-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathICA_astrocyte.py
182 lines (137 loc) · 8.58 KB
/
ICA_astrocyte.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()
import numpy as np
class gpu_astro:
def __init__(self, num_astros, max_syns_per_astro,input_morph_matrix
, glutamate_per_synapse = 0.2 # glutamate concentration
, ip3_init = 0.148465 # steady state GCh-I value
, C_init = 0.0698025 # steady state GCh-I value
, h_init = 0.793086 # steady state GCh-I value
):
self.num_astros = num_astros # compatible with only 1 astrocyte
self.syns_per_astro = max_syns_per_astro
self.shp = num_astros
self.input_shp = (num_astros,max_syns_per_astro)
# INITIAL GCH-I MODEL STATE VARIABLE VALUES
self.ip3_init = ip3_init
self.ca_init = C_init
self.h_init = h_init
# -------------- GCH-I MODEL PARAMETERS ---------------
self.r_C = 6.0
self.r_L = 0.11
self.C_0 = 2.0
self.c_1 = 0.185
self.K_ER = 0.05 # set to FM value
self.d_1 = 0.13
self.d_2 = 1.049
self.d_3 = 0.9434
self.d_5 = 0.08234
self.v_delta = 0.05 # set to FM value
self.K_PLC_delta = 0.1
self.k_delta = 1.5
self.v_ER = 0.9
self.r_5P = 0.05 # set to FM value
self.v_3K = 2.0
self.K_D = 0.7
self.K_3 = 1.0
self.a_2 = 0.2
self.K_R = 1.3
self.K_P = 10.0
self.K_pi = 0.6
# RESCALING RATE VARIABLES FROM SECONDS TO MILLISECONDS
self.r_C = self.r_C * (1 / 1000)
self.r_L = self.r_L * (1 / 1000)
self.v_ER = self.v_ER * (1 / 1000)
self.a_2 = self.a_2 * (1 / 1000)
self.v_delta = self.v_delta * (1 / 1000)
self.r_5P = self.r_5P * (1 / 1000)
self.v_3K = self.v_3K * (1 / 1000)
self.glutamate_per_syn = glutamate_per_synapse
# PRECOMPUTED VALUES FOR GCH-I SIMULATION
self.c_1_plus_1 = 1.0 + self.c_1
def initialize_vars(self):
# INITIALIZE TENSORFLOW DATA STRUCTURES FOR GCH-I
ip3_state = tf.Variable(self.ip3_init * np.ones(self.shp), dtype=tf.float32,
expected_shape=self.shp)
ip3_store = tf.Variable(np.ones(self.shp), dtype=tf.float32,
expected_shape=self.shp)
ca_state = tf.Variable(self.ca_init * np.ones(self.shp), dtype=tf.float32,
expected_shape=self.shp)
ca_store = tf.Variable(np.ones(self.shp), dtype=tf.float32, expected_shape=self.shp)
h_state = tf.Variable(self.h_init * np.ones(self.shp), dtype=tf.float32,
expected_shape=self.shp)
return ip3_state,ca_state,h_state,ip3_store,ca_store
def initialize_ph(self):
syn_input_feed = tf.placeholder(dtype=tf.float32,shape=self.input_shp)
input_morph_feed = tf.placeholder(dtype=tf.float32,shape=self.input_shp)
v_beta_feed = tf.placeholder(dtype=tf.float32,shape=self.input_shp)
return syn_input_feed,input_morph_feed,v_beta_feed
def reset_var(self,var,new_val_scalar):
return tf.assign(var,tf.add(new_val_scalar,tf.scalar_mul(0.0,var)))
def run_ip3_state_transition(self, L1_ip3_state, L1_ca_state, syn_inp,v_beta_var,input_morph):
'''
:param L1_ip3_state: IP3 CONCENTRATION OF PREVIOUS ms TIMESTEP
:param L1_ca_state: CA CONCENTRATION OF PREVIOUS ms TIMESTEP
:param syn_inp: GLUTAMATE CONCENTRATION AT CURRENT ms TIMESTEP FOR SET OF SYNAPTIC INPUTS
:param v_beta_var: GLUTAMATE RECEPTOR DENSITY (WEIGHT) OF EACH SYNAPTIC INPUT FOR SET OF SYNAPTIC INPUTS
:param input_morph: MASK FOR EXISTING TRIPARTITE CONNECTIONS
:return: :return: COMPUTES [IP3] FOR CURRENT ms TIMESTEP
'''
delta_ip3 = tf.add_n([
tf.reduce_sum(tf.multiply(input_morph,tf.multiply(v_beta_var,
tf.divide(tf.pow(syn_inp, 0.7), tf.add(tf.pow(syn_inp, 0.7),
tf.tile(tf.expand_dims(tf.pow(
tf.add(self.K_R,
tf.scalar_mul(self.K_P,
tf.divide(L1_ca_state, tf.add(L1_ca_state, self.K_pi)))),
0.7),axis=1),[1,self.syns_per_astro]))))),axis=1),
tf.multiply(tf.divide(self.v_delta, tf.add(tf.divide(L1_ip3_state, self.k_delta), 1.0)),
tf.divide(tf.pow(L1_ca_state, 2.0),
tf.add(tf.pow(L1_ca_state, 2.0), tf.pow(self.K_PLC_delta, 2.0)))),
tf.scalar_mul(tf.negative(self.v_3K),
tf.multiply(tf.divide(tf.pow(L1_ca_state, 4.0),
tf.add(tf.pow(L1_ca_state, 4.0), tf.pow(self.K_D, 4.0))),
tf.divide(L1_ip3_state, tf.add(L1_ip3_state, self.K_3)))),
tf.scalar_mul(tf.negative(self.r_5P), L1_ip3_state)])
return tf.add(L1_ip3_state, delta_ip3)
def run_ca_state_transition(self, L1_ip3_state, L1_ca_state, L1_h_state):
'''
:param L1_ip3_state: IP3 CONCENTRATION OF PREVIOUS ms TIMESTEP
:param L1_ca_state: CA CONCENTRATION OF PREVIOUS ms TIMESTEP
:param L1_h_state: H VALUE OF PREVIOUS ms TIMESTEP
:return: COMPUTES [CA] FOR CURRENT ms TIMESTEP
'''
delta_ca = tf.add_n([tf.multiply(self.r_C, tf.multiply(tf.pow(tf.divide(L1_ip3_state,
tf.add(L1_ip3_state,
self.d_1)), 3.0),
tf.multiply(tf.pow(tf.divide(L1_ca_state,
tf.add(L1_ca_state,
self.d_5)),
3.0),
tf.multiply(tf.pow(L1_h_state
, 3.0),
tf.subtract(self.C_0,
tf.scalar_mul(
self.c_1_plus_1,
L1_ca_state
)))))),
tf.scalar_mul(self.r_L,
tf.subtract(self.C_0, tf.scalar_mul(self.c_1_plus_1, L1_ca_state
))),
tf.multiply(tf.negative(self.v_ER), tf.divide(tf.pow(L1_ca_state
, 2.0), tf.add(tf.pow(L1_ca_state
, 2.0),
tf.pow(self.K_ER,
2.0))))])
return tf.add(L1_ca_state, delta_ca)
def run_h_state_transition(self, L1_ip3_state, L1_ca_state, L1_h_state):
'''
:param L1_ip3_state: IP3 CONCENTRATION OF PREVIOUS ms TIMESTEP
:param L1_ca_state: CA CONCENTRATION OF PREVIOUS ms TIMESTEP
:param L1_h_state: H VALUE OF PREVIOUS ms TIMESTEP
:return: :return: COMPUTES H FOR CURRENT ms TIMESTEP
'''
Q2_L1 = tf.scalar_mul(self.d_2, tf.divide(tf.add(L1_ip3_state, self.d_1), tf.add(L1_ip3_state, self.d_3)))
delta_h = tf.divide(tf.subtract(tf.divide(Q2_L1, tf.add(Q2_L1, L1_ca_state)), L1_h_state),
tf.divide(1.0, tf.scalar_mul(self.a_2, tf.add(Q2_L1, L1_ca_state))))
return tf.add(L1_h_state, delta_h)