static char *getEstDir(char *def, char *com)
/* Return EST direction as deduced from definition and comment lines. */
{
char *three = "3'";
char *five = "5'";
char *dir = NULL;
boolean gotThreeDef = FALSE, gotFiveDef = FALSE;
boolean gotThreeCom = FALSE, gotFiveCom = FALSE;
gotThreeDef = isThreePrime(def);
gotFiveDef = isFivePrime(def);
if (gotThreeDef ^ gotFiveDef)
dir = (gotThreeDef ? three : five);
if (dir == NULL)
{
gotThreeCom = isThreePrime(com);
gotFiveCom = isFivePrime(com);
if (gotThreeCom ^ gotFiveCom)
dir = (gotThreeCom ? three : five);
}
/* either didn't get 5' or 3' in def and com, or got both */
return dir;
}
boolean findIgRegion(struct entity *ent, int *retStart, int *retEnd, int *retCount)
/* Find intergenic region compute mean and variance of 3' ends, then
* return mean after discarding outlyers. Returns false if data looks funky. */
{
int totalCount = 0;
int totalPos = 0;
int insideCount = 0;
int insidePos = 0;
double mean;
double insideMean;
double dif;
double varience = 0;
double std;
struct cdaRef *ref;
struct cdaAli *ali;
int end;
boolean revStrand = (ent->strand == '-');
/* Calculate mean. */
for (ref = ent->cdaRefList; ref != NULL; ref = ref->next)
{
ali = ref->ali;
if (isThreePrime(ali))
{
++totalCount;
if (revStrand)
totalPos += ali->chromStart-1;
else
totalPos += ali->chromEnd;
}
}
if (totalCount <= 0)
return FALSE;
mean = (double)totalPos/totalCount;
/* Calculate square root of varience to estimate standard deviation. */
for (ref = ent->cdaRefList; ref != NULL; ref = ref->next)
{
ali = ref->ali;
if (isThreePrime(ali))
{
if (revStrand)
dif = ali->chromStart-1;
else
dif = ali->chromEnd;
dif -= mean;
varience += dif*dif;
}
}
varience /= totalCount;
std = paranoidSqrt(varience);
/* If varience too large, or curve not very bell shaped return FALSE.
* Figure out insideMean (mean of stuff within one standard deviation of outer mean.) */
if (std > 200)
{
return FALSE;
}
totalPos = 0;
for (ref = ent->cdaRefList; ref != NULL; ref = ref->next)
{
ali = ref->ali;
if (isThreePrime(ali))
{
++totalCount;
if (revStrand)
end = ali->chromStart-1;
else
end = ali->chromEnd;
dif = end - mean;
if (fabs(dif) <= 200)
{
++insideCount;
insidePos += end;
}
}
}
if (insideCount*2 < totalCount)
{
return FALSE;
}
insideMean = (double)insidePos/insideCount;
if (revStrand)
{
end = round(insideMean - std*0.5);
}
else
{
end = round(insideMean + std*0.5);
}
*retStart = end-3;
*retEnd = end+3;
*retCount = totalCount;
return TRUE;
}
