bw-python

Python wrapper to Devon Ryan's bigwig library using cffi. It is licensed under MIT (see LICENSE).

Installation

Currently you will need libcurl to be installed. To compile the source files type (to use from this directory)

python setup.py develop

Example

>>> from bw import BigWig
>>> b = BigWig("libBigWig/test/test.bw")
>>> b
BigWig('libBigWig/test/test.bw')
>>> for interval in b("1", 0, 99):
...     interval
Interval(chrom='1', start=0, end=1, value=0.10000000149011612)
Interval(chrom='1', start=1, end=2, value=0.20000000298023224)
Interval(chrom='1', start=2, end=3, value=0.30000001192092896)

# get an array.array() for large numbers of values.
# default is to return nan's for missing values
>>> b.values("1", 0, 9)
array('f', [0.10000000149011612, 0.20000000298023224, 0.30000001192092896, nan, nan, nan, nan, nan, nan])

# we can also get missing, but won't know the base it's associated with. 
>>> b.values("1", 0, 9, False)
array('f', [0.10000000149011612, 0.20000000298023224, 0.30000001192092896])


# stats are ("mean", "std", "max", "min", "coverage")
>>> b.stats("1", 0, 9, stat="mean")
0.2000000054637591

>>> b.stats("1", 0, 9, stat="std")
0.10000000521540645

>>> b.stats("1", 0, 4, stat="coverage")
0.75
>>> b.stats("1", 0, 4, stat="coverage", nBins=2)
array('d', [1.0, 0.5])

# get the chromosomes and lengths as a list of tuples:
>>> b.chroms
[('1', 195471971), ('10', 130694993)]

>>> b.close()

An array.array can be turned into a numpy array with np.frombuffer(a, dtype='f')

Unit Tests

The unit tests are located in the bw/tests/ directory. You can run them via pytest.