def plot_synchronies(data_cluster):
#% all rep together
data_diff = data_cluster.query('type == "diff"')
data_same = data_cluster.query('type == "same"')
results_mwu = pg.mwu(data_same['distance'].values,
data_diff['distance'].values, 'greater')
p_val = results_mwu['p-val'][0]
d = pg.compute_effsize(data_same['distance'],
data_diff['distance'],
eftype='cohen')
plt.hist(data_diff['distance'].values, density=True, color='r', alpha=0.25)
plt.hist(data_same['distance'].values, density=True, color='g', alpha=0.25)
plt.grid()
ca = plt.gca()
x_min, x_max = ca.get_xlim()
x = x_min + 0.5 * (x_max - x_min)
y_min, y_max = ca.get_ylim()
y = y_min + 0.85 * (y_max - y_min)
if p_val < 0.05:
plt.text(x,
y,
'p = {:.3f}, d = {:.3f}'.format(p_val, d),
ha='center',
fontweight='bold')
else:
plt.text(x, y, 'p = {:.3f}, d = {:.3f}'.format(p_val, d), ha='center')
def calculate_cohens_d(es, fr):
feature_values = {}
for feature in es:
feature_values[feature] = pn.compute_effsize(fr[feature],
es[feature],
eftype='cohen')
return feature_values
def get_d_p(data_cluster):
data_diff = data_cluster.query('type == "SEP"')
data_same = data_cluster.query('type == "TOG"')
same_boot = bootstrap(data_same['distance'].values)
diff_boot = bootstrap(data_diff['distance'].values)
results_mwu = pg.mwu(same_boot, diff_boot, 'greater')
p_val = results_mwu['p-val'][0]
d = pg.compute_effsize(same_boot, diff_boot, eftype='cohen')
return (d, p_val)
female_data_all_atlas.append(
female_residual
) #female_data_all_atlas stores the residuals for females for all atlas column wise
t_value, p_value = scipy.stats.ranksums(
np.arctanh(male_residual),
np.arctanh(female_residual)) #two tailed t test for corr(eFC, eSC)
#t_value, p_value = scipy.stats.ranksums(male_residual, female_residual) #for coupling strength
p_val.append(
p_value
) #p_val is a list that stores the p value for residuals for all atlas
#eff_size.append(pg.compute_effsize(male_residual, female_residual, eftype = 'hedges')) # for coupling strength
eff_size.append(
pg.compute_effsize(np.arctanh(male_residual),
np.arctanh(female_residual),
eftype='hedges')
) #eff_size is a list that stores the effect size for residuals for all atlas
r""""
plt.rcParams['font.size'] = '20'
plt.figure(figsize = (16, 8))
boxprops = {'linewidth': 2}
whiskerprops = {'linewidth': 2}
capprops = {'linewidth': 2}
male_plots = plt.boxplot(np.array(male_data_all_atlas).transpose(), positions = np.array(range(l))*2 - 0.3, boxprops = boxprops, whiskerprops = whiskerprops, capprops = capprops) #the ones after regression
female_plots = plt.boxplot(np.array(female_data_all_atlas).transpose(), positions = np.array(range(l))*2 + 0.3, boxprops = boxprops, whiskerprops = whiskerprops, capprops = capprops) #the ones after regression
male_data_br = pd.read_csv(r"E:\Shraddha\Data\male_data_sfc_efc_struc_fit_lc_br.csv", header = None).values #loading the ones before reg
female_data_br = pd.read_csv(r"E:\Shraddha\Data\female_data_sfc_efc_struc_fit_lc_br.csv", header = None).values #loading the ones before reg
####...end of MLR...####
residual = np.tanh(residual) #inverse fisher z transform
corr_efc_esc_male, corr_efc_esc_female = categorise_male_female(
residual
) #replace 'residual' with corr_efc_esc_list if you do not want to regress anything
male_data_all_atlas.append(corr_efc_esc_male)
female_data_all_atlas.append(corr_efc_esc_female)
t_value, p_value = scipy.stats.ranksums(
np.arctanh(corr_efc_esc_male),
np.arctanh(corr_efc_esc_female)) #two tailed t test for corr(eFC, eSC)
p_val.append(p_value)
eff_size.append(
pg.compute_effsize(np.arctanh(corr_efc_esc_male),
np.arctanh(corr_efc_esc_female)))
r"""
print('P value: ', p_val)
print('Effect size:', eff_size)
plt.plot(atlas, eff_size, marker = '.', markersize = 20, label = 'Effect size')
plt.plot(atlas, p_val, marker = '.', markersize = 20, label = 'Significance - p value')
plt.xlabel('Atlas')
plt.ylabel('Effect size (or) Significance')
plt.title('Effect size vs Atlas for Corr(eFC, eSC)')
plt.legend()
plt.show()
eff_size_p_val_arr = np.zeros([l, 2])
eff_size_p_val_arr[:, 0] = eff_size
eff_size_p_val_arr[:, 1] = p_val
DATA_scat.query('summary_state2=="State 2"')["value"].reset_index(
drop=True))
print(
stats.ttest_rel(
DATA_scat.query('summary_state2=="State 1"')["value"],
DATA_scat.query('summary_state2=="State 2"')["value"]), d)
d_VTC = CalculateEffect(
DATA_scat_VTC.query('variable=="State 1"')["value"].reset_index(drop=True),
DATA_scat_VTC.query('variable=="State 2"')["value"].reset_index(drop=True))
print(
stats.ttest_rel(
DATA_scat_VTC.query('variable=="State 1"')["value"],
DATA_scat_VTC.query('variable=="State 2"')["value"]), d_VTC)
pg.compute_effsize(
DATA_scat_VTC.query('variable=="State 1"')["value"].reset_index(drop=True),
DATA_scat_VTC.query('variable=="State 2"')["value"].reset_index(drop=True))
d_RT = CalculateEffect(
DATA_scat_RT.query('variable=="State 1"')["value"].reset_index(drop=True),
DATA_scat_RT.query('variable=="State 2"')["value"].reset_index(drop=True))
print(
stats.ttest_rel(
DATA_scat_RT.query('variable=="State 1"')["value"],
DATA_scat_RT.query('variable=="State 2"')["value"]), d_RT)
d_dprime = CalculateEffect(
DATA_scat_dprime.query('variable=="State 1"')["value"].reset_index(
drop=True),
DATA_scat_dprime.query('variable=="State 2"')["value"].reset_index(
drop=True))
residual = Y - Y_hat
#resdiual = np.tanh(residual)
#########.....end of MLR....#########
"""
shan_entr_eFC_male, shan_entr_eFC_female = categorise_male_female(
shan_entr_eFC)
shan_entr_eFC_male_all_atlas[:, i] = np.array(shan_entr_eFC_male)
shan_entr_eFC_female_all_atlas[:, i] = np.array(shan_entr_eFC_female)
t_value1, p_value1 = scipy.stats.ranksums(shan_entr_eFC_male,
shan_entr_eFC_female)
eff_size_shan_entr_eFC.append(
pg.compute_effsize(
shan_entr_eFC_male, shan_entr_eFC_female,
eftype='hedges')) #effect size for coupling strength
p_val_shan_entr_eFC.append(p_value1)
np.savetxt(
r"D:\Shraddha\Data\male_data_shan_entr_efc_br.csv",
shan_entr_eFC_male_all_atlas,
delimiter=',') #this storage is for the grey plots in the background
np.savetxt(
r"D:\Shraddha\Data\female_data_shan_entr_efc_br.csv",
shan_entr_eFC_female_all_atlas,
delimiter=',') #this storage is for the grey plots in the background
l = len(atlas)
plt.rcParams['font.size'] = '20'
plt.figure(figsize=(16, 8))
index = sub_num_list_phen.index(sub_num_list_351_ordered[i])
gen = gender_list[index]
if (gen == 'M'):
return_male.append(l1[i])
if (gen == 'F'):
return_female.append(l1[i])
return return_male, return_female
male_sfc_efc_list, female_sfc_efc_list = male_female_classify(
avg_corr_list_fc_351)
male_sfc_esc_list, female_sfc_esc_list = male_female_classify(
avg_corr_list_sc_351)
t, p = scipy.stats.ttest_ind(male_sfc_esc_list, female_sfc_esc_list)
eff_size = pingouin.compute_effsize(male_sfc_esc_list, female_sfc_esc_list)
#print("t = ", t)
#print("p = ", p)
#print("Effect size = ", eff_size)
r"""
fc_data = [male_sfc_efc_list, female_sfc_efc_list]
sc_data = [male_sfc_esc_list, female_sfc_esc_list]
fig, ax = plt.subplots(nrows = 1, ncols = 2)
ax[0].boxplot(fc_data)
ax[0].set_xticklabels(['Male','Female'])
ax[0].set_ylabel('Best fit correlation between sFC and eFC - 365 subjects')
ax[0].set_title('FC')
ax[1].boxplot(sc_data)
ax[1].set_xticklabels(['Male','Female'])
ax[1].set_ylabel('Best fit correlation between sFC and eSC')
ax[1].set_title('SC')
index = sub_num_list_phen.index(sub_num_list_old[i])
gen = gender_list[index]
if (gen == 'M'):
return_list1.append(x[i])
if (gen == 'F'):
return_list2.append(x[i])
return return_list1, return_list2
avg_male_sc_list, avg_female_sc_list = categorise_male_female(
avg_corr_sfc_esc_list)
#print(avg_female_sc_list)
t, p = scipy.stats.ttest_ind(avg_male_sc_list, avg_female_sc_list)
#print("t value: ", t)
#print("p value: ", p)
eff_size = pingouin.compute_effsize(avg_male_sc_list, avg_female_sc_list)
#print(eff_size)
#print(avg_female_fc_list)
#avg_male_sc_list, avg_female_sc_list = categorise_male_female(avg_corr_sfc_esc_list)
#print(max_corr_list_sc)
#print(corr_sfc_esc_list)
#male_fc_list, female_fc_list = categorise_male_female(corr_sfc_efc_list)
#print(female_fc_list)
#male_sc_list, female_sc_list = categorise_male_female(corr_sfc_esc_list)
#print(female_sc_list)
#male_delay_fc, female_delay_fc = categorise_male_female(max_delay_list_fc)
#male_delay_sc, female_delay_sc = categorise_male_female(max_delay_list_sc)
#male_coup_str_fc, female_coup_str_fc = categorise_male_female(max_coup_str_list_fc)
#male_coup_str_sc, female_coup_str_sc = categorise_male_female(max_coup_str_list_sc)
#print(len(male_fc_list))
#t_value_fc, p_value_fc = scipy.stats.ranksums(np.arctanh(corr_sfc_efc_male), np.arctanh(corr_sfc_efc_female)) #two tailed t test for corr(sFC, eFC)
#t_value_sc, p_value_sc = scipy.stats.ranksums(np.arctanh(corr_sfc_esc_male), np.arctanh(corr_sfc_esc_female)) #two tailed t test for corr(sFC, eSC)
#eff_size_sfc_efc.append(pg.compute_effsize(np.arctanh(corr_sfc_efc_male), np.arctanh(corr_sfc_efc_female), eftype = 'hedges'))
#eff_size_sfc_esc.append(pg.compute_effsize(np.arctanh(corr_sfc_esc_male), np.arctanh(corr_sfc_esc_female), eftype = 'hedges'))
#t_value, p_value = scipy.stats.ranksums(np.arctanh(corr_sfc_esc_func_fit_male), np.arctanh(corr_sfc_esc_func_fit_female))
#eff_size_sfc_esc_func_fit.append(pg.compute_effsize(np.arctanh(corr_sfc_esc_func_fit_male), np.arctanh(corr_sfc_esc_func_fit_female), eftype = 'hedges')) #effect size for coupling strength
t_value, p_value = scipy.stats.ranksums(
np.arctanh(corr_sfc_efc_struc_fit_male),
np.arctanh(corr_sfc_efc_struc_fit_female))
eff_size_sfc_efc_struc_fit.append(
pg.compute_effsize(
np.arctanh(corr_sfc_efc_struc_fit_male),
np.arctanh(corr_sfc_efc_struc_fit_female),
eftype='hedges')) #effect size for coupling strength
p_val.append(p_value)
r"""
np.savetxt(r"E:\Shraddha\Data\corr_sfc_efc_struc_fit_all_atlas_lc.csv", np.array(corr_sfc_efc_struc_fit_all_atlas).transpose(), delimiter = ',') #saving the corr(sFC, eSC) - functional fit for all atlas where each column represents one atlas in the .csv file
plt.rcParams['font.size'] = '20'
plt.figure(figsize = (16, 8))
male_plots = plt.boxplot(male_data_sfc_efc_struc_fit, positions = np.array(range(l))*2 - 0.3)
female_plots = plt.boxplot(female_data_sfc_efc_struc_fit, positions = np.array(range(l))*2 + 0.3)
set_box_color(male_plots, 'blue')
set_box_color(female_plots, 'red')