void doGenePreds(struct sqlConnection *conn, char *db, char *orthoDb, char *chrom,
char *netTable, char *geneFileName, char *geneTableName,
char *outBedName, char *selectedFileName, int *foundCount, int *notFoundCount)
/* Map over genePreds. */
{
FILE *bedOut = NULL;
FILE *selectedOut = NULL;
FILE *cdsErrorFp = NULL;
struct genePred *gene = NULL, *geneList = NULL;
struct bed *bed = NULL;
//init output files
if(optionExists("cdsErrorFile"))
{
cdsErrorFp = fopen( optionVal("cdsErrorFile", NULL), "w" );
fprintf( cdsErrorFp, "#name\tchrom\ttxStart\ttxEnd\tcdsStart\tcdsEnd\tstrand\texonCount\n" );
fclose(cdsErrorFp);
}
warn("Loading Gene Predictions.");
assert(outBedName);
if(geneFileName)
geneList=genePredLoadAll(geneFileName);
else
geneList=loadGeneFromTable(conn, geneTableName, chrom, 0, BIGNUM);
/* Convert genePreds. */
warn("Converting genes.");
bedOut = mustOpen(outBedName, "w");
if (selectedFileName != NULL)
selectedOut = mustOpen(selectedFileName, "w");
for(gene = geneList; gene != NULL; gene = gene->next)
{
struct genePred *synGene = NULL;
if(differentString(gene->chrom, chrom))
continue;
synGene = orthoBedFromGene(conn, db, orthoDb, netTable, gene);
occassionalDot();
if(synGene != NULL && synGene->exonCount > 0)
{
(*foundCount)++;
genePredTabOut(synGene, bedOut);
if (selectedOut != NULL)
genePredTabOut(gene, selectedOut);
}
else
(*notFoundCount)++;
genePredFree(&synGene);
}
carefulClose(&selectedOut);
carefulClose(&bedOut);
}
void mafToSnpBed(char *database, char *mafIn, char *gpIn, char *bedOut)
/* mafToSnpBed - finds SNPs in MAF and builds a bed with their functional consequence. */
{
struct mafFile *mafFile = mafReadAll(mafIn);
struct genePred *genePred = genePredLoadAll(gpIn);
FILE *f = mustOpen(bedOut, "w");
struct genePred *gp, *next;
for(gp = genePred; gp; gp = next)
{
next = gp->next;
gp->next = NULL;
parseOneGp(database, mafFile->alignments, gp, f);
}
}
int main(int argc, char *argv[])
{
struct genePred *gpList = NULL;
FILE *out = NULL;
optionInit(&argc, argv, NULL);
if(argc !=4)
usage();
warn("Loading gene predictions.");
char *chromSizesFile = argv[1];
if(optionExists("bedFormat"))
gpList = gpFromBedFile(argv[2]);
else
gpList = genePredLoadAll(argv[2]);
out = mustOpen(argv[3],"w");
warn("Doing conversion.");
pslListFromGenePred(chromSizesFile, gpList, out);
carefulClose(&out);
warn("Done.");
return 0;
}
void createAltSplices(char *db, char *outFile, boolean memTest)
/* Top level routine, gets genePredictions and runs through them to
build altSplice graphs. */
{
struct genePred *gp = NULL, *gpList = NULL;
struct altGraphX *ag=NULL;
FILE *out = NULL;
struct sqlConnection *conn = hAllocConn(db);
char *gpFile = NULL;
char *bedFile = NULL;
int count =0;
/* Figure out where to get coordinates from. */
bedFile = optionVal("beds", NULL);
gpFile = optionVal("genePreds", NULL);
if(bedFile != NULL)
gpList = convertBedsToGps(bedFile);
else if(gpFile != NULL)
gpList = genePredLoadAll(gpFile);
else
{
warn("Must specify target loci as either a bed file or a genePred file");
usage();
}
if (!gpAllSameChrom(gpList))
errAbort("Multiple chromosomes in bed or genePred file.");
/* Sanity check to make sure we got some loci to work
with. */
if(gpList == NULL)
errAbort("No gene boundaries were found.");
slSort(&gpList, genePredCmp);
setupTables(gpList->chrom);
/* If local memory get things going here. */
if(optionExists("localMem"))
{
warn("Using local memory. Setting up caches...");
useChromKeeper = TRUE;
setupChromKeeper(conn, optionVal("db", NULL), gpList->chrom);
if(!optionExists("skipTissues"))
{
if(optionExists("tissueLibFile"))
readTissueLibraryIntoCache(optionVal("tissueLibFile", NULL));
else
setupTissueLibraryCache(conn);
}
warn("Done setting up local caches.");
}
else /* Have to set up agxSeen binKeeper based on genePreds. */
{
int maxPos = 0;
int minPos = BIGNUM;
for(gp = gpList; gp != NULL; gp = gp->next)
{
maxPos = max(maxPos, gp->txEnd);
minPos = min(minPos, gp->txStart);
}
agxSeenBin = binKeeperNew(max(0, minPos-10000), min(BIGNUM,maxPos+10000));
}
dotForUserInit(max(slCount(gpList)/10, 1));
out = mustOpen(outFile, "w");
for(gp = gpList; gp != NULL && count < 5; )
{
dotForUser();
fflush(stderr);
ag = agFromGp(db, gp, conn, 5, out); /* memory held in binKeeper. Free
* later. */
if (memTest != TRUE)
gp = gp->next;
}
genePredFreeList(&gpList);
hFreeConn(&conn);
/* uglyf("%d genePredictions with %d clusters, %d cassette exons, %d of are not mod 3.\n", */
/* slCount(gpList), clusterCount, cassetteCount, misSense); */
}
