PBWT *pbwtSubRange (PBWT *pOld, int start, int end)
{
int M = pOld->M ;
PBWT *pNew = pbwtCreate (M) ;
int i, j, k ;
uchar *x ;
PbwtCursor *uOld = pbwtCursorCreate (pOld, TRUE, TRUE) ;
pNew->yz = arrayCreate (pNew->N*8, uchar) ;
PbwtCursor *uNew = pbwtCursorCreate (pNew, TRUE, TRUE) ;
if (!pOld || !pOld->yz) die ("subrange without an existing pbwt") ;
if (start < 0 || end > pOld->N || end <= start)
die ("subrange invalid start %d, end %d", start, end) ;
x = myalloc (M, uchar) ;
if (pOld->sites) pNew->sites = arrayCreate (4096, Site) ;
for (i = 0 ; i < end ; ++i)
{ if (i >= start)
{ for (j = 0 ; j < M ; ++j) x[uOld->a[j]] = uOld->y[j] ;
for (j = 0 ; j < M ; ++j) uNew->y[j] = x[uNew->a[j]] ;
pbwtCursorWriteForwards (uNew) ;
if (pOld->sites) array(pNew->sites, pNew->N, Site) = arr(pOld->sites, i, Site) ;
++pNew->N ;
}
pbwtCursorForwardsRead (uOld) ;
}
pNew->aFend = myalloc (pNew->M, int) ; memcpy (pNew->aFend, uNew->a, pNew->M*sizeof(int)) ;
pbwtDestroy (pOld) ; pbwtCursorDestroy (uOld) ; pbwtCursorDestroy (uNew) ;
free(x) ;
return pNew ;
}
void pbwtBuildReverse (PBWT *p)
{
int i, j, M = p->M ;
uchar *x = myalloc (M, uchar) ;
PbwtCursor *uF ;
if (p->aFend)
uF = pbwtCursorCreate (p, TRUE, FALSE) ;
else
{ uF = pbwtCursorCreate (p, TRUE, TRUE) ;
for (i = 0 ; i < p->N ; ++i) /* first run forwards to the end */
pbwtCursorForwardsRead (uF) ;
pbwtCursorToAFend (uF, p) ;
}
/* use p->aFend also to start the reverse cursor - this gives better performance */
if (!p->aRstart) p->aRstart = myalloc (M, int) ; memcpy (p->aRstart, uF->a, M * sizeof(int)) ;
p->zz = arrayReCreate (p->zz, arrayMax(p->yz), uchar) ;
PbwtCursor *uR = pbwtCursorCreate (p, FALSE, TRUE) ; /* will pick up aRstart */
for (i = p->N ; i-- ; )
{ pbwtCursorReadBackwards (uF) ;
for (j = 0 ; j < M ; ++j) x[uF->a[j]] = uF->y[j] ;
for (j = 0 ; j < M ; ++j) uR->y[j] = x[uR->a[j]] ;
pbwtCursorWriteForwards (uR) ;
}
/* save uR->a, which is the lexicographic order of the sequences */
if (!p->aRend) p->aRend = myalloc (M, int) ; memcpy (p->aRend, uR->a, M * sizeof(int)) ;
fprintf (logFile, "built reverse PBWT - size %ld\n", arrayMax(p->zz)) ;
if (isCheck) /* print out the reversed haplotypes */
{ FILE *fp = fopen ("rev.haps","w") ;
Array tz = p->yz ; p->yz = p->zz ;
int* ta = p->aFstart ; p->aFstart = p->aRstart ;
pbwtWriteHaplotypes (fp, p) ;
p->yz = tz ; p->aFstart = ta ;
}
free (x) ;
pbwtCursorDestroy (uF) ; pbwtCursorDestroy (uR) ;
}
PBWT *pbwtSubSites (PBWT *pOld, double fmin, double frac)
{
int M = pOld->M ;
PBWT *pNew = pbwtCreate (M, 0) ;
int i, j, k, thresh = M*(1-fmin) ;
double bit = 0.0 ;
uchar *x ;
PbwtCursor *uOld = pbwtCursorCreate (pOld, TRUE, TRUE) ;
PbwtCursor *uNew = pbwtCursorCreate (pNew, TRUE, TRUE) ;
if (!pOld || !pOld->yz) die ("subsites without an existing pbwt") ;
if (fmin < 0 || fmin >= 1 || frac <= 0 || frac > 1)
die ("fmin %f, frac %f for subsites out of range\n", fmin, frac) ;
x = myalloc (M, uchar) ;
if (pOld->sites) pNew->sites = arrayCreate (4096, Site) ;
for (i = 0 ; i < pOld->N ; ++i)
{ if ((uOld->c < thresh) && ((bit += frac) > 1.0))
{ for (j = 0 ; j < M ; ++j) x[uOld->a[j]] = uOld->y[j] ;
for (j = 0 ; j < M ; ++j) uNew->y[j] = x[uNew->a[j]] ;
pbwtCursorWriteForwards (uNew) ;
if (pOld->sites) array(pNew->sites, pNew->N, Site) = arr(pOld->sites, i, Site) ;
++pNew->N ;
bit -= 1.0 ;
}
pbwtCursorForwardsRead (uOld) ;
}
pbwtCursorToAFend (uNew, pNew) ;
fprintf (logFile, "subsites with fmin %f, frac %f leaves %d sites\n", fmin, frac, pNew->N) ;
pNew->chrom = pOld->chrom ; pOld->chrom = 0 ;
pNew->samples = pOld->samples ; pOld->samples = 0 ;
pNew->missingOffset = pOld->missingOffset ; pOld->missingOffset = 0 ;
pNew->zMissing = pOld->zMissing ; pOld->zMissing = 0 ;
pbwtDestroy (pOld) ; pbwtCursorDestroy (uOld) ; pbwtCursorDestroy (uNew) ;
free(x) ;
return pNew ;
}
PBWT *pbwtSubRange (PBWT *pOld, int start, int end)
{
int M = pOld->M ;
PBWT *pNew = pbwtCreate (M, 0) ;
int i, j, k ;
uchar *x ;
PbwtCursor *uOld = pbwtCursorCreate (pOld, TRUE, TRUE) ;
PbwtCursor *uNew = pbwtCursorCreate (pNew, TRUE, TRUE) ;
if (!pOld || !pOld->yz) die ("subrange without an existing pbwt") ;
if (start < 0 || end > pOld->N || end <= start)
die ("subrange invalid start %d, end %d", start, end) ;
x = myalloc (M, uchar) ;
if (pOld->sites) pNew->sites = arrayCreate (4096, Site) ;
for (i = 0 ; i < end ; ++i)
{ if (i >= start)
{ for (j = 0 ; j < M ; ++j) x[uOld->a[j]] = uOld->y[j] ;
for (j = 0 ; j < M ; ++j) uNew->y[j] = x[uNew->a[j]] ;
pbwtCursorWriteForwards (uNew) ;
if (pOld->sites) array(pNew->sites, pNew->N, Site) = arr(pOld->sites, i, Site) ;
++pNew->N ;
}
pbwtCursorForwardsRead (uOld) ;
}
pbwtCursorToAFend (uNew, pNew) ;
pNew->chrom = pOld->chrom ; pOld->chrom = 0 ;
pNew->samples = pOld->samples ; pOld->samples = 0 ;
pNew->missingOffset = pOld->missingOffset ; pOld->missingOffset = 0 ;
pNew->zMissing = pOld->zMissing ; pOld->zMissing = 0 ;
pbwtDestroy (pOld) ; pbwtCursorDestroy (uOld) ; pbwtCursorDestroy (uNew) ;
free(x) ;
return pNew ;
}