def merge_sudden_markov_fitness_sims(input_fnames, output_fname):
"""
Merge all fitness simulation data into one pickle file.
"""
print "merging fitness simulation files"
all_sims = OrderedDict()
for input_fname in input_fnames:
data = simulation.load_data(input_fname)
sim_cond = data["sim_cond"]
all_sims["sim%d" %(sim_cond)] = data
with open(output_fname, "w") as outfile:
pickle.dump(all_sims, outfile)
def plot_switch_ssm_fitness_simulations(input_fname, plot_fname, label):
"""
Plot switch SSM fitness simulations.
"""
print "plotting fitness simulation for switch ssm"
sim_info = simulation.load_data(input_fname)
results = sim_info["data"]
# make the plot here for three of the simulations
sims_to_plot = ["switch_ssm_fitness_sim1",
"switch_ssm_fitness_sim2",
"switch_ssm_fitness_sim3",
"switch_ssm_fitness_sim4"]
num_plots = len(sims_to_plot)
fig = plt.figure(figsize=(10, 6))
sns.set_style("ticks")
ystep = 10
yticks = np.arange(10, 40 + ystep, ystep)
print yticks, "yticks"
for n, sim_to_plot in enumerate(sims_to_plot):
if sim_to_plot not in results:
raise Exception, "No sim %s" %(sim_to_plot)
params = results[sim_to_plot]["params"]
params["policy_colors"] = \
{"Random": sns.color_palette("Set1")[1],
"Plastic": sns.color_palette("Set1")[2],
"Posterior predictive": sns.color_palette("Set1")[0],
"Glucose-only": sns.color_palette("Set1")[3]}
# "Posterior pred. (BH)": "red",
# "Random (BH)": "g"}
plt.subplot(2, int(round(num_plots / 2.)), n + 1)
gluc_val = params["nutr_labels"].index("glucose")
galac_val = params["nutr_labels"].index("galactose")
gluc_growth_rate = params["nutr_growth_rates"][gluc_val]
galac_growth_rate = params["nutr_growth_rates"][galac_val]
# title = r"$\mu_{\mathsf{Glu}} = %.2f, \mu_{\mathsf{Gal}} = %.3f$, " \
# r"$\mu_{\mathsf{Mis}} = %.1f$, " \
# r"$p_1 = %.2f, p_2 = %.2f$" \
# %(gluc_growth_rate,
# galac_growth_rate,
# params["mismatch_growth_rate"],
# params["p_switch_to_switch"],
# params["p_noswitch_to_switch"])
title = r"$p_1 = %.2f, p_2 = %.2f$" \
%(params["p_switch_to_switch"],
params["p_noswitch_to_switch"])
sim_results = results[sim_to_plot]["data"]
model_switch_ssm.plot_fitness_sim_results(sim_results, params,
title=title,
yticks=yticks)
fig.set_tight_layout(True)
plt.savefig(plot_fname)
def plot_sudden_markov_popsize(data_fname, plot_fname):
"""
Plot sudden Markov model population size for different growth
policies.
"""
print "plotting: %s" %(os.path.basename(plot_fname))
print " - data: %s" %(data_fname)
sim_data = simulation.load_data(data_fname)
sns.set_style("ticks")
df = sim_data["sim_results"]
params = sim_data["params"]
plt.figure(figsize=(6, 4))
model_sudden_markov.plot_popsize_by_policies(df, params)
plt.savefig(plot_fname)
def plot_sudden_ssm_filtering(input_fname, plot_fname, label):
"""
Plot SSM filtering predictions.
"""
print "plotting: %s" %(os.path.basename(plot_fname))
fig = plt.figure(figsize=(7, 5))
sns.set_style("ticks")
all_results = simulation.load_data(input_fname)
all_results = all_results["model"]
num_plots = len(all_results)
total_plots = num_plots * 2
gs = gridspec.GridSpec(total_plots, 1,
height_ratios=[1, 0.2]*num_plots)
curr_plot_num = 0
axes = {}
for n, data_label in enumerate(all_results.keys()):
data_set = all_results[data_label]
params = data_set["params"]
time_obj = time_unit.Time(params["t_start"],
params["t_end"],
step_size=params["step_size"])
c = 0.8
x_axis = time_obj.t[0::4]
xlims = [time_obj.t[0] - c, time_obj.t[-1] + c]
ax1 = plt.subplot(gs[curr_plot_num, 0])
pred_probs = [p[0] for p in data_set["preds_with_lag"]]
plt.plot(time_obj.t, pred_probs, "-o", color=plot_utils.red,
label="Prediction",
clip_on=False,
zorder=100)
plt.xlabel(r"Time step")
plt.ylabel(r"$P(C_{t+1} =\ \mathsf{Glu} \mid C_{0:t})$",
fontsize=11)
plt.title("lag = %d" %(params["decision_lag_time"]), fontsize=8)
plt.legend(loc="lower right")
plt.xticks(x_axis, fontsize=8)
plt.xlim(xlims)
plt.ylim([0, 1])
plt.yticks(np.arange(0, 1 + 0.2, 0.2))
ax2 = plt.subplot(gs[curr_plot_num + 1, 0])
data_to_labels = {0: "Glu", 1: "Gal"}
labels_to_colors = {"Glu": plot_utils.green, "Gal": plot_utils.blue}
data = params["data"]
ax2.get_yaxis().set_visible(False)
ax2.set_yticks([])
ax2.spines["left"].set_visible(False)
plot_utils.plot_sudden_switches(time_obj, data,
data_to_labels=data_to_labels,
labels_to_colors=labels_to_colors,
box_height=0.025,
y_val=0.02,
ax=ax2,
despine=False,
with_legend=True,
legend_outside=(0,1))
plt.xticks(x_axis, fontsize=8)
plt.xlim(xlims)
# despine axes
sns.despine(trim=True, left=True, ax=ax2)
sns.despine(trim=True, ax=ax1)
# advance number of plots by two
curr_plot_num += 2
ax1.spines["left"].set_visible(True)
#plt.tight_layout(h_pad=0.1)
fig.set_tight_layout(True)
plt.savefig(plot_fname)
##
##
##
import numpy as np
import pandas
import utils
import fitness
import simulation
import scipy
from scipy.interpolate import interp1d
import matplotlib.pylab as plt
f = "../simulations_data/sudden_switch/merged_switch_ssm_fitness_sims.data"
sim_info = simulation.load_data(f)
results = sim_info["data"]
# get the simulations that should be plotted based on their
# parameter values
params_to_plot = [{"p_switch_to_switch": 0.1,
"p_noswitch_to_switch": 0.1},
{"p_switch_to_switch": 0.1,
"p_noswitch_to_switch": 0.95},
{"p_switch_to_switch": 0.95,
"p_noswitch_to_switch": 0.1},
{"p_switch_to_switch": 0.95,
"p_noswitch_to_switch": 0.95}]
sims_to_plot = []
for sim_name in results:
# see if current simulation matches the parameters
# we're looking for
for curr_params in params_to_plot:
