added maxk: max unique k-mer at each position
This commit is contained in:
parent
1f99921b73
commit
163f949a33
6
Makefile
6
Makefile
|
|
@ -6,7 +6,7 @@ AR= ar
|
||||||
DFLAGS= -DHAVE_PTHREAD $(WRAP_MALLOC)
|
DFLAGS= -DHAVE_PTHREAD $(WRAP_MALLOC)
|
||||||
LOBJS= utils.o kthread.o kstring.o ksw.o bwt.o bntseq.o bwa.o bwamem.o bwamem_pair.o bwamem_extra.o malloc_wrap.o
|
LOBJS= utils.o kthread.o kstring.o ksw.o bwt.o bntseq.o bwa.o bwamem.o bwamem_pair.o bwamem_extra.o malloc_wrap.o
|
||||||
AOBJS= QSufSort.o bwt_gen.o bwashm.o bwase.o bwaseqio.o bwtgap.o bwtaln.o bamlite.o \
|
AOBJS= QSufSort.o bwt_gen.o bwashm.o bwase.o bwaseqio.o bwtgap.o bwtaln.o bamlite.o \
|
||||||
is.o bwtindex.o bwape.o kopen.o pemerge.o \
|
is.o bwtindex.o bwape.o kopen.o pemerge.o maxk.o \
|
||||||
bwtsw2_core.o bwtsw2_main.o bwtsw2_aux.o bwt_lite.o \
|
bwtsw2_core.o bwtsw2_main.o bwtsw2_aux.o bwt_lite.o \
|
||||||
bwtsw2_chain.o fastmap.o bwtsw2_pair.o
|
bwtsw2_chain.o fastmap.o bwtsw2_pair.o
|
||||||
PROG= bwa
|
PROG= bwa
|
||||||
|
|
@ -45,7 +45,8 @@ depend:
|
||||||
QSufSort.o: QSufSort.h
|
QSufSort.o: QSufSort.h
|
||||||
bamlite.o: bamlite.h malloc_wrap.h
|
bamlite.o: bamlite.h malloc_wrap.h
|
||||||
bntseq.o: bntseq.h utils.h kseq.h malloc_wrap.h khash.h
|
bntseq.o: bntseq.h utils.h kseq.h malloc_wrap.h khash.h
|
||||||
bwa.o: bntseq.h bwa.h bwt.h ksw.h utils.h kstring.h malloc_wrap.h kseq.h
|
bwa.o: bntseq.h bwa.h bwt.h ksw.h utils.h kstring.h malloc_wrap.h kvec.h
|
||||||
|
bwa.o: kseq.h
|
||||||
bwamem.o: kstring.h malloc_wrap.h bwamem.h bwt.h bntseq.h bwa.h ksw.h kvec.h
|
bwamem.o: kstring.h malloc_wrap.h bwamem.h bwt.h bntseq.h bwa.h ksw.h kvec.h
|
||||||
bwamem.o: ksort.h utils.h kbtree.h
|
bwamem.o: ksort.h utils.h kbtree.h
|
||||||
bwamem_extra.o: bwa.h bntseq.h bwt.h bwamem.h kstring.h malloc_wrap.h
|
bwamem_extra.o: bwa.h bntseq.h bwt.h bwamem.h kstring.h malloc_wrap.h
|
||||||
|
|
@ -79,5 +80,6 @@ kstring.o: kstring.h malloc_wrap.h
|
||||||
ksw.o: ksw.h malloc_wrap.h
|
ksw.o: ksw.h malloc_wrap.h
|
||||||
main.o: kstring.h malloc_wrap.h utils.h
|
main.o: kstring.h malloc_wrap.h utils.h
|
||||||
malloc_wrap.o: malloc_wrap.h
|
malloc_wrap.o: malloc_wrap.h
|
||||||
|
maxk.o: bwa.h bntseq.h bwt.h bwamem.h kseq.h malloc_wrap.h
|
||||||
pemerge.o: ksw.h kseq.h malloc_wrap.h kstring.h bwa.h bntseq.h bwt.h utils.h
|
pemerge.o: ksw.h kseq.h malloc_wrap.h kstring.h bwa.h bntseq.h bwt.h utils.h
|
||||||
utils.o: utils.h ksort.h malloc_wrap.h kseq.h
|
utils.o: utils.h ksort.h malloc_wrap.h kseq.h
|
||||||
|
|
|
||||||
2
main.c
2
main.c
|
|
@ -25,6 +25,7 @@ int main_mem(int argc, char *argv[]);
|
||||||
int main_shm(int argc, char *argv[]);
|
int main_shm(int argc, char *argv[]);
|
||||||
|
|
||||||
int main_pemerge(int argc, char *argv[]);
|
int main_pemerge(int argc, char *argv[]);
|
||||||
|
int main_maxk(int argc, char *argv[]);
|
||||||
|
|
||||||
static int usage()
|
static int usage()
|
||||||
{
|
{
|
||||||
|
|
@ -84,6 +85,7 @@ int main(int argc, char *argv[])
|
||||||
else if (strcmp(argv[1], "mem") == 0) ret = main_mem(argc-1, argv+1);
|
else if (strcmp(argv[1], "mem") == 0) ret = main_mem(argc-1, argv+1);
|
||||||
else if (strcmp(argv[1], "shm") == 0) ret = main_shm(argc-1, argv+1);
|
else if (strcmp(argv[1], "shm") == 0) ret = main_shm(argc-1, argv+1);
|
||||||
else if (strcmp(argv[1], "pemerge") == 0) ret = main_pemerge(argc-1, argv+1);
|
else if (strcmp(argv[1], "pemerge") == 0) ret = main_pemerge(argc-1, argv+1);
|
||||||
|
else if (strcmp(argv[1], "maxk") == 0) ret = main_maxk(argc-1, argv+1);
|
||||||
else {
|
else {
|
||||||
fprintf(stderr, "[main] unrecognized command '%s'\n", argv[1]);
|
fprintf(stderr, "[main] unrecognized command '%s'\n", argv[1]);
|
||||||
return 1;
|
return 1;
|
||||||
|
|
|
||||||
|
|
@ -0,0 +1,66 @@
|
||||||
|
#include <zlib.h>
|
||||||
|
#include <stdio.h>
|
||||||
|
#include <stdlib.h>
|
||||||
|
#include <limits.h>
|
||||||
|
#include <string.h>
|
||||||
|
#include "bwa.h"
|
||||||
|
#include "bwamem.h"
|
||||||
|
#include "kseq.h"
|
||||||
|
KSEQ_DECLARE(gzFile)
|
||||||
|
|
||||||
|
int main_maxk(int argc, char *argv[])
|
||||||
|
{
|
||||||
|
int i, c, self = 0, max_len = 0;
|
||||||
|
uint8_t *cnt = 0;
|
||||||
|
uint64_t hist[256];
|
||||||
|
bwt_t *bwt;
|
||||||
|
kseq_t *ks;
|
||||||
|
smem_i *itr;
|
||||||
|
gzFile fp;
|
||||||
|
|
||||||
|
while ((c = getopt(argc, argv, "s")) >= 0) {
|
||||||
|
if (c == 's') self = 1;
|
||||||
|
}
|
||||||
|
if (optind + 2 > argc) {
|
||||||
|
fprintf(stderr, "Usage: bwa maxk [-s] <index.prefix> <seq.fa>\n");
|
||||||
|
return 1;
|
||||||
|
}
|
||||||
|
fp = strcmp(argv[optind+1], "-")? gzopen(argv[optind+1], "rb") : gzdopen(fileno(stdin), "rb");
|
||||||
|
ks = kseq_init(fp);
|
||||||
|
bwt = bwt_restore_bwt(argv[optind]);
|
||||||
|
itr = smem_itr_init(bwt);
|
||||||
|
if (self) smem_config(itr, 2, INT_MAX, 0);
|
||||||
|
memset(hist, 0, 8 * 256);
|
||||||
|
|
||||||
|
while (kseq_read(ks) >= 0) {
|
||||||
|
const bwtintv_v *a;
|
||||||
|
if (ks->seq.l > max_len) {
|
||||||
|
max_len = ks->seq.l;
|
||||||
|
kroundup32(max_len);
|
||||||
|
cnt = realloc(cnt, max_len);
|
||||||
|
}
|
||||||
|
memset(cnt, 0, ks->seq.l);
|
||||||
|
for (i = 0; i < ks->seq.l; ++i)
|
||||||
|
ks->seq.s[i] = nst_nt4_table[(int)ks->seq.s[i]];
|
||||||
|
smem_set_query(itr, ks->seq.l, (uint8_t*)ks->seq.s);
|
||||||
|
while ((a = smem_next(itr)) != 0) {
|
||||||
|
for (i = 0; i < a->n; ++i) {
|
||||||
|
bwtintv_t *p = &a->a[i];
|
||||||
|
int j, l, start = p->info>>32, end = (uint32_t)p->info;
|
||||||
|
l = end - start < 255? end - start : 255;
|
||||||
|
for (j = start; j < end; ++j)
|
||||||
|
cnt[j] = cnt[j] > l? cnt[j] : l;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
for (i = 0; i < ks->seq.l; ++i) ++hist[cnt[i]];
|
||||||
|
}
|
||||||
|
for (i = 0; i < 256; ++i)
|
||||||
|
printf("%d\t%lld\n", i, (long long)hist[i]);
|
||||||
|
free(cnt);
|
||||||
|
|
||||||
|
smem_itr_destroy(itr);
|
||||||
|
bwt_destroy(bwt);
|
||||||
|
kseq_destroy(ks);
|
||||||
|
gzclose(fp);
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
Loading…
Reference in New Issue