def _from_java(jtt):
return tmatrix(dtype(jtt.globalType().toString()),
dtype(jtt.colType().toString()),
jiterable_to_list(jtt.colKey()),
dtype(jtt.rowType().toString()),
jiterable_to_list(jtt.rowKey()),
dtype(jtt.entryType().toString()))
def _from_java(jtt):
return tmatrix(
dtype(jtt.globalType().toString()),
dtype(jtt.colType().toString()),
jiterable_to_list(jtt.colKey()),
dtype(jtt.rowType().toString()),
jiterable_to_list(jtt.rowKey()),
dtype(jtt.entryType().toString()))
def _from_json(json):
return tmatrix(
dtype(json['global']),
dtype(json['col']),
json['col_key'],
dtype(json['row']),
json['row_key'],
dtype(json['entry']))
def register_reference_genome_functions(rg):
from hail.expr.types import dtype
register_session_function(f"isValidContig({rg})", (dtype("str"),), dtype("bool"))
register_session_function(f"isValidLocus({rg})", (dtype("str"),dtype("int32"),), dtype("bool"))
register_session_function(f"contigLength({rg})", (dtype("str"),), dtype("int32"))
register_session_function(f"getReferenceSequenceFromValidLocus({rg})", (dtype("str"),dtype("int32"),dtype("int32"),dtype("int32"),), dtype("str"))
register_session_function(f"getReferenceSequence({rg})", (dtype("str"),dtype("int32"),dtype("int32"),dtype("int32"),), dtype("str"))
def regression_test_type(test):
glm_fit_schema = dtype('struct{n_iterations:int32,converged:bool,exploded:bool}')
if test == 'wald':
return dtype(f'struct{{beta:float64,standard_error:float64,z_stat:float64,p_value:float64,fit:{glm_fit_schema}}}')
elif test == 'lrt':
return dtype(f'struct{{beta:float64,chi_sq_stat:float64,p_value:float64,fit:{glm_fit_schema}}}')
elif test == 'score':
return dtype('struct{chi_sq_stat:float64,p_value:float64}')
else:
assert test == 'firth', test
return dtype(f'struct{{beta:float64,chi_sq_stat:float64,p_value:float64,fit:{glm_fit_schema}}}')
def regression_test_type(test):
glm_fit_schema = dtype('struct{n_iterations:int32,converged:bool,exploded:bool}')
if test == 'wald':
return dtype(
f'struct{{beta:float64,standard_error:float64,z_stat:float64,p_value:float64,fit:{glm_fit_schema}}}')
elif test == 'lrt':
return dtype(f'struct{{beta:float64,chi_sq_stat:float64,p_value:float64,fit:{glm_fit_schema}}}')
elif test == 'score':
return dtype('struct{chi_sq_stat:float64,p_value:float64}')
else:
assert test == 'firth', test
return dtype(f'struct{{beta:float64,chi_sq_stat:float64,p_value:float64,fit:{glm_fit_schema}}}')
def _compute_type(self):
name = self.config['name']
child_typ = self.child.typ
if name == 'LinearRegressionRowsChained':
pass_through = self.config['passThrough']
chained_schema = hl.dtype('struct{n:array<int32>,sum_x:array<float64>,y_transpose_x:array<array<float64>>,beta:array<array<float64>>,standard_error:array<array<float64>>,t_stat:array<array<float64>>,p_value:array<array<float64>>}')
self._type = hl.ttable(
child_typ.global_type,
(child_typ.row_key_type
._insert_fields(**{f: child_typ.row_type[f] for f in pass_through})
._concat(chained_schema)),
child_typ.row_key)
elif name == 'LinearRegressionRowsSingle':
pass_through = self.config['passThrough']
chained_schema = hl.dtype('struct{n:int32,sum_x:float64,y_transpose_x:array<float64>,beta:array<float64>,standard_error:array<float64>,t_stat:array<float64>,p_value:array<float64>}')
self._type = hl.ttable(
child_typ.global_type,
(child_typ.row_key_type
._insert_fields(**{f: child_typ.row_type[f] for f in pass_through})
._concat(chained_schema)),
child_typ.row_key)
elif name == 'LogisticRegression':
pass_through = self.config['passThrough']
logreg_type = hl.tstruct(logistic_regression=hl.tarray(regression_test_type(self.config['test'])))
self._type = hl.ttable(
child_typ.global_type,
(child_typ.row_key_type
._insert_fields(**{f: child_typ.row_type[f] for f in pass_through})
._concat(logreg_type)),
child_typ.row_key)
elif name == 'PoissonRegression':
pass_through = self.config['passThrough']
poisreg_type = regression_test_type(self.config['test'])
self._type = hl.ttable(
child_typ.global_type,
(child_typ.row_key_type
._insert_fields(**{f: child_typ.row_type[f] for f in pass_through})
._concat(poisreg_type)),
child_typ.row_key)
elif name == 'Skat':
key_field = self.config['keyField']
key_type = child_typ.row_type[key_field]
skat_type = hl.dtype(f'struct{{id:{key_type},size:int32,q_stat:float64,p_value:float64,fault:int32}}')
self._type = hl.ttable(
hl.tstruct(),
skat_type,
['id'])
elif name == 'PCA':
self._type = hl.ttable(
hl.tstruct(eigenvalues=hl.tarray(hl.tfloat64),
scores=hl.tarray(child_typ.col_key_type._insert_field('scores', hl.tarray(hl.tfloat64)))),
child_typ.row_key_type._insert_field('loadings', dtype('array<float64>')),
child_typ.row_key)
else:
assert name == 'LocalLDPrune', name
self._type = hl.ttable(
hl.tstruct(),
child_typ.row_key_type._insert_fields(mean=hl.tfloat64, centered_length_rec=hl.tfloat64),
list(child_typ.row_key))
async def _async_value_type(self, ir):
async def inputs(infile, _):
await write_int(infile, ServiceBackend.VALUE_TYPE)
await write_str(infile, tmp_dir())
await write_str(infile, self.billing_project)
await write_str(infile, self.remote_tmpdir)
await write_str(infile, self.render(ir))
_, resp, _ = await self._rpc('value_type(...)', inputs)
return dtype(resp)
def execute(self, ir):
code = self._render(ir)
resp = requests.post(f'{self.url}/execute', json=code)
resp.raise_for_status()
resp_json = resp.json()
typ = dtype(resp_json['type'])
result = resp_json['value']
return typ._from_json(result)
async def _async_execute_untimed(self, ir):
token = secret_alnum_string()
with TemporaryDirectory(ensure_exists=False) as dir:
async def create_inputs():
with self.fs.open(dir + '/in', 'wb') as infile:
write_int(infile, ServiceBackend.EXECUTE)
write_str(infile, tmp_dir())
write_str(infile, self.billing_project)
write_str(infile, self.bucket)
write_str(infile, self.render(ir))
write_str(infile, token)
async def create_batch():
batch_attributes = self.batch_attributes
if 'name' not in batch_attributes:
batch_attributes = {**batch_attributes, 'name': 'execute(...)'}
bb = self.async_bc.create_batch(token=token, attributes=batch_attributes)
j = bb.create_jvm_job([
'is.hail.backend.service.ServiceBackendSocketAPI2',
os.environ['HAIL_SHA'],
os.environ['HAIL_JAR_URL'],
batch_attributes['name'],
dir + '/in',
dir + '/out',
], mount_tokens=True)
return (j, await bb.submit(disable_progress_bar=self.disable_progress_bar))
_, (j, b) = await asyncio.gather(create_inputs(), create_batch())
status = await b.wait(disable_progress_bar=self.disable_progress_bar)
if status['n_succeeded'] != 1:
raise ValueError(f'batch failed {status} {await j.log()}')
with self.fs.open(dir + '/out', 'rb') as outfile:
success = read_bool(outfile)
if success:
s = read_str(outfile)
try:
resp = json.loads(s)
except json.decoder.JSONDecodeError as err:
raise ValueError(f'could not decode {s}') from err
else:
jstacktrace = read_str(outfile)
raise FatalError(jstacktrace)
typ = dtype(resp['type'])
if typ == tvoid:
x = None
else:
x = typ._convert_from_json_na(resp['value'])
return x
def execute(self, ir):
code = self._render(ir)
resp = requests.post(f'{self.url}/execute', json=code)
if resp.status_code == 400:
resp_json = resp.json()
raise FatalError(resp_json['message'])
resp.raise_for_status()
resp_json = resp.json()
typ = dtype(resp_json['type'])
result = resp_json['value']
return typ._from_json(result)
def execute(self, ir, timed=False):
resp = self.socket.request('execute',
code=self._render(ir),
billing_project=self._billing_project,
bucket=self._bucket)
typ = dtype(resp['type'])
if typ == tvoid:
value = None
else:
value = typ._convert_from_json_na(resp['value'])
# FIXME put back timings
return (value, None) if timed else value
def execute(self, ir):
r = Renderer(stop_at_jir=True)
code = r(ir)
assert len(r.jirs) == 0
resp = requests.post(f'http://hail-apiserver:5000/execute', json=code)
resp.raise_for_status()
resp_json = resp.json()
typ = dtype(resp_json['type'])
result = resp_json['value']
return typ._from_json(result)
def execute(self, ir):
r = Renderer(stop_at_jir=True)
code = r(ir)
assert len(r.jirs) == 0
resp = requests.post(f'http://hail-apiserver:5000/execute', json=code)
resp.raise_for_status()
resp_json = resp.json()
typ = dtype(resp_json['type'])
result = resp_json['value']
return typ._from_json(result)
async def _async_execute(self, ir, timed=False):
async def inputs(infile, token):
await write_int(infile, ServiceBackend.EXECUTE)
await write_str(infile, tmp_dir())
await write_str(infile, self.billing_project)
await write_str(infile, self.remote_tmpdir)
await write_str(infile, self.render(ir))
await write_str(infile, token)
_, resp, timings = await self._rpc('execute(...)', inputs)
typ = dtype(resp['type'])
converted_value = typ._convert_from_json_na(resp['value'])
if timed:
return converted_value, timings
return converted_value
def execute(self, ir, timed=False):
code = self._render(ir)
resp = requests.post(f'{self.url}/execute', json=code, cookies=self.cookies)
if resp.status_code == 400:
resp_json = resp.json()
raise FatalError(resp_json['message'])
resp.raise_for_status()
resp_json = resp.json()
typ = dtype(resp_json['type'])
result = json.loads(resp_json['result'])
value = typ._from_json(result['value'])
timings = result['timings']
return (value, timings) if timed else value
def execute(self, ir, timed=False):
code = self._render(ir)
resp = retry_response_returning_functions(requests.post,
f'{self.url}/execute',
json=code,
headers=self.headers)
if resp.status_code == 400 or resp.status_code == 500:
raise FatalError(resp.text)
resp.raise_for_status()
resp_json = resp.json()
typ = dtype(resp_json['type'])
value = typ._convert_from_json_na(resp_json['value'])
# FIXME put back timings
return (value, None) if timed else value
def execute(self, ir, timed=False):
code = self._render(ir)
resp = requests.post(f'{self.url}/execute', json=code, cookies=self.cookies)
if resp.status_code == 400:
resp_json = resp.json()
raise FatalError(resp_json['message'])
resp.raise_for_status()
resp_json = resp.json()
typ = dtype(resp_json['type'])
result = json.loads(resp_json['result'])
value = typ._from_json(result['value'])
timings = result['timings']
return (value, timings) if timed else value
def register_liftover_functions(rg, dest_rg):
from hail.expr.types import dtype
register_session_function(
f"liftoverLocus({rg})({dest_rg})", (
dtype(f"locus<{rg}>"),
dtype('float64'),
), dtype(f"struct{{result:locus<{dest_rg}>,is_negative_strand:bool}}"))
register_session_function(
f"liftoverLocusInterval({rg})({dest_rg})", (
dtype(f"interval<locus<{rg}>>"),
dtype('float64'),
),
dtype(
f"struct{{result:interval<locus<{dest_rg}>>,is_negative_strand:bool}}"
))
def value_type(self, ir):
token = secret_alnum_string()
with TemporaryDirectory(ensure_exists=False) as dir:
with self.fs.open(dir + '/in', 'wb') as infile:
write_int(infile, ServiceBackend.VALUE_TYPE)
write_str(infile, tmp_dir())
write_str(infile, self.render(ir))
batch_attributes = self.batch_attributes
if 'name' not in batch_attributes:
batch_attributes = {**batch_attributes, 'name': 'value_type(...)'}
bb = self.bc.create_batch(token=token, attributes=batch_attributes)
j = bb.create_jvm_job([
'is.hail.backend.service.ServiceBackendSocketAPI2',
os.environ['HAIL_SHA'],
os.environ['HAIL_JAR_URL'],
batch_attributes['name'],
dir + '/in',
dir + '/out',
], mount_tokens=True)
b = bb.submit(disable_progress_bar=self.disable_progress_bar)
status = b.wait(disable_progress_bar=self.disable_progress_bar)
if status['n_succeeded'] != 1:
raise ValueError(f'batch failed {status} {j.log()}')
with self.fs.open(dir + '/out', 'rb') as outfile:
success = read_bool(outfile)
if success:
s = read_str(outfile)
try:
return dtype(json.loads(s))
except json.decoder.JSONDecodeError as err:
raise ValueError(f'could not decode {s}') from err
else:
jstacktrace = read_str(outfile)
raise FatalError(jstacktrace)
def register_liftover_functions(rg, dest_rg):
from hail.expr.types import dtype
register_function(f"liftoverLocus({rg})({dest_rg})", (dtype(f"locus<{rg}>"), dtype('float64'),), dtype(f"struct{{result:locus<{dest_rg}>,is_negative_strand:bool}}"))
register_function(f"liftoverLocusInterval({rg})({dest_rg})", (dtype(f"interval<locus<{rg}>>"), dtype('float64'),), dtype(f"struct{{result:interval<locus<{dest_rg}>>,is_negative_strand:bool}}"))
def _compute_type(self):
name = self.config['name']
child_typ = self.child.typ
if name == 'LinearRegressionRowsChained':
pass_through = self.config['passThrough']
chained_schema = hl.dtype(
'struct{n:array<int32>,sum_x:array<float64>,y_transpose_x:array<array<float64>>,beta:array<array<float64>>,standard_error:array<array<float64>>,t_stat:array<array<float64>>,p_value:array<array<float64>>}')
self._type = hl.ttable(
child_typ.global_type,
(child_typ.row_key_type
._insert_fields(**{f: child_typ.row_type[f] for f in pass_through})
._concat(chained_schema)),
child_typ.row_key)
elif name == 'LinearRegressionRowsSingle':
pass_through = self.config['passThrough']
chained_schema = hl.dtype(
'struct{n:int32,sum_x:float64,y_transpose_x:array<float64>,beta:array<float64>,standard_error:array<float64>,t_stat:array<float64>,p_value:array<float64>}')
self._type = hl.ttable(
child_typ.global_type,
(child_typ.row_key_type
._insert_fields(**{f: child_typ.row_type[f] for f in pass_through})
._concat(chained_schema)),
child_typ.row_key)
elif name == 'LogisticRegression':
pass_through = self.config['passThrough']
logreg_type = hl.tstruct(logistic_regression=hl.tarray(regression_test_type(self.config['test'])))
self._type = hl.ttable(
child_typ.global_type,
(child_typ.row_key_type
._insert_fields(**{f: child_typ.row_type[f] for f in pass_through})
._concat(logreg_type)),
child_typ.row_key)
elif name == 'PoissonRegression':
pass_through = self.config['passThrough']
poisreg_type = regression_test_type(self.config['test'])
self._type = hl.ttable(
child_typ.global_type,
(child_typ.row_key_type
._insert_fields(**{f: child_typ.row_type[f] for f in pass_through})
._concat(poisreg_type)),
child_typ.row_key)
elif name == 'Skat':
key_field = self.config['keyField']
key_type = child_typ.row_type[key_field]
skat_type = hl.dtype(f'struct{{id:{key_type},size:int32,q_stat:float64,p_value:float64,fault:int32}}')
self._type = hl.ttable(
hl.tstruct(),
skat_type,
['id'])
elif name == 'PCA':
self._type = hl.ttable(
hl.tstruct(eigenvalues=hl.tarray(hl.tfloat64),
scores=hl.tarray(child_typ.col_key_type._insert_field('scores', hl.tarray(hl.tfloat64)))),
child_typ.row_key_type._insert_field('loadings', dtype('array<float64>')),
child_typ.row_key)
else:
assert name == 'LocalLDPrune', name
self._type = hl.ttable(
hl.tstruct(),
child_typ.row_key_type._insert_fields(mean=hl.tfloat64, centered_length_rec=hl.tfloat64),
list(child_typ.row_key))
def register_functions():
from hail.expr.types import dtype
register_function("flatten", (dtype("array<array<?T>>"), ),
dtype("array<?T>"))
register_function("difference", (
dtype("set<?T>"),
dtype("set<?T>"),
), dtype("set<?T>"))
register_function("median", (dtype("set<?T:numeric>"), ), dtype("?T"))
register_function("median", (dtype("array<?T:numeric>"), ), dtype("?T"))
register_function("uniqueMinIndex", (dtype("array<?T>"), ), dtype("int32"))
register_function("mean", (dtype("set<?T:numeric>"), ), dtype("float64"))
register_function("mean", (dtype("array<?T:numeric>"), ), dtype("float64"))
register_function("toFloat32", (dtype("?T:numeric"), ), dtype("float32"))
register_function("uniqueMaxIndex", (dtype("array<?T>"), ), dtype("int32"))
register_function("toSet", (dtype("array<?T>"), ), dtype("set<?T>"))
def array_floating_point_divide(arg_type, ret_type):
register_function("/", (
arg_type,
hl.tarray(arg_type),
), hl.tarray(ret_type))
register_function("/", (hl.tarray(arg_type), arg_type),
hl.tarray(ret_type))
register_function("/", (hl.tarray(arg_type), hl.tarray(arg_type)),
hl.tarray(ret_type))
array_floating_point_divide(hl.tint32, hl.tfloat32)
array_floating_point_divide(hl.tint64, hl.tfloat32)
array_floating_point_divide(hl.tfloat32, hl.tfloat32)
array_floating_point_divide(hl.tfloat64, hl.tfloat64)
def ndarray_floating_point_divide(arg_type, ret_type):
register_function("/", (
arg_type,
hl.tndarray(arg_type, NatVariable()),
), hl.tndarray(ret_type, NatVariable()))
register_function("/",
(hl.tndarray(arg_type, NatVariable()), arg_type),
hl.tndarray(ret_type, NatVariable()))
register_function("/", (hl.tndarray(
arg_type, NatVariable()), hl.tndarray(arg_type, NatVariable())),
hl.tndarray(ret_type, NatVariable()))
ndarray_floating_point_divide(hl.tint32, hl.tfloat32)
ndarray_floating_point_divide(hl.tint64, hl.tfloat32)
ndarray_floating_point_divide(hl.tfloat32, hl.tfloat32)
ndarray_floating_point_divide(hl.tfloat64, hl.tfloat64)
register_function("values", (dtype("dict<?key, ?value>"), ),
dtype("array<?value>"))
register_function("[*:]", (
dtype("array<?T>"),
dtype("int32"),
), dtype("array<?T>"))
register_function("[*:]", (
dtype("str"),
dtype("int32"),
), dtype("str"))
register_function("get", (
dtype("dict<?key, ?value>"),
dtype("?key"),
), dtype("?value"))
register_function("get", (
dtype("dict<?key, ?value>"),
dtype("?key"),
dtype("?value"),
), dtype("?value"))
register_function("max", (dtype("array<?T:numeric>"), ), dtype("?T"))
register_function("nanmax", (dtype("array<?T:numeric>"), ), dtype("?T"))
register_function("max", (
dtype("?T"),
dtype("?T"),
), dtype("?T"))
register_function("nanmax", (
dtype("?T"),
dtype("?T"),
), dtype("?T"))
register_function("max_ignore_missing", (
dtype("?T"),
dtype("?T"),
), dtype("?T"))
register_function("nanmax_ignore_missing", (
dtype("?T"),
dtype("?T"),
), dtype("?T"))
register_function("product", (dtype("set<?T:numeric>"), ), dtype("?T"))
register_function("product", (dtype("array<?T:numeric>"), ), dtype("?T"))
register_function("toInt32", (dtype("?T:numeric"), ), dtype("int32"))
register_function("extend", (
dtype("array<?T>"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("argmin", (dtype("array<?T>"), ), dtype("int32"))
register_function("toFloat64", (dtype("?T:numeric"), ), dtype("float64"))
register_function("sort", (dtype("array<?T>"), ), dtype("array<?T>"))
register_function("sort", (
dtype("array<?T>"),
dtype("bool"),
), dtype("array<?T>"))
register_function("isSubset", (
dtype("set<?T>"),
dtype("set<?T>"),
), dtype("bool"))
register_function("[*:*]", (
dtype("str"),
dtype("int32"),
dtype("int32"),
), dtype("str"))
register_function("[*:*]", (
dtype("array<?T>"),
dtype("int32"),
dtype("int32"),
), dtype("array<?T>"))
register_function("+", (
dtype("array<?T:numeric>"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("+", (
dtype("array<?T:numeric>"),
dtype("?T"),
), dtype("array<?T>"))
register_function("+", (
dtype("?T:numeric"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("+", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("+", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("?T"),
), dtype("ndarray<?T, ?nat>"))
register_function("+", (
dtype("?T:numeric"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("**", (
dtype("array<?T:numeric>"),
dtype("array<?T>"),
), dtype("array<float64>"))
register_function("**", (
dtype("array<?T:numeric>"),
dtype("?T"),
), dtype("array<float64>"))
register_function("**", (
dtype("?T:numeric"),
dtype("array<?T>"),
), dtype("array<float64>"))
register_function("**", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("**", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("?T"),
), dtype("ndarray<?T, ?nat>"))
register_function("**", (
dtype("?T:numeric"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("append", (
dtype("array<?T>"),
dtype("?T"),
), dtype("array<?T>"))
register_function("[:*]", (
dtype("str"),
dtype("int32"),
), dtype("str"))
register_function("[:*]", (
dtype("array<?T>"),
dtype("int32"),
), dtype("array<?T>"))
register_function("remove", (
dtype("set<?T>"),
dtype("?T"),
), dtype("set<?T>"))
register_function("[]", (
dtype("str"),
dtype("int32"),
), dtype("str"))
register_function("indexArray", (
dtype("array<?T>"),
dtype("int32"),
), dtype("?T"))
register_function("[]", (
dtype("dict<?key, ?value>"),
dtype("?key"),
), dtype("?value"))
register_function("dictToArray", (dtype("dict<?key, ?value>"), ),
dtype("array<tuple(?key, ?value)>"))
register_function("%", (
dtype("array<?T:numeric>"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("%", (
dtype("array<?T:numeric>"),
dtype("?T"),
), dtype("array<?T>"))
register_function("%", (
dtype("?T:numeric"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("%", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("%", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("?T"),
), dtype("ndarray<?T, ?nat>"))
register_function("%", (
dtype("?T:numeric"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("dict", (dtype("array<tuple(?key, ?value)>"), ),
dtype("dict<?key, ?value>"))
register_function("dict", (dtype("set<tuple(?key, ?value)>"), ),
dtype("dict<?key, ?value>"))
register_function("keys", (dtype("dict<?key, ?value>"), ),
dtype("array<?key>"))
register_function("min", (dtype("array<?T:numeric>"), ), dtype("?T"))
register_function("nanmin", (dtype("array<?T:numeric>"), ), dtype("?T"))
register_function("min", (
dtype("?T"),
dtype("?T"),
), dtype("?T"))
register_function("nanmin", (
dtype("?T"),
dtype("?T"),
), dtype("?T"))
register_function("min_ignore_missing", (
dtype("?T"),
dtype("?T"),
), dtype("?T"))
register_function("nanmin_ignore_missing", (
dtype("?T"),
dtype("?T"),
), dtype("?T"))
register_function("sum", (dtype("set<?T:numeric>"), ), dtype("?T"))
register_function("sum", (dtype("array<?T:numeric>"), ), dtype("?T"))
register_function("toInt64", (dtype("?T:numeric"), ), dtype("int64"))
register_function("contains", (
dtype("dict<?key, ?value>"),
dtype("?key"),
), dtype("bool"))
register_function("contains", (
dtype("array<?T>"),
dtype("?T"),
), dtype("bool"))
register_function("contains", (
dtype("set<?T>"),
dtype("?T"),
), dtype("bool"))
register_function("-", (
dtype("array<?T:numeric>"),
dtype("?T"),
), dtype("array<?T>"))
register_function("-", (
dtype("array<?T:numeric>"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("-", (
dtype("?T:numeric"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("-", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("-", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("?T"),
), dtype("ndarray<?T, ?nat>"))
register_function("-", (
dtype("?T:numeric"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("addone", (dtype("int32"), ), dtype("int32"))
register_function("isEmpty", (dtype("dict<?key, ?value>"), ),
dtype("bool"))
register_function("isEmpty", (dtype("array<?T>"), ), dtype("bool"))
register_function("isEmpty", (dtype("set<?T>"), ), dtype("bool"))
register_function("[:]", (dtype("array<?T>"), ), dtype("array<?T>"))
register_function("[:]", (dtype("str"), ), dtype("str"))
register_function("union", (
dtype("set<?T>"),
dtype("set<?T>"),
), dtype("set<?T>"))
register_function("*", (
dtype("array<?T:numeric>"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("*", (
dtype("array<?T:numeric>"),
dtype("?T"),
), dtype("array<?T>"))
register_function("*", (
dtype("?T:numeric"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("*", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("*", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("?T"),
), dtype("ndarray<?T, ?nat>"))
register_function("*", (
dtype("?T:numeric"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("intersection", (
dtype("set<?T>"),
dtype("set<?T>"),
), dtype("set<?T>"))
register_function("add", (
dtype("set<?T>"),
dtype("?T"),
), dtype("set<?T>"))
register_function("argmax", (dtype("array<?T>"), ), dtype("int32"))
register_function("//", (
dtype("array<?T:numeric>"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("//", (
dtype("array<?T:numeric>"),
dtype("?T"),
), dtype("array<?T>"))
register_function("//", (
dtype("?T:numeric"),
dtype("array<?T>"),
), dtype("array<?T>"))
register_function("//", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("//", (
dtype("ndarray<?T:numeric, ?nat>"),
dtype("?T"),
), dtype("ndarray<?T, ?nat>"))
register_function("//", (
dtype("?T:numeric"),
dtype("ndarray<?T, ?nat>"),
), dtype("ndarray<?T, ?nat>"))
register_function("keySet", (dtype("dict<?key, ?value>"), ),
dtype("set<?key>"))
register_function("qnorm", (dtype("float64"), ), dtype("float64"))
register_function("oneHotAlleles", (
dtype("call"),
dtype("int32"),
), dtype("array<int32>"))
register_function("dpois", (
dtype("float64"),
dtype("float64"),
dtype("bool"),
), dtype("float64"))
register_function("dpois", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("ploidy", (dtype("call"), ), dtype("int32"))
register_function("||", (
dtype("bool"),
dtype("bool"),
), dtype("bool"))
register_function("ppois", (
dtype("float64"),
dtype("float64"),
dtype("bool"),
dtype("bool"),
), dtype("float64"))
register_function("ppois", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("log10", (dtype("float64"), ), dtype("float64"))
register_function("isHet", (dtype("call"), ), dtype("bool"))
register_function("isAutosomalOrPseudoAutosomal", (dtype("?T:locus"), ),
dtype("bool"))
register_function("testCodeUnification", (
dtype("?x:numeric"),
dtype("?x:int32"),
), dtype("?x"))
register_seeded_function("rand_pois", (dtype("float64"), ),
dtype("float64"))
register_seeded_function("rand_pois", (
dtype("int32"),
dtype("float64"),
), dtype("array<float64>"))
register_function("toFloat32", (dtype("str"), ), dtype("float32"))
register_function("toFloat32", (dtype("bool"), ), dtype("float32"))
register_function("isAutosomal", (dtype("?T:locus"), ), dtype("bool"))
register_function("isPhased", (dtype("call"), ), dtype("bool"))
register_function("isHomVar", (dtype("call"), ), dtype("bool"))
register_function("corr", (
dtype("array<float64>"),
dtype("array<float64>"),
), dtype("float64"))
register_function("log", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("log", (dtype("float64"), ), dtype("float64"))
register_function("foobar2", (), dtype("int32"))
register_function("approxEqual", (
dtype("float64"),
dtype("float64"),
dtype("float64"),
dtype("bool"),
dtype("bool"),
), dtype("bool"))
register_function("plDosage", (dtype("array<?N:int32>"), ),
dtype("float64"))
register_function("includesEnd", (dtype("interval<?T>"), ), dtype("bool"))
register_function("position", (dtype("?T:locus"), ), dtype("int32"))
register_seeded_function("rand_unif", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("str", (dtype("?T"), ), dtype("str"))
register_function("valuesSimilar", (
dtype("?T"),
dtype("?T"),
dtype('float64'),
dtype('bool'),
), dtype("bool"))
register_function("replace", (
dtype("str"),
dtype("str"),
dtype("str"),
), dtype("str"))
register_function("exp", (dtype("float64"), ), dtype("float64"))
register_function("&&", (
dtype("bool"),
dtype("bool"),
), dtype("bool"))
register_function("compare", (
dtype("int32"),
dtype("int32"),
), dtype("int32"))
register_function("triangle", (dtype("int32"), ), dtype("int32"))
register_function("Interval", (
dtype("?T"),
dtype("?T"),
dtype("bool"),
dtype("bool"),
), dtype("interval<?T>"))
register_function("contig", (dtype("?T:locus"), ), dtype("str"))
register_function("Call", (dtype("bool"), ), dtype("call"))
register_function("Call", (dtype("str"), ), dtype("call"))
register_function("Call", (
dtype("int32"),
dtype("bool"),
), dtype("call"))
register_function("Call", (
dtype("int32"),
dtype("int32"),
dtype("bool"),
), dtype("call"))
register_function("Call", (
dtype("array<int32>"),
dtype("bool"),
), dtype("call"))
register_function("qchisqtail", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("binomTest", (
dtype("int32"),
dtype("int32"),
dtype("float64"),
dtype("int32"),
), dtype("float64"))
register_function("qpois", (
dtype("float64"),
dtype("float64"),
), dtype("int32"))
register_function("qpois", (
dtype("float64"),
dtype("float64"),
dtype("bool"),
dtype("bool"),
), dtype("int32"))
register_function("is_finite", (dtype("float32"), ), dtype("bool"))
register_function("is_finite", (dtype("float64"), ), dtype("bool"))
register_function("inYPar", (dtype("?T:locus"), ), dtype("bool"))
register_function("contingency_table_test", (
dtype("int32"),
dtype("int32"),
dtype("int32"),
dtype("int32"),
dtype("int32"),
), dtype("struct{p_value: float64, odds_ratio: float64}"))
register_function("toInt32", (dtype("bool"), ), dtype("int32"))
register_function("toInt32", (dtype("str"), ), dtype("int32"))
register_function("foobar1", (), dtype("int32"))
register_function("toFloat64", (dtype("str"), ), dtype("float64"))
register_function("toFloat64", (dtype("bool"), ), dtype("float64"))
register_function("dbeta", (
dtype("float64"),
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("min_rep", (
dtype("?T:locus"),
dtype("array<str>"),
), dtype("struct{locus: ?T, alleles: array<str>}"))
register_function("locus_windows_per_contig", (
dtype("array<array<float64>>"),
dtype("float64"),
), dtype("tuple(array<int32>, array<int32>)"))
register_function("toBoolean", (dtype("str"), ), dtype("bool"))
register_seeded_function("rand_bool", (dtype("float64"), ), dtype("bool"))
register_function("pchisqtail", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_seeded_function("rand_cat", (dtype("array<float64>"), ),
dtype("int32"))
register_function("inYNonPar", (dtype("?T:locus"), ), dtype("bool"))
register_function("+", (
dtype("str"),
dtype("str"),
), dtype("str"))
register_function("**", (
dtype("float32"),
dtype("float32"),
), dtype("float64"))
register_function("**", (
dtype("int32"),
dtype("int32"),
), dtype("float64"))
register_function("**", (
dtype("int64"),
dtype("int64"),
), dtype("float64"))
register_function("**", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("length", (dtype("str"), ), dtype("int32"))
register_function("slice", (
dtype("str"),
dtype("int32"),
dtype("int32"),
), dtype("str"))
register_function("split", (
dtype("str"),
dtype("str"),
dtype("int32"),
), dtype("array<str>"))
register_function("split", (
dtype("str"),
dtype("str"),
), dtype("array<str>"))
register_seeded_function("rand_gamma", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("UnphasedDiploidGtIndexCall", (dtype("int32"), ),
dtype("call"))
register_function("[]", (
dtype("call"),
dtype("int32"),
), dtype("int32"))
register_function("sign", (dtype("int64"), ), dtype("int64"))
register_function("sign", (dtype("float64"), ), dtype("float64"))
register_function("sign", (dtype("float32"), ), dtype("float32"))
register_function("sign", (dtype("int32"), ), dtype("int32"))
register_function("unphasedDiploidGtIndex", (dtype("call"), ),
dtype("int32"))
register_function("gamma", (dtype("float64"), ), dtype("float64"))
register_function("%", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("%", (
dtype("int64"),
dtype("int64"),
), dtype("int64"))
register_function("%", (
dtype("float32"),
dtype("float32"),
), dtype("float32"))
register_function("%", (
dtype("int32"),
dtype("int32"),
), dtype("int32"))
register_function(
"fisher_exact_test", (
dtype("int32"),
dtype("int32"),
dtype("int32"),
dtype("int32"),
),
dtype(
"struct{p_value: float64, odds_ratio: float64, ci_95_lower: float64, ci_95_upper: float64}"
))
register_function("floor", (dtype("float64"), ), dtype("float64"))
register_function("floor", (dtype("float32"), ), dtype("float32"))
register_function("isNonRef", (dtype("call"), ), dtype("bool"))
register_function("includesStart", (dtype("interval<?T>"), ),
dtype("bool"))
register_function("isHetNonRef", (dtype("call"), ), dtype("bool"))
register_function("hardy_weinberg_test", (
dtype("int32"),
dtype("int32"),
dtype("int32"),
), dtype("struct{het_freq_hwe: float64, p_value: float64}"))
register_function("haplotype_freq_em", (dtype("array<int32>"), ),
dtype("array<float64>"))
register_function("nNonRefAlleles", (dtype("call"), ), dtype("int32"))
register_function("abs", (dtype("float64"), ), dtype("float64"))
register_function("abs", (dtype("float32"), ), dtype("float32"))
register_function("abs", (dtype("int64"), ), dtype("int64"))
register_function("abs", (dtype("int32"), ), dtype("int32"))
register_function("endswith", (
dtype("str"),
dtype("str"),
), dtype("bool"))
register_function("sqrt", (dtype("float64"), ), dtype("float64"))
register_function("isnan", (dtype("float32"), ), dtype("bool"))
register_function("isnan", (dtype("float64"), ), dtype("bool"))
register_function("lower", (dtype("str"), ), dtype("str"))
register_seeded_function("rand_beta", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_seeded_function("rand_beta", (
dtype("float64"),
dtype("float64"),
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("toInt64", (dtype("bool"), ), dtype("int64"))
register_function("toInt64", (dtype("str"), ), dtype("int64"))
register_function("testCodeUnification2", (dtype("?x"), ), dtype("?x"))
register_function("contains", (
dtype("str"),
dtype("str"),
), dtype("bool"))
register_function("contains", (
dtype("interval<?T>"),
dtype("?T"),
), dtype("bool"))
register_function("entropy", (dtype("str"), ), dtype("float64"))
register_function("filtering_allele_frequency", (
dtype("int32"),
dtype("int32"),
dtype("float64"),
), dtype("float64"))
register_function("gqFromPL", (dtype("array<?N:int32>"), ), dtype("int32"))
register_function("startswith", (
dtype("str"),
dtype("str"),
), dtype("bool"))
register_function("ceil", (dtype("float32"), ), dtype("float32"))
register_function("ceil", (dtype("float64"), ), dtype("float64"))
register_function("json", (dtype("?T"), ), dtype("str"))
register_function("strip", (dtype("str"), ), dtype("str"))
register_function("firstMatchIn", (
dtype("str"),
dtype("str"),
), dtype("array<str>"))
register_function("isEmpty", (dtype("interval<?T>"), ), dtype("bool"))
register_function("~", (
dtype("str"),
dtype("str"),
), dtype("bool"))
register_function("mkString", (
dtype("set<str>"),
dtype("str"),
), dtype("str"))
register_function("mkString", (
dtype("array<str>"),
dtype("str"),
), dtype("str"))
register_function("dosage", (dtype("array<?N:float64>"), ),
dtype("float64"))
register_function("upper", (dtype("str"), ), dtype("str"))
register_function("overlaps", (
dtype("interval<?T>"),
dtype("interval<?T>"),
), dtype("bool"))
register_function("downcode", (
dtype("call"),
dtype("int32"),
), dtype("call"))
register_function("inXPar", (dtype("?T:locus"), ), dtype("bool"))
register_function("format", (
dtype("str"),
dtype("?T:tuple"),
), dtype("str"))
register_function("pnorm", (dtype("float64"), ), dtype("float64"))
register_function("is_infinite", (dtype("float32"), ), dtype("bool"))
register_function("is_infinite", (dtype("float64"), ), dtype("bool"))
register_function("isHetRef", (dtype("call"), ), dtype("bool"))
register_function("isMitochondrial", (dtype("?T:locus"), ), dtype("bool"))
register_function("hamming", (
dtype("str"),
dtype("str"),
), dtype("int32"))
register_function("end", (dtype("interval<?T>"), ), dtype("?T"))
register_function("start", (dtype("interval<?T>"), ), dtype("?T"))
register_function("inXNonPar", (dtype("?T:locus"), ), dtype("bool"))
register_function("escapeString", (dtype("str"), ), dtype("str"))
register_function("isHomRef", (dtype("call"), ), dtype("bool"))
register_seeded_function("rand_norm", (
dtype("float64"),
dtype("float64"),
), dtype("float64"))
register_function("chi_squared_test", (
dtype("int32"),
dtype("int32"),
dtype("int32"),
dtype("int32"),
), dtype("struct{p_value: float64, odds_ratio: float64}"))
def value_type(self, ir):
resp = self._request_type(ir, 'value')
return dtype(resp)
def register_aggregators():
from hail.expr.types import dtype
register_aggregator('ApproxCDF', (dtype('int32'),), (dtype('int32'),),
dtype('struct{values:array<int32>,ranks:array<int64>,_compaction_counts:array<int32>}'))
register_aggregator('ApproxCDF', (dtype('int32'),), (dtype('int64'),),
dtype('struct{values:array<int64>,ranks:array<int64>,_compaction_counts:array<int32>}'))
register_aggregator('ApproxCDF', (dtype('int32'),), (dtype('float32'),),
dtype('struct{values:array<float32>,ranks:array<int64>,_compaction_counts:array<int32>}'))
register_aggregator('ApproxCDF', (dtype('int32'),), (dtype('float64'),),
dtype('struct{values:array<float64>,ranks:array<int64>,_compaction_counts:array<int32>}'))
register_aggregator('Collect', (), (dtype("?in"),), dtype('array<?in>'))
info_score_aggregator_type = dtype('struct{score:float64,n_included:tint32}')
register_aggregator('InfoScore', (), (dtype('array<float64>'),), info_score_aggregator_type)
register_aggregator('Sum', (), (dtype('int64'),), dtype('int64'))
register_aggregator('Sum', (), (dtype('float64'),), dtype('float64'))
register_aggregator('Sum', (), (dtype('array<int64>'),), dtype('array<int64>'))
register_aggregator('Sum', (), (dtype('array<float64>'),), dtype('array<float64>'))
register_aggregator('CollectAsSet', (), (dtype("?in"),), dtype('set<?in>'))
register_aggregator('Product', (), (dtype('int64'),), dtype('int64'))
register_aggregator('Product', (), (dtype('float64'),), dtype('float64'))
hwe_aggregator_type = dtype('struct { het_freq_hwe: float64, p_value: float64 }')
register_aggregator('HardyWeinberg', (), (dtype('call'),), hwe_aggregator_type)
register_aggregator('Max', (), (dtype('bool'),), dtype('bool'))
register_aggregator('Max', (), (dtype('int32'),), dtype('int32'))
register_aggregator('Max', (), (dtype('int64'),), dtype('int64'))
register_aggregator('Max', (), (dtype('float32'),), dtype('float32'))
register_aggregator('Max', (), (dtype('float64'),), dtype('float64'))
register_aggregator('Min', (), (dtype('bool'),), dtype('bool'))
register_aggregator('Min', (), (dtype('int32'),), dtype('int32'))
register_aggregator('Min', (), (dtype('int64'),), dtype('int64'))
register_aggregator('Min', (), (dtype('float32'),), dtype('float32'))
register_aggregator('Min', (), (dtype('float64'),), dtype('float64'))
register_aggregator('Count', (), (), dtype('int64'))
register_aggregator('Counter', (), (dtype('?in'),), dtype('dict<?in, int64>'))
register_aggregator('Take', (dtype('int32'),), (dtype('?in'),), dtype('array<?in>'))
register_aggregator('TakeBy', (dtype('int32'),), (dtype('?in'), dtype('?key'),), dtype('array<?in>'))
downsample_aggregator_type = dtype('array<tuple(float64, float64, array<str>)>')
register_aggregator('Downsample', (dtype('int32'),), (dtype('float64'), dtype('float64'), dtype('array<?T>'),), downsample_aggregator_type)
call_stats_aggregator_type = dtype('struct{AC: array<int32>,AF:array<float64>,AN:int32,homozygote_count:array<int32>}')
register_aggregator('CallStats', (dtype('int32'),), (dtype('call'),), call_stats_aggregator_type)
inbreeding_aggregator_type = dtype('struct{f_stat:float64,n_called:int64,expected_homs:float64,observed_homs:int64}')
register_aggregator('Inbreeding', (), (dtype('call'), dtype('float64'),), inbreeding_aggregator_type)
linreg_aggregator_type = dtype('struct{xty:array<float64>,beta:array<float64>,diag_inv:array<float64>,beta0:array<float64>}')
register_aggregator('LinearRegression', (dtype('int32'), dtype('int32'),), (dtype('float64'), dtype('array<float64>'),), linreg_aggregator_type)
register_aggregator('PrevNonnull', (), (dtype('?in'),), dtype('?in'))
def _from_json(json):
return tblockmatrix(dtype(json['elementType']), json['shape'],
json['blockSize'], json['dimsPartitioned'])
def value_type(self, ir):
jir = self._to_java_ir(ir)
return dtype(jir.typ().toString())
def value_type(self, ir):
resp = self._request_type(ir, 'value')
return dtype(resp)
def _from_java(jtbm):
return tblockmatrix(
dtype(jtbm.elementType().toString()),
jiterable_to_list(jtbm.shape()),
jtbm.isRowVector(),
jtbm.blockSize())
def _from_json(json):
return ttable(
dtype(json['global']),
dtype(json['row']),
json['row_key'])
def _from_java(jtbm):
return tblockmatrix(dtype(jtbm.elementType().toString()),
jiterable_to_list(jtbm.shape()),
jtbm.isRowVector(), jtbm.blockSize())
def _from_json(json):
return tblockmatrix(dtype(json['element_type']), json['shape'],
json['is_row_vector'], json['block_size'])
def _from_json(json):
return tblockmatrix(dtype(json['element_type']),
json['shape'],
json['is_row_vector'],
json['block_size'])
def register_aggregators():
from hail.expr.types import dtype
register_aggregator('Fraction', (), None, (dtype('bool'), ),
dtype('float64'))
stats_aggregator_type = dtype(
'struct{mean:float64,stdev:float64,min:float64,max:float64,n:int64,sum:float64}'
)
register_aggregator('Statistics', (), None, (dtype('float64'), ),
stats_aggregator_type)
register_aggregator('Collect', (), None, (dtype("?in"), ),
dtype('array<?in>'))
info_score_aggregator_type = dtype(
'struct{score:float64,n_included:tint32}')
register_aggregator('InfoScore', (), None, (dtype('array<float64>'), ),
info_score_aggregator_type)
register_aggregator('Sum', (), None, (dtype('int64'), ), dtype('int64'))
register_aggregator('Sum', (), None, (dtype('float64'), ),
dtype('float64'))
register_aggregator('Sum', (), None, (dtype('array<int64>'), ),
dtype('array<int64>'))
register_aggregator('Sum', (), None, (dtype('array<float64>'), ),
dtype('array<float64>'))
register_aggregator('CollectAsSet', (), None, (dtype("?in"), ),
dtype('set<?in>'))
register_aggregator('Product', (), None, (dtype('int64'), ),
dtype('int64'))
register_aggregator('Product', (), None, (dtype('float64'), ),
dtype('float64'))
hwe_aggregator_type = dtype(
'struct { het_freq_hwe: float64, p_value: float64 }')
register_aggregator('HardyWeinberg', (), None, (dtype('call'), ),
hwe_aggregator_type)
register_aggregator('Max', (), None, (dtype('bool'), ), dtype('bool'))
register_aggregator('Max', (), None, (dtype('int32'), ), dtype('int32'))
register_aggregator('Max', (), None, (dtype('int64'), ), dtype('int64'))
register_aggregator('Max', (), None, (dtype('float32'), ),
dtype('float32'))
register_aggregator('Max', (), None, (dtype('float64'), ),
dtype('float64'))
register_aggregator('Min', (), None, (dtype('bool'), ), dtype('bool'))
register_aggregator('Min', (), None, (dtype('int32'), ), dtype('int32'))
register_aggregator('Min', (), None, (dtype('int64'), ), dtype('int64'))
register_aggregator('Min', (), None, (dtype('float32'), ),
dtype('float32'))
register_aggregator('Min', (), None, (dtype('float64'), ),
dtype('float64'))
register_aggregator('Count', (), None, (), dtype('int64'))
register_aggregator('Counter', (), None, (dtype('?in'), ),
dtype('dict<?in, int64>'))
register_aggregator('Take', (dtype('int32'), ), None, (dtype('?in'), ),
dtype('array<?in>'))
register_aggregator('TakeBy', (dtype('int32'), ), None, (
dtype('?in'),
dtype('?key'),
), dtype('array<?in>'))
histogram_aggregator_type = dtype(
'struct{bin_edges:array<float64>,bin_freq:array<int64>,n_smaller:int64,n_larger:int64}'
)
register_aggregator('Histogram', (
dtype('float64'),
dtype('float64'),
dtype('int32'),
), None, (dtype('float64'), ), histogram_aggregator_type)
downsample_aggregator_type = dtype(
'array<tuple(float64, float64, array<str>)>')
register_aggregator('Downsample', (dtype('int32'), ), None, (
dtype('float64'),
dtype('float64'),
dtype('array<?T>'),
), downsample_aggregator_type)
call_stats_aggregator_type = dtype(
'struct{AC: array<int32>,AF:array<float64>,AN:int32,homozygote_count:array<int32>}'
)
register_aggregator('CallStats', (), (dtype('int32'), ), (dtype('call'), ),
call_stats_aggregator_type)
inbreeding_aggregator_type = dtype(
'struct{f_stat:float64,n_called:int64,expected_homs:float64,observed_homs:int64}'
)
register_aggregator('Inbreeding', (), None, (
dtype('call'),
dtype('float64'),
), inbreeding_aggregator_type)
linreg_aggregator_type = dtype(
'struct{beta:array<float64>,standard_error:array<float64>,t_stat:array<float64>,p_value:array<float64>,multiple_standard_error:float64,multiple_r_squared:float64,adjusted_r_squared:float64,f_stat:float64,multiple_p_value:float64,n:int64}'
)
register_aggregator('LinearRegression', (
dtype('int32'),
dtype('int32'),
), None, (
dtype('float64'),
dtype('array<float64>'),
), linreg_aggregator_type)
register_aggregator('PearsonCorrelation', (), None, (
dtype('tfloat64'),
dtype('float64'),
), dtype('float64'))
register_aggregator('PrevNonnull', (), None, (dtype('?in'), ),
dtype('?in'))
def value_type(self, ir):
jir = self._to_java_ir(ir)
return dtype(jir.typ().toString())
def register_aggregators():
from hail.expr.types import dtype
register_aggregator('ApproxCDF', (dtype('int32'),), None, (dtype('int32'),),
dtype('struct{values:array<int32>,ranks:array<int64>}'))
register_aggregator('ApproxCDF', (dtype('int32'),), None, (dtype('int64'),),
dtype('struct{values:array<int64>,ranks:array<int64>}'))
register_aggregator('ApproxCDF', (dtype('int32'),), None, (dtype('float32'),),
dtype('struct{values:array<float32>,ranks:array<int64>}'))
register_aggregator('ApproxCDF', (dtype('int32'),), None, (dtype('float64'),),
dtype('struct{values:array<float64>,ranks:array<int64>}'))
register_aggregator('Fraction', (), None, (dtype('bool'),), dtype('float64'))
stats_aggregator_type = dtype('struct{mean:float64,stdev:float64,min:float64,max:float64,n:int64,sum:float64}')
register_aggregator('Statistics', (), None, (dtype('float64'),), stats_aggregator_type)
register_aggregator('Collect', (), None, (dtype("?in"),), dtype('array<?in>'))
info_score_aggregator_type = dtype('struct{score:float64,n_included:tint32}')
register_aggregator('InfoScore', (), None, (dtype('array<float64>'),), info_score_aggregator_type)
register_aggregator('Sum', (), None, (dtype('int64'),), dtype('int64'))
register_aggregator('Sum', (), None, (dtype('float64'),), dtype('float64'))
register_aggregator('Sum', (), None, (dtype('array<int64>'),), dtype('array<int64>'))
register_aggregator('Sum', (), None, (dtype('array<float64>'),), dtype('array<float64>'))
register_aggregator('CollectAsSet', (), None, (dtype("?in"),), dtype('set<?in>'))
register_aggregator('Product', (), None, (dtype('int64'),), dtype('int64'))
register_aggregator('Product', (), None, (dtype('float64'),), dtype('float64'))
hwe_aggregator_type = dtype('struct { het_freq_hwe: float64, p_value: float64 }')
register_aggregator('HardyWeinberg', (), None, (dtype('call'),), hwe_aggregator_type)
register_aggregator('Max', (), None, (dtype('bool'),), dtype('bool'))
register_aggregator('Max', (), None, (dtype('int32'),), dtype('int32'))
register_aggregator('Max', (), None, (dtype('int64'),), dtype('int64'))
register_aggregator('Max', (), None, (dtype('float32'),), dtype('float32'))
register_aggregator('Max', (), None, (dtype('float64'),), dtype('float64'))
register_aggregator('Min', (), None, (dtype('bool'),), dtype('bool'))
register_aggregator('Min', (), None, (dtype('int32'),), dtype('int32'))
register_aggregator('Min', (), None, (dtype('int64'),), dtype('int64'))
register_aggregator('Min', (), None, (dtype('float32'),), dtype('float32'))
register_aggregator('Min', (), None, (dtype('float64'),), dtype('float64'))
register_aggregator('Count', (), None, (), dtype('int64'))
register_aggregator('Counter', (), None, (dtype('?in'),), dtype('dict<?in, int64>'))
register_aggregator('Take', (dtype('int32'),), None, (dtype('?in'),), dtype('array<?in>'))
register_aggregator('TakeBy', (dtype('int32'),), None, (dtype('?in'), dtype('?key'),), dtype('array<?in>'))
histogram_aggregator_type = dtype('struct{bin_edges:array<float64>,bin_freq:array<int64>,n_smaller:int64,n_larger:int64}')
register_aggregator('Histogram', (dtype('float64'), dtype('float64'), dtype('int32'),), None, (dtype('float64'),), histogram_aggregator_type)
downsample_aggregator_type = dtype('array<tuple(float64, float64, array<str>)>')
register_aggregator('Downsample', (dtype('int32'),), None, (dtype('float64'), dtype('float64'), dtype('array<?T>'),), downsample_aggregator_type)
call_stats_aggregator_type = dtype('struct{AC: array<int32>,AF:array<float64>,AN:int32,homozygote_count:array<int32>}')
register_aggregator('CallStats', (), (dtype('int32'),), (dtype('call'),), call_stats_aggregator_type)
inbreeding_aggregator_type = dtype('struct{f_stat:float64,n_called:int64,expected_homs:float64,observed_homs:int64}')
register_aggregator('Inbreeding', (), None, (dtype('call'), dtype('float64'),), inbreeding_aggregator_type)
linreg_aggregator_type = dtype('struct{beta:array<float64>,standard_error:array<float64>,t_stat:array<float64>,p_value:array<float64>,multiple_standard_error:float64,multiple_r_squared:float64,adjusted_r_squared:float64,f_stat:float64,multiple_p_value:float64,n:int64}')
register_aggregator('LinearRegression', (dtype('int32'), dtype('int32'),), None, (dtype('float64'), dtype('array<float64>'),), linreg_aggregator_type)
register_aggregator('PearsonCorrelation', (), None, (dtype('tfloat64'), dtype('float64'),), dtype('float64'))
register_aggregator('PrevNonnull', (), None, (dtype('?in'),), dtype('?in'))
def register_reference_genome_functions(rg):
from hail.expr.types import dtype
tvariant = dtype(f"struct{{locus:locus<{rg}>,alleles:array<str>}}")
tinterval = dtype(f"interval<locus<{rg}>>")
register_session_function(f"Locus({rg})", (dtype("str"), ),
dtype(f"locus<{rg}>"))
register_session_function(f"Locus({rg})", (
dtype("str"),
dtype("int32"),
), dtype(f"locus<{rg}>"))
register_session_function(f"LocusAlleles({rg})", (dtype("str"), ),
tvariant)
register_session_function(f"LocusInterval({rg})", (
dtype("str"),
dtype("bool"),
), tinterval)
register_session_function(f"LocusInterval({rg})", (
dtype("str"),
dtype("int32"),
dtype("int32"),
dtype("bool"),
dtype("bool"),
dtype("bool"),
), tinterval)
register_session_function(f"isValidContig({rg})", (dtype("str"), ),
dtype("bool"))
register_session_function(f"isValidLocus({rg})", (
dtype("str"),
dtype("int32"),
), dtype("bool"))
register_session_function(f"getReferenceSequenceFromValidLocus({rg})", (
dtype("str"),
dtype("int32"),
dtype("int32"),
dtype("int32"),
), dtype("str"))
register_session_function(f"getReferenceSequence({rg})", (
dtype("str"),
dtype("int32"),
dtype("int32"),
dtype("int32"),
), dtype("str"))
register_session_function(f"globalPosToLocus({rg})", (dtype("int64"), ),
dtype(f"locus<{rg}>"))
register_session_function(f"locusToGlobalPos({rg})",
(dtype(f"locus<{rg}>"), ), dtype("int64"))
def register_functions():
from hail.expr.types import dtype
register_function("flatten", (dtype("array<array<?T>>"),), dtype("array<?T>"))
register_function("difference", (dtype("set<?T>"),dtype("set<?T>"),), dtype("set<?T>"))
register_function("median", (dtype("set<?T:numeric>"),), dtype("?T"))
register_function("median", (dtype("array<?T:numeric>"),), dtype("?T"))
register_function("uniqueMinIndex", (dtype("array<?T>"),), dtype("int32"))
register_function("mean", (dtype("set<?T:numeric>"),), dtype("float64"))
register_function("mean", (dtype("array<?T:numeric>"),), dtype("float64"))
register_function("toFloat32", (dtype("?T:numeric"),), dtype("float32"))
register_function("uniqueMaxIndex", (dtype("array<?T>"),), dtype("int32"))
register_function("toSet", (dtype("array<?T>"),), dtype("set<?T>"))
def floating_point_divide(arg_type, ret_type):
register_function("/", (arg_type, hl.tarray(arg_type),), hl.tarray(ret_type))
register_function("/", (hl.tarray(arg_type),arg_type), hl.tarray(ret_type))
register_function("/", (hl.tarray(arg_type),hl.tarray(arg_type)), hl.tarray(ret_type))
floating_point_divide(hl.tint32, hl.tfloat32)
floating_point_divide(hl.tint64, hl.tfloat32)
floating_point_divide(hl.tfloat32, hl.tfloat32)
floating_point_divide(hl.tfloat64, hl.tfloat64)
register_function("values", (dtype("dict<?key, ?value>"),), dtype("array<?value>"))
register_function("[*:]", (dtype("array<?T>"),dtype("int32"),), dtype("array<?T>"))
register_function("[*:]", (dtype("str"),dtype("int32"),), dtype("str"))
register_function("get", (dtype("dict<?key, ?value>"),dtype("?key"),), dtype("?value"))
register_function("get", (dtype("dict<?key, ?value>"),dtype("?key"),dtype("?value"),), dtype("?value"))
register_function("max", (dtype("array<?T:numeric>"),), dtype("?T"))
register_function("max", (dtype("set<?T:numeric>"),), dtype("?T"))
register_function("max", (dtype("?T"),dtype("?T"),), dtype("?T"))
register_function("product", (dtype("set<?T:numeric>"),), dtype("?T"))
register_function("product", (dtype("array<?T:numeric>"),), dtype("?T"))
register_function("toInt32", (dtype("?T:numeric"),), dtype("int32"))
register_function("extend", (dtype("array<?T>"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("argmin", (dtype("array<?T>"),), dtype("int32"))
register_function("toFloat64", (dtype("?T:numeric"),), dtype("float64"))
register_function("sort", (dtype("array<?T>"),), dtype("array<?T>"))
register_function("sort", (dtype("array<?T>"),dtype("bool"),), dtype("array<?T>"))
register_function("isSubset", (dtype("set<?T>"),dtype("set<?T>"),), dtype("bool"))
register_function("[*:*]", (dtype("str"),dtype("int32"),dtype("int32"),), dtype("str"))
register_function("[*:*]", (dtype("array<?T>"),dtype("int32"),dtype("int32"),), dtype("array<?T>"))
register_function("+", (dtype("array<?T:numeric>"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("+", (dtype("array<?T:numeric>"),dtype("?T"),), dtype("array<?T>"))
register_function("+", (dtype("?T:numeric"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("**", (dtype("?T:numeric"),dtype("array<?T>"),), dtype("array<float64>"))
register_function("**", (dtype("array<?T:numeric>"),dtype("array<?T>"),), dtype("array<float64>"))
register_function("**", (dtype("array<?T:numeric>"),dtype("?T"),), dtype("array<float64>"))
register_function("append", (dtype("array<?T>"),dtype("?T"),), dtype("array<?T>"))
register_function("[:*]", (dtype("str"),dtype("int32"),), dtype("str"))
register_function("[:*]", (dtype("array<?T>"),dtype("int32"),), dtype("array<?T>"))
register_function("remove", (dtype("set<?T>"),dtype("?T"),), dtype("set<?T>"))
register_function("[]", (dtype("str"),dtype("int32"),), dtype("str"))
register_function("indexArray", (dtype("array<?T>"),dtype("int32"),), dtype("?T"))
register_function("[]", (dtype("dict<?key, ?value>"),dtype("?key"),), dtype("?value"))
register_function("dictToArray", (dtype("dict<?key, ?value>"),), dtype("array<tuple(?key, ?value)>"))
register_function("%", (dtype("array<?T:numeric>"),dtype("?T"),), dtype("array<?T>"))
register_function("%", (dtype("?T:numeric"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("%", (dtype("array<?T:numeric>"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("dict", (dtype("array<tuple(?key, ?value)>"),), dtype("dict<?key, ?value>"))
register_function("dict", (dtype("set<tuple(?key, ?value)>"),), dtype("dict<?key, ?value>"))
register_function("keys", (dtype("dict<?key, ?value>"),), dtype("array<?key>"))
register_function("min", (dtype("array<?T:numeric>"),), dtype("?T"))
register_function("min", (dtype("set<?T:numeric>"),), dtype("?T"))
register_function("min", (dtype("?T"),dtype("?T"),), dtype("?T"))
register_function("sum", (dtype("set<?T:numeric>"),), dtype("?T"))
register_function("sum", (dtype("array<?T:numeric>"),), dtype("?T"))
register_function("toInt64", (dtype("?T:numeric"),), dtype("int64"))
register_function("contains", (dtype("dict<?key, ?value>"),dtype("?key"),), dtype("bool"))
register_function("contains", (dtype("array<?T>"),dtype("?T"),), dtype("bool"))
register_function("contains", (dtype("set<?T>"),dtype("?T"),), dtype("bool"))
register_function("-", (dtype("?T:numeric"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("-", (dtype("array<?T:numeric>"),dtype("?T"),), dtype("array<?T>"))
register_function("-", (dtype("array<?T:numeric>"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("addone", (dtype("int32"),), dtype("int32"))
register_function("isEmpty", (dtype("dict<?key, ?value>"),), dtype("bool"))
register_function("isEmpty", (dtype("array<?T>"),), dtype("bool"))
register_function("isEmpty", (dtype("set<?T>"),), dtype("bool"))
register_function("[:]", (dtype("array<?T>"),), dtype("array<?T>"))
register_function("[:]", (dtype("str"),), dtype("str"))
register_function("union", (dtype("set<?T>"),dtype("set<?T>"),), dtype("set<?T>"))
register_function("*", (dtype("array<?T:numeric>"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("*", (dtype("?T:numeric"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("*", (dtype("array<?T:numeric>"),dtype("?T"),), dtype("array<?T>"))
register_function("intersection", (dtype("set<?T>"),dtype("set<?T>"),), dtype("set<?T>"))
register_function("add", (dtype("set<?T>"),dtype("?T"),), dtype("set<?T>"))
register_function("argmax", (dtype("array<?T>"),), dtype("int32"))
register_function("//", (dtype("array<?T:numeric>"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("//", (dtype("array<?T:numeric>"),dtype("?T"),), dtype("array<?T>"))
register_function("//", (dtype("?T:numeric"),dtype("array<?T>"),), dtype("array<?T>"))
register_function("keySet", (dtype("dict<?key, ?value>"),), dtype("set<?key>"))
register_function("qnorm", (dtype("float64"),), dtype("float64"))
register_function("oneHotAlleles", (dtype("call"),dtype("int32"),), dtype("array<int32>"))
register_function("dpois", (dtype("float64"),dtype("float64"),dtype("bool"),), dtype("float64"))
register_function("dpois", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("ploidy", (dtype("call"),), dtype("int32"))
register_function("||", (dtype("bool"),dtype("bool"),), dtype("bool"))
register_function("ppois", (dtype("float64"),dtype("float64"),dtype("bool"),dtype("bool"),), dtype("float64"))
register_function("ppois", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("log10", (dtype("float64"),), dtype("float64"))
register_function("isHet", (dtype("call"),), dtype("bool"))
register_function("isAutosomalOrPseudoAutosomal", (dtype("?T:locus"),), dtype("bool"))
register_function("testCodeUnification", (dtype("?x:numeric"),dtype("?x:int32"),), dtype("?x"))
register_seeded_function("rand_pois", (dtype("float64"),), dtype("float64"))
register_seeded_function("rand_pois", (dtype("int32"),dtype("float64"),), dtype("array<float64>"))
register_function("toFloat32", (dtype("str"),), dtype("float32"))
register_function("toFloat32", (dtype("bool"),), dtype("float32"))
register_function("isAutosomal", (dtype("?T:locus"),), dtype("bool"))
register_function("isPhased", (dtype("call"),), dtype("bool"))
register_function("isHomVar", (dtype("call"),), dtype("bool"))
register_function("corr", (dtype("array<float64>"),dtype("array<float64>"),), dtype("float64"))
register_function("log", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("log", (dtype("float64"),), dtype("float64"))
register_function("foobar2", (), dtype("int32"))
register_function("approxEqual", (dtype("float64"),dtype("float64"),dtype("float64"),dtype("bool"),dtype("bool"),), dtype("bool"))
register_function("plDosage", (dtype("array<?N:int32>"),), dtype("float64"))
register_function("includesEnd", (dtype("interval<?T>"),), dtype("bool"))
register_function("position", (dtype("?T:locus"),), dtype("int32"))
register_seeded_function("rand_unif", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("str", (dtype("?T"),), dtype("str"))
register_function("valuesSimilar", (dtype("?T"),dtype("?T"),dtype('float64'),dtype('bool'),), dtype("bool"))
register_function("replace", (dtype("str"),dtype("str"),dtype("str"),), dtype("str"))
register_function("exp", (dtype("float64"),), dtype("float64"))
register_function("&&", (dtype("bool"),dtype("bool"),), dtype("bool"))
register_function("compare", (dtype("int32"),dtype("int32"),), dtype("int32"))
register_function("triangle", (dtype("int32"),), dtype("int32"))
register_function("Interval", (dtype("?T"),dtype("?T"),dtype("bool"),dtype("bool"),), dtype("interval<?T>"))
register_function("contig", (dtype("?T:locus"),), dtype("str"))
register_function("Call", (dtype("bool"),), dtype("call"))
register_function("Call", (dtype("str"),), dtype("call"))
register_function("Call", (dtype("int32"),dtype("bool"),), dtype("call"))
register_function("Call", (dtype("int32"),dtype("int32"),dtype("bool"),), dtype("call"))
register_function("Call", (dtype("array<int32>"),dtype("bool"),), dtype("call"))
register_function("qchisqtail", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("binomTest", (dtype("int32"),dtype("int32"),dtype("float64"),dtype("int32"),), dtype("float64"))
register_function("qpois", (dtype("float64"),dtype("float64"),), dtype("int32"))
register_function("qpois", (dtype("float64"),dtype("float64"),dtype("bool"),dtype("bool"),), dtype("int32"))
register_function("is_finite", (dtype("float32"),), dtype("bool"))
register_function("is_finite", (dtype("float64"),), dtype("bool"))
register_function("inYPar", (dtype("?T:locus"),), dtype("bool"))
register_function("contingency_table_test", (dtype("int32"),dtype("int32"),dtype("int32"),dtype("int32"),dtype("int32"),), dtype("struct{p_value: float64, odds_ratio: float64}"))
register_function("toInt32", (dtype("bool"),), dtype("int32"))
register_function("toInt32", (dtype("str"),), dtype("int32"))
register_function("foobar1", (), dtype("int32"))
register_function("toFloat64", (dtype("str"),), dtype("float64"))
register_function("toFloat64", (dtype("bool"),), dtype("float64"))
register_function("dbeta", (dtype("float64"),dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("min_rep", (dtype("?T:locus"),dtype("array<str>"),), dtype("struct{locus: ?T, alleles: array<str>}"))
register_function("toBoolean", (dtype("str"),), dtype("bool"))
register_seeded_function("rand_bool", (dtype("float64"),), dtype("bool"))
register_function("pchisqtail", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_seeded_function("rand_cat", (dtype("array<float64>"),), dtype("int32"))
register_function("inYNonPar", (dtype("?T:locus"),), dtype("bool"))
register_function("+", (dtype("str"),dtype("str"),), dtype("str"))
register_function("**", (dtype("float32"),dtype("float32"),), dtype("float64"))
register_function("**", (dtype("int32"),dtype("int32"),), dtype("float64"))
register_function("**", (dtype("int64"),dtype("int64"),), dtype("float64"))
register_function("**", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("length", (dtype("str"),), dtype("int32"))
register_function("slice", (dtype("str"),dtype("int32"),dtype("int32"),), dtype("str"))
register_function("split", (dtype("str"),dtype("str"),dtype("int32"),), dtype("array<str>"))
register_function("split", (dtype("str"),dtype("str"),), dtype("array<str>"))
register_seeded_function("rand_gamma", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("UnphasedDiploidGtIndexCall", (dtype("int32"),), dtype("call"))
register_function("[]", (dtype("call"),dtype("int32"),), dtype("int32"))
register_function("sign", (dtype("int64"),), dtype("int64"))
register_function("sign", (dtype("float64"),), dtype("float64"))
register_function("sign", (dtype("float32"),), dtype("float32"))
register_function("sign", (dtype("int32"),), dtype("int32"))
register_function("unphasedDiploidGtIndex", (dtype("call"),), dtype("int32"))
register_function("gamma", (dtype("float64"),), dtype("float64"))
register_function("%", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("%", (dtype("int64"),dtype("int64"),), dtype("int64"))
register_function("%", (dtype("float32"),dtype("float32"),), dtype("float32"))
register_function("%", (dtype("int32"),dtype("int32"),), dtype("int32"))
register_function("fisher_exact_test", (dtype("int32"),dtype("int32"),dtype("int32"),dtype("int32"),), dtype("struct{p_value: float64, odds_ratio: float64, ci_95_lower: float64, ci_95_upper: float64}"))
register_function("floor", (dtype("float64"),), dtype("float64"))
register_function("floor", (dtype("float32"),), dtype("float32"))
register_function("isNonRef", (dtype("call"),), dtype("bool"))
register_function("includesStart", (dtype("interval<?T>"),), dtype("bool"))
register_function("isHetNonRef", (dtype("call"),), dtype("bool"))
register_function("hardy_weinberg_test", (dtype("int32"),dtype("int32"),dtype("int32"),), dtype("struct{het_freq_hwe: float64, p_value: float64}"))
register_function("haplotype_freq_em", (dtype("array<int32>"),), dtype("array<float64>"))
register_function("nNonRefAlleles", (dtype("call"),), dtype("int32"))
register_function("abs", (dtype("float64"),), dtype("float64"))
register_function("abs", (dtype("float32"),), dtype("float32"))
register_function("abs", (dtype("int64"),), dtype("int64"))
register_function("abs", (dtype("int32"),), dtype("int32"))
register_function("endswith", (dtype("str"),dtype("str"),), dtype("bool"))
register_function("sqrt", (dtype("float64"),), dtype("float64"))
register_function("isnan", (dtype("float32"),), dtype("bool"))
register_function("isnan", (dtype("float64"),), dtype("bool"))
register_function("lower", (dtype("str"),), dtype("str"))
register_seeded_function("rand_beta", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_seeded_function("rand_beta", (dtype("float64"),dtype("float64"),dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("toInt64", (dtype("bool"),), dtype("int64"))
register_function("toInt64", (dtype("str"),), dtype("int64"))
register_function("testCodeUnification2", (dtype("?x"),), dtype("?x"))
register_function("contains", (dtype("str"),dtype("str"),), dtype("bool"))
register_function("contains", (dtype("interval<?T>"),dtype("?T"),), dtype("bool"))
register_function("entropy", (dtype("str"),), dtype("float64"))
register_function("filtering_allele_frequency", (dtype("int32"),dtype("int32"),dtype("float64"),), dtype("float64"))
register_function("gqFromPL", (dtype("array<?N:int32>"),), dtype("int32"))
register_function("startswith", (dtype("str"),dtype("str"),), dtype("bool"))
register_function("ceil", (dtype("float32"),), dtype("float32"))
register_function("ceil", (dtype("float64"),), dtype("float64"))
register_function("json", (dtype("?T"),), dtype("str"))
register_function("strip", (dtype("str"),), dtype("str"))
register_function("firstMatchIn", (dtype("str"),dtype("str"),), dtype("array<str>"))
register_function("isEmpty", (dtype("interval<?T>"),), dtype("bool"))
register_function("~", (dtype("str"),dtype("str"),), dtype("bool"))
register_function("mkString", (dtype("set<str>"),dtype("str"),), dtype("str"))
register_function("mkString", (dtype("array<str>"),dtype("str"),), dtype("str"))
register_function("dosage", (dtype("array<?N:float64>"),), dtype("float64"))
register_function("upper", (dtype("str"),), dtype("str"))
register_function("overlaps", (dtype("interval<?T>"),dtype("interval<?T>"),), dtype("bool"))
register_function("downcode", (dtype("call"),dtype("int32"),), dtype("call"))
register_function("inXPar", (dtype("?T:locus"),), dtype("bool"))
register_function("format", (dtype("str"),dtype("?T:tuple"),), dtype("str"))
register_function("pnorm", (dtype("float64"),), dtype("float64"))
register_function("is_infinite", (dtype("float32"),), dtype("bool"))
register_function("is_infinite", (dtype("float64"),), dtype("bool"))
register_function("isHetRef", (dtype("call"),), dtype("bool"))
register_function("isMitochondrial", (dtype("?T:locus"),), dtype("bool"))
register_function("hamming", (dtype("str"),dtype("str"),), dtype("int32"))
register_function("end", (dtype("interval<?T>"),), dtype("?T"))
register_function("start", (dtype("interval<?T>"),), dtype("?T"))
register_function("inXNonPar", (dtype("?T:locus"),), dtype("bool"))
register_function("escapeString", (dtype("str"),), dtype("str"))
register_function("isHomRef", (dtype("call"),), dtype("bool"))
register_seeded_function("rand_norm", (dtype("float64"),dtype("float64"),), dtype("float64"))
register_function("chi_squared_test", (dtype("int32"),dtype("int32"),dtype("int32"),dtype("int32"),), dtype("struct{p_value: float64, odds_ratio: float64}"))
def _from_java(jtt):
return ttable(
dtype(jtt.globalType().toString()),
dtype(jtt.rowType().toString()),
jiterable_to_list(jtt.key()))
def _from_java(jtbm):
return tblockmatrix(dtype(jtbm.elementType().toString()),
jiterable_to_list(jtbm.shape()), jtbm.blockSize(),
jiterable_to_list(jtbm.dimsPartitioned()))
def register_reference_genome_functions(rg):
from hail.expr.types import dtype
tvariant = dtype(f"struct{{locus:locus<{rg}>,alleles:array<str>}}")
tinterval = dtype(f"interval<locus<{rg}>>")
register_function(f"Locus({rg})", (dtype("str"),), dtype(f"locus<{rg}>"))
register_function(f"Locus({rg})", (dtype("str"),dtype("int32"),), dtype(f"locus<{rg}>"))
register_function(f"LocusAlleles({rg})", (dtype("str"),), tvariant)
register_function(f"LocusInterval({rg})", (dtype("str"),), tinterval)
register_function(f"LocusInterval({rg})", (dtype("str"),dtype("int32"),dtype("int32"),dtype("bool"),dtype("bool"),), tinterval)
register_function(f"isValidContig({rg})", (dtype("str"),), dtype("bool"))
register_function(f"isValidLocus({rg})", (dtype("str"),dtype("int32"),), dtype("bool"))
register_function(f"getReferenceSequenceFromValidLocus({rg})", (dtype("str"),dtype("int32"),dtype("int32"),dtype("int32"),), dtype("str"))
register_function(f"getReferenceSequence({rg})", (dtype("str"),dtype("int32"),dtype("int32"),dtype("int32"),), dtype("str"))
register_function(f"globalPosToLocus({rg})", (dtype("int64"),), dtype(f"locus<{rg}>"))
register_function(f"locusToGlobalPos({rg})", (dtype(f"locus<{rg}>"),), dtype("int64"))
