def plot_fits():
params_dict = {'c0_to_c1_k': 2e-2,
'c1_scaling': 20.,
'c1_to_c2_k': 1e-4,
'c2_scaling': 6.,
'c1_to_c2_k': 1e-3,
'c3_scaling': 0.5}
builder = Builder(params_dict=params_dict)
builder.build_model_multiconf(3, data_dict['c126_nbd'][0],
normalized_data=True,
reversible=False)
"""
k1 = builder.model.parameters['c0_to_c1_k']
k2 = builder.model.parameters['c1_to_c2_k']
k1_index = builder.model.parameters.index(k1)
k2_index = builder.model.parameters.index(k2)
k1_est_index = builder.estimate_params.index(k1)
k2_est_index = builder.estimate_params.index(k2)
"""
pysb_fit = fitting.fit_pysb_builder(builder, 'NBD', data_dict['c126_nbd_time'],
data_dict['c126_nbd'])
plt.figure()
plt.plot(data_dict['c126_nbd_time'], data_dict['c126_nbd'], linestyle='', marker='.')
plt.plot(data_dict['c126_nbd_time'], pysb_fit.ypred)
plt.xlabel('Time (sec)')
def plot_2conf_fits(activator):
plt.ion()
#nbd_sites = ['15']
nbd_sites = ['3', '5', '15', '36', '47', '54', '62', '68', '79', '120',
'122', '126', '138', '151', '175', '179', '184', '188']
replicates = range(1, 4)
count = 0
fit_results = []
plt.figure("Fitted k1")
for nbd_index, nbd_site in enumerate(nbd_sites):
k1_means = []
k1_sds = []
for rep_index in replicates:
nt = df[(activator, 'NBD', nbd_site, rep_index, 'TIME')].values
ny = df[(activator, 'NBD', nbd_site, rep_index, 'VALUE')].values
plt.figure('%s, NBD-%s-Bax Fits' % (activator, nbd_site),
figsize=(6, 5))
builder = Builder(params_dict=params_dict)
builder.build_model_multiconf(2, ny[0], normalized_data=True)
pysb_fit = fitting.fit_pysb_builder(builder, 'NBD', nt, ny)
plt.plot(nt, ny, linestyle='', marker='.',
color=rep_colors[rep_index])
plt.plot(nt, pysb_fit.ypred,
label='%s Rep %d' % (activator, rep_index),
color=rep_colors[rep_index])
plt.xlabel('Time (sec)')
plt.ylabel('$F/F_0$')
plt.title('NBD $F/F_0$ fits, NBD-%s-Bax' % nbd_site)
plt.legend(loc='lower right')
# Calculate stderr of parameters (doesn't account for covariance)
(k1_mean, k1_sd) = _mean_sd('c0_to_c1_k', builder, pysb_fit)
k1_means.append(k1_mean)
k1_sds.append(k1_sd)
(c1_mean, c1_sd) = _mean_sd('c1_scaling', builder, pysb_fit)
fit_results.append(FitResult(builder, activator, nbd_site,
rep_index, 'NBD', 'c0_to_c1_k',
k1_mean, k1_sd))
fit_results.append(FitResult(builder, activator, nbd_site,
rep_index, 'NBD', 'c1_scaling',
c1_mean, c1_sd))
count += 1
plt.figure('%s, NBD-%s-Bax Fits' % (activator, nbd_site))
plt.tight_layout()
plt.figure("Fitted k1")
plt.bar(range(nbd_index*4, (nbd_index*4) + 3), k1_means,
yerr=k1_sds, width=1, color='r', ecolor='k')
num_sites = len(nbd_sites)
plt.figure("Fitted k1")
plt.ylabel('Fitted k1 ($sec^{-1}$)')
ax = plt.gca()
ax.set_xticks(np.arange(1.5, 1.5 + num_sites * 4, 4))
ax.set_xticklabels(nbd_sites)
return fit_results
a = fitting.Parameter(1e-3)
b = fitting.Parameter(1e-4)
c = fitting.Parameter(1e-4)
def gompertz(t):
return a() * np.exp(-b() * np.exp(-c()* t))
fitting.fit(gompertz, [a, b, c], rp, rt)
plt.plot(rt, gompertz(rt))
plt.figure()
plt.plot(rt, ry, color='r')
plt.title('Release')
#t = np.linspace(0, 10000, 10000)
t = rt
solver = Solver(builder.model, t)
solver.run()
plt.figure()
plt.plot(rt, ry, color='r')
plt.plot(t, solver.yexpr['Release'])
plt.plot(t, solver.yexpr['iBax_frac'])
plt.plot(t, solver.yexpr['cBax_frac'])
pysb_fit = fitting.fit_pysb_builder(builder, 'Release', rt, ry)
plt.figure()
plt.plot(rt, ry, color='r')
plt.plot(rt, pysb_fit.ypred)
def plot_3conf_fits(activator):
plt.ion()
#nbd_sites = ['15']
nbd_sites = ['3', '5', '15', '36', '47', '54', '62', '68', '79', '120',
'122', '126', '138', '151', '175', '179', '184', '188']
replicates = range(1, 4)
count = 0
fit_results = []
plt.figure("Fitted k1/k2")
for nbd_index, nbd_site in enumerate(nbd_sites):
k1_means = []
k2_means = []
k1_sds = []
k2_sds = []
for rep_index in replicates:
rt = df[(activator, 'Release', nbd_site, rep_index, 'TIME')].values
ry = df[(activator, 'Release', nbd_site, rep_index, 'VALUE')].values
nt = df[(activator, 'NBD', nbd_site, rep_index, 'TIME')].values
ny = df[(activator, 'NBD', nbd_site, rep_index, 'VALUE')].values
"""
plt.figure()
plt.plot(rt, ry)
plt.figure()
plt.plot(nt, ny)
plt.figure()
plt.plot(nt, ry / ny)
ry_norm = (ry - np.min(ry)) / (np.max(ry) - np.min(ry))
ny_norm = (ny - np.min(ny)) / (np.max(ny) - np.min(ny))
plt.figure()
plt.plot(rt, ry_norm, color='g')
plt.plot(rt, ny_norm, color='b')
"""
plt.figure('%s, NBD-%s-Bax Fits' % (activator, nbd_site),
figsize=(12, 5))
plt.subplot(1, 2, 1)
plt.plot(rt, ry,
linestyle='', marker='.',
color=rep_colors[rep_index])
twoexp = tf.TwoExpLinear()
#twoexp = tf.TwoExp()
params = twoexp.fit_timecourse(rt, ry)
plt.plot(rt, twoexp.fit_func(rt, params),
label='%s Rep %d' % (activator, rep_index),
color=rep_colors[rep_index])
plt.xlabel('Time (sec)')
plt.ylabel('% Tb release')
plt.title('%% Tb release fits, NBD-%s-Bax' % nbd_site)
plt.legend(loc='lower right')
builder = Builder(params_dict=params_dict)
builder.build_model_multiconf(3, ny[0], normalized_data=True)
pysb_fit = fitting.fit_pysb_builder(builder, 'NBD', nt, ny)
plt.subplot(1, 2, 2)
plt.plot(nt, ny, linestyle='', marker='.',
color=rep_colors[rep_index])
plt.plot(nt, pysb_fit.ypred,
label='%s Rep %d' % (activator, rep_index),
color=rep_colors[rep_index])
plt.xlabel('Time (sec)')
plt.ylabel('$F/F_0$')
plt.title('NBD $F/F_0$ fits, NBD-%s-Bax' % nbd_site)
plt.legend(loc='lower right')
# Calculate stderr of parameters (doesn't account for covariance)
(k1_mean, k1_sd) = _mean_sd('c0_to_c1_k', builder, pysb_fit)
(k2_mean, k2_sd) = _mean_sd('c1_to_c2_k', builder, pysb_fit)
(c1_mean, c1_sd) = _mean_sd('c1_scaling', builder, pysb_fit)
(c2_mean, c2_sd) = _mean_sd('c2_scaling', builder, pysb_fit)
fit_results.append(FitResult(builder, activator, nbd_site,
rep_index, 'NBD', 'c0_to_c1_k',
k1_mean, k1_sd))
fit_results.append(FitResult(builder, activator, nbd_site,
rep_index, 'NBD', 'c1_scaling',
c1_mean, c1_sd))
fit_results.append(FitResult(builder, activator, nbd_site,
rep_index, 'NBD', 'c1_to_c2_k',
k2_mean, k2_sd))
fit_results.append(FitResult(builder, activator, nbd_site,
rep_index, 'NBD', 'c2_scaling',
c2_mean, c2_sd))
k1_means.append(k1_mean)
k2_means.append(k2_mean)
k1_sds.append(k1_sd)
k2_sds.append(k2_sd)
"""
plt.figure()
s = Solver(builder.model, nt)
s.run(param_values=pysb_fit.params)
plt.plot(nt, s.yobs['Bax_c0'])
plt.plot(nt, s.yobs['Bax_c1'])
plt.plot(nt, s.yobs['Bax_c2'])
"""
count += 1
#plt.title(nbd_site)
#plt.xlabel('Time (sec)')
#plt.ylabel('$F/F_0$')
plt.figure('%s, NBD-%s-Bax Fits' % (activator, nbd_site))
plt.tight_layout()
plt.figure("Fitted k1/k2")
plt.bar(range(nbd_index*7, (nbd_index*7) + 3), k1_means,
yerr=k1_sds, width=1, color='r', ecolor='k')
plt.bar(range(nbd_index*7+3, (nbd_index*7) + 6), k2_means,
yerr=k2_sds, width=1, color='g', ecolor='k')
num_sites = len(nbd_sites)
plt.figure("Fitted k1/k2")
plt.ylabel('Fitted k1/k2 ($sec^{-1}$)')
ax = plt.gca()
ax.set_xticks(np.arange(3, 3 + num_sites * 7, 7))
ax.set_xticklabels(nbd_sites)
return fit_results
