void
Sim4::free_align(edit_script_list *aligns) {
edit_script_list *head;
head = aligns;
while ((head=aligns)!=NULL) {
aligns = aligns->next_script;
Free_script(head->script);
ckfree(head);
}
}
static void
print_align_lat(uchar *seq1, uchar *seq2, result_p_t r)
{
int *S;
edit_script_list_p_t head, aligns;
if (r->sList == NULL)
return;
aligns = r->sList;
while (aligns != NULL) {
head = aligns;
aligns = aligns->next_script;
S = (int *) xmalloc((2 * head->len2 + 1 + 1) * sizeof(int));
S++;
S2A(head->script, S, 0);
Free_script(head->script);
IDISPLAY(seq1 + head->offset1 - 1 - 1, seq2 + head->offset2 - 1 - 1,
head->len1, head->len2, S,
head->offset1, head->offset2, &r->eCol, r->direction);
free(S - 1);
free(head);
}
r->sList = NULL;
}
/* s2 is the cdna */
static void trim_polyT_align(struct edit_script_list **Sptr, Exon **exons, const int ec, int *pT, uchar *s1, uchar *s2)
{
edit_script_list *head = *Sptr;
edit_script *tp;
int tmpi = 0, num, idents = 0;
uchar *a, *b;
Exon *t;
int i, j; /* i index in the cDNA */
if (ec<head->offset2) {
*pT = ec;
return;
}
if (ec==head->offset2+head->len2-1) {
/* cDNA gap: remove the entire script */
*Sptr = head->next_script;
Free_script(head->script);
free(head);
while ((*exons)->from2<ec) {
t = *exons; *exons = t->next_exon; free(t);
}
*pT = ec;
return;
}
i = head->offset2;
j = head->offset1;
tp = head->script;
while (i<=ec && tp) {
num = tp->num;
switch (tp->op_type) {
case INSERT:
if (i<=ec && ec<i+num-1) {
tmpi += ec-i+1; tp->num -= ec-i+1; i = ec+1;
} else {
i += num; tmpi += num; head->script = tp->next;
free(tp); tp = head->script;
}
break;
case DELETE:
j += num; tmpi += num; head->script = tp->next;
free(tp); tp = head->script;
break;
case SUBSTITUTE:
if (i<=ec && ec<i+num-1) {
a = s2+i-1; b = s1+j-1;
while (a<s2+ec) {
if (*a++!=*b++) tmpi++; else idents++;
}
j += ec-i+1; tp->num -= ec-i+1; i = ec+1;
} else {
/* at most 1 nt remaining */
a = s2+i-1; b = s1+j-1;
while (a<s2+i+tp->num-1) {
if (*a++!=*b++) tmpi++; else idents++;
}
i +=num; j += num;
head->script = tp->next;
free(tp); tp = head->script;
}
break;
default: fatalf("Unrecognized opcode %d.\n",tp->op_type);
}
/* indel walk */
}
assert(i==ec+1);
while (tp->op_type!=SUBSTITUTE && j-1<=(*exons)->to1) {
if (tp->op_type==INSERT) {
i += tp->num; tmpi += tp->num;
} else if (j>=(*exons)->to1 && i>=(*exons)->to2) {
j += tp->num;
} else {
j += tp->num; tmpi += tp->num;
}
head->script = tp->next;
free(tp); tp = head->script;
}
if (head->script==NULL) {
*Sptr = head->next_script;
free(head);
} else {
head->len1 -= j-head->offset1;
head->len2 -= i-head->offset2;
head->offset2 = i;
head->offset1 = j;
head->score -= tmpi;
}
if ((*exons)->to2<i) {
t = *exons; *exons = t->next_exon; free(t);
} else {
double tmp_matches;
(*exons)->from2 = i;
(*exons)->from1 = j;
(*exons)->length = (*exons)->to2-(*exons)->from2+1;
tmp_matches = (*exons)->nmatches - idents;
(*exons)->alen -= tmpi+idents;
(*exons)->match = (int)(100*tmp_matches/(*exons)->alen);
}
*pT = i-1;
return;
}
static void trim_polyA_align(struct edit_script_list **Sptr, Exon *lblock, Exon **exons, const int bc, int *pA, uchar *s1,uchar *s2)
{
edit_script_list *head = *Sptr;
edit_script *tp;
int tmpi = 0, num, idents = 0;
uchar *a, *b;
Exon *prev;
int i, j; /* i index in the cDNA */
if (bc>head->offset2+head->len2-1) {
*pA = bc;
return;
}
if (bc==head->offset2) {
/* cDNA gap: remove the entire script; this should be properly sorted */
*Sptr = head->next_script;
Free_script(head->script);
free(head);
while ((*exons)->from2>=bc) {
prev = find_previous(lblock,*exons);
prev->next_exon = (*exons)->next_exon;
free(*exons); *exons = prev;
}
*pA = bc;
return;
}
Flip_script(&(head->script));
i = head->offset2 + head->len2 -1;
j = head->offset1 + head->len1 -1;
tp = head->script;
while (i>=bc && tp) {
num = tp->num;
switch (tp->op_type) {
case INSERT:
if (i>=bc && bc>i-num+1) {
tmpi += i-bc+1; tp->num -= i-bc+1; i = bc-1;
} else {
i -= num; tmpi += num; head->script = tp->next;
free(tp); tp = head->script;
}
break;
case DELETE:
j -= num; tmpi += num; head->script = tp->next;
free(tp); tp = head->script;
break;
case SUBSTITUTE:
if (i>=bc && bc>i-num+1) {
a = s2+i-1; b = s1+j-1;
while (a>=s2+bc-1) {
if (*a--!=*b--) tmpi++; else idents++;
}
j -= i-bc+1; tp->num -= i-bc+1; i = bc-1;
} else {
/* at most 1 nt remaining */
a = s2+i-1; b = s1+j-1;
while (a>=s2+i-num) {
if (*a--!=*b--) tmpi++; else idents++;
}
i -= num; j -= num;
head->script = tp->next;
free(tp); tp = head->script;
}
break;
default: fatalf("Unrecognized opcode %d.\n",tp->op_type);
}
/* indel walk */
}
assert(i==bc-1);
while (tp->op_type!=SUBSTITUTE && j+1>=(*exons)->from1) {
if (tp->op_type==INSERT) {
i -= tp->num; tmpi += tp->num;
} else if (j<(*exons)->from1 && i<(*exons)->from2) {
j -= tp->num;
} else {
j -= tp->num; tmpi += tp->num;
}
head->script = tp->next;
free(tp); tp = head->script;
}
if (head->script==NULL) {
*Sptr = head->next_script;
free(head);
} else {
head->len1 = j-head->offset1+1;
head->len2 = i-head->offset2+1;
head->score -= tmpi;
Flip_script(&(head->script));
}
if ((*exons)->from2>i) {
prev = find_previous(lblock,*exons);
prev->next_exon = (*exons)->next_exon;
free(*exons); *exons = prev;
} else {
double tmp_matches;
(*exons)->to2 = i;
(*exons)->to1 = j;
(*exons)->length = (*exons)->to2-(*exons)->from2+1;
tmp_matches = (*exons)->nmatches - idents;
(*exons)->alen -= tmpi+idents;
(*exons)->match = (int)(100*tmp_matches/(*exons)->alen);
}
*pA = i+1;
return;
}
void
Sim4::trim_polyA_align(struct edit_script_list **Sptr, Exon *lblock, Exon **exons, const int bc, int *pA, char *s1,char *s2)
{
edit_script_list *head = *Sptr;
edit_script *tp;
int tmpi = 0, num, idents = 0, identsN = 0;
char *a, *b;
Exon *prev;
int i, j; /* i index in the cDNA */
if (bc>head->offset2+head->len2-1) {
*pA = bc;
return;
}
if (bc==head->offset2) {
/* cDNA gap: remove the entire script; is this properly sorted? LLL */
*Sptr = head->next_script;
Free_script(head->script);
ckfree(head);
while ((*exons)->frEST>=bc) {
prev = find_previous(lblock,*exons);
if (prev == 0L) {
fprintf(stderr, "trim_polyA_align(): Corrupted exon list, cannot find the previous exon (remove entire script).\n");
for (; lblock; lblock = lblock->next_exon)
fprintf(stderr, " GEN f=%8d t=%8d EST f=%8d t=%8d flag=%d\n",
lblock->frGEN, lblock->toGEN, lblock->frEST, lblock->toEST, lblock->flag);
kill(getpid(), SIGKILL);
}
prev->next_exon = (*exons)->next_exon;
//freeExon(*exons); garbage collected
*exons = prev;
}
*pA = bc;
return;
}
Flip_script(&(head->script));
i = head->offset2 + head->len2 -1;
j = head->offset1 + head->len1 -1;
tp = head->script;
while (i>=bc && tp) {
num = tp->num;
switch (tp->op_type) {
case INSERT:
if (i>=bc && bc>i-num+1) {
(*exons)->numInDel -= i - bc + 1;
(*exons)->numEdits -= i - bc + 1;
tmpi += i-bc+1;
tp->num -= i-bc+1;
i = bc-1;
} else {
(*exons)->numInDel -= num;
(*exons)->numEdits -= num;
tmpi += num;
i -= num;
head->script = tp->next;
ckfree(tp);
tp = head->script;
}
break;
case DELETE:
(*exons)->numInDel -= num;
(*exons)->numEdits -= num;
j -= num;
tmpi += num;
head->script = tp->next;
ckfree(tp);
tp = head->script;
break;
case SUBSTITUTE:
if (i>=bc && bc>i-num+1) {
a = s2+i-1; b = s1+j-1;
while (a>=s2+bc-1) {
if (*a != *b) {
(*exons)->numEdits--;
tmpi++;
} else {
if (*a == 'N') {
(*exons)->numNs--;
identsN++;
} else {
(*exons)->numMatches--;
idents++;
}
}
a--;
b--;
}
j -= i-bc+1; tp->num -= i-bc+1; i = bc-1;
} else {
/* at most 1 nt remaining */
a = s2+i-1; b = s1+j-1;
while (a>=s2+i-num) {
if (*a != *b) {
(*exons)->numEdits--;
tmpi++;
} else {
if (*a == 'N') {
(*exons)->numNs--;
identsN++;
} else {
(*exons)->numMatches--;
idents++;
}
}
a--;
b--;
}
i -= num; j -= num;
head->script = tp->next;
ckfree(tp);
tp = head->script;
}
break;
#if 0
default:
fatalf("Unrecognized opcode %d.\n",tp->op_type);
#endif
}
/* indel walk */
}
assert(i==bc-1);
while ((tp != 0L) &&
(tp->op_type != SUBSTITUTE) && (j+1 >= (*exons)->frGEN)) {
if (tp->op_type==INSERT) {
i -= tp->num;
tmpi += tp->num;
(*exons)->numInDel -= tp->num;
(*exons)->numEdits -= tp->num;
} else if (j<(*exons)->frGEN && i<(*exons)->frEST) {
j -= tp->num;
} else {
j -= tp->num;
tmpi += tp->num;
(*exons)->numInDel -= tp->num;
(*exons)->numEdits -= tp->num;
}
head->script = tp->next;
ckfree(tp);
tp = head->script;
}
if (head->script==NULL) {
*Sptr = head->next_script;
ckfree(head);
} else {
head->len1 = j-head->offset1+1;
head->len2 = i-head->offset2+1;
head->score -= tmpi;
Flip_script(&(head->script));
}
if ((*exons)->frEST>i) {
prev = find_previous(lblock,*exons);
if (prev == 0L) {
fprintf(stderr, "trim_polyA_align(): Corrupted exon list, cannot find the previous exon (frEST).\n");
for (; lblock; lblock = lblock->next_exon)
fprintf(stderr, " GEN f=%8d t=%8d EST f=%8d t=%8d flag=%d\n",
lblock->frGEN, lblock->toGEN, lblock->frEST, lblock->toEST, lblock->flag);
kill(getpid(), SIGKILL);
}
prev->next_exon = (*exons)->next_exon;
//freeExon(*exons); garbage collected
*exons = prev;
} else {
(*exons)->toEST = i;
(*exons)->toGEN = j;
(*exons)->length = (*exons)->toEST-(*exons)->frEST+1;
(*exons)->alignmentLength = ((*exons)->toGEN - (*exons)->frGEN + 1 +
(*exons)->toEST - (*exons)->frEST + 1 +
(*exons)->numInDel);
(*exons)->percentID = computePercentIdentity((*exons)->numEdits,
(*exons)->alignmentLength);
}
*pA = i+1;
return;
}
/* s2 is the cdna */
void
Sim4::trim_polyT_align(struct edit_script_list **Sptr, Exon **exons, const int ec, int *pT, char *s1, char *s2)
{
edit_script_list *head = *Sptr;
edit_script *tp;
int tmpi = 0, num, idents = 0, identsN = 0;
char *a, *b;
Exon *t;
int i, j; /* i index in the cDNA */
if (ec<head->offset2) {
*pT = ec;
return;
}
if (ec==head->offset2+head->len2-1) {
/* cDNA gap: remove the entire script */
*Sptr = head->next_script;
Free_script(head->script);
ckfree(head);
while ((*exons)->frEST<ec) {
t = *exons;
*exons = t->next_exon;
//freeExon(t); garbage collected
}
*pT = ec;
return;
}
i = head->offset2;
j = head->offset1;
tp = head->script;
while (i<=ec && tp) {
num = tp->num;
switch (tp->op_type) {
case INSERT:
if (i<=ec && ec<i+num-1) {
(*exons)->numInDel -= ec - i + 1;
(*exons)->numEdits -= ec - i + 1;
tmpi += ec-i+1;
tp->num -= ec-i+1;
i = ec+1;
} else {
(*exons)->numInDel -= num;
(*exons)->numEdits -= num;
tmpi += num;
i += num;
head->script = tp->next;
ckfree(tp);
tp = head->script;
}
break;
case DELETE:
(*exons)->numInDel -= num;
(*exons)->numEdits -= num;
j += num;
tmpi += num;
head->script = tp->next;
ckfree(tp);
tp = head->script;
break;
case SUBSTITUTE:
if (i<=ec && ec<i+num-1) {
a = s2+i-1; b = s1+j-1;
while (a<s2+ec) {
if (*a != *b) {
(*exons)->numEdits--;
tmpi++;
} else {
if (*a == 'N') {
(*exons)->numNs--;
identsN++;
} else {
(*exons)->numMatches--;
idents++;
}
}
a++;
b++;
}
j += ec-i+1; tp->num -= ec-i+1; i = ec+1;
} else {
/* at most 1 nt remaining */
a = s2+i-1; b = s1+j-1;
while (a<s2+i+tp->num-1) {
if (*a != *b) {
(*exons)->numEdits--;
tmpi++;
} else {
if (*a == 'N') {
(*exons)->numNs--;
identsN++;
} else {
(*exons)->numMatches--;
idents++;
}
}
a++;
b++;
}
i +=num; j += num;
head->script = tp->next;
ckfree(tp);
tp = head->script;
}
break;
}
/* indel walk */
}
assert(i==ec+1);
while ((tp != 0L) &&
(tp->op_type!=SUBSTITUTE) && (j-1<=(*exons)->toGEN)) {
if (tp->op_type==INSERT) {
i += tp->num;
tmpi += tp->num;
(*exons)->numInDel -= tp->num;
(*exons)->numEdits -= tp->num;
} else if (j>=(*exons)->toGEN && i>=(*exons)->toEST) {
j += tp->num;
} else {
j += tp->num;
tmpi += tp->num;
(*exons)->numInDel -= tp->num;
(*exons)->numEdits -= tp->num;
}
head->script = tp->next;
ckfree(tp);
tp = head->script;
}
if (head->script==NULL) {
*Sptr = head->next_script;
ckfree(head);
} else {
head->len1 -= j-head->offset1;
head->len2 -= i-head->offset2;
head->offset2 = i;
head->offset1 = j;
head->score -= tmpi;
}
if ((*exons)->toEST<i) {
t = *exons;
*exons = t->next_exon;
//freeExon(t); garbage collected
} else {
(*exons)->frEST = i;
(*exons)->frGEN = j;
(*exons)->length = (*exons)->toEST-(*exons)->frEST+1;
(*exons)->alignmentLength = ((*exons)->toGEN - (*exons)->frGEN + 1 +
(*exons)->toEST - (*exons)->frEST + 1 +
(*exons)->numInDel);
(*exons)->percentID = computePercentIdentity((*exons)->numEdits,
(*exons)->alignmentLength);
}
*pT = i-1;
return;
}