def __init__(self,
filename,
count_A1=None,
iid=None,
sid=None,
pos=None,
skip_format_check=False
): #!!!document these new optionals. they are here
super(Bed, self).__init__()
self._ran_once = False
self._file_pointer = None
self.filename = filename
if count_A1 is None:
warnings.warn(
"'count_A1' was not set. For now it will default to 'False', but in the future it will default to 'True'",
FutureWarning)
count_A1 = False
self.count_A1 = count_A1
self.skip_format_check = skip_format_check
if iid is not None:
self._row = PstData._fixup_input(
iid, empty_creator=lambda ignore: np.empty([0, 2], dtype=str))
if sid is not None:
self._col = PstData._fixup_input(
sid, empty_creator=lambda ignore: np.empty([0], dtype=str))
if pos is not None:
self._col_property = PstData._fixup_input(
pos,
count=len(self._col),
empty_creator=lambda count: np.array([[np.nan, np.nan, np.nan]]
* count))
def _empty_inner(self, row, col, filename, row_property, col_property, order, dtype, val_shape):
self._ran_once = True
self._dtype = np.dtype(dtype)
self._order = order
row = PstData._fixup_input(row)
col = PstData._fixup_input(col)
row_property = PstData._fixup_input(row_property,count=len(row))
col_property = PstData._fixup_input(col_property,count=len(col))
with open(filename,'wb') as fp:
np.save(fp, np.array([_magic_number]))
np.save(fp, np.array(["pstmemmap"])) #name of file format
np.save(fp, np.array([2])) #file format version
np.save(fp, row)
np.save(fp, col)
np.save(fp, row_property)
np.save(fp, col_property)
np.save(fp, np.array([self._dtype]))
np.save(fp, np.array([self._order]))
np.save(fp, np.array([val_shape]))
self._offset = fp.tell()
logging.info("About to start allocating memmap '{0}'".format(filename))
shape = (len(row),len(col)) if val_shape is None else (len(row),len(col),val_shape)
val = np.memmap(filename, offset=self._offset, dtype=dtype, mode="r+", order=order, shape=shape)
logging.info("Finished allocating memmap '{0}'. Size is {1:,}".format(filename,os.path.getsize(filename)))
PstData.__init__(self,row,col,val,row_property,col_property,name="np.memmap('{0}')".format(filename))
def __init__(self, iid, test=None): #!!! add docs and test for test
super(Identity, self).__init__()
if test is None:
test = iid
if test is iid:
iid = PstData._fixup_input(
iid,
empty_creator=lambda ignore: np.empty([0, 2], dtype='str'),
dtype='str')
test = iid
else:
iid = PstData._fixup_input(
iid,
empty_creator=lambda ignore: np.empty([0, 2], dtype='str'),
dtype='str')
test = PstData._fixup_input(
test,
empty_creator=lambda ignore: np.empty([0, 2], dtype='str'),
dtype='str')
if len(iid) > 0:
self._row0 = iid
else:
self._row0 = self._empty
if len(test) > 0:
self._row1 = test
else:
self._row1 = self._empty
def __init__(self,
iid=None,
iid0=None,
iid1=None,
val=None,
name=None,
parent_string=None,
xp=None
): #!!!autodoc doesn't generate good doc for this constructor
#We don't have a 'super(KernelData, self).__init__()' here because KernelData takes full responsibility for initializing both its superclasses
xp = pstutil.array_module(xp)
self._val = None
#!!why does SnpData __init__ have a copy_inputs, but KernelData doesn't?
assert (iid is None) != (
iid0 is None and iid1 is None
), "Either 'iid' or both 'iid0' 'iid1' must be provided."
assert name is None or parent_string is None, "Can't set both 'name' and the deprecated 'parent_string'"
if parent_string is not None:
warnings.warn("'parent_string' is deprecated. Use 'name'",
DeprecationWarning)
if iid is not None:
self._row = PstData._fixup_input(
iid,
empty_creator=lambda ignore: np.empty([0, 2], dtype='str'),
dtype='str')
self._col = self._row
else:
self._row = PstData._fixup_input(
iid0,
empty_creator=lambda ignore: np.empty([0, 2], dtype='str'),
dtype='str')
self._col = PstData._fixup_input(
iid1,
empty_creator=lambda ignore: np.empty([0, 2], dtype='str'),
dtype='str')
self._row_property = PstData._fixup_input(
None,
count=len(self._row),
empty_creator=lambda count: np.empty([count, 0], dtype='str'),
dtype='str')
self._col_property = PstData._fixup_input(
None,
count=len(self._col),
empty_creator=lambda count: np.empty([count, 0], dtype='str'),
dtype='str')
self._val = PstData._fixup_input_val(
val,
row_count=len(self._row),
col_count=len(self._col),
empty_creator=lambda row_count, col_count: xp.empty(
[row_count, col_count], dtype=xp.float64),
xp=xp)
self._assert_iid0_iid1(check_val=True)
self._name = name or parent_string or ""
self._std_string_list = []
self._xp = xp
def __init__(self, iid, sid, val, pos=None, name=None, parent_string=None, copyinputs_function=None):
if parent_string is not None:
warnings.warn("'parent_string' is deprecated. Use 'name'", DeprecationWarning)
self._row = PstData._fixup_input(iid,empty_creator=lambda ignore:np.empty([0,2],dtype=str))
self._col = PstData._fixup_input(sid,empty_creator=lambda ignore:np.empty([0],dtype=str))
self._row_property = PstData._fixup_input(None,count=len(self._row),empty_creator=lambda count:np.empty([count,0],dtype=str))
self._col_property = PstData._fixup_input(pos,count=len(self._col),empty_creator=lambda count:np.array([[np.nan, np.nan, np.nan]]*count))
self.val = PstData._fixup_input_val(val,row_count=len(self._row),col_count=len(self._col),empty_creator=lambda row_count,col_count:np.empty([row_count,col_count],dtype=np.float64))
self._assert_iid_sid_pos()
self._name = name or parent_string or ""
self._std_string_list = []
def __init__(self, iid, sid, val, pos=None, name=None, copyinputs_function=None):
#We don't have a 'super(DistData, self).__init__()' here because DistData takes full responsibility for initializing both its superclasses
self._val = None
self._row = PstData._fixup_input(iid,empty_creator=lambda ignore:np.empty([0,2],dtype='str'),dtype='str')
self._col = PstData._fixup_input(sid,empty_creator=lambda ignore:np.empty([0],dtype='str'),dtype='str')
self._row_property = PstData._fixup_input(None,count=len(self._row),empty_creator=lambda count:np.empty([count,0],dtype='str'),dtype='str')
self._col_property = PstData._fixup_input(pos,count=len(self._col),empty_creator=lambda count:np.full([count, 3], np.nan))
self._val = PstData._fixup_input_val(val,row_count=len(self._row),col_count=len(self._col),empty_creator=lambda row_count,col_count:np.empty([row_count,col_count,3],dtype=np.float64))#!!!Replace empty with my FillNA method?
self._assert_iid_sid_pos(check_val=True)
self._name = name or ""
self._std_string_list = []
def __init__(self, iid, sid, val, pos=None, name=None, parent_string=None, copyinputs_function=None):
#We don't have a 'super(SnpData, self).__init__()' here because SnpData takes full responsiblity for initializing both its superclasses
self.val = None
if parent_string is not None:
warnings.warn("'parent_string' is deprecated. Use 'name'", DeprecationWarning)
self._row = PstData._fixup_input(iid,empty_creator=lambda ignore:np.empty([0,2],dtype='S'),dtype='S')
self._col = PstData._fixup_input(sid,empty_creator=lambda ignore:np.empty([0],dtype='S'),dtype='S')
self._row_property = PstData._fixup_input(None,count=len(self._row),empty_creator=lambda count:np.empty([count,0],dtype='S'),dtype='S')
self._col_property = PstData._fixup_input(pos,count=len(self._col),empty_creator=lambda count:np.array([[np.nan, np.nan, np.nan]]*count))
self.val = PstData._fixup_input_val(val,row_count=len(self._row),col_count=len(self._col),empty_creator=lambda row_count,col_count:np.empty([row_count,col_count],dtype=np.float64))
self._assert_iid_sid_pos()
self._name = name or parent_string or ""
self._std_string_list = []
def __init__(self, iid, sid, val, pos=None, name=None, parent_string=None, copyinputs_function=None):
#We don't have a 'super(SnpData, self).__init__()' here because SnpData takes full responsiblity for initializing both its superclasses
self.val = None
if parent_string is not None:
warnings.warn("'parent_string' is deprecated. Use 'name'", DeprecationWarning)
self._row = PstData._fixup_input(iid,empty_creator=lambda ignore:np.empty([0,2],dtype=str))
self._col = PstData._fixup_input(sid,empty_creator=lambda ignore:np.empty([0],dtype=str))
self._row_property = PstData._fixup_input(None,count=len(self._row),empty_creator=lambda count:np.empty([count,0],dtype=str))
self._col_property = PstData._fixup_input(pos,count=len(self._col),empty_creator=lambda count:np.array([[np.nan, np.nan, np.nan]]*count))
self.val = PstData._fixup_input_val(val,row_count=len(self._row),col_count=len(self._col),empty_creator=lambda row_count,col_count:np.empty([row_count,col_count],dtype=np.float64))
self._assert_iid_sid_pos()
self._name = name or parent_string or ""
self._std_string_list = []
def __init__(self,
iid,
sid,
val,
pos=None,
name=None,
parent_string=None,
copyinputs_function=None,
xp=None,
_require_float32_64=True):
#We don't have a 'super(SnpData, self).__init__()' here because SnpData takes full responsibility for initializing both its superclasses
xp = pstutil.array_module(xp)
self._val = None
if parent_string is not None:
warnings.warn("'parent_string' is deprecated. Use 'name'",
DeprecationWarning)
self._row = PstData._fixup_input(
iid,
empty_creator=lambda ignore: np.empty([0, 2], dtype='str'),
dtype='str')
self._col = PstData._fixup_input(
sid,
empty_creator=lambda ignore: np.empty([0], dtype='str'),
dtype='str')
self._row_property = PstData._fixup_input(
None,
count=len(self._row),
empty_creator=lambda count: np.empty([count, 0], dtype='str'),
dtype='str')
self._col_property = PstData._fixup_input(
pos,
count=len(self._col),
empty_creator=lambda count: np.full([count, 3], np.nan))
self._val = PstData._fixup_input_val(
val,
row_count=len(self._row),
col_count=len(self._col),
empty_creator=lambda row_count, col_count: np.empty(
[row_count, col_count], dtype=np.float64),
_require_float32_64=_require_float32_64,
xp=xp)
self._assert_iid_sid_pos(check_val=True)
self._name = name or parent_string or ""
self._std_string_list = []
self._xp = xp
def __init__(self, filename, count_A1=None, iid=None, sid=None, pos=None, skip_format_check=False): #!!!document these new optionals. they are here
super(Bed, self).__init__()
self._ran_once = False
self._file_pointer = None
self.filename = filename
if count_A1 is None:
warnings.warn("'count_A1' was not set. For now it will default to 'False', but in the future it will default to 'True'", FutureWarning)
count_A1 = False
self.count_A1 =count_A1
self.skip_format_check = skip_format_check
if iid is not None:
self._row = PstData._fixup_input(iid,empty_creator=lambda ignore:np.empty([0,2],dtype='S'),dtype='S')
if sid is not None:
self._col = PstData._fixup_input(sid,empty_creator=lambda ignore:np.empty([0],dtype='S'),dtype='S')
if pos is not None:
self._col_property = PstData._fixup_input(pos,count=len(self._col),empty_creator=lambda count:np.array([[np.nan, np.nan, np.nan]]*count))
def __init__(self, iid=None, iid0=None, iid1=None, val=None, name=None, parent_string=None): #!!!autodoc doesn't generate good doc for this constructor
#!!why does SnpData __init__ have a copy_inputs, but KernelData doesn't?
assert (iid is None) != (iid0 is None and iid1 is None), "Either 'iid' or both 'iid0' 'iid1' must be provided."
assert name is None or parent_string is None, "Can't set both 'name' and the deprecated 'parent_string'"
if parent_string is not None:
warnings.warn("'parent_string' is deprecated. Use 'name'", DeprecationWarning)
if iid is not None:
self._row = PstData._fixup_input(iid,empty_creator=lambda ignore:np.empty([0,2],dtype=str))
self._col = self._row
else:
self._row = PstData._fixup_input(iid0,empty_creator=lambda ignore:np.empty([0,2],dtype=str))
self._col = PstData._fixup_input(iid1,empty_creator=lambda ignore:np.empty([0,2],dtype=str))
self._row_property = PstData._fixup_input(None,count=len(self._row),empty_creator=lambda count:np.empty([count,0],dtype=str))
self._col_property = PstData._fixup_input(None,count=len(self._col),empty_creator=lambda count:np.empty([count,0],dtype=str))
self.val = PstData._fixup_input_val(val,row_count=len(self._row),col_count=len(self._col),empty_creator=lambda row_count,col_count:np.empty([row_count,col_count],dtype=np.float64))
self._assert_iid0_iid1()
self._name = name or parent_string or ""
self._std_string_list = []