Homec4science
Diffusion G(roup)-RENAISSANCE (master) Browse

grenaissance/docker/work/configfile.ini
57772ff3a7f5master

Last Change
Raw File

configfile.ini
View Options

[MLP]
## Neural network hyperparameters
# Number of dimensions in generator seed (Input dimension)
latent_dim = 99
# Number of Km parameters in the model (No of parameters to be generated, Output dimension)
no_kms = 384
# Number of parameter sets to be generated each time
n_samples = 100
# Number of neurons in each layer of the generator
layer_1 = 256
layer_2 = 512
layer_3 = 1024
[PATHS]
# Metabolic model being analyzed. All data are loaded accordingly from the folder '/Models' in supplementary data
met_model = varma_ecoli_shikki
# Output folder for renaissance (stores reward)
output_path = output/rnsc_opt/
[CONSTRAINTS]
## Custom constraints
# Minimum value of generated Kms = exp(-25)
min_km = -25
# Maximum value of generated Kms = exp(3)
max_km = 3
[EVOSTRAT]
# Number of statistical repeats
repeats = 10
# Saving step (How often to save weights during optimisation process)
save_step = 1
# Number of evolution generations
generations = 25
# Population size of generators
pop_size = 20
# Noise (Search Radius)
noise = 0.01
# Learning rate
lr = 1e-3
#learning rate decay per generation (0.95 means 5% decay per gen)
decay = 0.95
# Indexes of the steady state to generate kinetic models around (there are 5000 steady states in the data provided)
ss_idx = 4882, 4552, 444, 4414, 3466, 2349, 285, 609, 1517
# 2445, 3873, 4666, 4149, 2624, 1023, 3195, 1719, 744, 4015, 1749, 2081
# 2445, 4882, 3873, 4552, 444, 4666, 4149, 4414, 3466, 2624, 1023, 3195, 1719, 744, 2349, 4015, 1749, 285, 609, 2081, 1517
# 2424, 4833, 3841, 4530, 459, 4679, 4140, 4465, 3404, 1665, 958, 3236, 1724, 767, 2360, 4016, 1733, 251, 620, 2123, 1473 median representatives
# 2445, 4882, 3873, 4552, 444, 4666, 4149, 4414, 3466, 2624, 1023, 3195, 1719, 744, 2349, 4015, 1749, 285, 609, 2081, 1517 # most representatives
# 1011, 11, 1938, 1093, 4352, 82, 1288
# 1011, 4843, 11, 744, 2164, 3466, 3117, 2627, 1498, 2445, 4358, 3871, 1938, 3733, 444, 2920, 1093, 4532, 82, 3250, 1288
# 2435, 2732, 4721, 1074, 3571, 3928, 3987, 983, 3288, 4120
# 709, 733, 734, 741, 755, 3995, 4000, 4002, 4005, 4015
# 800, 767, 812, 734, 789, 733, 755, 797, 709, 741
# 4008, 4025, 4015, 4000, 3995, 3999, 4014, 4002, 4005, 4023
# 1195, 1196, 1197, 1198, 1200
# 3865, 3866, 3867, 3869, 3870
# Number of threads
n_threads = 32
# Printing step (How often to print current state of the optimisation process)
print_step = 5
[REWARDS]
# Desired optimisation objective ( Flag 0 - Push for max incidence of valid models; Flag 1 - Push for fastest dynamic possible)
reward_flag = 0
# Number of fastest eigenvalues to consider when Flag = 1
n_consider = 10
# eigenvalue partition between valid and invalid models
eig_partition = -2.5
[PARAMETER_FIXING]
# Parameter fixing flag ( 1 - Enabled/ 0 - Disabled)
pf_flag = 1
# Path to fixed parameter names (must be .csv, must follow naming convention of kinetic model being used)
fixed_parameter_names = data/param_fixing/param_fixing/io_files/brenda_names.csv
# Path to fixed parameter ranges (must be numpy array of min and max km of every fixed km in mM)
fixed_parameter_ranges = data/param_fixing/param_fixing/io_files/brenda_ranges.npy
[LOAD_AND_GENERATE]
# path to saved weights #output/toy_data/saved_weights.pkl#
path_to_weights =
# Number of parameter sets to generate (1 million sets takes ~20s)
n_sets = 10
# Output path yo save as .hdf5 format
path_output = output/toy_data/
# Output name
name_output = gen_x_parameters
#
[ODE_INTEGRATION]
# Path to full parameter set (must be .hdf5 format, see step above)
path_to_parameter_set = output/toy_data/gen_x_parameters.hdf5
# Percent perturbation (perturbation to steady state, 0.1 means steady state conditions will be pertubed upto +/- 10% before integration and used as initial condition)
pc_perturbed = 0.25
# Output path yo save as .hdf5 format
path_output = output/toy_data/
# Output name
name_output = gen_x_parameters
[BIOREACTOR_SIMULATION]
path_to_parameter_set = output/toy_data/gen_x_parameters.hdf5
# File containing all the initial conditions for bioreactors from experimental data
path_to_initiator = models/bioreactor_initiator.yaml
# Output path
path_output = output/toy_data/
# Output name
name_output = gen_x_parameters
c4science · Help