ALNoverlap *
Optimal_Overlap_AS_forCNS(char *a, char *b,
int begUNUSED, int endUNUSED,
int ahang, int bhang,
int opposite,
double erate, double thresh, int minlen,
CompareOptions what) {
static Optimal_Overlap_Data *ood = NULL;
if (ood == NULL)
ood = new Optimal_Overlap_Data;
memset(ood->h_alignA, 0, sizeof(char) * (AS_MAX_READLEN + AS_MAX_READLEN + 2));
memset(ood->h_alignB, 0, sizeof(char) * (AS_MAX_READLEN + AS_MAX_READLEN + 2));
memset(ood->h_trace, 0, sizeof(int) * (AS_MAX_READLEN + AS_MAX_READLEN + 2));
alignLinker_s al;
//if (VERBOSE_MULTIALIGN_OUTPUT >= 3)
// fprintf(stderr, "Optimal_Overlap_AS_forCNS()-- Begins\n");
#if 0
if (erate > AS_MAX_ERROR_RATE) {
//fprintf(stderr, "Optimal_Overlap_AS_forCNS()-- erate=%f >= AS_MAX_ERROR_RATE=%f, reset to max\n", erate, (double)AS_MAX_ERROR_RATE);
erate = AS_MAX_ERROR_RATE;
}
assert((0.0 <= erate) && (erate <= AS_MAX_ERROR_RATE));
#endif
if (opposite)
reverseComplementSequence(b, strlen(b));
//if (VERBOSE_MULTIALIGN_OUTPUT >= 3) {
// fprintf(stderr, "ALIGN %s\n", a);
// fprintf(stderr, "ALIGN %s\n", b);
//}
alignLinker(ood->h_alignA,
ood->h_alignB,
a,
b,
&al,
true, // Looking for global end-to-end alignments
false, // Count matches to N as matches
ahang, bhang);
if (al.alignLen == 0) {
return NULL;
}
//if (VERBOSE_MULTIALIGN_OUTPUT >= 3) {
// fprintf(stderr, "ALIGN %d %d-%d %d-%d opposite=%d\n", al.alignLen, al.begI, al.endI, al.begJ, al.endJ, opposite);
// fprintf(stderr, "ALIGN '%s'\n", ood->h_alignA);
// fprintf(stderr, "ALIGN '%s'\n", ood->h_alignB);
//}
if (opposite) {
reverseComplementSequence(b, strlen(b));
reverseComplementSequence(ood->h_alignA, al.alignLen);
reverseComplementSequence(ood->h_alignB, al.alignLen);
int x = al.begJ;
al.begJ = al.lenB - al.endJ;
al.endJ = al.lenB - x;
}
// We don't expect partial overlaps here. At least one fragment
// must have an alignment to the very start.
//
// ECR depends on this return value; it is allowed to fail
// when building a new unitig multialign. For example:
//
// <-----------------------
// ------>
//
// When ECR tries to extend the second fragment, it checks that
// the extended fragment overlaps the next contig. It does not
// check that the extended bits agree with the first fragment,
// leaving that up to "does the unitig rebuild".
//
if ((al.begJ != 0) && (al.begI != 0))
return(NULL);
ood->o.begpos = (al.begI > 0) ? (al.begI) : -(al.begJ);
ood->o.endpos = (al.lenB - al.endJ > 0) ? (al.lenB - al.endJ) : -(al.lenA - al.endI);
ood->o.length = al.alignLen;
ood->o.diffs = 0;
ood->o.comp = opposite;
ood->o.trace = ood->h_trace;
{
int x=0;
int tp = 0;
int ap = al.begI;
int bp = al.begJ;
for (x=0; x<al.alignLen; x++) {
if (ood->h_alignA[x] == '-') {
ood->h_trace[tp++] = -(ap + 1);
ap--;
}
if (ood->h_alignB[x] == '-') {
ood->h_trace[tp++] = bp + 1;
bp--;
}
// Count the differences.
//
// But allow N's and lowercase as matches. If either letter is N, then the other letter is
// NOT N (if both letters were N, both would be lowercase n, representing a match). This
// just subtracts out the diff we added in above.
//
bool diff = false;
bool ignore = false;
if (toupper(ood->h_alignA[x]) != toupper(ood->h_alignB[x]))
diff = true;
if ((ood->h_alignA[x] == 'N') || (ood->h_alignA[x] == 'n') ||
(ood->h_alignB[x] == 'N') || (ood->h_alignB[x] == 'n'))
ignore = true;
if (islower(ood->h_alignA[x]) && (ood->h_alignB[x] == '-'))
ignore = true;
if ((diff == true) && (ignore == false))
ood->o.diffs++;
bp++;
ap++;
}
ood->h_trace[tp] = 0;
//if (VERBOSE_MULTIALIGN_OUTPUT >= 4) {
// fprintf(stderr, "trace");
// for (x=0; x<tp; x++)
// fprintf(stderr, " %d", ood->h_trace[x]);
// fprintf(stderr, "\n");
// fprintf(stderr, "A: %4d-%4d %4d %s\n", al.begI, al.endI, al.lenA, ood->h_alignA);
// fprintf(stderr, "B: %4d-%4d %4d %s\n", al.begJ, al.endJ, al.lenB, ood->h_alignB);
//}
}
//if (VERBOSE_MULTIALIGN_OUTPUT >= 3) {
// fprintf(stderr, "ERATE: diffs=%d / length=%d = %f\n", ood->o.diffs, ood->o.length, (double)ood->o.diffs / ood->o.length);
// fprintf(stderr, "Optimal_Overlap_AS_forCNS()-- Ends\n");
//}
if ((double)ood->o.diffs / ood->o.length <= erate)
return(&ood->o);
return(NULL);
}
int foundLinker(char *seq,char *link1,char *link2,int *st,int *en,float *pId,
int allowTail) {
int score,alnlen,id,start,end,coverage;
int score1,alnlen1,id1,start1,end1,coverage1;
int score2,alnlen2,id2,start2,end2,coverage2;
int found,found1,found2;
float percentId,percentId1,percentId2;
char *link;
found1 = found2 = found = 0;
score1 = alignLinker(seq,link1,&alnlen1,&id1,&start1,&end1);
score2 = alignLinker(seq,link2,&alnlen2,&id2,&start2,&end2);
percentId1 = ((float) id1) / ((float) alnlen1);
percentId2 = ((float) id2) / ((float) alnlen2);
coverage1 = end1-start1+1;
coverage2 = end2-start2+1;
if (percentId1 >= identity &&
(coverage1 >= matchLen || (allowTail && strlen(seq)-end1 <= 2 &&
coverage1 >= minLengthToCallLinker))) {
found1 = 1;
}
if (percentId2 >= identity &&
(coverage2 >= matchLen || (allowTail && strlen(seq)-end2 <= 2 &&
coverage2 >= minLengthToCallLinker))) {
found2 = 1;
}
if (verbose > 1) {
if (found1) {
fprintf(stderr,"found1: %s score=%d id=%d alnlen=%d\n",seq,score1,id1,alnlen1);
fprintf(stderr," %*s%s\n",start1,"",link1);
}
if (found2) {
fprintf(stderr,"found2: %s score=%d id=%d alnlen=%d\n",seq,score2,id2,alnlen2);
fprintf(stderr," %*s%s\n",start2,"",link2);
}
}
if (found1 && (!found2 || score1 > score2)) {
found = 1;
score = score1;
alnlen = alnlen1;
id = id1;
start = start1;
end = end1;
coverage = coverage1;
percentId = percentId1;
link = link1;
} else if ((!found1 || score2 >= score1) && found2) {
found = 1;
score = score2;
alnlen = alnlen2;
id = id2;
start = start2;
end = end2;
coverage = coverage2;
percentId = percentId2;
link = link2;
} else if (!found1 && !found2) {
found = 0;
} else {
fprintf(stderr,"ERROR: ambiguous case in foundLinker\n");
fprintf(stderr," seq: %s\n",seq);
fprintf(stderr," lnk1: %s score1=%d id1=%d start1=%d end1=%d alnlen1=%d\n",
link1,score1,id1,start1,end1,alnlen1);
fprintf(stderr," lnk2: %s score2=%d id2=%d start2=%d end2=%d alnlen2=%d\n",
link2,score2,id2,start2,end2,alnlen2);
found = 0;
}
if (found) {
*pId = percentId;
*st = start;
*en = end;
if (verbose) {
fprintf(stderr,"final: %s score=%d id=%d alnlen=%d\n",seq,score,id,alnlen);
fprintf(stderr," %*s%s\n",start,"",link);
}
} else {
if (verbose) fprintf(stderr,"no match %s\n",seq);
}
return found;
}