void pslIntronsOnly(char *inPslName, char *genoFile, char *outPslName)
/* pslIntronsOnly - Filter psl files to only include those with introns. */
{
struct lineFile *lf = NULL;
FILE *outFile = NULL;
struct hash *genoHash = loadGeno(genoFile);
struct psl *psl;
int count = 0, intronCount = 0;
lf = pslFileOpen(inPslName);
outFile = mustOpen(outPslName, "w");
while ((psl = pslNext(lf)) != NULL)
{
struct dnaSeq *geno = hashMustFindVal(genoHash, psl->tName);
if (pslHasIntron(psl, geno, 0))
{
++intronCount;
pslTabOut(psl, outFile);
}
pslFree(&psl);
++count;
}
carefulClose(&outFile);
lineFileClose(&lf);
printf("%d of %d in %s have introns\n", intronCount, count, inPslName);
}
struct altGraphX *agFromGp(char *db, struct genePred *gp, struct sqlConnection *conn,
int maxGap, FILE *out)
/** Create an altGraphX record by clustering psl records within coordinates
specified by genePred record. */
{
struct altGraphX *ag = NULL;
struct dnaSeq *genoSeq = NULL;
struct ggMrnaAli *maList=NULL, *ma=NULL, *maNext=NULL, *maSameStrand=NULL;
struct psl *pslList = NULL, *psl = NULL, *pslCluster = NULL, *pslNext = NULL;
char *chrom = gp->chrom;
int chromStart = BIGNUM;
int chromEnd = -1;
verbose(2, "agFromGp on %s %s:%d-%d\n", gp->name, gp->chrom, gp->txStart, gp->txEnd);
pslList = getPsls(gp, conn);
verbose(3, " got %d psls\n", slCount(pslList));
if(slCount(pslList) == 0)
{
verbose(2, "No available alignments for %s.", gp->name);
return NULL;
}
/* expand to find the furthest boundaries of alignments */
expandToMaxAlignment(pslList, chrom, &chromStart, &chromEnd);
verbose(3, " expanded to %s:%d-%d\n", chrom, chromStart, chromEnd);
/* get the sequence */
genoSeq = dnaFromChrom(db, chrom, chromStart, chromEnd, dnaLower);
for(psl = pslList; psl != NULL; psl = pslNext)
{
pslNext = psl->next;
if(singleExonOk || pslHasIntron(psl, genoSeq, chromStart))
{
slAddHead(&pslCluster, psl);
}
else
{
if(!useChromKeeper)
pslFree(&psl);
}
}
verbose(3, " got %d psls after intron/singleExon check\n", slCount(pslCluster));
/* load and check the alignments */
maList = pslListToGgMrnaAliList(pslCluster, gp->chrom, chromStart, chromEnd, genoSeq, maxGap);
verbose(3, " got %d in maList\n", slCount(maList));
for(ma = maList; ma != NULL; ma = maNext)
{
maNext = ma->next;
verbose(4, " ma->strand %s, gp->strand %s\n", ma->strand, gp->strand);
if(ma->strand[0] == gp->strand[0])
{
slSafeAddHead(&maSameStrand, ma);
}
else
ggMrnaAliFree(&ma);
}
slReverse(&maSameStrand);
verbose(3, " got %d in ma on same strand\n", slCount(maSameStrand));
/* If there is a cluster to work with create an geneGraph */
if(maSameStrand != NULL)
{
ag = agFromAlignments(db, maSameStrand, genoSeq, conn, chromStart, chromEnd, out);
}
else
{
dnaSeqFree(&genoSeq);
ggMrnaAliFreeList(&maSameStrand);
}
/* Only free psls if not using cache... */
if(!useChromKeeper)
pslFreeList(&pslCluster);
return ag;
}
