find all the combos in each grid!!
fprintf(debug,"AND NOW THE SEARCH IS ON!\n");
for (uint32_t i = 0; i < nuniq_h1; ++i) {
char *ref = H1Str[i], *ptr = ref;
while (ptr = strchr(ptr,';')) {
//printf("Searching: %s until %lu...\n",ref,ptr-ref-1);
int64_t find = binsearch_str_L(H1Str,ref,nuniq_h1,ptr-ref-1);
int nsemi = 0; for (char *p = ptr; *p; ++p) nsemi += *p==';';
fprintf(debug,"%.*s\t%ld\t%ld\n",(int)(ptr-ref),ref,find,Tbins_h1[find+1]-nsemi);
//printf("%.*s\t%ld\n",(int)(ptr-ref-1),ref,find);
++ptr;
}
//int64_t find = binsearch_str(H1Str,ref,nuniq_h1);
int64_t find = binsearch_str_L(H1Str,ref,nuniq_h1,9999);
int64_t find2 = binsearch_str(H1Str,ref,nuniq_h1);
if (find2 != find) {puts("ERROR FIND"); exit(6);}
fprintf(debug,"%s\t%ld\t%ld\n",ref,find,Tbins_h1[find+1]);
}
exit(1); */
#ifdef WSL
uint16_t wsum = 0;
for (uint64_t i = 0; i < numK; i+=4096/sizeof(*KGrid))
wsum += KGrid[i];
printf("4k Checksum: %d\n",wsum);
#endif
uint32_t *QueryAligns = calloc(numK,sizeof(*QueryAligns)),
*FullQueryAligns = calloc((uint64_t)numRef,sizeof(*FullQueryAligns));
// Open those queries up and align 'em
KPod *EndPod = KGrid + Nibs[nbins] + 1;
uint64_t n_raw = 0, n_filt = 0, n_matchedF = 0, n_matchedR = 0;
uint32_t mask32 = ((uint64_t)1 << (2*PL)) - 1;
kmer_t maskK = ((kmer_t)1 << (2*SL)) - 1;
if (SL==sizeof(kmer_t)*4) maskK = -1;
uint32_t preposit = PL*2-2, pstposit = SL*2-2;
wtime = omp_get_wtime();
printf("\n* Beginning alignment...\n");
gzFile in;
if (!strcmp(seqPath,"-")) in = gzdopen(fileno(stdin),"rb");
else in = gzopen(seqPath,"rb");
if (!in) {puts("ERROR: bad input fast[a/q][.gz]"); exit(2);}
uint8_t *lineO = calloc(16 + (1 << 28),1), *line = lineO+16,
*head = malloc(1 << 20);
uint32_t qChunk = 1<<16;
uint8_t **QBucket = malloc(sizeof(*QBucket)*qChunk);
char **HBucket = malloc(sizeof(*HBucket)*qChunk);
*QBucket = calloc((uint64_t)131072*qChunk,sizeof(**QBucket));
*HBucket = calloc((uint64_t)131072*qChunk,sizeof(**HBucket));
// Make a bucket for temporary LCA & ref votes, per query
uint64_t maxQsz = 1 << 27; //27; // arbitrary? Set a limit?
uint64_t masterBnSz = (uint64_t)1 << 32;
uint64_t masterLstSz = (uint64_t)1 << 31;
typedef struct {uint32_t p; uint64_t s;} SBin_t;
SBin_t **SBins = malloc(sizeof(*SBins)*threads); // For ixing matches
int32_t **RBins = malloc(sizeof(*RBins)*threads), // For tally stores
**TBins = malloc(sizeof(*TBins)*threads);
#pragma omp parallel
{
int tid = omp_get_thread_num();
//printf("I'm thread %d / %d\n",tid, threads);
int counter = 0;
counter += !!(SBins[tid] = huge_malloc(maxQsz*sizeof(*SBins)));
counter += !!(TBins[tid] = huge_malloc(maxQsz*sizeof(*TBins)));
counter += !!(RBins[tid] = huge_calloc((uint64_t)numRef*sizeof(*RBins)));
if (counter != 3) printf("Error allocating bins on thread %d (counter = %d)\n",tid,counter);
}
// For the capitalist bins (variable-length bins depending on hits). Thread-local.
uint64_t **C_ixs = huge_malloc(sizeof(*C_ixs)*threads); //[3]; // for rix, h1, h2 cap arrays
uint64_t **C_szs = huge_malloc(sizeof(*C_szs)*threads); //[3]; // the current size of the bins
uint64_t ***C_bins = huge_malloc(sizeof(*C_bins)*threads); //[3]; // the bin arrays themselves
//uint64_t C_MAX = (uint64_t)1 << 21;
uint64_t C_INIT = (uint64_t)1 << 12;
#pragma omp parallel
{
int tid = omp_get_thread_num();
C_ixs[tid] = calloc(3,sizeof(**C_ixs));
C_szs[tid] = calloc(3,sizeof(**C_szs)); // enables thread resizing
C_bins[tid] = calloc(3,sizeof(**C_bins));
for (int i = 0; i < 3; ++i)
C_bins[tid][i] = malloc(C_INIT*sizeof(***C_bins)),
C_szs[tid][i] = C_INIT-1;
}
// For "master" bins -- number of bins equal to number of queries. Single global storage.
//typedef union {uint32_t a[4]; __uint128_t n; uint8_t b[16];} MasterBin_t;
MasterBin_t *MasterBin = calloc(masterBnSz,sizeof(*MasterBin)); // bin IX // 32
if (!MasterBin) {printf("ERROR: failed to allocate master bin!\n"); exit(3);}
// a[0] = rix, a[1] = h1, a[2] = h2, a[3] = ...?
uint8_t **MasterIx = calloc(3,sizeof(*MasterIx)); // which thread's bin
uint64_t **MasterList = calloc(3,sizeof(*MasterList));
for (int i = 0; i < 3; ++i) {
int counter = 0;
counter += !!(MasterList[i] = calloc(masterLstSz,sizeof(*MasterList[i]))); //31
counter += !!(MasterIx[i] = calloc(masterLstSz,sizeof(*MasterIx[i]))); //31
if (counter != 2) printf("Error allocating master list (counter = %d)\n",counter);
}
FILE *outq = 0;
if (perqPath) {
outq = fopen(perqPath,"wb");
if (!outq) {puts("ERROR: can't open per-q output file!"); exit(2);}
}
uint64_t nq, NQ = 0, nAligns = 0;