def plotCost(number_points=0):
rand_samples = []
for model_index in range(num_models_regions):
region = model_regions[model_index]
if number_points:
samples = region.points[np.random.choice(
region.points.shape[0], number_points, replace=False), :]
else:
samples = region.points
rand_samples.append(samples)
all_costs = []
_, axes = plt.subplots(1, 3)
for box_plot in range(num_models_fitness):
ax = axes.flat[box_plot]
#plt.subplot(1,3,box_plot+1)
ax.set_title(str(box_plot + 1) + '-bit model')
ax.set_xlabel('region id')
if box_plot == 0:
ax.set_ylabel('cost')
model = BioProc(np.array([
"protein_production", "protein_production", "protein_production",
"protein_production", "protein_degradation", "protein_degradation",
"Kd", "hill", "protein_production", "protein_degradation", "Kd",
"hill"
]),
model_mode=models[box_plot])
model_evals = []
for model_index in range(num_models_regions):
evals = []
for sample in rand_samples[model_index]:
evals.append(-model.eval(sample)[0])
model_evals.append(evals)
#plt.boxplot(model_evals)
ax.violinplot(model_evals)
ax.boxplot(model_evals)
ax.set_xticks([1, 2, 3])
all_costs.append(model_evals)
#dump data to file
f = open(os.path.join(base_path_robustness, "cost_distrib.p"), "wb")
pickle.dump(all_costs, f)
f.close()
plt.savefig(os.path.join(base_path_robustness, 'cost_distrib.pdf'),
bbox_inches='tight')
plt.show()
def calculateVolumes(model_indexes=None):
if model_indexes == None:
model_indexes = range(num_models_regions)
elif type(model_indexes) == int:
model_indexes = [model_indexes]
for model_index in model_indexes:
model = BioProc(np.array([
"protein_production", "protein_production", "protein_production",
"protein_production", "protein_degradation", "protein_degradation",
"Kd", "hill", "protein_production", "protein_degradation", "Kd",
"hill"
]),
model_mode=models[model_index],
avg_dev=30)
print(model.threshold)
solver = Solver(model)
_, vol_desc = solver.getViableVolume([model_regions[model_index]])
model_str = '0' + str(model_index + 1) + '_'
f = open(
os.path.join(base_path_robustness,
model_str + "viable_volume.txt"), "w")
f.write(vol_desc)
f.close()
def calculateVolumesRepeats(model_indexes=None, repeats=10):
if model_indexes == None:
model_indexes = range(num_models_regions)
elif type(model_indexes) == int:
model_indexes = [model_indexes]
df = pd.DataFrame(columns=["Model id", "Volume", "Total", "Ratio"])
for model_index in model_indexes:
model = BioProc(np.array([
"protein_production", "protein_production", "protein_production",
"protein_production", "protein_degradation", "protein_degradation",
"Kd", "hill", "protein_production", "protein_degradation", "Kd",
"hill"
]),
model_mode=models[model_index],
avg_dev=30)
print(model.threshold)
solver = Solver(model)
for _ in range(repeats):
(volume, total,
ratio), _ = solver.getViableVolume([model_regions[model_index]])
df = df.append(
{
"Model id": model_index + 1,
"Volume": volume,
"Total": total,
"Ratio": ratio
},
ignore_index=True)
df.to_csv(os.path.join(base_path_robustness, 'volumes.csv'), index=False)
def getCosts(number_points=0, file_name=""):
rand_samples = []
for model_index in range(num_models_regions):
region = model_regions[model_index]
if number_points:
samples = region.points[np.random.choice(
region.points.shape[0], number_points, replace=False), :]
else:
samples = region.points
rand_samples.append(samples)
df = pd.DataFrame(columns=['Model id', 'Region id', 'cost'])
for model_id in range(num_models_fitness):
model = BioProc(np.array([
"protein_production", "protein_production", "protein_production",
"protein_production", "protein_degradation", "protein_degradation",
"Kd", "hill", "protein_production", "protein_degradation", "Kd",
"hill"
]),
model_mode=models[model_id])
for region_id in range(num_models_regions):
for sample in rand_samples[region_id]:
cost = -model.eval(sample)[0]
df = df.append(
{
'Model id': model_id + 1,
'Region id': region_id + 1,
'cost': cost
},
ignore_index=True)
if file_name:
df.to_csv(file_name, index=False)
return df
}
}
#model_mode = one_bit_processor_ext
#model_mode = two_bit_processor_ext
#model_mode = three_bit_processor_ext
model_mode = four_bit_processor_ext
folder_name = "results_opt"
if model_mode == one_bit_processor_ext:
file_name = "01_bioproc"
elif model_mode == two_bit_processor_ext:
file_name = "02_bioproc"
elif model_mode == three_bit_processor_ext:
file_name = "03_bioproc"
else:
file_name = "04_bioproc"
filename = os.path.join(".", folder_name, file_name)
print(filename)
model = BioProc(np.array([
"protein_production", "protein_production", "protein_production",
"protein_production", "protein_degradation", "protein_degradation",
"Kd", "hill", "protein_production", "protein_degradation", "Kd", "hill"
]),
model_mode=model_mode,
parameter_values=param_values,
avg_dev=30)
sol = solver.Solver(model)
sol.run(filename, maxDepth=1) #do not cluster
def plotStochasticSimulations(from_file=True,
number_points=3,
plotFlipflops=False,
pickle_dump=False):
sns.set_style("white")
#flatui = ['#f7f7f7','#d9d9d9','#bdbdbd','#969696','#737373','#525252','#252525']
#flatui = ['#bdbdbd','#969696','#737373','#525252','#252525','#000000']
flatui = ['#d9d9d9', '#bdbdbd', '#969696', '#737373', '#525252', '#252525']
#sns.palplot(sns.color_palette(flatui))
sns.set_palette(flatui)
print("Plotting stochastic simulations")
fig, axs = plt.subplots(3, number_points, sharey='row')
for model_index in range(3):
print(model_index)
region = model_regions[model_index]
model = BioProc(np.array([
"protein_production", "protein_production", "protein_production",
"protein_production", "protein_degradation", "protein_degradation",
"Kd", "hill", "protein_production", "protein_degradation", "Kd",
"hill"
]),
model_mode=models[model_index],
plot_fitness=False,
plot_devs=False)
samples = []
if not from_file:
samples = region.points[np.random.choice(
region.points.shape[0], number_points, replace=False), :]
else:
for i in range(number_points):
samples.append(
pickle.load(
open(
"selected_points\\model" + str(model_index + 1) +
"sample" + str(i + 1) + ".p", "rb")))
sample_num = 0
for sample in samples:
print(sample)
T, Y = model.simulateStochastic(sample)
Y_ode = model.simulate(sample)
colors = plt.rcParams["axes.prop_cycle"]() #get color iterator
for i in range(model_index + 1):
if plotFlipflops:
c = next(colors)["color"]
axs[model_index, sample_num].plot(T,
Y[:, i * 4 + 2],
label='q' + str(i + 1),
color=c,
alpha=0.7)
axs[model_index, sample_num].plot(model.ts,
Y_ode[:, i * 4 + 2],
color=c,
linestyle="--")
#axs[model_index, sample_num].get_xaxis().set_visible(False)
for i in range(2 * (model_index + 1)):
c = next(colors)["color"]
axs[model_index,
sample_num].plot(T,
Y[:, -(2 * (model_index + 1) - i)],
label="i" + str(i + 1),
color=c,
alpha=0.7)
axs[model_index,
sample_num].plot(model.ts,
Y_ode[:, -(2 * (model_index + 1) - i)],
color=c,
linestyle="--")
if not from_file:
pickle.dump(
sample,
open(
"selected_points\\test_model" + str(model_index + 1) +
"sample" + str(sample_num + 1) + ".p", "wb+"))
sample_num += 1
for i in range(len(samples)):
axs[2, i].set(xlabel='Time [h]')
for i in range(3):
axs[i, 0].set(ylabel='Concentration [nM]')
#plt.legend(loc="upper left")
#plt.legend()
plt.legend(ncol=10,
loc='upper center',
bbox_to_anchor=(0.5, 0.95),
bbox_transform=plt.gcf().transFigure)
plt.gcf().set_size_inches(15, 12)
plt.savefig(os.path.join(base_path_robustness, 'ssa.pdf'),
bbox_inches='tight')
if pickle_dump:
pickle.dump(fig, open("selected_points\\plot_SSA.pickle", "wb"))
plt.show()
#
models = [
one_bit_processor_ext, two_bit_processor_ext, three_bit_processor_ext,
four_bit_processor_ext
]
#folders = [os.path.join(base_path, "one_bit_model"), os.path.join(base_path, "two_bit_model"), os.path.join(base_path, "three_bit_model")]
model_regions = []
for model_index in range(num_models_regions):
#folder = folders[model_index]
model = BioProc(np.array([
"protein_production", "protein_production", "protein_production",
"protein_production", "protein_degradation", "protein_degradation",
"Kd", "hill", "protein_production", "protein_degradation", "Kd",
"hill"
]),
model_mode=models[model_index])
solver = Solver(model)
model_str = '0' + str(model_index + 1) + '_'
region_files = []
for base_path_opt in base_paths_opt:
if ga_solutions:
region_files.append(
os.path.join(base_path_opt,
model_str + "bioprocViableSet_IterGA.p"))
if local_solutions:
for i in range(10):
region_files.append(
"Kd": {
"min": 0.01,
"max": 250
},
"protease_concentration": {
"min": 10,
"max": 1000
}
}
#flip flop with instructions external clock
filename = os.path.join(".", "bioproc", "three_bit_model_new_new",
"bioproc")
print(filename)
model = BioProc(np.array([
"protein_production", "protein_production", "protein_production",
"protein_production", "protein_degradation", "protein_degradation",
"Kd", "hill", "protein_production", "protein_degradation", "Kd", "hill"
]),
model_mode=three_bit_processor_ext,
parameter_values=param_values,
avg_dev=30)
solver = Solver(model)
solver.run(filename, maxDepth=1) #do not cluster
"""
model modes:
- one_bit_processor_ext
- two_bit_processor_ext
- three_bit_processor_ext
"""