void bigWigFileCreate(
char *inName, /* Input file in ascii wiggle format. */
char *chromSizes, /* Two column tab-separated file: <chromosome> <size>. */
int blockSize, /* Number of items to bundle in r-tree. 1024 is good. */
int itemsPerSlot, /* Number of items in lowest level of tree. 512 is good. */
boolean clipDontDie, /* If TRUE then clip items off end of chrom rather than dying. */
boolean compress, /* If TRUE then compress data. */
boolean keepAllChromosomes, /* If TRUE then store all chromosomes in chromosomal b-tree. */
boolean fixedSummaries, /* If TRUE then impose fixed summary levels. */
char *outName)
/* Convert ascii format wig file (in fixedStep, variableStep or bedGraph format)
* to binary big wig format. */
{
/* This code needs to agree with code in two other places currently - bigBedFileCreate,
* and bbiFileOpen. I'm thinking of refactoring to share at least between
* bigBedFileCreate and bigWigFileCreate. It'd be great so it could be structured
* so that it could send the input in one chromosome at a time, and send in the zoom
* stuff only after all the chromosomes are done. This'd potentially reduce the memory
* footprint by a factor of 2 or 4. Still, for now it works. -JK */
struct hash *chromSizeHash = bbiChromSizesFromFile(chromSizes);
struct lm *lm = lmInit(0);
struct bwgSection *sectionList = bwgParseWig(inName, clipDontDie, chromSizeHash, itemsPerSlot, lm);
if (sectionList == NULL)
errAbort("%s is empty of data", inName);
bwgCreate(sectionList, chromSizeHash, blockSize, itemsPerSlot, compress, keepAllChromosomes, fixedSummaries, outName);
lmCleanup(&lm);
}
void writeBw(char *inName, char *outName, struct hash *chromSizeHash)
/* copied from bwtool_shared.c */
{
struct lm *lm = lmInit(0);
struct bwgSection *sectionList = bwgParseWig(inName, TRUE, chromSizeHash, 1024, lm);
if (sectionList == NULL)
errAbort("%s is empty of data", inName);
bwgCreate(sectionList, chromSizeHash, 256, 1024, TRUE, outName);
lmCleanup(&lm);
}
SEXP BWGFile_fromWIG(SEXP r_infile, SEXP r_seqlengths, SEXP r_outfile) {
pushRHandlers();
struct lm *lm = lmInit(0);
struct hash *lenHash = createIntHash(r_seqlengths);
struct bwgSection *sections =
bwgParseWig((char *)CHAR(asChar(r_infile)), FALSE, lenHash, itemsPerSlot,
lm);
bwgCreate(sections, lenHash, blockSize, itemsPerSlot, TRUE, TRUE, TRUE,
(char *)CHAR(asChar(r_outfile)));
lmCleanup(&lm);
freeHash(&lenHash);
popRHandlers();
return r_outfile;
}
/* --- .Call ENTRY POINT --- */
SEXP BWGSectionList_write(SEXP r_sections, SEXP r_seqlengths, SEXP r_compress,
SEXP r_file)
{
struct bwgSection *sections = NULL;
struct hash *lenHash = createIntHash(r_seqlengths);
if (r_sections != R_NilValue) {
sections = R_ExternalPtrAddr(r_sections);
slReverse(§ions);
}
pushRHandlers();
bwgCreate(sections, lenHash, blockSize, itemsPerSlot, asLogical(r_compress),
(char *)CHAR(asChar(r_file)));
freeHash(&lenHash);
popRHandlers();
return r_file;
}
void bigWigFileCreate2(
char *inName, /* Input file in ascii wiggle format. */
struct hash *chromSizeHash, /* chrom size hash. */
char *outName)
/* Convert ascii format wig file (in fixedStep, variableStep or bedGraph format)
* to binary big wig format. */
{
/* This code needs to agree with code in two other places currently - bigBedFileCreate,
* and bbiFileOpen. I'm thinking of refactoring to share at least between
* bigBedFileCreate and bigWigFileCreate. It'd be great so it could be structured
* so that it could send the input in one chromosome at a time, and send in the zoom
* stuff only after all the chromosomes are done. This'd potentially reduce the memory
* footprint by a factor of 2 or 4. Still, for now it works. -JK */
struct lm *lm = lmInit(0);
struct bwgSection *sectionList = bwgParseWig(inName, clipDontDie, chromSizeHash, itemsPerSlot, lm);
if (sectionList == NULL)
errAbort("%s is empty of data", inName);
bwgCreate(sectionList, chromSizeHash, blockSize, itemsPerSlot, compress, outName);
lmCleanup(&lm);
}
