seg_len = SEG_LEN_IN_SEC / (1. / fs)
sig = np.array(sig)
sig_use = sig[:, channel_ind]
out = ecg.ecg(signal=sig_use, sampling_rate=fs, show=False)
ecg_filtered, peaks = out[1], out[2]
segs = mf.get_segs(sig_use, peaks, seg_len, fs=fs)
sigs_all = [sig[:, 0], sig[:, 1], ecg_filtered]
desp_all = [
'original signal', 'primary filtered signal', 'secondary filtered signal'
]
i = 0
for curr_sig in sigs_all:
fig = plt.figure(figsize=(8, 6))
plt.plot(curr_sig)
mf.config_plot('time', desp_all[i])
plt.show()
curr_segs = mf.get_segs(curr_sig, peaks, seg_len, fs=fs)
fig, axes = plt.subplots(nrows=2, ncols=3, figsize=(10, 8))
j = 0
for ax in axes.flat:
ax.plot(curr_segs[i])
#ax.axis('off')
j += 1
plt.show()
i += 1
records += segs
fs = fields['fs']
sig_use = sig[:,1]
dt = 1./fs
seg_len = SEG_LEN_MS/1000. / (1./fs) # unit: num of bins
#print seg_len
TO_PLOT_ORIGIN_SIG = False
if TO_PLOT_ORIGIN_SIG and 'Person_01/rec_7' in filename:
ts = np.linspace(0,dt*len(sig_use),len(sig_use))
fig = plt.figure(figsize=(8,6))
plt.plot(ts, sig[:,0])
#my_funcs.config_plot('Time /s', 'Amplitude /mV')
#plt.show()
plt.plot(ts, sig[:,1])
my_funcs.config_plot('Time /s', 'Amplitude /mV')
print 'filename: '+filename
plt.show()
## baseline drift handle
x = sig[:,0]
my_funcs.write_to_file(x)
quit()
y = my_funcs.my_butter(x,freqs=0.95,type='high')
plt.plot(x,y)
plt.show()
##
quit()
out = ecg.ecg(signal=sig_use, sampling_rate=fs, show=False)
print all_train_times
print all_accuBySeg_Vs
print all_accuBySeg_Ns
for i in range(len(all_accuBySeg_Vs)):
plt.scatter(ratios,
all_accuBySeg_Vs[i],
color=color_methods[i],
marker=marker_methods[i],
s=150,
label=clsfy_methods[i],
facecolors=facecolors_methods[i],
linewidth=1.5)
plt.ylim([.5, 1.])
mf.config_plot('Ratio of V segments in Test',
'Accuracy of V by segment',
legend_loc=(.9, .55))
plt.show()
for i in range(len(all_accuBySeg_Ns)):
plt.scatter(ratios,
all_accuBySeg_Ns[i],
color=color_methods[i],
marker=marker_methods[i],
s=150,
label=clsfy_methods[i],
facecolors=facecolors_methods[i],
linewidth=1.5)
plt.ylim([.4, 1.])
mf.config_plot('Ratio of V segments in Test', 'Accuracy of N by segment') #,
#legend_loc=(.9,.55))
def plot_pt_MSE_avgstd(use_saved_files=False):
L = 24.0 #23.9
r2 = 2.5
[Mss_EGFP, sample_ratio, months] = rs.get_monkey_parameters(monkey_ID=3)
data_expr = rs.get_expr_data(monkey_ID=3)
row, col = data_expr.shape
def get_avgstds(data):
return np.mean(data, axis=1), np.std(data, axis=1)
fz_expr_avg, fz_expr_std = get_avgstds(data_expr)
if not os.path.exists(res_pt_avgstd_folderpath):
os.mkdir(res_pt_avgstd_folderpath)
if use_saved_files and os.path.isfile(res_pt_avgstd_folderpath +
'samples_theo_' + str(L) + '.txt'):
samples_theo = np.loadtxt(res_pt_avgstd_folderpath + 'samples_theo_' +
str(L) + '.txt')
else:
filename = simulated_folderpath + '[r2,L]=' + "[%.1f,%.1f]" % (r2, L)
print filename
mi_data = np.loadtxt(filename)
mi_row, mi_col = mi_data.shape
sample_days_inds = range(3 * 30, mi_col, 60)
sample_big = rs.get_samples(mi_s=mi_data,
sample_days_inds=sample_days_inds,
sample_ratio=sample_ratio)
repeat_num = 1
all_stats = []
for k in range(repeat_num):
samples_theo = rs.resample(sample_big, col)
all_stats += rs.two_sample_test(samples_theo, data_expr)[:-1]
all_stats = np.array(all_stats)
print sum(all_stats >= 0.05) / float(len(all_stats))
np.savetxt(
res_pt_avgstd_folderpath + 'samples_theo_' + str(L) + '.txt',
samples_theo)
TO_PLOT_ABUND = False
if TO_PLOT_ABUND:
import my_funcs as mf
cumu_fhi = []
for j in range(col):
curr_cumu = np.cumsum(samples_theo[:, j])
cumu_fhi.append(curr_cumu)
cumu_fhi = np.array(cumu_fhi)
PRE_CLASSIFY_CLONES = True
avg_par = []
for i in range(row):
avg_par.append(data_expr[i, 1:].mean())
avg_par = np.array(avg_par)
large_marks_pre = avg_par > 0.012
small_marks_pre = avg_par < 0.0005
# inter_marks_pre = (avg_par >= small_marks_pre) and (avg_par <= large_marks_pre)
fig = plt.figure(figsize=(8, 6))
num_abund_clones = cumu_fhi.shape[1]
linewidths = np.array([.5] *
row) # pre-set to intermediate clones setting
linewidths[large_marks_pre] = np.array([1.] * sum(large_marks_pre))
linewidths[small_marks_pre] = np.array([.1] * sum(small_marks_pre))
# [exp(x) for x in avgs_all] #[exp(x) for x in np.linspace(log(1.),log(.1),num_abund_clones)]
# print linewidths
for j in range(num_abund_clones):
plt.plot(mf.months, cumu_fhi[:, j], linewidth=linewidths[j])
# plt.plot(mf.months, cumu_fhi, linewidth=.7)
mf.config_plot('$t_j$ (in month)', '$\hat{f}_{i}(t_j)$ (stacked)')
mf.use_sci_nota('x', usesci=False)
mf.use_sci_nota('y', usesci=False)
plt.xlim(mf.months[0], mf.months[-1])
plt.ylim(0, 1)
# plt.axis('off')
plt.show()
def is_large_clone(data):
res = []
row, col = data.shape
for i in range(row):
res.append(
sum(data[i, :] == 0) == 0) # if clone i has some absence
return np.array(res)
fz_theo_avg, fz_theo_std = get_avgstds(samples_theo)
large_mrks = is_large_clone(samples_theo)
other_mrks = [not x for x in large_mrks]
import my_funcs as mf
plt.scatter(np.log(fz_expr_avg), fz_expr_std, c='b', label='experiment')
ax = plt.subplot(111)
for_ppt = False
if for_ppt:
curr_label = 'model + sampling'
else:
curr_label = 'model + \n sampling' # 'simulated \n sampling'
plt.scatter(np.log(fz_theo_avg),
fz_theo_std,
label=curr_label,
linewidth=1.5,
edgecolors='g',
facecolors='none',
marker='^',
s=120)
# plt.scatter(np.log(fz_theo_avg[large_mrks]), fz_theo_std[large_mrks], label='simulated \n sampling',linewidth=1.5,
# edgecolors='r',facecolors='none', marker='^', s=120)
plt.xlim([-8.5, -2.1])
plt.ylim([-0.005, 0.05]) # plt.ylim([-0.001,0.01]) #
# mf.config_plot(r'$\ln y_i$',r'$\sigma_i$',legend_loc=(.65,.8))
if not for_ppt:
mf.config_plot('', '', legend_loc=(.65, .8))
else:
mf.config_plot(r'ln($y_i$)', r'$\sigma_i$', legend_loc=(.7, .8))
mf.use_sci_nota('x', False)
mf.use_sci_nota('y', True)
# plt.show()
plt.savefig(res_pt_avgstd_folderpath +
'fig9b.png') #, format='svg', dpi=600
def plot_pt_MSE(use_saved_files=True):
L = 24.0 # 23.4
r2 = 2.5
[Mss_EGFP, sample_ratio, months] = rs.get_monkey_parameters(monkey_ID=3)
data_expr = rs.get_expr_data(monkey_ID=3)
row, col = data_expr.shape
fz_expr_avg, fz_expr_std = rs.get_fz_stats(data_expr)
if not os.path.exists(res_pt_folderpath):
os.mkdir(res_pt_folderpath)
if not use_saved_files:
all_dists = []
all_avgs = []
all_stds = []
filename = 'simulated_clones/' + '[r2,L]=' + "[%.1f,%.1f]" % (r2, L)
print filename
mi_data = np.loadtxt(filename)
mi_row, mi_col = mi_data.shape
sample_days_inds = range(3 * 30, mi_col, 60)
sample_big = rs.get_samples(mi_s=mi_data,
sample_days_inds=sample_days_inds,
sample_ratio=sample_ratio)
repeat_num = 200
all_stats = []
for k in range(repeat_num):
samples_theo = rs.resample(sample_big, col)
all_stats += rs.two_sample_test(samples_theo, data_expr)[:-1]
all_stats = np.array(all_stats)
print sum(all_stats >= 0.05) / float(len(all_stats))
fz_theo_avg, fz_theo_std = rs.get_fz_stats(samples_theo)
np.savetxt(res_pt_folderpath + 'fz_theo_avg.txt', fz_theo_avg)
np.savetxt(res_pt_folderpath + 'fz_theo_std.txt', fz_theo_std)
else:
fz_theo_avg = np.loadtxt(res_pt_folderpath + 'fz_theo_avg.txt')
fz_theo_std = np.loadtxt(res_pt_folderpath + 'fz_theo_std.txt')
import my_funcs as mf
fig = plt.figure(figsize=(8, 6))
plt.errorbar(range(1, 8),
fz_expr_avg[1:],
yerr=fz_expr_std[1:],
color='k',
linewidth=2)
eb_theo = plt.errorbar(range(1, 8),
fz_theo_avg[1:],
yerr=fz_theo_std[1:],
color='k',
linewidth=2,
linestyle='--')
eb_theo[-1][0].set_linestyle('--')
mf.config_plot(r'$z$', r'$\qquad Y_{z}(L_{\rm e}), ~~~~~~\hat{Y}_{z}$')
mf.use_sci_nota('x', usesci=False)
mf.use_sci_nota('y', usesci=True)
plt.xlim([0.5, 7.5])
plt.ylim([-0.0005, 0.007])
# plt.yticks(np.linspace(0.005,0.025,num=5).tolist())
# plt.tight_layout()
plt.savefig(res_pt_folderpath + 'fig7b.png') #, format='svg', dpi=600
def plot_1D_line(use_saved_files=True):
r2 = 2.5
from main_simulations import L_s, res_folderpath
[Mss_EGFP, sample_ratio, months] = rs.get_monkey_parameters(monkey_ID=3)
change_Mss = False
if change_Mss:
Mss_EGFP /= 5.
sample_ratio *= 5
data_expr = rs.get_expr_data(monkey_ID=3)
fz_expr_avg, fz_expr_std = rs.get_fz_stats(data_expr)
if not use_saved_files:
repeat_num = 200
all_dists = []
all_avgs = []
all_stds = []
for j in range(len(L_s)):
L = L_s[j]
filename = '[r2,L]=' + "[%.1f,%.1f]" % (r2, L)
print filename
mi_data = np.loadtxt(res_folderpath + filename)
mi_row, mi_col = mi_data.shape
sample_days_inds = range(3 * 30, mi_col, 60)
sample_big = rs.get_samples(mi_s=mi_data,
sample_days_inds=sample_days_inds,
sample_ratio=sample_ratio)
curr_dists = []
for k in range(repeat_num):
samples_theo = rs.resample(sample_big, 8)
fz_theo_avg, fz_theo_std = rs.get_fz_stats(samples_theo)
curr_dists.append(
np.square(rs.my_euclid(fz_theo_avg[1:], fz_expr_avg[1:])))
all_dists.append(curr_dists)
curr_dists = np.array(curr_dists)
all_avgs.append(curr_dists.mean())
all_stds.append(curr_dists.std())
if not change_Mss:
np.savetxt(res_1D_folderpath + 'all_avgs_fix_r2', all_avgs)
np.savetxt(res_1D_folderpath + 'all_stds_fix_r2', all_stds)
else:
np.savetxt(res_1D_folderpath + 'all_avgs_fix_r2', all_avgs)
np.savetxt(res_1D_folderpath + 'all_stds_fix_r2', all_stds)
else:
if not change_Mss:
all_avgs = np.loadtxt(res_1D_folderpath + 'all_avgs_fix_r2')
all_stds = np.loadtxt(res_1D_folderpath + 'all_stds_fix_r2')
else:
all_avgs = np.loadtxt(res_1D_folderpath + 'all_avgs_fix_r2')
all_stds = np.loadtxt(res_1D_folderpath + 'all_stds_fix_r2')
import my_funcs as mf
plt.errorbar(L_s[:-4],
all_avgs[:-4],
yerr=all_stds[:-4],
color='k',
linewidth=2)
mf.config_plot(r'$L_{\rm e}$', 'MSE')
mf.use_sci_nota('x', usesci=False)
mf.use_sci_nota('y', usesci=True)
print L_s[:-4][np.argmin(all_avgs[:-4])]
plt.ylim([0, 0.00032]) # plt.ylim([0,0.00025]) #
plt.xlim([18.8, 26.2])
# plt.yticks(np.linspace(0.005,0.025,num=5).tolist())
# plt.tight_layout()
plt.savefig(res_1D_folderpath + 'fig7a.png') # , format='svg', dpi=600
# plt.show()
ind_min = np.argmin(all_avgs)
avg_min = all_avgs[ind_min]
std_min = all_stds[ind_min]
L_min = L_s[ind_min]
print avg_min, std_min, L_min
fs = fields['fs']
sig_use = sig[:, 1] ## TODO: treated by wfdb
dt = 1. / fs
seg_len = SEG_LEN_MS / 1000. / (1. / fs) # unit: num of bins
#print seg_len
TO_PLOT_ORIGIN_SIG = False
if TO_PLOT_ORIGIN_SIG and 'Person_01/rec_7' in filename:
ts = np.linspace(0, dt * len(sig_use), len(sig_use))
fig = plt.figure(figsize=(8, 6))
plt.plot(ts, sig[:, 0])
#my_funcs.config_plot('Time /s', 'Amplitude /mV')
#plt.show()
plt.plot(ts, sig[:, 1])
my_funcs.config_plot('Time /s', 'Amplitude /mV')
print 'filename: ' + filename
plt.show()
## baseline drift handle
x = sig[:, 0]
my_funcs.write_to_file(x)
quit()
y = my_funcs.my_butter(x, freqs=0.95, type='high')
plt.plot(x, y)
plt.show()
##
quit()
out = ecg.ecg(signal=sig_use, sampling_rate=fs, show=False)