-
Notifications
You must be signed in to change notification settings - Fork 0
/
alanine_simulation.py
163 lines (126 loc) · 6.02 KB
/
alanine_simulation.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
"""Forward simulation of alanine dipeptide."""
import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
import warnings
warnings.filterwarnings('ignore') # disable warnings about float64 usage
import cloudpickle as pickle
from pathlib import Path
import time
from jax import vmap, random, tree_util, numpy as jnp
from jax_md import space
import numpy as onp
from chemtrain.jax_md_mod import io, custom_space, custom_quantity
from chemtrain import data_processing, traj_util
from util import Initialization
import visualization
Path('output/trajectories').mkdir(parents=True, exist_ok=True)
# user input
save_name = 'FM_2fs_100ns'
folder_name = 'FM_2fs_100ns/'
labels = ['Reference', 'Predicted']
file_topology = 'data/confs/heavy_2_7nm.gro'
configuration_str = 'data/dataset/confs_heavy_100ns.npy'
used_dataset_size = 500000
n_trajectory = 50
model = 'CGDimeNet'
saved_params_path = ('output/force_matching/'
'trained_params_alanine_FM_model_alanine.pkl')
# saved_params_path = ('output/rel_entropy/'
# 'trained_params_alanine_RE_model_alanine.pkl')
# simulation parameters
system_temperature = 300 # Kelvin
boltzmann_constant = 0.0083145107 # in kJ / mol K
kbt = system_temperature * boltzmann_constant
time_step = 0.002
total_time = 110000
t_equilib = 10000.
print_every = 0.2
###############################
box, _, masses, _ = io.load_box(file_topology)
priors = ['bond', 'angle', 'dihedral']
species, prior_idxs, prior_constants = Initialization.select_protein(
'heavy_alanine_dipeptide', priors)
# random starting points
position_data = data_processing.get_dataset(configuration_str)[1:]
key = random.PRNGKey(0)
r_init = random.choice(key, position_data, (n_trajectory,), replace=False)
simulation_data = Initialization.InitializationClass(
r_init=r_init, box=box, kbt=kbt, masses=masses, dt=time_step,
species=species)
timings = traj_util.process_printouts(time_step, total_time, t_equilib,
print_every)
reference_state, energy_params, simulation_fns, compute_fns, targets = \
Initialization.initialize_simulation(simulation_data,
model,
integrator='Langevin',
prior_constants=prior_constants,
prior_idxs=prior_idxs)
simulator_template, energy_fn_template, neighbor_fn = simulation_fns
if saved_params_path is not None:
print('Using saved params')
with open(saved_params_path, 'rb') as pickle_file:
params = pickle.load(pickle_file)
energy_params = tree_util.tree_map(jnp.array, params)
trajectory_generator = traj_util.trajectory_generator_init(simulator_template,
energy_fn_template,
timings)
t_start = time.time()
traj_state = trajectory_generator(energy_params, reference_state)
t_end = time.time() - t_start
print('total runtime in min:', t_end / 60.)
assert not traj_state.overflow, ('Neighborlist overflow during trajectory '
'generation. Increase capacity and re-run.')
# postprocessing
traj_positions = traj_state.trajectory.position
jnp.save(f'output/trajectories/confs_alanine_{save_name}', traj_positions)
box_tensor, _ = custom_space.init_fractional_coordinates(box)
displacement, _ = space.periodic_general(box_tensor,
fractional_coordinates=True)
position_data = data_processing.scale_dataset_fractional(position_data, box)
# dihedrals
dihedral_idxs = jnp.array([[1, 3, 4, 6], [3, 4, 6, 8]]) # 0: phi 1: psi
batched_dihedrals = vmap(custom_quantity.dihedral_displacement, (0, None, None))
dihedrals_ref = batched_dihedrals(position_data, displacement, dihedral_idxs)
dihedral_angles = batched_dihedrals(traj_positions, displacement, dihedral_idxs)
phi = dihedral_angles[:, 0].reshape((n_trajectory, -1))
psi = dihedral_angles[:, 1].reshape((n_trajectory, -1))
# mean squared error (in rad)
nbins = 60
dihedrals_ref_rad = jnp.deg2rad(dihedrals_ref)
dihedral_angles_rad = jnp.deg2rad(dihedral_angles)
h_ref, y1, y2 = onp.histogram2d(
dihedrals_ref_rad[:, 0], dihedrals_ref_rad[:, 1], bins=nbins, density=True)
h_pred, _, _ = onp.histogram2d(
dihedral_angles_rad[:, 0], dihedral_angles_rad[:, 1], bins=nbins,
density=True)
mse = onp.mean((h_ref - h_pred)**2)
print('MSE of the phi-psi dihedral density histrogram: ', mse)
# unstack parallel trajectories
dihedral_angles_split = dihedral_angles.reshape((n_trajectory, -1, 2))
# Plots
phi_angles_ref = onp.load('data/dataset/phi_angles_r100ns.npy')
psi_angles_ref = onp.load('data/dataset/psi_angles_r100ns.npy')
# dihedral histograms
Path(f'output/postprocessing/{folder_name}').mkdir(parents=True, exist_ok=True)
visualization.plot_histogram_density(dihedral_angles_split[0, :],
save_name + '_first_predicted_',
folder=folder_name)
visualization.plot_histogram_density(dihedrals_ref, save_name + '_REF',
folder=folder_name)
visualization.plot_1d_dihedral(
[phi_angles_ref, phi], 'phi_' + save_name, labels=labels[0:2],
folder=folder_name)
visualization.plot_1d_dihedral(
[psi_angles_ref, psi], 'psi_' + save_name, location='upper left',
labels=labels[0:2], xlabel='$\psi$ in deg', folder=folder_name)
visualization.plot_histogram_free_energy(dihedral_angles_split[0, :],
save_name + '_pred',
kbt, folder=folder_name)
visualization.plot_histogram_free_energy(dihedrals_ref, save_name + '_REF',
kbt, folder=folder_name)
visualization.plot_compare_histogram_free_energy(
[dihedrals_ref, dihedral_angles_split[0, :]], save_name, kbt, titles=labels,
folder=folder_name)
visualization.plot_compare_histogram_density(
[dihedrals_ref, dihedral_angles_split[0, :]], save_name, titles=labels,
folder=folder_name)