A C library for reading/parsing local and remote bigWig and bigBed files. While Kent's source code is free to use for these purposes, it's really inappropriate as library code since it has the unfortunate habit of calling exit() whenever there's an error. If that's then used inside of something like python then the python interpreter gets killed. This library is aimed at resolving these sorts of issues and should also use more standard things like curl and has a friendlier license to boot.
Documentation is automatically generated by doxygen and can be found under docs/html or online here.
The only functions and structures that end users need to care about are in "bigWig.h". Below is a commented example. You can see the files under test/ for further examples.
#include "bigWig.h"
int main(int argc, char *argv[]) {
bigWigFile_t *fp = NULL;
bwOverlappingIntervals_t *intervals = NULL;
double *stats = NULL;
if(argc != 2) {
fprintf(stderr, "Usage: %s {file.bw|URL://path/file.bw}\n", argv[0]);
return 1;
}
//Initialize enough space to hold 128KiB (1<<17) of data at a time
if(bwInit(1<<17) != 0) {
fprintf(stderr, "Received an error in bwInit\n");
return 1;
}
//Open the local/remote file
fp = bwOpen(argv[1], NULL, "r");
if(!fp) {
fprintf(stderr, "An error occured while opening %s\n", argv[1]);
return 1;
}
//Get values in a range (0-based, half open) without NAs
intervals = bwGetValues(fp, "chr1", 10000000, 10000100, 0);
bwDestroyOverlappingIntervals(intervals); //Free allocated memory
//Get values in a range (0-based, half open) with NAs
intervals = bwGetValues(fp, "chr1", 10000000, 10000100, 1);
bwDestroyOverlappingIntervals(intervals); //Free allocated memory
//Get the full intervals that overlap
intervals = bwGetOverlappingIntervals(fp, "chr1", 10000000, 10000100);
bwDestroyOverlappingIntervals(intervals);
//Get an example statistic - standard deviation
//We want ~4 bins in the range
stats = bwStats(fp, "chr1", 10000000, 10000100, 4, dev);
if(stats) {
printf("chr1:10000000-10000100 std. dev.: %f %f %f %f\n", stats[0], stats[1], stats[2], stats[3]);
free(stats);
}
bwClose(fp);
bwCleanup();
return 0;
}N.B., creation of bigBed files is not supported (there are no plans to change this).
Below is an example of how to write bigWig files. You can also find this file under test/exampleWrite.c. Unlike with Kent's tools, you can create bigWig files entry by entry without needing an intermediate wiggle or bedGraph file. Entries in bigWig files are stored in blocks with each entry in a block referring to the same chromosome and having the same type, of which there are three (see the wiggle specification for more information on this).
#include "bigWig.h"
int main(int argc, char *argv[]) {
bigWigFile_t *fp = NULL;
char *chroms[] = {"1", "2"};
char *chromsUse[] = {"1", "1", "1"};
uint32_t chrLens[] = {1000000, 1500000};
uint32_t starts[] = {0, 100, 125,
200, 220, 230,
500, 600, 625,
700, 800, 850};
uint32_t ends[] = {5, 120, 126,
205, 226, 231};
float values[] = {0.0f, 1.0f, 200.0f,
-2.0f, 150.0f, 25.0f,
0.0f, 1.0f, 200.0f,
-2.0f, 150.0f, 25.0f,
-5.0f, -20.0f, 25.0f,
-5.0f, -20.0f, 25.0f};
if(bwInit(1<<17) != 0) {
fprintf(stderr, "Received an error in bwInit\n");
return 1;
}
fp = bwOpen("example_output.bw", NULL, "w");
if(!fp) {
fprintf(stderr, "An error occurred while opening example_output.bw for writingn\n");
return 1;
}
//Allow up to 10 zoom levels, though fewer will be used in practice
if(bwCreateHdr(fp, 10)) goto error;
//Create the chromosome lists
fp->cl = bwCreateChromList(chroms, chrLens, 2);
if(!fp->cl) goto error;
//Write the header
if(bwWriteHdr(fp)) goto error;
//Some example bedGraph-like entries
if(bwAddIntervals(fp, chromsUse, starts, ends, values, 3)) goto error;
//We can continue appending similarly formatted entries
//N.B. you can't append a different chromosome (those always go into different
if(bwAppendIntervals(fp, starts+3, ends+3, values+3, 3)) goto error;
//Add a new block of entries with a span. Since bwAdd/AppendIntervals was just used we MUST create a new block
if(bwAddIntervalSpans(fp, "1", starts+6, 20, values+6, 3)) goto error;
//We can continue appending similarly formatted entries
if(bwAppendIntervalSpans(fp, starts+9, values+9, 3)) goto error;
//Add a new block of fixed-step entries
if(bwAddIntervalSpanSteps(fp, "1", 900, 20, 30, values+12, 3)) goto error;
//The start is then 760, since that's where the previous step ended
if(bwAppendIntervalSpanSteps(fp, values+15, 3)) goto error;
//Add a new chromosome
chromsUse[0] = "2";
chromsUse[1] = "2";
chromsUse[2] = "2";
if(bwAddIntervals(fp, chromsUse, starts, ends, values, 3)) goto error;
//Closing the file causes the zoom levels to be created
bwClose(fp);
bwCleanup();
return 0;
error:
fprintf(stderr, "Received an error somewhere!\n");
bwClose(fp);
bwCleanup();
return 1;
}As of version 0.3.0, this library supports accessing bigBed files, which are related to bigWig files. Applications that need to support both bigWig and bigBed input can use the bwIsBigWig and bbIsBigBed functions to determine if their inputs are bigWig/bigBed files:
...code...
if(bwIsBigWig(input_file_name, NULL)) {
//do something
} else if(bbIsBigBed(input_file_name, NULL)) {
//do something else
} else {
//handle unknown input
}Note that these two functions rely on the "magic number" at the beginning of each file, which differs between bigWig and bigBed files.
Support for accessing bigBed files was added in version 0.3.0. The function names used for accessing bigBed files are similar to those used for bigWig files.
Function | Use
--- | ---
bbOpen | Opens a bigBed file
bbGetSQL | Returns the SQL string (if it exists) in a bigBed file
bbGetOverlappingEntries | Returns all entries overlapping an interval (either with or without their associated strings
bbDestroyOverlappingEntries | Free memory allocated by the above command
Other functions, such as bwClose and bwInit, are shared between bigWig and bigBed files. See test/testBigBed.c for a full example.
Inside bigBed files, entries are stored as chromosome, start, and end coordinates with an (optional) associated string. For example, a "bedRNAElements" file from Encode has name, score, strand, "level", "significance", and "score2" values associated with each entry. These are stored inside the bigBed files as a single tab-separated character vector (char *), which makes parsing difficult. The names of the various fields inside of bigBed files is stored as an SQL string, for example:
table RnaElements
"BED6 + 3 scores for RNA Elements data "
(
string chrom; "Reference sequence chromosome or scaffold"
uint chromStart; "Start position in chromosome"
uint chromEnd; "End position in chromosome"
string name; "Name of item"
uint score; "Normalized score from 0-1000"
char[1] strand; "+ or - or . for unknown"
float level; "Expression level such as RPKM or FPKM. Set to -1 for no data."
float signif; "Statistical significance such as IDR. Set to -1 for no data."
uint score2; "Additional measurement/count e.g. number of reads. Set to 0 for no data."
)
Entries will then be of the form (one per line):
59426 115 - 0.021 0.48 218
51 209 + 0.071 0.74 130
52 170 + 0.045 0.61 171
59433 178 - 0.049 0.34 296
53 156 + 0.038 0.19 593
59436 186 - 0.054 0.15 1010
59437 506 - 1.560 0.00 430611
Note that chromosome and start/end intervals are stored separately, so there's no need to parse them out of string. libBigWig can return these entries, either with or without the above associated strings. Parsing these string is left to the application requiring them and is currently outside the scope of this library.
Sometimes it is desirable to request a large number of intervals from a bigWig file or entries from a bigBed file, but not hold them all in memory at once (e.g., due to saving memory). To support this, libBigWig (since version 0.3.0) supports two kinds of iterators. The general process of using iterators is: (1) iterator creation, (2) traversal, and finally (3) iterator destruction. Only iterator creation differs between bigWig and bigBed files.
Importantly, iterators return results by one or more blocks. This is for convenience, since bigWig intervals and bigBed entries are stored in together in fixed-size groups, called blocks. The number of blocks of entries returned, therefore, is an option that can be specified to balance performance and memory usage.
For bigwig files, iterators are created with the bwOverlappingIntervalsIterator(). This function takes chromosomal bounds (chromosome name, start, and end position) as well as a number of blocks. The equivalent function for bigBed files is bbOverlappingEntriesIterator(), which additionally takes a withString argutment, which dictates whether the returned entries include the associated string values or not.
Each of the aforementioned files returns a pointer to a bwOverlapIterator_t object. The only important parts of this structure for end users are the following members: entries, intervals, and data. entries is a pointer to a bbOverlappingEntries_t object, or NULL if a bigWig file is being used. Likewise, intervals is a pointer to a bwOverlappingIntervals_t object, or NULL if a bigBed file is being used. data is a special pointer, used to signify the end of iteration. Thus, when data is a NULL pointer, iteration has ended.
Regardless of whether a bigWig or bigBed file is being used, the bwIteratorNext() function will free currently used memory and load the appropriate intervals or entries for the next block(s). On error, this will return a NULL pointer (memory is already internally freed in this case).
bwOverlapIterator_t objects MUST be destroyed after use. This can be done with the bwIteratorDestroy() function.
A full example is provided in tests/testIterator.c, but a small example of iterating over all bigWig intervals in chr1:0-10000000 in chunks of 5 blocks follows:
iter = bwOverlappingIntervalsIterator(fp, "chr1", 0, 10000000, 5);
while(iter->data) {
//Do stuff with iter->intervals
iter = bwIteratorNext(iter);
}
bwIteratorDestroy(iter);The results of min, max, and mean should be the same as those from BigWigSummary. stdev and coverage, however, may differ due to Kent's tools producing incorrect results (at least for coverage, though the same appears to be the case for stdev). The sum method doesn't exist in Kent's tools, so note that if zoom levels are used, that it will multiply the block average by the lesser of the number of bases covered in the block and the number of bases in a block overlapping the desired region.
There are currently two python interfaces that make use of libBigWig: pyBigWig by me and bw-python by Brent Pederson. Those interested are encouraged to give both a try!
If you want to compile without remote file access (e.g., you don't have curl installed), then you can append -DNOCURL to the CFLAGS line in the Makefile. You will also need to remove -lcurl from the LIBS line.
If you are building libBigWig using CMake you can instead pass -DWITH_CURL=OFF when calling CMake at configuration time.