void
Sim4::remove_polyT_front(struct edit_script_list **Sptr, Exon *Exons, char *s1, char *s2, int *lastT)
{
Exon *t, *exons_head; /* start from Lblock */
char *b, *end;
int numT, dummy, trim_p, reverse_script=0, pT;
int startPos=0, cutAmount=0;
*lastT = pT = 0;
if (!Exons || !Exons->next_exon || !Exons->next_exon->toGEN) return;
if ((*Sptr)->next_script &&
(*Sptr)->offset1>(*Sptr)->next_script->offset1) {
script_flip_list(Sptr);
reverse_script = 1;
}
exons_head = Exons->next_exon; trim_p = 1;
if (exons_head) {
startPos = exons_head->frEST;
while ((t=exons_head)!=NULL && t->toGEN && trim_p) {
/* compute the 'T' contents of the exon */
b = s2 + t->frEST-1; end = s2+t->toEST; numT = 0;
while (b<end && (numT+t->toEST-(b-s2)>=globalParams->_polyTailPercent*t->length)) {
numT += (*b++=='T');
}
// Determine how much of the cut stuff was actually
// poly-containing. The first method below returns the number of
// bases cut from the end of the est, while the second return the
// number of bases cut from the end of the alignment.
//
//cutAmount = l2 - *lastT + 1;
if (numT>=globalParams->_polyTailPercent*t->length) {
/* remove the entire exon */
trim_polyT_align(Sptr,&exons_head,t->toEST,lastT,s1,s2);
cutAmount = *lastT - startPos + 1;
} else {
get_polyAT(s2+(*Sptr)->offset2-1,(*Sptr)->len2,&pT,&dummy,T_ONLY);
if (pT) {
int ct_pT;
ct_pT = pT + (*Sptr)->offset2-1;
ct_pT = (t->toEST-ct_pT>MIN_EXON) ? ct_pT : t->toEST;
trim_polyT_align(Sptr,&exons_head,ct_pT,lastT,s1,s2);
cutAmount = *lastT - startPos + 1;
}
if (t==exons_head) trim_p = 0;
}
}
}
Exons->next_exon = exons_head;
*lastT = cutAmount;
if (reverse_script) script_flip_list(Sptr);
}
static void remove_polyA_back(struct edit_script_list **Sptr, Exon *Exons, uchar *s1, uchar *s2, int len2, int *lastA) {
Exon *t, *exons_tail, *prev; /* start from Lblock */
uchar *b, *end;
int numA, pA, dummy, trim_p, reverse_script=0;
*lastA = len2+1; pA = 0;
if (!Exons || ! Exons->next_exon || ! Exons->next_exon->to1) return;
if ((*Sptr)->next_script &&
(*Sptr)->offset1<(*Sptr)->next_script->offset1) {
reverse_script = 1;
script_flip_list(Sptr);
}
exons_tail = Exons->next_exon; prev = Exons;
for ( ; exons_tail->next_exon && exons_tail->next_exon->to1;
prev=exons_tail, exons_tail=exons_tail->next_exon);
trim_p = TRUE;
while ((t=exons_tail)!=NULL && t->to1 && trim_p) {
/* compute the 'A' contents of the exon */
b = s2 + t->to2-1; end = s2+t->from2-1; numA = 0;
while (b>=end && numA+(b-s2)>=.60*t->length) {
numA += (*b--=='A');
}
if (numA>=.60*t->length) {
/* remove the entire exon */
trim_polyA_align(Sptr,Exons,&exons_tail,t->from2,lastA,s1,s2);
/* assert(*lastA==t->from2); t was removed */
} else {
get_polyAT(s2+(*Sptr)->offset2-1,(*Sptr)->len2,&dummy,&pA,A_ONLY);
if (pA) {
int ct_pA;
/* first position to be removed */
ct_pA = t->to2-pA+1;
ct_pA = (ct_pA-t->from2>=MIN_EXON) ? ct_pA : t->from2;
/* note: pA is the last (innermost) position in the tail */
trim_polyA_align(Sptr,Exons,&exons_tail,ct_pA,lastA,s1,s2);
}
if (t==exons_tail) trim_p = FALSE;
}
}
if (reverse_script) script_flip_list(Sptr);
}
static void remove_polyT_front(struct edit_script_list **Sptr, Exon *Exons, uchar *s1, uchar *s2, int *lastT)
{
Exon *t, *exons_head; /* start from Lblock */
uchar *b, *end;
int numT, dummy, trim_p, reverse_script=0, pT;
*lastT = pT = 0;
if (!Exons || !Exons->next_exon || !Exons->next_exon->to1) return;
if ((*Sptr)->next_script &&
(*Sptr)->offset1>(*Sptr)->next_script->offset1) {
script_flip_list(Sptr);
reverse_script = 1;
}
exons_head = Exons->next_exon; trim_p = TRUE;
while ((t=exons_head)!=NULL && t->to1 && trim_p) {
/* compute the 'T' contents of the exon */
b = s2 + t->from2-1; end = s2+t->to2; numT = 0;
while (b<end && (numT+t->to2-(b-s2-t->from2+1)>=.60*t->length)) {
numT += (*b++=='T');
}
if (numT>=.60*t->length) {
/* remove the entire exon */
trim_polyT_align(Sptr,&exons_head,t->to2,lastT,s1,s2);
/* assert(*lastT==t->to2); t was removed */
} else {
get_polyAT(s2+(*Sptr)->offset2-1,(*Sptr)->len2,&pT,&dummy,T_ONLY);
if (pT) {
int ct_pT;
ct_pT = pT + (*Sptr)->offset2-1;
ct_pT = (t->to2-ct_pT>=MIN_EXON) ? ct_pT : t->to2;
trim_polyT_align(Sptr,&exons_head,ct_pT,lastT,s1,s2);
}
if (t==exons_head) trim_p = FALSE;
}
}
Exons->next_exon = exons_head;
if (reverse_script) script_flip_list(Sptr);
}
void
Sim4::remove_polyA_back(struct edit_script_list **Sptr, Exon *Exons,
char *s1, char *s2,
int l2, int *lastA) {
Exon *t;
Exon *exons_tail;
char *b, *end;
int numA, pA, dummy, trim_p, reverse_script=0;
int startPos=0, cutAmount=0;
*lastA = l2+1; pA = 0;
if (!Exons || ! Exons->next_exon || ! Exons->next_exon->toGEN) return;
if ((*Sptr)->next_script &&
(*Sptr)->offset1<(*Sptr)->next_script->offset1) {
reverse_script = 1;
script_flip_list(Sptr);
}
exons_tail = Exons->next_exon;
while (exons_tail->next_exon && exons_tail->next_exon->toGEN)
exons_tail=exons_tail->next_exon;
trim_p = 1;
if (exons_tail) {
startPos = exons_tail->toEST;
while ((t=exons_tail)!=NULL && t->toGEN && trim_p) {
/* compute the 'A' contents of the exon */
b = s2 + t->toEST-1; end = s2+t->frEST-1; numA = 0;
while (b>=end && numA+(b-end)>=globalParams->_polyTailPercent*t->length) {
numA += (*b--=='A');
}
// Determine how much of the cut stuff was actually
// poly-containing. The first method below returns the number of
// bases cut from the end of the est, while the second return the
// number of bases cut from the end of the alignment.
//
//cutAmount = l2 - *lastA + 1;
if (numA>=globalParams->_polyTailPercent*t->length) {
/* remove the entire exon */
trim_polyA_align(Sptr,Exons,&exons_tail,t->frEST,lastA,s1,s2);
cutAmount = startPos - *lastA + 1;
} else {
get_polyAT(s2+(*Sptr)->offset2-1,(*Sptr)->len2,&dummy,&pA,A_ONLY);
if (pA) {
int ct_pA;
/* first position to be removed */
ct_pA = t->toEST-pA+1;
ct_pA = (ct_pA-t->frEST>MIN_EXON) ? ct_pA : t->frEST;
/* note: pA is the last (innermost) position in the tail */
trim_polyA_align(Sptr,Exons,&exons_tail,ct_pA,lastA,s1,s2);
cutAmount = startPos - *lastA + 1;
}
if (t==exons_tail) trim_p = 0;
}
}
}
*lastA = cutAmount;
if (reverse_script) script_flip_list(Sptr);
}
