def __init__(self, chroms, # pylint: disable=too-many-locals
stranded=True,
conditions=None,
typecode='d',
datatags=None,
resolution=1,
order=1,
store_whole_genome=True,
cache=True,
overwrite=False, loader=None, loader_args=None):
super(NPGenomicArray, self).__init__(stranded, conditions, typecode,
resolution,
order, store_whole_genome)
if stranded:
datatags = datatags + ['stranded'] if datatags else ['stranded']
memmap_dir = _get_output_data_location(datatags)
filename = 'storage.npz'
if cache and not os.path.exists(memmap_dir):
os.makedirs(memmap_dir)
if cache and not os.path.exists(os.path.join(memmap_dir, filename)) \
or overwrite or not cache:
data = {chrom: numpy.zeros(shape=(int(numpy.ceil(chroms[chrom] / self.resolution)),
2 if stranded else 1,
len(self.condition)),
dtype=self.typecode) for chrom in chroms}
self.handle = data
# invoke the loader
if loader:
loader(self, *loader_args)
condition = [numpy.string_(x) for x in self.condition]
names = [x for x in data]
data['conditions'] = condition
data['order'] = order
data['resolution'] = resolution
if cache:
numpy.savez(os.path.join(memmap_dir, filename), **data)
if cache:
print('reload {}'.format(os.path.join(memmap_dir, filename)))
data = numpy.load(os.path.join(memmap_dir, filename))
names = [x for x in data.files if x not in ['conditions', 'order', 'resolution']]
condition = data['conditions']
order = data['order']
resolution = data['resolution']
# here we get either the freshly loaded data or the reloaded
# data from numpy.load.
self.handle = {key: data[key] for key in names}
self.condition = condition
self.resolution = resolution
self.order = order
def _get_cachefile(cachestr, tags, fileending):
""" Determine cache file location """
filename = None
if cachestr is not None:
memmap_dir = _get_output_data_location(tags)
if not os.path.exists(memmap_dir):
os.makedirs(memmap_dir)
filename = str(cachestr) + fileending
filename = os.path.join(memmap_dir, filename)
return filename
return None
def __init__(self, chroms, # pylint: disable=too-many-locals
stranded=True,
conditions=None,
typecode='d',
datatags=None,
resolution=1,
order=1,
store_whole_genome=True,
cache=True,
overwrite=False, loader=None, loader_args=None):
super(HDF5GenomicArray, self).__init__(stranded, conditions, typecode,
resolution,
order, store_whole_genome)
if not cache:
raise ValueError('HDF5 format requires cache=True')
if stranded:
datatags = datatags + ['stranded'] if datatags else ['stranded']
memmap_dir = _get_output_data_location(datatags)
filename = 'storage.h5'
if not os.path.exists(memmap_dir):
os.makedirs(memmap_dir)
if not os.path.exists(os.path.join(memmap_dir, filename)) or overwrite:
self.handle = h5py.File(os.path.join(memmap_dir, filename), 'w')
for chrom in chroms:
shape = (int(numpy.ceil(chroms[chrom] / self.resolution)),
2 if stranded else 1, len(self.condition))
self.handle.create_dataset(chrom, shape,
dtype=self.typecode, compression='gzip',
data=numpy.zeros(shape, dtype=self.typecode))
self.handle.attrs['conditions'] = [numpy.string_(x) for x in self.condition]
self.handle.attrs['order'] = self.order
self.handle.attrs['resolution'] = self.resolution
# invoke the loader
if loader:
loader(self, *loader_args)
self.handle.close()
print('reload {}'.format(os.path.join(memmap_dir, filename)))
self.handle = h5py.File(os.path.join(memmap_dir, filename), 'r',
driver='stdio')
self.condition = self.handle.attrs['conditions']
self.order = self.handle.attrs['order']
self.resolution = self.handle.attrs['resolution']
def __init__(
self,
chroms, # pylint: disable=too-many-locals
stranded=True,
conditions=None,
typecode='d',
datatags=None,
resolution=1,
order=1,
store_whole_genome=True,
cache=True,
overwrite=False,
loader=None,
collapser=None):
super(SparseGenomicArray,
self).__init__(stranded, conditions, typecode, resolution, order,
store_whole_genome, collapser)
if stranded:
datatags = datatags + ['stranded'] if datatags else ['stranded']
memmap_dir = _get_output_data_location(datatags)
filename = 'sparse.npz'
if not os.path.exists(memmap_dir):
os.makedirs(memmap_dir)
if cache and not os.path.exists(os.path.join(memmap_dir, filename)) \
or overwrite or not cache:
data = {
chrom: sparse.dok_matrix(
(_get_iv_length(chroms[chrom], self.resolution),
(2 if stranded else 1) * len(self.condition)),
dtype=self.typecode)
for chrom in chroms
}
self.handle = data
# invoke the loader
if loader:
loader(self)
data = self.handle
data = {chrom: data[chrom].tocoo() for chrom in data}
condition = [np.string_(x) for x in self.condition]
names = [x for x in data]
storage = {chrom: np.column_stack([data[chrom].data,
data[chrom].row,
data[chrom].col]) \
for chrom in data}
storage.update({'shape.'+chrom: \
np.asarray(data[chrom].shape) for chrom in data})
storage['conditions'] = condition
storage['order'] = order
storage['resolution'] = resolution if resolution is not None else 0
if cache:
np.savez(os.path.join(memmap_dir, filename), **storage)
if cache:
print('reload {}'.format(os.path.join(memmap_dir, filename)))
storage = np.load(os.path.join(memmap_dir, filename))
names = [
x for x in storage.files
if x not in ['conditions', 'order', 'resolution']
and x[:6] != 'shape.'
]
condition = storage['conditions']
order = storage['order']
resolution = storage[
'resolution'] if storage['resolution'] > 0 else None
self.handle = {
key: sparse.coo_matrix(
(storage[key][:, 0], (storage[key][:, 1].astype('int'),
storage[key][:, 2].astype('int'))),
shape=tuple(storage['shape.' + key])).tocsr()
for key in names
}
self.condition = condition
self.resolution = resolution
self.order = order