X, Y, sulci_df_qc = load_residualized_bmi_data(cache=False)
colnames = sulci_df_qc.columns
penalty_start = 11
# # if add tiv² as an additionnal cofound
# penalty_start = 12
# Initialize beta_map
beta_map = np.zeros((X.shape[1], Y.shape[1]))
print "##############################################################"
print ("# Perform Mass-Univariate Linear Modeling "
"based Ordinary Least Squares #")
print "##############################################################"
#MUOLS
bigols = MUOLS()
bigols.fit(X, Y)
t, p, df = bigols.stats_t_coefficients(X, Y,
contrast=[0.] * penalty_start +
[1.] * (X.shape[1] - penalty_start),
# # if add tiv² as an additionnal cofound
# contrast=[0.] * penalty_start +
# [1.] * (X.shape[1] - penalty_start),
# # if want the contrast associated to mean_pds
# contrast=[0.] * (penalty_start - 2) + [1.]
# + [0.] * (X.shape[1] - penalty_start + 1),
pval=True)
proba = []
for i in np.arange(0, p.shape[0]):
if (p[i] > 0.95):
# Load data
X, Y, sulci_df_qc = load_residualized_bmi_data(cache=False)
colnames = sulci_df_qc.columns
penalty_start = 12
# Initialize beta_map
beta_map = np.zeros((X.shape[1], Y.shape[1]))
print "##############################################################"
print(
"# Perform Mass-Univariate Linear Modeling "
"based Ordinary Least Squares #")
print "##############################################################"
#MUOLS
bigols = MUOLS()
bigols.fit(X, Y)
t, p, df = bigols.stats_t_coefficients(
X,
Y,
# if add tiv² as an additionnal cofound
contrast=[0.] * penalty_start + [1.] * (X.shape[1] - penalty_start),
pval=True)
proba = []
for i in np.arange(0, p.shape[0]):
if (p[i] > 0.95):
p[i] = 1 - p[i]
proba.append('%.15f' % p[i])
# Beta values: coefficients of the fit
SHARED_DIR = os.path.join(BASE_SHARED_DIR, 'residualized_bmi_cache_IMAGEN')
if not os.path.exists(SHARED_DIR):
os.makedirs(SHARED_DIR)
X, Y = load_residualized_bmi_data(cache=False)
n, p = Y.shape
np.save(os.path.join(WD, 'X.npy'), X)
np.save(os.path.join(WD, 'Y.npy'), Y)
print "################################"
print "# Perform Mass-Univariate Linear Modeling based Ordinary Least Squares #"
print "################################"
#MUOLS
beta_map = np.zeros(p)
bigols = MUOLS()
bigols.fit(X, Y)
s, p = bigols.stats_t_coefficients(
X,
Y,
contrast=[0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 1.],
pval=True)
beta_map[:] = s[:]
template_for_size = os.path.join(DATA_PATH, 'mask', 'mask.nii')
template_for_size_img = ni.load(template_for_size)
mask_data = template_for_size_img.get_data()
masked_data_index = (mask_data == 1.0)
image = np.zeros(template_for_size_img.get_data().shape)
image[masked_data_index] = beta_map
sulci_depthMax_df.to_csv(os.path.join(QC_PATH, 'sulci_depthMax_df.csv'))
print "Dataframe containing sulci maximal depth after quality control has been saved."
colnames = sulci_depthMax_df.columns
penalty_start = 11
# Initialize beta_map
beta_map = np.zeros((X.shape[1], Y.shape[1]))
print "##############################################################"
print ("# Perform Mass-Univariate Linear Modeling "
"based Ordinary Least Squares #")
print "##############################################################"
#MUOLS
bigols = MUOLS()
bigols.fit(X, Y)
t, p, df = bigols.stats_t_coefficients(X, Y,
contrast=[0.] * penalty_start +
[1.] * (X.shape[1] - penalty_start),
pval=True)
proba = []
for i in np.arange(0, p.shape[0]):
if (p[i] > 0.95):
p[i] = 1 - p[i]
proba.append('%.15f' % p[i])
# Beta values: coefficients of the fit
beta_map = bigols.coef_