static struct hash* bbiChromSizes(struct bbiFile* bbi)
/* return the hash of chrom sizes from the bigBed/bigWig */
{
struct bbiChromInfo* cList = bbiChromList(bbi);
struct bbiChromInfo* c;
struct hash* cHash = newHash(10);
for (c = cList; c != NULL; c = c->next)
hashAddInt(cHash, c->name, (int)c->size);
bbiChromInfoFreeList(&cList);
return cHash;
}
void bigBedToBed(char *inFile, char *outFile)
/* bigBedToBed - Convert from bigBed to ascii bed format.. */
{
struct bbiFile *bbi = bigBedFileOpen(inFile);
FILE *f = mustOpen(outFile, "w");
struct bbiChromInfo *chrom, *chromList = bbiChromList(bbi);
int itemCount = 0;
for (chrom = chromList; chrom != NULL; chrom = chrom->next)
{
if (clChrom != NULL && !sameString(clChrom, chrom->name))
continue;
char *chromName = chrom->name;
int start = 0, end = chrom->size;
if (clStart > 0)
start = clStart;
if (clEnd > 0)
end = clEnd;
int itemsLeft = 0; // Zero actually means no limit....
if (maxItems != 0)
{
itemsLeft = maxItems - itemCount;
if (itemsLeft <= 0)
break;
}
struct lm *lm = lmInit(0);
struct bigBedInterval *interval, *intervalList = bigBedIntervalQuery(bbi, chromName,
start, end, itemsLeft, lm);
for (interval = intervalList; interval != NULL; interval = interval->next)
{
fprintf(f, "%s\t%u\t%u", chromName, interval->start, interval->end);
char *rest = interval->rest;
if (rest != NULL)
fprintf(f, "\t%s\n", rest);
else
fprintf(f, "\n");
}
lmCleanup(&lm);
}
bbiChromInfoFreeList(&chromList);
carefulClose(&f);
bbiFileClose(&bbi);
}
struct genomeRangeTree *edwGrtFromBigBed(char *fileName)
/* Return genome range tree for simple (unblocked) bed */
{
struct bbiFile *bbi = bigBedFileOpen(fileName);
struct bbiChromInfo *chrom, *chromList = bbiChromList(bbi);
struct genomeRangeTree *grt = genomeRangeTreeNew();
for (chrom = chromList; chrom != NULL; chrom = chrom->next)
{
struct rbTree *tree = genomeRangeTreeFindOrAddRangeTree(grt, chrom->name);
struct lm *lm = lmInit(0);
struct bigBedInterval *iv, *ivList = NULL;
ivList = bigBedIntervalQuery(bbi, chrom->name, 0, chrom->size, 0, lm);
for (iv = ivList; iv != NULL; iv = iv->next)
rangeTreeAdd(tree, iv->start, iv->end);
lmCleanup(&lm);
}
bigBedFileClose(&bbi);
bbiChromInfoFreeList(&chromList);
return grt;
}
/* --- .Call ENTRY POINT --- */
SEXP BWGFile_seqlengths(SEXP r_filename) {
pushRHandlers();
struct bbiFile * file = bigWigFileOpen((char *)CHAR(asChar(r_filename)));
struct bbiChromInfo *chromList = bbiChromList(file);
struct bbiChromInfo *chrom = chromList;
SEXP seqlengths, seqlengthNames;
PROTECT(seqlengths = allocVector(INTSXP, slCount(chromList)));
seqlengthNames = allocVector(STRSXP, length(seqlengths));
setAttrib(seqlengths, R_NamesSymbol, seqlengthNames);
for(int i = 0; i < length(seqlengths); i++) {
INTEGER(seqlengths)[i] = chrom->size;
SET_STRING_ELT(seqlengthNames, i, mkChar(chrom->name));
chrom = chrom->next;
}
bbiChromInfoFreeList(&chromList);
popRHandlers();
UNPROTECT(1);
return seqlengths;
}
void bigWigToWig(char *inFile, char *outFile)
/* bigWigToWig - Convert bigWig to wig. This will keep more of the same structure of the
* original wig than bigWigToBedGraph does, but still will break up large stepped sections into
* smaller ones. */
{
struct bbiFile *bwf = bigWigFileOpen(inFile);
FILE *f = mustOpen(outFile, "w");
struct bbiChromInfo *chrom, *chromList = bbiChromList(bwf);
for (chrom = chromList; chrom != NULL; chrom = chrom->next)
{
if (clChrom != NULL && !sameString(clChrom, chrom->name))
continue;
char *chromName = chrom->name;
int start = 0, end = chrom->size;
if (clStart > 0)
start = clStart;
if (clEnd > 0)
end = clEnd;
bigWigIntervalDump(bwf, chromName, start, end, 0, f);
}
bbiChromInfoFreeList(&chromList);
carefulClose(&f);
bbiFileClose(&bwf);
}
void doEnrichmentsFromBigWig(struct sqlConnection *conn,
struct cdwFile *ef, struct cdwValidFile *vf,
struct cdwAssembly *assembly, struct target *targetList)
/* Figure out enrichments from a bigBed file. */
{
/* Get path to bigBed, open it, and read all chromosomes. */
char *bigWigPath = cdwPathForFileId(conn, ef->id);
struct bbiFile *bbi = bigWigFileOpen(bigWigPath);
struct bbiChromInfo *chrom, *chromList = bbiChromList(bbi);
struct bigWigValsOnChrom *valsOnChrom = bigWigValsOnChromNew();
/* This takes a while, so let's figure out what parts take the time. */
long totalBigQueryTime = 0;
long totalOverlapTime = 0;
/* Do a pretty complex loop that just aims to set target->overlapBases and ->uniqOverlapBases
* for all targets. This is complicated by just wanting to keep one chromosome worth of
* bigWig data in memory. Also just for performance we do a lookup of target range tree to
* get chromosome specific one to use, which avoids a hash lookup in the inner loop. */
for (chrom = chromList; chrom != NULL; chrom = chrom->next)
{
long startBigQueryTime = clock1000();
boolean gotData = bigWigValsOnChromFetchData(valsOnChrom, chrom->name, bbi);
long endBigQueryTime = clock1000();
totalBigQueryTime += endBigQueryTime - startBigQueryTime;
if (gotData)
{
double *valBuf = valsOnChrom->valBuf;
Bits *covBuf = valsOnChrom->covBuf;
/* Loop through all targets adding overlaps from ivList */
long startOverlapTime = clock1000();
struct target *target;
for (target = targetList; target != NULL; target = target->next)
{
if (target->skip)
continue;
struct genomeRangeTree *grt = target->grt;
struct rbTree *targetTree = genomeRangeTreeFindRangeTree(grt, chrom->name);
if (targetTree != NULL)
{
struct range *range, *rangeList = rangeTreeList(targetTree);
for (range = rangeList; range != NULL; range = range->next)
{
int s = range->start, e = range->end, i;
for (i=s; i<=e; ++i)
{
if (bitReadOne(covBuf, i))
{
double x = valBuf[i];
target->uniqOverlapBases += 1;
target->overlapBases += x;
}
}
}
}
}
long endOverlapTime = clock1000();
totalOverlapTime += endOverlapTime - startOverlapTime;
}
}
verbose(1, "totalBig %0.3f, totalOverlap %0.3f\n", 0.001*totalBigQueryTime, 0.001*totalOverlapTime);
/* Now loop through targets and save enrichment info to database */
struct target *target;
for (target = targetList; target != NULL; target = target->next)
{
if (target->skip)
continue;
struct cdwQaEnrich *enrich = enrichFromOverlaps(ef, vf, assembly, target,
target->overlapBases, target->uniqOverlapBases);
cdwQaEnrichSaveToDb(conn, enrich, "cdwQaEnrich", 128);
cdwQaEnrichFree(&enrich);
}
bigWigValsOnChromFree(&valsOnChrom);
bbiChromInfoFreeList(&chromList);
bigWigFileClose(&bbi);
freez(&bigWigPath);
}
/* This old way is ~3 times as slow */
void doEnrichmentsFromBigWig(struct sqlConnection *conn,
struct cdwFile *ef, struct cdwValidFile *vf,
struct cdwAssembly *assembly, struct target *targetList)
/* Figure out enrichments from a bigBed file. */
{
/* Get path to bigBed, open it, and read all chromosomes. */
char *bigWigPath = cdwPathForFileId(conn, ef->id);
struct bbiFile *bbi = bigWigFileOpen(bigWigPath);
struct bbiChromInfo *chrom, *chromList = bbiChromList(bbi);
/* This takes a while, so let's figure out what parts take the time. */
long totalBigQueryTime = 0;
long totalOverlapTime = 0;
/* Do a pretty complex loop that just aims to set target->overlapBases and ->uniqOverlapBases
* for all targets. This is complicated by just wanting to keep one chromosome worth of
* bigWig data in memory. Also just for performance we do a lookup of target range tree to
* get chromosome specific one to use, which avoids a hash lookup in the inner loop. */
for (chrom = chromList; chrom != NULL; chrom = chrom->next)
{
/* Get list of intervals in bigWig for this chromosome, and feed it to a rangeTree. */
struct lm *lm = lmInit(0);
long startBigQueryTime = clock1000();
struct bbiInterval *ivList = bigWigIntervalQuery(bbi, chrom->name, 0, chrom->size, lm);
long endBigQueryTime = clock1000();
totalBigQueryTime += endBigQueryTime - startBigQueryTime;
struct bbiInterval *iv;
/* Loop through all targets adding overlaps from ivList */
long startOverlapTime = clock1000();
struct target *target;
for (target = targetList; target != NULL; target = target->next)
{
struct genomeRangeTree *grt = target->grt;
struct rbTree *targetTree = genomeRangeTreeFindRangeTree(grt, chrom->name);
if (targetTree != NULL)
{
for (iv = ivList; iv != NULL; iv = iv->next)
{
int overlap = rangeTreeOverlapSize(targetTree, iv->start, iv->end);
target->uniqOverlapBases += overlap;
target->overlapBases += overlap * iv->val;
}
}
}
long endOverlapTime = clock1000();
totalOverlapTime += endOverlapTime - startOverlapTime;
lmCleanup(&lm);
}
verbose(1, "totalBig %0.3f, totalOverlap %0.3f\n", 0.001*totalBigQueryTime, 0.001*totalOverlapTime);
/* Now loop through targets and save enrichment info to database */
struct target *target;
for (target = targetList; target != NULL; target = target->next)
{
struct cdwQaEnrich *enrich = enrichFromOverlaps(ef, vf, assembly, target,
target->overlapBases, target->uniqOverlapBases);
cdwQaEnrichSaveToDb(conn, enrich, "cdwQaEnrich", 128);
cdwQaEnrichFree(&enrich);
}
bbiChromInfoFreeList(&chromList);
bigWigFileClose(&bbi);
freez(&bigWigPath);
}
void doEnrichmentsFromBigBed(struct sqlConnection *conn,
struct cdwFile *ef, struct cdwValidFile *vf,
struct cdwAssembly *assembly, struct target *targetList)
/* Figure out enrichments from a bigBed file. */
{
/* Get path to bigBed, open it, and read all chromosomes. */
char *bigBedPath = cdwPathForFileId(conn, ef->id);
struct bbiFile *bbi = bigBedFileOpen(bigBedPath);
struct bbiChromInfo *chrom, *chromList = bbiChromList(bbi);
/* Do a pretty complex loop that just aims to set target->overlapBases and ->uniqOverlapBases
* for all targets. This is complicated by just wanting to keep one chromosome worth of
* bigBed data in memory. */
for (chrom = chromList; chrom != NULL; chrom = chrom->next)
{
/* Get list of intervals in bigBed for this chromosome, and feed it to a rangeTree. */
struct lm *lm = lmInit(0);
struct bigBedInterval *ivList = bigBedIntervalQuery(bbi, chrom->name, 0, chrom->size, 0, lm);
struct bigBedInterval *iv;
struct rbTree *bbTree = rangeTreeNew();
for (iv = ivList; iv != NULL; iv = iv->next)
rangeTreeAdd(bbTree, iv->start, iv->end);
struct range *bbRange, *bbRangeList = rangeTreeList(bbTree);
/* Loop through all targets adding overlaps from ivList and unique overlaps from bbRangeList */
struct target *target;
for (target = targetList; target != NULL; target = target->next)
{
if (target->skip)
continue;
struct genomeRangeTree *grt = target->grt;
struct rbTree *targetTree = genomeRangeTreeFindRangeTree(grt, chrom->name);
if (targetTree != NULL)
{
struct bigBedInterval *iv;
for (iv = ivList; iv != NULL; iv = iv->next)
{
int overlap = rangeTreeOverlapSize(targetTree, iv->start, iv->end);
target->overlapBases += overlap;
}
for (bbRange = bbRangeList; bbRange != NULL; bbRange = bbRange->next)
{
int overlap = rangeTreeOverlapSize(targetTree, bbRange->start, bbRange->end);
target->uniqOverlapBases += overlap;
}
}
}
rangeTreeFree(&bbTree);
lmCleanup(&lm);
}
/* Now loop through targets and save enrichment info to database */
struct target *target;
for (target = targetList; target != NULL; target = target->next)
{
if (target->skip)
continue;
struct cdwQaEnrich *enrich = enrichFromOverlaps(ef, vf, assembly, target,
target->overlapBases, target->uniqOverlapBases);
cdwQaEnrichSaveToDb(conn, enrich, "cdwQaEnrich", 128);
cdwQaEnrichFree(&enrich);
}
bbiChromInfoFreeList(&chromList);
bigBedFileClose(&bbi);
freez(&bigBedPath);
}
