forked from samtools/samtools
-
Notifications
You must be signed in to change notification settings - Fork 0
/
amplicon_stats.c
1756 lines (1551 loc) · 62.6 KB
/
amplicon_stats.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
/* stats.c -- This is the former bamcheck integrated into samtools/htslib.
Copyright (C) 2020-2021 Genome Research Ltd.
Author: James Bonfield <[email protected]>
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE. */
/*
* This tool is designed to give "samtools stats" style output, but dedicated
* to small amplicon sequencing projects. It gathers stats on the
* distribution of reads across amplicons.
*/
/*
* TODO:
* - Cope with multiple references. What do we do here? Just request one?
* - Permit regions rather than consuming whole file (maybe solves above).
*/
#include <config.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <inttypes.h>
#include <getopt.h>
#include <unistd.h>
#include <math.h>
#include <htslib/sam.h>
#include <htslib/khash.h>
#include "samtools.h"
#include "sam_opts.h"
#include "bam_ampliconclip.h"
KHASH_MAP_INIT_INT64(tcoord, int64_t)
KHASH_MAP_INIT_STR(qname, int64_t)
#ifndef MIN
#define MIN(a,b) ((a)<(b)?(a):(b))
#endif
#ifndef MAX
#define MAX(a,b) ((a)>(b)?(a):(b))
#endif
#ifndef ABS
#define ABS(a) ((a)>=0?(a):-(a))
#endif
#define TCOORD_MIN_COUNT 10
#define MAX_AMP 1000 // Default maximum number of amplicons
#define MAX_AMP_LEN 1000 // Default maximum length of any single amplicon
#define MAX_PRIMER_PER_AMPLICON 4 // Max primers per LEFT/RIGHT
#define MAX_DEPTH 5 // Number of different depths permitted
typedef struct {
sam_global_args ga;
uint32_t flag_require;
uint32_t flag_filter;
int max_delta; // Used for matching read to amplicon primer loc
int min_depth[MAX_DEPTH]; // Used for coverage; must be >= min_depth deep
int use_sample_name;
int max_amp; // Total number of amplicons
int max_amp_len; // Maximum length of an individual amplicon
double depth_bin;// aggregate depth within this fraction
int tlen_adj; // Adjust tlen by this amount, due to clip but no fixmate
FILE *out_fp;
char *argv;
int tcoord_min_count;
int tcoord_bin;
int multi_ref;
} astats_args_t;
typedef struct {
int nseq; // total sequence count
int nfiltered; // sequence filtered
int nfailprimer;// count of sequences not matching the primer locations
// Sizes of memory allocated below, to permit reset
int max_amp, max_amp_len, max_len;
// Summary across all samples, sum(x) plus sum(x^2) for s.d. calc
int64_t *nreads, *nreads2; // [max_amp]
double *nfull_reads; // [max_amp]; 0.5/read if paired.
double *nrperc, *nrperc2; // [max_amp]
int64_t *nbases, *nbases2; // [max_amp]
int64_t *coverage; // [max_amp][max_amp_len]
double (*covered_perc)[MAX_DEPTH]; // [max_amp][MAX_DEPTH]
double (*covered_perc2)[MAX_DEPTH];// [max_amp][MAX_DEPTH];
khash_t(tcoord) **tcoord; // [max_amp+1]
// 0 is correct pair, 1 is incorrect pair, 2 is unidentified
int (*amp_dist)[3]; // [MAX_AMP][3];
int *depth_valid; // [max_len]
int *depth_all; // [max_len]
khash_t(qname) *qend; // queryname end, for overlap removal
} astats_t;
// We can have multiple primers for LEFT / RIGHT, so this
// permits detection by any compatible combination.
// One reference:
typedef struct {
int64_t left[MAX_PRIMER_PER_AMPLICON];
int nleft;
int64_t right[MAX_PRIMER_PER_AMPLICON];
int nright;
int64_t max_left, min_right; // inner dimensions
int64_t min_left, max_right; // outer dimensions
} amplicon_t;
// Multiple references, we have an array of amplicons_t - one per used ref.
// We have per reference local and global stats here, as some of the stats
// are coordinate based. However we report them combined together as a single
// list across all references.
// "namp" is the number of amplicons in this reference, but they're
// numbered first_amp to first_amp+namp-1 inclusively.
typedef struct {
int tid, namp;
int64_t len;
bed_entry_list_t *sites;
amplicon_t *amp;
astats_t *lstats, *gstats; // local (1 file) and global (all file) stats
const char *ref; // ref name (pointer to the bed hash table key)
int first_amp; // first amplicon number for this ref
} amplicons_t;
// Reinitialised for each new reference/chromosome.
// Counts from 1 to namp, -1 for no match and 0 for ?.
static int *pos2start = NULL;
static int *pos2end = NULL;
static int pos2size = 0; // allocated size of pos2start/end
// Lookup table to go from position to amplicon based on
// read start / end.
static int initialise_amp_pos_lookup(astats_args_t *args,
amplicons_t *amps,
int ref) {
int64_t i, j;
amplicon_t *amp = amps[ref].amp;
int64_t max_len = amps[ref].len;
int namp = amps[ref].namp;
if (max_len+1 > pos2size) {
if (!(pos2start = realloc(pos2start, (max_len+1)*sizeof(*pos2start))))
return -1;
if (!(pos2end = realloc(pos2end, (max_len+1)*sizeof(*pos2end))))
return -1;
pos2size = max_len;
}
for (i = 0; i < max_len; i++)
pos2start[i] = pos2end[i] = -1;
for (i = 0; i < namp; i++) {
for (j = 0; j < amp[i].nleft; j++) {
int64_t p;
for (p = amp[i].left[j] - args->max_delta;
p <= amp[i].left[j] + args->max_delta; p++) {
if (p < 1 || p > max_len)
continue;
pos2start[p-1] = i;
}
}
for (j = 0; j < amp[i].nright; j++) {
int64_t p;
for (p = amp[i].right[j] - args->max_delta;
p <= amp[i].right[j] + args->max_delta; p++) {
if (p < 1 || p > max_len)
continue;
pos2end[p-1] = i;
}
}
}
return 0;
}
// Counts amplicons.
// Assumption: input BED file alternates between LEFT and RIGHT primers
// per amplicon, thus we can count the number based on the switching
// orientation.
static int count_amplicon(bed_entry_list_t *sites) {
int i, namp, last_rev = 0;
for (i = namp = 0; i < sites->length; i++) {
if (sites->bp[i].rev == 0 && last_rev)
namp++;
last_rev = sites->bp[i].rev;
}
return ++namp;
}
// We're only interest in the internal part of the amplicon.
// Our bed file has LEFT start/end followed by RIGHT start/end,
// so collapse these to LEFT end / RIGHT start.
//
// Returns right most amplicon position on success,
// < 0 on error
static int64_t bed2amplicon(astats_args_t *args, bed_entry_list_t *sites,
amplicon_t *amp, int *namp, int do_title,
const char *ref, int first_amp) {
int i, j;
int64_t max_right = 0;
FILE *ofp = args->out_fp;
*namp = 0;
// Assume all primers for the same amplicon are adjacent in BED
// with all + followed by all -. Thus - to + signifies next primer set.
int last_rev = 0;
amp[0].max_left = 0;
amp[0].min_right = INT64_MAX;
amp[0].min_left = INT64_MAX;
amp[0].max_right = 0;
if (do_title) {
fprintf(ofp, "# Amplicon locations from BED file.\n");
fprintf(ofp, "# LEFT/RIGHT are <start>-<end> format and "
"comma-separated for alt-primers.\n");
if (args->multi_ref)
fprintf(ofp, "#\n# AMPLICON\tREF\tNUMBER\tLEFT\tRIGHT\n");
else
fprintf(ofp, "#\n# AMPLICON\tNUMBER\tLEFT\tRIGHT\n");
}
for (i = j = 0; i < sites->length; i++) {
if (i == 0 && sites->bp[i].rev != 0) {
fprintf(stderr, "[ampliconstats] error: BED file should start"
" with the + strand primer\n");
return -1;
}
if (sites->bp[i].rev == 0 && last_rev) {
j++;
if (j >= args->max_amp) {
fprintf(stderr, "[ampliconstats] error: too many amplicons"
" (%d). Use -a option to raise this.\n", j);
return -1;
}
amp[j].max_left = 0;
amp[j].min_right = INT64_MAX;
amp[j].min_left = INT64_MAX;
amp[j].max_right = 0;
}
if (sites->bp[i].rev == 0) {
if (i == 0 || last_rev) {
if (j>0) fprintf(ofp, "\n");
if (args->multi_ref)
fprintf(ofp, "AMPLICON\t%s\t%d", ref, j+1 + first_amp);
else
fprintf(ofp, "AMPLICON\t%d", j+1);
}
if (amp[j].nleft >= MAX_PRIMER_PER_AMPLICON) {
print_error_errno("ampliconstats",
"too many primers per amplicon (%d).\n",
MAX_PRIMER_PER_AMPLICON);
return -1;
}
amp[j].left[amp[j].nleft++] = sites->bp[i].right;
if (amp[j].max_left < sites->bp[i].right+1)
amp[j].max_left = sites->bp[i].right+1;
if (amp[j].min_left > sites->bp[i].right+1)
amp[j].min_left = sites->bp[i].right+1;
// BED file, so left+1 as zero based. right(+1-1) as
// BED goes one beyond end (and we want inclusive range).
fprintf(ofp, "%c%"PRId64"-%"PRId64, "\t,"[amp[j].nleft > 1],
sites->bp[i].left+1, sites->bp[i].right);
} else {
if (amp[j].nright >= MAX_PRIMER_PER_AMPLICON) {
print_error_errno("ampliconstats",
"too many primers per amplicon (%d)",
MAX_PRIMER_PER_AMPLICON);
return -1;
}
amp[j].right[amp[j].nright++] = sites->bp[i].left;
if (amp[j].min_right > sites->bp[i].left-1)
amp[j].min_right = sites->bp[i].left-1;
if (amp[j].max_right < sites->bp[i].left-1) {
amp[j].max_right = sites->bp[i].left-1;
if (amp[j].max_right - amp[j].min_left + 1 >=
args->max_amp_len) {
fprintf(stderr, "[ampliconstats] error: amplicon "
"longer (%d) than max_amp_len option (%d)\n",
(int)(amp[j].max_right - amp[j].min_left + 2),
args->max_amp_len);
return -1;
}
if (max_right < amp[j].max_right)
max_right = amp[j].max_right;
}
fprintf(ofp, "%c%"PRId64"-%"PRId64, "\t,"[amp[j].nright > 1],
sites->bp[i].left+1, sites->bp[i].right);
}
last_rev = sites->bp[i].rev;
}
if (last_rev != 1) {
fprintf(ofp, "\n"); // useful if going to stdout
fprintf(stderr, "[ampliconstats] error: bed file does not end on"
" a reverse strand primer.\n");
return -1;
}
*namp = ++j;
if (j) fprintf(ofp, "\n");
if (j >= args->max_amp) {
fprintf(stderr, "[ampliconstats] error: "
"too many amplicons (%d). Use -a option to raise this.", j);
return -1;
}
// for (i = 0; i < *namp; i++) {
// printf("%d\t%ld", i, amp[i].length);
// for (j = 0; j < amp[i].nleft; j++)
// printf("%c%ld", "\t,"[j>0], amp[i].left[j]);
// for (j = 0; j < amp[i].nright; j++)
// printf("%c%ld", "\t,"[j>0], amp[i].right[j]);
// printf("\n");
// }
return max_right;
}
void stats_free(astats_t *st) {
if (!st)
return;
free(st->nreads);
free(st->nreads2);
free(st->nfull_reads);
free(st->nrperc);
free(st->nrperc2);
free(st->nbases);
free(st->nbases2);
free(st->coverage);
free(st->covered_perc);
free(st->covered_perc2);
free(st->amp_dist);
free(st->depth_valid);
free(st->depth_all);
if (st->tcoord) {
int i;
for (i = 0; i <= st->max_amp; i++) {
if (st->tcoord[i])
kh_destroy(tcoord, st->tcoord[i]);
}
free(st->tcoord);
}
khiter_t k;
for (k = kh_begin(st->qend); k != kh_end(st->qend); k++)
if (kh_exist(st->qend, k))
free((void *)kh_key(st->qend, k));
kh_destroy(qname, st->qend);
free(st);
}
astats_t *stats_alloc(int64_t max_len, int max_amp, int max_amp_len) {
astats_t *st = calloc(1, sizeof(*st));
if (!st)
return NULL;
st->max_amp = max_amp;
st->max_amp_len = max_amp_len;
st->max_len = max_len;
if (!(st->nreads = calloc(max_amp, sizeof(*st->nreads)))) goto err;
if (!(st->nreads2 = calloc(max_amp, sizeof(*st->nreads2)))) goto err;
if (!(st->nrperc = calloc(max_amp, sizeof(*st->nrperc)))) goto err;
if (!(st->nrperc2 = calloc(max_amp, sizeof(*st->nrperc2)))) goto err;
if (!(st->nbases = calloc(max_amp, sizeof(*st->nbases)))) goto err;
if (!(st->nbases2 = calloc(max_amp, sizeof(*st->nbases2)))) goto err;
if (!(st->nfull_reads = calloc(max_amp, sizeof(*st->nfull_reads))))
goto err;
if (!(st->coverage = calloc(max_amp*max_amp_len, sizeof(*st->coverage))))
goto err;
if (!(st->covered_perc = calloc(max_amp, sizeof(*st->covered_perc))))
goto err;
if (!(st->covered_perc2 = calloc(max_amp, sizeof(*st->covered_perc2))))
goto err;
if (!(st->tcoord = calloc(max_amp+1, sizeof(*st->tcoord)))) goto err;
int i;
for (i = 0; i <= st->max_amp; i++)
if (!(st->tcoord[i] = kh_init(tcoord)))
goto err;
if (!(st->qend = kh_init(qname)))
goto err;
if (!(st->depth_valid = calloc(max_len, sizeof(*st->depth_valid))))
goto err;
if (!(st->depth_all = calloc(max_len, sizeof(*st->depth_all))))
goto err;
if (!(st->amp_dist = calloc(max_amp, sizeof(*st->amp_dist)))) goto err;
return st;
err:
stats_free(st);
return NULL;
}
static void stats_reset(astats_t *st) {
st->nseq = 0;
st->nfiltered = 0;
st->nfailprimer = 0;
memset(st->nreads, 0, st->max_amp * sizeof(*st->nreads));
memset(st->nreads2, 0, st->max_amp * sizeof(*st->nreads2));
memset(st->nfull_reads, 0, st->max_amp * sizeof(*st->nfull_reads));
memset(st->nrperc, 0, st->max_amp * sizeof(*st->nrperc));
memset(st->nrperc2, 0, st->max_amp * sizeof(*st->nrperc2));
memset(st->nbases, 0, st->max_amp * sizeof(*st->nbases));
memset(st->nbases2, 0, st->max_amp * sizeof(*st->nbases2));
memset(st->coverage, 0, st->max_amp * st->max_amp_len
* sizeof(*st->coverage));
memset(st->covered_perc, 0, st->max_amp * sizeof(*st->covered_perc));
memset(st->covered_perc2, 0, st->max_amp * sizeof(*st->covered_perc2));
// Keep the allocated entries as it's likely all files will share
// the same keys. Instead we reset counters to zero for common ones
// and delete rare ones.
int i;
for (i = 0; i <= st->max_amp; i++) {
khiter_t k;
for (k = kh_begin(st->tcoord[i]);
k != kh_end(st->tcoord[i]); k++)
if (kh_exist(st->tcoord[i], k)) {
if (kh_value(st->tcoord[i], k) < 5)
kh_del(tcoord, st->tcoord[i], k);
else
kh_value(st->tcoord[i], k) = 0;
}
}
khiter_t k;
for (k = kh_begin(st->qend); k != kh_end(st->qend); k++)
if (kh_exist(st->qend, k))
free((void *)kh_key(st->qend, k));
kh_clear(qname, st->qend);
memset(st->depth_valid, 0, st->max_len * sizeof(*st->depth_valid));
memset(st->depth_all, 0, st->max_len * sizeof(*st->depth_all));
memset(st->amp_dist, 0, st->max_amp * sizeof(*st->amp_dist));
}
static void amp_stats_reset(amplicons_t *amps, int nref) {
int i;
for (i = 0; i < nref; i++) {
if (!amps[i].sites)
continue;
stats_reset(amps[i].lstats);
}
}
static int accumulate_stats(astats_args_t *args, amplicons_t *amps,
bam1_t *b) {
int ref = b->core.tid;
amplicon_t *amp = amps[ref].amp;
astats_t *stats = amps[ref].lstats;
int len = amps[ref].len;
if (!stats)
return 0;
stats->nseq++;
if ((b->core.flag & args->flag_require) != args->flag_require ||
(b->core.flag & args->flag_filter) != 0) {
stats->nfiltered++;
return 0;
}
int64_t start = b->core.pos, mstart = start; // modified start
int64_t end = bam_endpos(b), i;
// Compute all-template-depth and valid-template-depth.
// We track current end location per read name so we can remove overlaps.
// Potentially we could use this data for a better amplicon-depth
// count too, but for now it's purely for the per-base plots.
int ret;
khiter_t k;
int prev_start = 0, prev_end = 0;
if ((b->core.flag & BAM_FPAIRED)
&& !(b->core.flag & (BAM_FSUPPLEMENTARY | BAM_FSECONDARY))) {
k = kh_put(qname, stats->qend, bam_get_qname(b), &ret);
if (ret == 0) {
prev_start = kh_value(stats->qend, k) & 0xffffffff;
prev_end = kh_value(stats->qend, k)>>32;
mstart = MAX(mstart, prev_end);
// Ideally we'd reuse strings so we don't thrash free/malloc.
// However let's see if the official way of doing that (malloc
// itself) is fast enough first.
free((void *)kh_key(stats->qend, k));
kh_del(qname, stats->qend, k);
//fprintf(stderr, "remove overlap %d to %d\n", (int)start, (int)mstart);
} else {
if (!(kh_key(stats->qend, k) = strdup(bam_get_qname(b))))
return -1;
kh_value(stats->qend, k) = start | (end << 32);
}
}
for (i = mstart; i < end && i < len; i++)
stats->depth_all[i]++;
if (i < end) {
print_error("ampliconstats", "record %s overhangs end of reference",
bam_get_qname(b));
// But keep going, as it's harmless.
}
// On single ended runs, eg ONT or PacBio, we just use the start/end
// of the template to assign.
int anum = (b->core.flag & BAM_FREVERSE) || !(b->core.flag & BAM_FPAIRED)
? (end-1 >= 0 && end-1 < len ? pos2end[end-1] : -1)
: (start >= 0 && start < len ? pos2start[start] : -1);
// ivar sometimes soft-clips 100% of the bases.
// This is essentially unmapped
if (end == start && (args->flag_filter & BAM_FUNMAP)) {
stats->nfiltered++;
return 0;
}
if (anum == -1)
stats->nfailprimer++;
if (anum >= 0) {
int64_t c = MIN(end,amp[anum].min_right+1) - MAX(start,amp[anum].max_left);
if (c > 0) {
stats->nreads[anum]++;
// NB: ref bases rather than read bases
stats->nbases[anum] += c;
int64_t i;
if (start < 0) start = 0;
if (end > len) end = len;
int64_t ostart = MAX(start, amp[anum].min_left-1);
int64_t oend = MIN(end, amp[anum].max_right);
int64_t offset = amp[anum].min_left-1;
for (i = ostart; i < oend; i++)
stats->coverage[anum*stats->max_amp_len + i-offset]++;
} else {
stats->nfailprimer++;
}
}
// Template length in terms of amplicon number to amplicon number.
// We expect left to right of same amplicon (len 0), but it may go
// to next amplicon (len 1) or prev (len -1), etc.
int64_t t_end;
int oth_anum = -1;
if (b->core.flag & BAM_FPAIRED) {
t_end = (b->core.flag & BAM_FREVERSE ? end : start)
+ b->core.isize;
// If we've clipped the primers but not followed up with a fixmates
// then our start+TLEN will take us to a location which is
// length(LEFT_PRIMER) + length(RIGHT_PRIMER) too far away.
//
// The correct solution is to run samtools fixmate so TLEN is correct.
// The hacky solution is to fudge the expected tlen by double the
// average primer length (e.g. 50).
t_end += b->core.isize > 0 ? -args->tlen_adj : +args->tlen_adj;
if (t_end > 0 && t_end < len && b->core.isize != 0)
oth_anum = (b->core.flag & BAM_FREVERSE)
? pos2start[t_end]
: pos2end[t_end];
} else {
// Not paired (see int anum = (REV || !PAIR) ?en :st expr above)
oth_anum = pos2start[start];
t_end = end;
}
// We don't want to count our pairs twice.
// If both left/right are known, count it on left only.
// If only one is known, we'll only get to this code once
// so we can also count it.
int astatus = 2;
if (anum != -1 && oth_anum != -1) {
astatus = oth_anum == anum ? 0 : 1;
if (start <= t_end)
stats->amp_dist[anum][astatus]++;
} else if (anum >= 0) {
stats->amp_dist[anum][astatus = 2]++;
}
if (astatus == 0 && !(b->core.flag & (BAM_FUNMAP | BAM_FMUNMAP))) {
if (prev_end && mstart > prev_end) {
// 2nd read with gap to 1st; undo previous increment.
for (i = prev_start; i < prev_end; i++)
stats->depth_valid[i]--;
stats->nfull_reads[anum] -= (b->core.flag & BAM_FPAIRED) ? 0.5 : 1;
} else {
// 1st read, or 2nd read that overlaps 1st
for (i = mstart; i < end; i++)
stats->depth_valid[i]++;
stats->nfull_reads[anum] += (b->core.flag & BAM_FPAIRED) ? 0.5 : 1;
}
}
// Track template start,end frequencies, so we can give stats on
// amplicon primer usage.
if ((b->core.flag & BAM_FPAIRED) && b->core.isize <= 0)
// left to right only, so we don't double count template positions.
return 0;
start = b->core.pos;
t_end = b->core.flag & BAM_FPAIRED
? start + b->core.isize-1
: end;
uint64_t tcoord = MIN(start+1, UINT32_MAX) | (MIN(t_end+1, UINT32_MAX)<<32);
k = kh_put(tcoord, stats->tcoord[anum+1], tcoord, &ret);
if (ret < 0)
return -1;
if (ret == 0)
kh_value(stats->tcoord[anum+1], k)++;
else
kh_value(stats->tcoord[anum+1], k)=1;
kh_value(stats->tcoord[anum+1], k) |= ((int64_t)astatus<<32);
return 0;
}
// Append file local stats to global stats
int append_lstats(astats_t *lstats, astats_t *gstats, int namp, int all_nseq) {
gstats->nseq += lstats->nseq;
gstats->nfiltered += lstats->nfiltered;
gstats->nfailprimer += lstats->nfailprimer;
int a;
for (a = -1; a < namp; a++) {
// Add khash local (kl) to khash global (kg)
khiter_t kl, kg;
for (kl = kh_begin(lstats->tcoord[a+1]);
kl != kh_end(lstats->tcoord[a+1]); kl++) {
if (!kh_exist(lstats->tcoord[a+1], kl) ||
kh_value(lstats->tcoord[a+1], kl) == 0)
continue;
int ret;
kg = kh_put(tcoord, gstats->tcoord[a+1],
kh_key(lstats->tcoord[a+1], kl),
&ret);
if (ret < 0)
return -1;
kh_value(gstats->tcoord[a+1], kg) =
(ret == 0
? (kh_value(gstats->tcoord[a+1], kg) & 0xFFFFFFFF)
: 0)
+ kh_value(lstats->tcoord[a+1], kl);
}
if (a == -1) continue;
gstats->nreads[a] += lstats->nreads[a];
gstats->nreads2[a] += lstats->nreads[a] * lstats->nreads[a];
gstats->nfull_reads[a] += lstats->nfull_reads[a];
// To get mean & sd for amplicon read percentage, we need
// to do the divisions here as nseq differs for each sample.
double nrperc = all_nseq ? 100.0 * lstats->nreads[a] / all_nseq : 0;
gstats->nrperc[a] += nrperc;
gstats->nrperc2[a] += nrperc*nrperc;
gstats->nbases[a] += lstats->nbases[a];
gstats->nbases2[a] += lstats->nbases[a] * lstats->nbases[a];
int d;
for (d = 0; d < MAX_DEPTH; d++) {
gstats->covered_perc[a][d] += lstats->covered_perc[a][d];
gstats->covered_perc2[a][d] += lstats->covered_perc[a][d]
* lstats->covered_perc[a][d];
}
for (d = 0; d < 3; d++)
gstats->amp_dist[a][d] += lstats->amp_dist[a][d];
}
for (a = 0; a < lstats->max_len; a++) {
gstats->depth_valid[a] += lstats->depth_valid[a];
gstats->depth_all[a] += lstats->depth_all[a];
}
return 0;
}
int append_stats(amplicons_t *amps, int nref) {
int i, r, all_nseq = 0;
for (r = 0; r < nref; r++) {
if (!amps[r].sites)
continue;
astats_t *stats = amps[r].lstats;
all_nseq += stats->nseq - stats->nfiltered - stats->nfailprimer;
}
for (i = 0; i < nref; i++) {
if (!amps[i].sites)
continue;
if (append_lstats(amps[i].lstats, amps[i].gstats, amps[i].namp,
all_nseq) < 0)
return -1;
}
return 0;
}
typedef struct {
int32_t start, end;
uint32_t freq;
uint32_t status;
} tcoord_t;
// Sort tcoord by descending frequency and then ascending start and end.
static int tcoord_freq_sort(const void *vp1, const void *vp2) {
const tcoord_t *t1 = (const tcoord_t *)vp1;
const tcoord_t *t2 = (const tcoord_t *)vp2;
if (t1->freq != t2->freq)
return t2->freq - t1->freq;
if (t1->start != t2->start)
return t1->start - t2->start;
return t1->end - t2->end;
}
/*
* Merges tcoord start,end,freq,status tuples if their coordinates are
* close together. We aim to keep the start,end for the most frequent
* value and assume that is the correct coordinate and all others are
* minor fluctuations due to errors or variants.
*
* We sort by frequency first and then merge later items in the list into
* the earlier more frequent ones. It's O(N^2), but sufficient for now
* given current scale of projects.
*
* If we ever need to resolve that then consider sorting by start
* coordinate and scanning the list to find all items within X, find
* the most frequent of those, and then cluster that way. (I'd have
* done that had I thought of it at the time!)
*/
static void aggregate_tcoord(astats_args_t *args, tcoord_t *tpos, size_t *np){
size_t n = *np, j, j2, j3, k;
// Sort by frequency and cluster infrequent coords into frequent
// ones provided they're close by.
// This is O(N^2), but we've already binned by tcoord_bin/2 so
// the list isn't intended to be vast at this point.
qsort(tpos, n, sizeof(*tpos), tcoord_freq_sort);
// For frequency ties, find mid start coord, and then find mid end
// coord of those matching start.
// We make that the first item so we merge into that mid point.
for (j = 0; j < n; j++) {
for (j2 = j+1; j2 < n; j2++) {
if (tpos[j].freq != tpos[j2].freq)
break;
if (tpos[j2].start - tpos[j].start >= args->tcoord_bin)
break;
}
// j to j2 all within bin of a common start,
// m is the mid start.
if (j2-1 > j) {
size_t m = (j2-1 + j)/2;
// Find mid end for this same start
while (m > 1 && tpos[m].start == tpos[m-1].start)
m--;
for (j3 = m+1; j3 < j2; j3++) {
if (tpos[m].start != tpos[j3].start)
break;
if (tpos[m].end - tpos[j3].end >= args->tcoord_bin)
break;
}
if (j3-1 > m)
m = (j3-1 + m)/2;
// Swap with first item.
tcoord_t tmp = tpos[j];
tpos[j] = tpos[m];
tpos[m] = tmp;
j = j2-1;
}
}
// Now merge in coordinates.
// This bit is O(N^2), so consider binning first to reduce the
// size of the list if we have excessive positional variation.
for (k = j = 0; j < n; j++) {
if (!tpos[j].freq)
continue;
if (k < j)
tpos[k] = tpos[j];
for (j2 = j+1; j2 < n; j2++) {
if (ABS(tpos[j].start-tpos[j2].start) < args->tcoord_bin/2 &&
ABS(tpos[j].end -tpos[j2].end) < args->tcoord_bin/2 &&
tpos[j].status == tpos[j2].status) {
tpos[k].freq += tpos[j2].freq;
tpos[j2].freq = 0;
}
}
k++;
}
*np = k;
}
int dump_stats(astats_args_t *args, char type, char *name, int nfile,
amplicons_t *amps, int nref, int local) {
int i, r;
FILE *ofp = args->out_fp;
tcoord_t *tpos = NULL;
size_t ntcoord = 0;
// summary stats for this sample (or for all samples)
fprintf(ofp, "# Summary stats.\n");
fprintf(ofp, "# Use 'grep ^%cSS | cut -f 2-' to extract this part.\n", type);
for (r = 0; r < nref; r++) {
if (!amps[r].sites)
continue;
astats_t *stats = local ? amps[r].lstats : amps[r].gstats;
int nmatch = stats->nseq - stats->nfiltered - stats->nfailprimer;
char *name_ref = malloc(strlen(name) + strlen(amps[r].ref) + 2);
if (!name_ref)
return -1;
if (args->multi_ref)
sprintf(name_ref, "%s\t%s", name, amps[r].ref);
else
sprintf(name_ref, "%s", name);
fprintf(ofp, "%cSS\t%s\traw total sequences:\t%d\n",
type, name_ref, stats->nseq);
fprintf(ofp, "%cSS\t%s\tfiltered sequences:\t%d\n",
type, name_ref, stats->nfiltered);
fprintf(ofp, "%cSS\t%s\tfailed primer match:\t%d\n",
type, name_ref, stats->nfailprimer);
fprintf(ofp, "%cSS\t%s\tmatching sequences:\t%d\n",
type, name_ref, nmatch);
int d = 0;
do {
// From first to last amplicon only, so not entire consensus.
// If contig length is known, maybe we want to add the missing
// count to < DEPTH figures?
int64_t start = 0, covered = 0, total = 0;
amplicon_t *amp = amps[r].amp;
for (i = 0; i < amps[r].namp; i++) {
int64_t j, offset = amp[i].min_left-1;
if (amp[i].min_right - amp[i].min_left > stats->max_amp_len) {
fprintf(stderr, "[ampliconstats] error: "
"Maximum amplicon length (%d) exceeded for '%s'\n",
stats->max_amp, name);
return -1;
}
for (j = MAX(start, amp[i].max_left-1);
j < MAX(start, amp[i].min_right); j++) {
if (stats->coverage[i*stats->max_amp_len + j-offset]
>= args->min_depth[d])
covered++;
total++;
}
start = MAX(start, amp[i].min_right);
}
fprintf(ofp, "%cSS\t%s\tconsensus depth count < %d and >= %d:\t%"
PRId64"\t%"PRId64"\n", type, name_ref,
args->min_depth[d], args->min_depth[d],
total-covered, covered);
} while (++d < MAX_DEPTH && args->min_depth[d]);
free(name_ref);
}
// Read count
fprintf(ofp, "# Absolute matching read counts per amplicon.\n");
fprintf(ofp, "# Use 'grep ^%cREADS | cut -f 2-' to extract this part.\n", type);
fprintf(ofp, "%cREADS\t%s", type, name);
for (r = 0; r < nref; r++) {
if (!amps[r].sites)
continue;
astats_t *stats = local ? amps[r].lstats : amps[r].gstats;
for (i = 0; i < amps[r].namp; i++) {
fprintf(ofp, "\t%"PRId64, stats->nreads[i]);
}
}
fprintf(ofp, "\n");
// Valid depth is the number of full length reads (already divided
// by the number we expect to cover), so +0.5 per read in pair.
// A.k.a "usable depth" in the plots.
fprintf(ofp, "%cVDEPTH\t%s", type, name);
for (r = 0; r < nref; r++) {
if (!amps[r].sites)
continue;
astats_t *stats = local ? amps[r].lstats : amps[r].gstats;
for (i = 0; i < amps[r].namp; i++)
fprintf(ofp, "\t%d", (int)stats->nfull_reads[i]);
}
fprintf(ofp, "\n");
if (type == 'C') {
// For combined we can compute mean & standard deviation too
fprintf(ofp, "CREADS\tMEAN");
for (r = 0; r < nref; r++) {
if (!amps[r].sites)
continue;
astats_t *stats = local ? amps[r].lstats : amps[r].gstats;
for (i = 0; i < amps[r].namp; i++) {
fprintf(ofp, "\t%.1f", stats->nreads[i] / (double)nfile);
}
}
fprintf(ofp, "\n");
fprintf(ofp, "CREADS\tSTDDEV");
for (r = 0; r < nref; r++) {
if (!amps[r].sites)
continue;
astats_t *stats = local ? amps[r].lstats : amps[r].gstats;
for (i = 0; i < amps[r].namp; i++) {
double n1 = stats->nreads[i];
fprintf(ofp, "\t%.1f", nfile > 1 && stats->nreads2[i] > 0
? sqrt(stats->nreads2[i]/(double)nfile
- (n1/nfile)*(n1/nfile))
: 0);
}
}
fprintf(ofp, "\n");
}
fprintf(ofp, "# Read percentage of distribution between amplicons.\n");
fprintf(ofp, "# Use 'grep ^%cRPERC | cut -f 2-' to extract this part.\n", type);
fprintf(ofp, "%cRPERC\t%s", type, name);
int all_nseq = 0;
for (r = 0; r < nref; r++) {
if (!amps[r].sites)
continue;
astats_t *stats = local ? amps[r].lstats : amps[r].gstats;
all_nseq += stats->nseq - stats->nfiltered - stats->nfailprimer;
}
for (r = 0; r < nref; r++) {
if (!amps[r].sites)
continue;
astats_t *stats = local ? amps[r].lstats : amps[r].gstats;
for (i = 0; i < amps[r].namp; i++) {
if (type == 'C') {
fprintf(ofp, "\t%.3f", (double)stats->nrperc[i] / nfile);
} else {
fprintf(ofp, "\t%.3f",
all_nseq ? 100.0 * stats->nreads[i] / all_nseq : 0);
}
}
}
fprintf(ofp, "\n");
if (type == 'C') {
// For combined we compute mean and standard deviation too
fprintf(ofp, "CRPERC\tMEAN");
for (r = 0; r < nref; r++) {
if (!amps[r].sites)
continue;
astats_t *stats = local ? amps[r].lstats : amps[r].gstats;
for (i = 0; i < amps[r].namp; i++) {
fprintf(ofp, "\t%.3f", stats->nrperc[i] / nfile);
}
}
fprintf(ofp, "\n");
fprintf(ofp, "CRPERC\tSTDDEV");
for (r = 0; r < nref; r++) {