void liftAcross(char *liftAcross, char *srcFile, char *dstOut)
/* liftAcross - convert one coordinate system to another, no overlapping items. */
{
struct hash *lftHash = readLift(liftAcross);
struct genePred *gpList = genePredExtLoadAll(srcFile);
struct genePred *gp = NULL;
FILE *out = mustOpen(dstOut, "w");
if (bedOut)
bedRegionOutput(lftHash);
int genePredItemCount = 0;
for (gp = gpList; gp != NULL; gp = gp->next)
{
struct liftSpec *lsFound = hashFindVal(lftHash, gp->chrom);
if (lsFound)
{
struct genePred *gpLifted = liftGenePred(gp, lsFound);
struct genePred *gpl;
for (gpl = gpLifted; gpl != NULL; gpl = gpl->next)
genePredTabOut(gpl, out);
genePredFreeList(&gpLifted);
}
else
{
genePredTabOut(gp, out);
}
++genePredItemCount;
}
/* lftHash and gpList are left allocated to disappear at exit */
verbose(2,"#\tgene pred item count: %d\n", genePredItemCount);
}
int palOutPredsInBeds(struct sqlConnection *conn, struct cart *cart,
struct bed *beds, char *table )
/* output the alignments whose names and coords match a bed*/
{
struct genePred *list = NULL;
for(; beds; beds = beds->next)
{
char where[10 * 1024];
sqlSafefFrag(where, sizeof where,
"name = '%s' and chrom='%s' and txEnd > %d and txStart <= %d",
beds->name, beds->chrom, beds->chromStart, beds->chromEnd);
struct genePredReader *reader = genePredReaderQuery( conn, table, where);
struct genePred *pred;
while ((pred = genePredReaderNext(reader)) != NULL)
slAddHead(&list, pred);
genePredReaderFree(&reader);
}
int outCount = 0;
if (list != NULL)
{
slReverse(&list);
outCount = palOutPredList( conn, cart, list);
genePredFreeList(&list);
}
return outCount;
}
static void getGeneAnns(struct sqlConnection *conn, struct hash *refSeqVerInfoTbl, char *outFile)
/* get request genePred annotations from database */
{
struct genePred *gps = genePredReaderLoadQuery(conn, "refGene", NULL);
slSort(&gps, genePredNameCmp);
FILE *fh = mustOpen(outFile, "w");
struct genePred *gp;
for (gp = gps; gp != NULL; gp = gp->next)
processGenePred(fh, refSeqVerInfoTbl, gp);
carefulClose(&fh);
genePredFreeList(&gps);
}
void hgLoadGenePred(char *db, char *table, int numGenePreds, char **genePredFiles)
/* hgLoadGenePred - Load up a mySQL database genePred table. */
{
struct genePred *genes = loadGenes(numGenePreds, genePredFiles);
struct sqlConnection *conn = sqlConnect(db);
char *tmpDir = ".";
FILE *tabFh = hgCreateTabFile(tmpDir, table);
mkTabFile(db, genes, tabFh);
genePredFreeList(&genes);
setupTable(db, conn, table);
hgLoadTabFile(conn, tmpDir, table, &tabFh);
sqlDisconnect(&conn);
hgRemoveTabFile(tmpDir, table);
}
static void createCcdsGene(struct sqlConnection *conn, char *ccdsGeneFile,
struct genomeInfo *genome, struct hash* ignoreTbl,
struct hash *gotCcds)
/* create the ccdsGene tab file from the ccds database */
{
struct ccdsLocationsJoin *locs = loadLocations(conn, genome, ignoreTbl, gotCcds);
struct genePred *gp, *genes = buildCcdsGene(&locs);
FILE *genesFh;
genesFh = mustOpen(ccdsGeneFile, "w");
for (gp = genes; gp != NULL; gp = gp->next)
{
if (loadDb)
fprintf(genesFh, "%d\t", binFromRange(gp->txStart, gp->txEnd));
genePredTabOut(gp, genesFh);
}
carefulClose(&genesFh);
genePredFreeList(&genes);
}
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); */
}
void oneChromInput(char *database, char *chrom, int chromSize,
char *rangeTrack, char *expTrack,
struct hash *refLinkHash, struct hash *erHash, FILE *f)
/* Read in info for one chromosome. */
{
struct binKeeper *rangeBk = binKeeperNew(0, chromSize);
struct binKeeper *expBk = binKeeperNew(0, chromSize);
struct binKeeper *knownBk = binKeeperNew(0, chromSize);
struct bed *rangeList = NULL, *range;
struct bed *expList = NULL;
struct genePred *knownList = NULL;
struct rangeInfo *riList = NULL, *ri;
struct hash *riHash = hashNew(0); /* rangeInfo values. */
struct binElement *rangeBeList = NULL, *rangeBe, *beList = NULL, *be;
/* Load up data from database. */
rangeList = loadBed(database, chrom, rangeTrack, 12, rangeBk);
expList = loadBed(database, chrom, expTrack, 15, expBk);
knownList = loadGenePred(database, chrom, "refGene", knownBk);
/* Build range info basics. */
rangeBeList = binKeeperFindAll(rangeBk);
for (rangeBe = rangeBeList; rangeBe != NULL; rangeBe = rangeBe->next)
{
range = rangeBe->val;
AllocVar(ri);
slAddHead(&riList, ri);
hashAddSaveName(riHash, range->name, ri, &ri->id);
ri->range = range;
ri->commonName = findCommonName(range, knownBk, refLinkHash);
}
slReverse(&riList);
/* Mark split ones. */
beList = binKeeperFindAll(expBk);
for (be = beList; be != NULL; be = be->next)
{
struct bed *exp = be->val;
struct binElement *subList = binKeeperFind(rangeBk,
exp->chromStart, exp->chromEnd);
if (slCount(subList) > 1)
{
struct binElement *sub;
for (sub = subList; sub != NULL; sub = sub->next)
{
struct bed *range = sub->val;
struct rangeInfo *ri = hashMustFindVal(riHash, range->name);
ri->isSplit = TRUE;
}
}
slFreeList(&subList);
}
/* Output the nice ones: not split and having some expression info. */
for (ri = riList; ri != NULL; ri = ri->next)
{
if (!ri->isSplit)
{
struct bed *range = ri->range;
beList = binKeeperFind(expBk, range->chromStart, range->chromEnd);
if (beList != NULL)
outputAveraged(f, ri, erHash, beList);
slFreeList(&beList);
}
}
/* Clean up time! */
freeHash(&riHash);
genePredFreeList(&knownList);
bedFree(&rangeList);
bedFree(&expList);
slFreeList(&rangeBeList);
slFreeList(&beList);
slFreeList(&riList);
binKeeperFree(&rangeBk);
binKeeperFree(&expBk);
binKeeperFree(&knownBk);
}