# Copyright 2020. Allen Institute. All rights reserved
#
# Redistribution and use in source and binary forms, with or without modification, are permitted provided that the
# following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following
# disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following
# disclaimer in the documentation and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote
# products derived from this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
# INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
import types
import six
import warnings
from functools import wraps
class _PyFunctions(object):
"""Structure that allows users to add functions that will overwrite certain parts of bmtk, in particular; how
cell-objects are created, how cells are processed, how synapatic objects are created, and how synaptic weights are
calculated. To use the user must write a custom function in their own scripts and register it using the add()
method or decorator. ex run_bionet.py::
from bmtk.simulator.bionet import synaptic_weight
@synaptic_weight
def calc_custom_weight(edge_props, src_cell, trg_cell):
syn_weight = ... # calulate updated weight using edge and cells props
return syn_weight
This will register the user function "calc_custom_weight(...)" in the py_modules object. Any edge object that has
attribute weight_function=="calc_custom_weight" will use this new user-defined function to readjust weights. ie
in the bmtk code where syn_weight value is calculate for each edge::
from bmtk.simulator.bionet.pyfunction_cache import py_modules
if 'weight_function' in edge_props:
fnc_name = edge_props['weight_function']
assert(py_modules.has_synaptic_weight(name=fnc_name)
weight_fnc = py_modules.synaptic_weight(name=fnc_name)
syn_weight = weight_fnc(edge_props, src_cell, trg_cell)
else:
syn_weight = ... # default way to get weight
The following functionality can be overwritten by the user.
cell_model: function to create a NEURON/NEST/DiPDE cell object(s). Uses SONATA properties "model_template" and
"model_type".
cell_processor: Used to do some post-processing of cell object after being created, ex fixing part of morphology
or updating channel variable parameters. Uses "model_processing" property.
synapse_model: Used to create a synapse between two (sets of) cells. Uses SONATA edges "model_template"
property.
synaptic_weight: Used to update/recalculate synaptic weight of an edge (synapse or junction). Uses
"weight_function" (not a default SONATA variable).
"""
def __init__(self):
self.__syn_weights = {}
self.__cell_models = {}
self.__synapse_models = {}
self.__cell_processors = {}
def clear(self):
self.__syn_weights.clear()
self.__cell_models.clear()
self.__synapse_models.clear()
self.__cell_processors.clear()
def add_synaptic_weight(self, name, func, overwrite=True):
"""stores synaptic function for given name"""
if overwrite or name not in self.__syn_weights:
self.__syn_weights[name] = func
@property
def synaptic_weights(self):
"""return list of the names of all available synaptic weight functions"""
return self.__syn_weights.keys()
def synaptic_weight(self, name):
"""return the synaptic weight function"""
return self.__syn_weights[name]
def has_synaptic_weight(self, name):
return name in self.__syn_weights
def __cell_model_key(self, directive, model_type):
return (directive, model_type)
def add_cell_model(self, directive, model_type, func, overwrite=True):
if not model_type:
# use * to indicate "model_type" is not specified and therefore a wildcard option.
model_type = '*'
key = self.__cell_model_key(directive, model_type)
if overwrite or key not in self.__cell_models:
self.__cell_models[key] = func
@property
def cell_models(self):
return self.__cell_models.keys()
def cell_model(self, directive, model_type='*'):
# Check to see if function exists with corresponding "directive" and "model_type". If not see if there is a
# function with "directive" but "model_type=*" to act as fall-through option
if self.has_cell_model(directive=directive, model_type=model_type):
return self.__cell_models[self.__cell_model_key(directive, model_type)]
elif model_type != '*' and self.has_cell_model(directive=directive, model_type='*'):
return self.__cell_models[self.__cell_model_key(directive, '*')]
else:
raise ValueError('Could not find cell_model() function with directive="{}" and model_type="{}"'.format(
directive, model_type
))
def has_cell_model(self, directive, model_type='*'):
if not model_type:
model_type = '*'
return self.__cell_model_key(directive, model_type) in self.__cell_models
def add_synapse_model(self, name, func, overwrite=True):
if overwrite or name not in self.__synapse_models:
self.__synapse_models[name] = func
def has_synapse_model(self, name):
return name in self.__synapse_models
@property
def synapse_models(self):
return self.__synapse_models.keys()
def synapse_model(self, name):
return self.__synapse_models[name]
def has_cell_processor(self, name):
return name in self.__cell_processors
@property
def cell_processors(self):
return self.__cell_processors.keys()
def cell_processor(self, name):
return self.__cell_processors[name]
def add_cell_processor(self, name, func, overwrite=True):
if overwrite or name not in self.__syn_weights:
self.__cell_processors[name] = func
def __repr__(self):
rstr = '{}: {}\n'.format('cell_models', self.cell_models)
rstr += '{}: {}\n'.format('synapse_models', self.synapse_models)
rstr += '{}: {}\n'.format('synaptic_weights', self.synaptic_weights)
rstr += '{}: {}'.format('cell_processors', self.cell_processors)
return rstr
py_modules = _PyFunctions()
[docs]def synaptic_weight(*wargs, **wkwargs):
"""A decorator for registering a function as a synaptic weight function.
To use either::
@synaptic_weight
def weight_function():
...
or::
@synaptic_weight(name='name_in_edge_types')
def weight_function():
...
Once the decorator has been attached and imported the functions will automatically be added to py_modules and BMTK
will when assigning synaptic/gap junction weights for edges with matching "weight_function" attribute
"""
if len(wargs) == 1 and callable(wargs[0]):
# for the case without decorator arguments, grab the function object in wargs and create a decorator
func = wargs[0]
py_modules.add_synaptic_weight(func.__name__, func) # add function assigned to its original name
@wraps(func)
def func_wrapper(*args, **kwargs):
return func(*args, **kwargs)
return func_wrapper
else:
# for the case with decorator arguments
assert(all(k in ['name'] for k in wkwargs.keys()))
def decorator(func):
# store the function in py_modules but under the name given in the decorator arguments
py_modules.add_synaptic_weight(wkwargs['name'], func)
@wraps(func)
def func_wrapper(*args, **kwargs):
return func(*args, **kwargs)
return func_wrapper
return decorator
[docs]def cell_model(*wargs, **wkwargs):
"""A decorator for registering NEURON cell loader functions."""
if len(wargs) == 1 and callable(wargs[0]):
# for the case without decorator arguments, grab the function object in wargs and create a decorator
func = wargs[0]
py_modules.add_cell_model(
directive=func.__name__, # add function assigned to its original name
model_type='*',
func=func
)
@wraps(func)
def func_wrapper(*args, **kwargs):
return func(*args, **kwargs)
return func_wrapper
else:
# look for directive/name argument inside decorator arguments, otherwise default
if 'directive' in wkwargs:
directive = wkwargs['directive']
elif 'name' in wkwargs:
directive = wkwargs['name']
elif len(wargs) >= 1 and isinstance(wargs[0], six.string_types):
directive = wargs[0]
elif len(wargs) >= 1 and callable(wargs[0]):
directive = wargs[0].__name__
else:
raise ValueError('Please specify "directive" name in call_model() arguments')
model_type = wkwargs.get('model_type', '*')
overwrite = wkwargs.get('overwrite', True)
def decorator(func):
# store the function in py_modules but under the name given in the decorator arguments
py_modules.add_cell_model(directive=directive, model_type=model_type, func=func, overwrite=overwrite)
@wraps(func)
def func_wrapper(*args, **kwargs):
return func(*args, **kwargs)
return func_wrapper
return decorator
[docs]def synapse_model(*wargs, **wkwargs):
"""A decorator for registering NEURON synapse loader functions."""
if len(wargs) == 1 and callable(wargs[0]):
# for the case without decorator arguments, grab the function object in wargs and create a decorator
func = wargs[0]
py_modules.add_synapse_model(func.__name__, func) # add function assigned to its original name
@wraps(func)
def func_wrapper(*args, **kwargs):
return func(*args, **kwargs)
return func_wrapper
else:
# for the case with decorator arguments
assert(all(k in ['name'] for k in wkwargs.keys()))
def decorator(func):
# store the function in py_modules but under the name given in the decorator arguments
py_modules.add_synapse_model(wkwargs['name'], func)
@wraps(func)
def func_wrapper(*args, **kwargs):
return func(*args, **kwargs)
return func_wrapper
return decorator
[docs]def add_weight_function(func, name=None, overwrite=True):
assert(callable(func))
func_name = name if name is not None else func.__name__
py_modules.add_synaptic_weight(func_name, func, overwrite)
[docs]def add_cell_model(func, directive, model_type='*', overwrite=True):
assert(callable(func))
py_modules.add_cell_model(directive=directive, model_type=model_type, func=func, overwrite=overwrite)
[docs]def add_cell_processor(func, name=None, overwrite=True):
assert(callable(func))
func_name = name if name is not None else func.__name__
py_modules.add_cell_processor(func_name, func, overwrite)
[docs]def add_synapse_model(func, name=None, overwrite=True):
assert (callable(func))
func_name = name if name is not None else func.__name__
py_modules.add_synapse_model(func_name, func, overwrite)
[docs]def load_py_modules(cell_models=None, syn_models=None, syn_weights=None, cell_processors=None):
# py_modules.clear()
warnings.warn('Do not call this method directly', DeprecationWarning)
if cell_models is not None:
assert(isinstance(cell_models, types.ModuleType))
for f in [cell_models.__dict__.get(f) for f in dir(cell_models)]:
if isinstance(f, types.FunctionType):
py_modules.add_cell_model(f.__name__, f)
if syn_models is not None:
assert(isinstance(syn_models, types.ModuleType))
for f in [syn_models.__dict__.get(f) for f in dir(syn_models)]:
if isinstance(f, types.FunctionType):
py_modules.add_synapse_model(f.__name__, f)
if syn_weights is not None:
assert(isinstance(syn_weights, types.ModuleType))
for f in [syn_weights.__dict__.get(f) for f in dir(syn_weights)]:
if isinstance(f, types.FunctionType):
py_modules.add_synaptic_weight(f.__name__, f)
if cell_processors is not None:
assert(isinstance(cell_processors, types.ModuleType))
for f in [cell_processors.__dict__.get(f) for f in dir(cell_processors)]:
if isinstance(f, types.FunctionType):
py_modules.add_cell_processor(f.__name__, f)
