def test_setslice():
s = SparseArray(100, 0)
s[1] = 2
s[99] = 2
s[5:8] = 3
assert len(s.container) == 5
assert s.keys() == [1, 5, 6, 7, 99]
assert s.values() == [2, 3, 3, 3, 2]
s = SparseArray(100, 0)
s[1] = 2
s[99] = 2
s[5:8] = [3,3,3]
assert len(s.container) == 5
assert s.keys() == [1, 5, 6, 7, 99]
assert s.values() == [2, 3, 3, 3, 2]
s = SparseArray(100, 0)
t = SparseArray(100,0)
t[5:10] = 1
s[5:10] = t[5:10]
assert len(t.container) == 5
assert t.values() == [1]*5
assert t.keys() == [5, 6, 7, 8, 9]
s = SparseArray(10, 0)
s[2:5] = SparseArray.from_dense([1]*10, 0)[2:5]
assert s.tolist() == [0,0,1,1,1,0,0,0,0,0]
s = SparseArray(100,0)
t = SparseArray(100,0)
t[20:50] = 1
s[20:50] = t[20:50]
assert all(sv == 1 for sv in s[20:50].tolist())
def test_setitem():
s = SparseArray(100, 0)
# Add a first element
s[10] = 1
# Add something before the first non-sparse element
s[5] = 1
# Add something after the last non-sparse element
s[20] = 1
# Add sometihng in between
s[15] = 2
assert s.keys() == [5, 10, 15, 20]
assert s.values() == [1, 1, 2, 1]
s[15] = 0
assert s.keys() == [5, 10, 20]
assert s.values() == [1, 1, 1]
s[20] = 2
assert s.keys() == [5, 10, 20]
assert s.values() == [1, 1, 2]
class SparseAlleles(AlleleContainer):
'''
An object representing a set of haploid genotypes efficiently by
storing allele differences from a reference. Useful for manipulating
genotypes from sequence data (e.g. VCF files)
In the interest of conserving memory for sequencing data, all alleles must
be represented by a signed 8-bit integer (i.e. between -128 and 127).
Negative values are interpreted as missing.
'''
def __init__(self, data=None, refcode=0, missingcode=-1, size=None, template=None):
self.template = template
self.refcode = refcode
if data is None:
if template is None and size is None:
raise ValueError('No template or size')
elif template is not None and size is None:
size = self.template.nmark()
self.container = SparseArray(size, refcode)
return
if type(data) is SparseArray:
raise NotImplementedError
else:
if not isinstance(data, np.ndarray):
data = np.array(data)
self.container = SparseArray.from_dense(data, refcode)
self.size = len(self.container)
def __getitem__(self, key):
return self.container[key]
def __setitem__(self, key, value):
self.container[key] = value
def keys(self):
return self.container.keys()
def values(self):
return self.container.values()
@property
def missingcode(self):
return -1
@property
def missing(self):
" Returns a numpy array indicating which markers have missing data "
missingindices = [i for i,v in self.container.items() if v == self.missingcode]
base = np.zeros(self.size, dtype=np.bool_)
base[missingindices] = 1
return base
def __eq__(self, other):
if type(other) is SparseAlleles:
return self.container == other.container
else:
return self.container == other
def __ne__(self, other):
if type(other) is SparseAlleles:
return self.container != other.container
else:
return self.container != other
def nmark(self):
'''
Return the number of markers (both reference and non-reference)
represented by the SparseAlleles object
'''
return self.container.size
def todense(self):
dense = Alleles(self.container.tolist(), template=self.template)
return dense
def empty_like(self):
output = SparseAlleles(template=self.template,
missingcode=self.missingcode,
refcode=self.refcode, size=self.nmark())
return output
def copy_span(self, template, copy_start, copy_stop):
if isinstance(template, SparseAlleles):
self.container[copy_start:copy_stop] = template.container[copy_start:copy_stop]
else:
self.container = template[copy_start:copy_stop]
def copy(self):
outp = self.empty_like()
outp.container = self.container.copy()
return outp
@staticmethod
def empty(reference=None, template=None, missingcode=''):
out = SparseAlleles(size, template=template, missingcode=missingcode)
return out
