mirror of
https://github.com/metabarcoding/obitools4.git
synced 2025-06-29 16:20:46 +00:00
Add the possibility to look for patterns allowing indels.
Former-commit-id: 0599c2b0ad16df086dbdb08e491503870d8904be
This commit is contained in:
@ -151,10 +151,7 @@ int32_t ManberNoErr (Seq *pseq , Pattern *ppat, int32_t patnum,int32_t beg
|
||||
int32_t ManberSub (Seq *pseq , Pattern *ppat, int32_t patnum,int32_t begin,int32_t length);
|
||||
int32_t ManberIndel (Seq *pseq , Pattern *ppat, int32_t patnum,int32_t begin,int32_t length);
|
||||
int32_t ManberAll (Seq *pseq , Pattern *ppat, int32_t patnum,int32_t begin,int32_t length);
|
||||
int32_t NwsPatAlign (Seq *pseq , Pattern *ppat, int32_t nerr ,
|
||||
int32_t *reslen , int32_t *reserr);
|
||||
|
||||
/* apat_sys.c */
|
||||
int32_t NwsPatAlign (Seq *pseq , Pattern *ppat, int32_t nerr, int32_t begin, int32_t *reslen, int32_t *reserr); /* apat_sys.c */
|
||||
|
||||
float UserCpuTime (int32_t reset);
|
||||
float SysCpuTime (int32_t reset);
|
||||
|
@ -23,7 +23,8 @@
|
||||
#define TOPCURS CursiToTop
|
||||
#define DOWNREAD ReadiDown
|
||||
|
||||
#define KRONECK(x, msk) ((~x & msk) ? 0 : 1)
|
||||
//#define KRONECK(x, msk) ((~x & msk) ? 0 : 1)
|
||||
#define KRONECK(x, msk) ((x & msk) ? 0 : 1)
|
||||
#define MIN(x, y) ((x) < (y) ? (x) : (y))
|
||||
|
||||
/* -------------------------------------------- */
|
||||
@ -192,8 +193,8 @@ int32_t ManberIndel(Seq *pseq, Pattern *ppat, int patnum,int begin,int length)
|
||||
{
|
||||
int e, emax, found;
|
||||
uint32_t pos;
|
||||
uint32_t smask, cmask, sindx;
|
||||
uint32_t *pr, r[2 * MAX_PAT_ERR + 2];
|
||||
patword_t smask, cmask, sindx;
|
||||
patword_t *pr, r[2 * MAX_PAT_ERR + 2];
|
||||
uint8_t *data;
|
||||
StackiPtr *stkpos, *stkerr;
|
||||
uint32_t end;
|
||||
@ -220,6 +221,9 @@ int32_t ManberIndel(Seq *pseq, Pattern *ppat, int patnum,int begin,int length)
|
||||
stkpos = pseq->hitpos + patnum;
|
||||
stkerr = pseq->hiterr + patnum;
|
||||
|
||||
EmptyStacki(stkpos[0]);
|
||||
EmptyStacki(stkerr[0]);
|
||||
|
||||
/* loop on text data */
|
||||
|
||||
for (pos = begin ; pos < end ; pos++) {
|
||||
@ -256,12 +260,14 @@ int32_t ManberIndel(Seq *pseq, Pattern *ppat, int patnum,int begin,int length)
|
||||
/* -------------------------------------------- */
|
||||
int32_t ManberAll(Seq *pseq, Pattern *ppat, int patnum,int begin,int length)
|
||||
{
|
||||
if (ppat->maxerr == 0)
|
||||
return ManberNoErr(pseq, ppat, patnum, begin, length);
|
||||
else if (ppat->hasIndel)
|
||||
if (ppat->maxerr == 0){
|
||||
return ManberNoErr(pseq, ppat, patnum, begin, length);}
|
||||
else if (ppat->hasIndel) {
|
||||
return ManberIndel(pseq, ppat, patnum, begin, length);
|
||||
else
|
||||
}
|
||||
else {
|
||||
return ManberSub(pseq, ppat, patnum, begin, length);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@ -271,11 +277,9 @@ int32_t ManberAll(Seq *pseq, Pattern *ppat, int patnum,int begin,int length)
|
||||
/* (avec substitution obligatoire aux bords) */
|
||||
/* -------------------------------------------- */
|
||||
|
||||
int32_t NwsPatAlign(pseq, ppat, nerr, reslen, reserr)
|
||||
Seq *pseq;
|
||||
Pattern *ppat;
|
||||
int32_t nerr, *reslen, *reserr;
|
||||
{
|
||||
int32_t NwsPatAlign(Seq *pseq, Pattern *ppat,
|
||||
int32_t nerr, int32_t begin,
|
||||
int32_t *reslen, int32_t *reserr) {
|
||||
uint8_t *sseq, *px;
|
||||
int32_t i, j, lseq, lpat, npos, dindel, dsub,
|
||||
*pc, *pi, *pd, *ps;
|
||||
@ -283,7 +287,9 @@ int32_t NwsPatAlign(pseq, ppat, nerr, reslen, reserr)
|
||||
|
||||
static int32_t sTab[(MAX_PAT_LEN+MAX_PAT_ERR+1) * (MAX_PAT_LEN+1)];
|
||||
|
||||
lseq = pseq->seqlen;
|
||||
lpat = ppat->patlen;
|
||||
lseq = MIN(lpat + MAX_PAT_ERR+1, pseq->seqlen - begin);
|
||||
sseq = pseq->data + begin - 1;
|
||||
|
||||
pc = sTab; /* |----|----| --> i */
|
||||
pi = pc - 1; /* | ps | pd | | */
|
||||
@ -291,36 +297,39 @@ int32_t NwsPatAlign(pseq, ppat, nerr, reslen, reserr)
|
||||
ps = pd - 1; /* | pi | pc | v j */
|
||||
/* |---------| */
|
||||
|
||||
lseq = pseq->seqlen;
|
||||
lpat = ppat->patlen;
|
||||
|
||||
sseq = pseq->data - 1;
|
||||
|
||||
amask = ONEMASK >> lpat;
|
||||
//amask = ONEMASK >> lpat;
|
||||
amask = 0x1L << (ppat->patlen);
|
||||
|
||||
for (j = 0 ; j <= lpat ; j++) {
|
||||
|
||||
|
||||
for (i = 0 , px = sseq ; i <= lseq ; i++, px++) {
|
||||
|
||||
if (i && j) {
|
||||
dindel = MIN(*pi, *pd) + 1;
|
||||
if (j == lpat) dindel--;
|
||||
dsub = *ps + KRONECK(ppat->smat[*px], amask);
|
||||
// fprintf(stderr, "mismatch : %d %d %d %d\n",j,*px,KRONECK(ppat->smat[*px], amask),amask);
|
||||
*pc = MIN(dindel, dsub);
|
||||
}
|
||||
else if (i) /* j == 0 */
|
||||
*pc = *pi + 1;
|
||||
else if (j) /* i == 0 */
|
||||
*pc = *pd + 1;
|
||||
*pc = *pd;
|
||||
else /* root */
|
||||
*pc = 0;
|
||||
|
||||
fprintf(stderr," %02d",*pc);
|
||||
pc++;
|
||||
pi++;
|
||||
pd++;
|
||||
ps++;
|
||||
}
|
||||
fprintf(stderr,"\n");
|
||||
|
||||
amask <<= 1;
|
||||
// amask <<= 1;
|
||||
amask >>= 1;
|
||||
}
|
||||
|
||||
pc--;
|
||||
@ -331,6 +340,7 @@ int32_t NwsPatAlign(pseq, ppat, nerr, reslen, reserr)
|
||||
*reserr++ = *pc;
|
||||
npos++;
|
||||
}
|
||||
fprintf(stderr,"i=%d *pc = %d<%d, reserr = %d npos = %d\n",i,*pc,nerr,*(reserr-1),npos);
|
||||
}
|
||||
|
||||
return npos;
|
||||
|
@ -337,7 +337,7 @@ int32_t delete_apatseq(SeqPtr pseq,
|
||||
return 1;
|
||||
}
|
||||
|
||||
PatternPtr buildPattern(const char *pat, int32_t error_max,
|
||||
PatternPtr buildPattern(const char *pat, int32_t error_max, uint8_t hasIndel,
|
||||
int *errno, char **errmsg)
|
||||
{
|
||||
PatternPtr pattern;
|
||||
@ -355,7 +355,7 @@ PatternPtr buildPattern(const char *pat, int32_t error_max,
|
||||
errno,errmsg);
|
||||
|
||||
pattern->ok = true;
|
||||
pattern->hasIndel= false;
|
||||
pattern->hasIndel= hasIndel;
|
||||
pattern->maxerr = error_max;
|
||||
|
||||
pattern->cpat = (char*)pattern + sizeof(Pattern);
|
||||
|
@ -118,7 +118,7 @@ ecoseq_t *new_ecoseq_with_data( char *AC,
|
||||
|
||||
int32_t delete_apatseq(SeqPtr pseq,
|
||||
int *errno, char **errmsg);
|
||||
PatternPtr buildPattern(const char *pat, int32_t error_max, int *errno, char **errmsg);
|
||||
PatternPtr buildPattern(const char *pat, int32_t error_max, uint8_t hasIndel, int *errno, char **errmsg);
|
||||
PatternPtr complementPattern(PatternPtr pat, int *errno, char **errmsg);
|
||||
|
||||
SeqPtr new_apatseq(const char *in,int32_t circular, int32_t seqlen,
|
||||
|
@ -13,6 +13,8 @@ import (
|
||||
|
||||
log "github.com/sirupsen/logrus"
|
||||
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/goutils"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obialign"
|
||||
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
|
||||
)
|
||||
|
||||
@ -65,14 +67,20 @@ var NilApatSequence = ApatSequence{nil}
|
||||
// the errormax parameter. Some positions can be marked as not
|
||||
// allowed for mismatches. They have to be signaled using a '#'
|
||||
// sign after the corresponding nucleotide.
|
||||
func MakeApatPattern(pattern string, errormax int) (ApatPattern, error) {
|
||||
func MakeApatPattern(pattern string, errormax int, allowsIndel bool) (ApatPattern, error) {
|
||||
cpattern := C.CString(pattern)
|
||||
defer C.free(unsafe.Pointer(cpattern))
|
||||
cerrormax := C.int32_t(errormax)
|
||||
|
||||
callosindel := C.uint8_t(0)
|
||||
if allowsIndel {
|
||||
callosindel = C.uint8_t(1)
|
||||
}
|
||||
|
||||
var errno C.int32_t
|
||||
var errmsg *C.char
|
||||
|
||||
apc := C.buildPattern(cpattern, cerrormax, &errno, &errmsg)
|
||||
apc := C.buildPattern(cpattern, cerrormax, callosindel, &errno, &errmsg)
|
||||
|
||||
if apc == nil {
|
||||
message := C.GoString(errmsg)
|
||||
@ -281,16 +289,13 @@ func (sequence ApatSequence) Free() {
|
||||
// values of the [3]int indicate respectively the start and the end position of
|
||||
// the match. Following the GO convention the end position is not included in the
|
||||
// match. The third value indicates the number of error detected for this occurrence.
|
||||
func (pattern ApatPattern) FindAllIndex(sequence ApatSequence, limits ...int) (loc [][3]int) {
|
||||
begin := 0
|
||||
length := sequence.Len()
|
||||
|
||||
if len(limits) > 0 {
|
||||
begin = limits[0]
|
||||
func (pattern ApatPattern) FindAllIndex(sequence ApatSequence, begin, length int) (loc [][3]int) {
|
||||
if begin < 0 {
|
||||
begin = 0
|
||||
}
|
||||
|
||||
if len(limits) > 1 {
|
||||
length = limits[1]
|
||||
if length < 0 {
|
||||
length = sequence.Len()
|
||||
}
|
||||
|
||||
nhits := int(C.ManberAll(sequence.pointer.pointer,
|
||||
@ -310,13 +315,70 @@ func (pattern ApatPattern) FindAllIndex(sequence ApatSequence, limits ...int) (l
|
||||
for i := 0; i < nhits; i++ {
|
||||
start := int(stktmp[i])
|
||||
err := int(errtmp[i])
|
||||
|
||||
log.Debugln(C.GoString(pattern.pointer.pointer.cpat), start, err)
|
||||
loc = append(loc, [3]int{start, start + patlen, err})
|
||||
}
|
||||
|
||||
log.Debugln("------------")
|
||||
return loc
|
||||
}
|
||||
|
||||
func (pattern ApatPattern) BestMatch(sequence ApatSequence, begin, length int) (start int, end int, nerr int, matched bool) {
|
||||
res := pattern.FindAllIndex(sequence, begin, length)
|
||||
|
||||
sbuffer := [(int(C.MAX_PAT_LEN) + int(C.MAX_PAT_ERR) + 1) * (int(C.MAX_PAT_LEN) + 1)]uint64{}
|
||||
buffer := sbuffer[:]
|
||||
|
||||
if len(res) == 0 {
|
||||
matched = false
|
||||
return
|
||||
}
|
||||
|
||||
matched = true
|
||||
|
||||
best := [3]int{0, 0, 10000}
|
||||
for _, m := range res {
|
||||
if m[2] < best[2] {
|
||||
best = m
|
||||
log.Debugln(best)
|
||||
}
|
||||
}
|
||||
|
||||
nerr = best[2]
|
||||
end = best[1]
|
||||
|
||||
if nerr == 0 || !pattern.pointer.pointer.hasIndel {
|
||||
start = best[0]
|
||||
log.Debugln("No nws ", start, nerr)
|
||||
return
|
||||
}
|
||||
|
||||
start = best[0] - nerr
|
||||
end = best[0] + int(pattern.pointer.pointer.patlen) + nerr
|
||||
start = goutils.MaxInt(start, 0)
|
||||
end = goutils.MinInt(end, sequence.Len())
|
||||
|
||||
cpattern := (*[1 << 30]byte)(unsafe.Pointer(pattern.pointer.pointer.cpat))
|
||||
cseq := (*[1 << 30]byte)(unsafe.Pointer(sequence.pointer.pointer.cseq))
|
||||
|
||||
log.Debugln(
|
||||
string((*cseq)[start:end]),
|
||||
string((*cpattern)[0:int(pattern.pointer.pointer.patlen)]),
|
||||
best[0], nerr, int(pattern.pointer.pointer.patlen),
|
||||
sequence.Len(), start, end)
|
||||
|
||||
score, lali := obialign.FastLCSEGFScoreByte(
|
||||
(*cseq)[start:end],
|
||||
(*cpattern)[0:int(pattern.pointer.pointer.patlen)],
|
||||
nerr*2, true, &buffer)
|
||||
|
||||
nerr = lali - score
|
||||
start = best[0] + int(pattern.pointer.pointer.patlen) - lali
|
||||
log.Println("results", score, lali, start, nerr)
|
||||
return
|
||||
}
|
||||
|
||||
// tagaacaggctcctctag
|
||||
// func AllocatedApaSequences() int {
|
||||
// return int(_AllocatedApaSequences)
|
||||
// }
|
||||
|
@ -133,7 +133,7 @@ func OptionForwardPrimer(primer string, max int) WithOption {
|
||||
f := WithOption(func(opt Options) {
|
||||
var err error
|
||||
|
||||
opt.pointer.forward, err = MakeApatPattern(primer, max)
|
||||
opt.pointer.forward, err = MakeApatPattern(primer, max, false)
|
||||
if err != nil {
|
||||
log.Fatalf("error : %v\n", err)
|
||||
}
|
||||
@ -155,7 +155,7 @@ func OptionReversePrimer(primer string, max int) WithOption {
|
||||
f := WithOption(func(opt Options) {
|
||||
var err error
|
||||
|
||||
opt.pointer.reverse, err = MakeApatPattern(primer, max)
|
||||
opt.pointer.reverse, err = MakeApatPattern(primer, max, false)
|
||||
if err != nil {
|
||||
log.Fatalf("error : %v\n", err)
|
||||
}
|
||||
@ -210,7 +210,7 @@ func _Pcr(seq ApatSequence,
|
||||
reverse := opt.pointer.reverse
|
||||
crev := opt.pointer.crev
|
||||
|
||||
forwardMatches := forward.FindAllIndex(seq)
|
||||
forwardMatches := forward.FindAllIndex(seq,0,-1)
|
||||
|
||||
if len(forwardMatches) > 0 {
|
||||
|
||||
@ -284,7 +284,7 @@ func _Pcr(seq ApatSequence,
|
||||
}
|
||||
}
|
||||
|
||||
forwardMatches = reverse.FindAllIndex(seq)
|
||||
forwardMatches = reverse.FindAllIndex(seq,0,-1)
|
||||
if forwardMatches != nil {
|
||||
|
||||
begin := forwardMatches[0][0]
|
||||
|
Reference in New Issue
Block a user