def test_retest_icc(atlas_name='MPM'):
import time
import numpy as np
import pickle as pkl
from cxy_hcp_ffa.lib.heritability import icc
# inputs
hemis = ('lh', 'rh')
rois = ('pFus-face', 'mFus-face')
test_file = pjoin(work_dir, f'rsfc_{atlas_name}2Cole_'
'{hemi}_test_rm-subj.pkl')
retest_file = pjoin(work_dir, f'rsfc_{atlas_name}2Cole_'
'{hemi}_rm-subj.pkl')
# outputs
out_file = pjoin(work_dir, f'{atlas_name}_rm-subj_icc.pkl')
# prepare
data = {}
# calculate
for hemi in hemis:
rsfc_test = pkl.load(open(test_file.format(hemi=hemi), 'rb'))
rsfc_retest = pkl.load(open(retest_file.format(hemi=hemi), 'rb'))
for roi in rois:
time1 = time.time()
test_vec = np.mean(rsfc_test[roi], 1)
retest_vec = np.mean(rsfc_retest[roi], 1)
k = f"{hemi}_{roi.split('-')[0]}"
data[k] = icc(test_vec, retest_vec, 10000, 95)
print(f'Finished {k}: cost {time.time()-time1} seconds.')
# save
pkl.dump(data, open(out_file, 'wb'))
def test_retest_icc(meas_name='activ', atlas_name='MPM'):
import time
import pickle as pkl
from cxy_hcp_ffa.lib.heritability import icc
# inputs
hemis = ('lh', 'rh')
rois = ('pFus', 'mFus')
meas_file = pjoin(work_dir, f'{meas_name}_{atlas_name}_rm-subj.pkl')
# outputs
out_file = pjoin(work_dir, f'{meas_name}_{atlas_name}_rm-subj_icc.pkl')
# prepare
meas = pkl.load(open(meas_file, 'rb'))
data = {}
# calculate
for hemi in hemis:
for roi in rois:
time1 = time.time()
k = f'{hemi}_{roi}'
data[k] = icc(meas[f'{k}_test'], meas[f'{k}_retest'], 10000, 95)
print(f'Finished {k}: cost {time.time()-time1} seconds.')
# save
pkl.dump(data, open(out_file, 'wb'))
def calc_Falconer_h2():
"""
Calculate Falconer's heritability by using ICC.
"""
import time
import numpy as np
import pandas as pd
n_bootstrap = 10000
confidence = 95
data_file = pjoin(work_dir, 'pre-heritability_rsfc-mean.csv')
out_file = pjoin(work_dir, 'heritability_icc_rsfc-mean.csv')
data = pd.read_csv(data_file)
mz_indices = data['zyg'] == 1
dz_indices = data['zyg'] == 3
var_names = [_[:-1] for _ in data.columns if _[-1] == '1']
n_var = len(var_names)
indices = [
'ICC_MZ', 'ICC_DZ', 'r_mz_lb', 'r_mz', 'r_mz_ub', 'r_dz_lb', 'r_dz',
'r_dz_ub', 'H2', 'h2_lb', 'h2', 'h2_ub'
]
out_dict = {}
for k in indices:
out_dict[k] = np.zeros(n_var)
for var_idx, var_name in enumerate(var_names):
time1 = time.time()
pair_name = [var_name + '1', var_name + '2']
mz = np.array(data.loc[mz_indices, pair_name]).T
mz = mz[:, ~np.any(np.isnan(mz), axis=0)]
dz = np.array(data.loc[dz_indices, pair_name]).T
dz = dz[:, ~np.any(np.isnan(dz), axis=0)]
print(f'#MZ in {var_name}:', mz.shape[1])
print(f'#DZ in {var_name}:', dz.shape[1])
out_dict['ICC_MZ'][var_idx] = h2.icc(mz)
out_dict['ICC_DZ'][var_idx] = h2.icc(dz)
r_mz_lb, r_mz, r_mz_ub = h2.icc(mz,
n_bootstrap=n_bootstrap,
confidence=confidence)
r_dz_lb, r_dz, r_dz_ub = h2.icc(dz,
n_bootstrap=n_bootstrap,
confidence=confidence)
out_dict['r_mz_lb'][var_idx] = r_mz_lb
out_dict['r_mz'][var_idx] = r_mz
out_dict['r_mz_ub'][var_idx] = r_mz_ub
out_dict['r_dz_lb'][var_idx] = r_dz_lb
out_dict['r_dz'][var_idx] = r_dz
out_dict['r_dz_ub'][var_idx] = r_dz_ub
out_dict['H2'][var_idx] = h2.heritability(mz, dz)
h2_lb, h, h2_ub = h2.heritability(mz,
dz,
n_bootstrap=n_bootstrap,
confidence=confidence)
out_dict['h2_lb'][var_idx] = h2_lb
out_dict['h2'][var_idx] = h
out_dict['h2_ub'][var_idx] = h2_ub
print(f'Finish {var_idx+1}/{n_var}, spend {time.time()-time1} seconds')
out_df = pd.DataFrame(np.array(list(out_dict.values())),
index=out_dict.keys(),
columns=var_names)
out_df.to_csv(out_file, index=True)
def test_retest_icc(hemi='lh'):
import numpy as np
import pickle as pkl
import nibabel as nib
from cxy_hcp_ffa.lib.predefine import roi2label, hemi2stru
from cxy_hcp_ffa.lib.heritability import icc
from magicbox.io.io import CiftiReader
# inputs
roi_names = ('pFus-face', 'mFus-face')
test_file = pjoin(proj_dir, 'analysis/s2/activation.dscalar.nii')
retest_file = pjoin(work_dir, 'activation.dscalar.nii')
subj_test_file = pjoin(proj_dir, 'analysis/s2/subject_id')
subj_retest_file = pjoin(work_dir, 'subject_id')
mpm_file = pjoin(
proj_dir, 'analysis/s2/1080_fROI/refined_with_Kevin/'
f'MPM_v3_{hemi}_0.25_FFA.nii.gz')
individual_file = pjoin(
proj_dir, 'analysis/s2/1080_fROI/refined_with_Kevin/'
f'rois_v3_{hemi}.nii.gz')
# outputs
out_file = pjoin(work_dir, f'ICC_{hemi}.pkl')
# prepare
subj_ids_test = open(subj_test_file).read().splitlines()
subj_ids_retest = open(subj_retest_file).read().splitlines()
retest_indices = [subj_ids_test.index(i) for i in subj_ids_retest]
test_maps = CiftiReader(test_file).get_data(hemi2stru[hemi],
True)[retest_indices]
retest_maps = CiftiReader(retest_file).get_data(hemi2stru[hemi], True)
mpm_mask = nib.load(mpm_file).get_fdata().squeeze()
individual_mask = nib.load(
individual_file).get_fdata().squeeze().T[retest_indices]
# calculate
n_roi = len(roi_names)
mpm_iccs = np.ones((3, n_roi)) * np.nan
individual_iccs = np.ones((3, n_roi)) * np.nan
for i, roi_name in enumerate(roi_names):
label = roi2label[roi_name]
mpm_roi_idx_arr = mpm_mask == label
individual_roi_idx_arr = individual_mask == label
valid_subj_idx_vec = np.any(individual_roi_idx_arr, 1)
individual_roi_idx_arr = individual_roi_idx_arr[valid_subj_idx_vec]
test_maps_tmp = test_maps[valid_subj_idx_vec]
retest_maps_tmp = retest_maps[valid_subj_idx_vec]
n_subj_valid = np.sum(valid_subj_idx_vec)
print(f'{hemi}_{roi_name}: {n_subj_valid} valid subjects')
# calculate ICC for MPM
test_series_mpm = np.mean(test_maps_tmp[:, mpm_roi_idx_arr], 1)
retest_series_mpm = np.mean(retest_maps_tmp[:, mpm_roi_idx_arr], 1)
mpm_iccs[:, i] = icc(test_series_mpm, retest_series_mpm, 10000, 95)
# calculate ICC for individual
test_series_ind = np.zeros((n_subj_valid, ))
retest_series_ind = np.zeros((n_subj_valid, ))
for j in range(n_subj_valid):
individual_roi_idx_vec = individual_roi_idx_arr[j]
test_series_ind[j] = np.mean(
test_maps_tmp[j][individual_roi_idx_vec])
retest_series_ind[j] = np.mean(
retest_maps_tmp[j][individual_roi_idx_vec])
individual_iccs[:, i] = icc(test_series_ind, retest_series_ind, 10000,
95)
# save
out_data = {
'roi_name': roi_names,
'mpm': mpm_iccs,
'individual': individual_iccs
}
pkl.dump(out_data, open(out_file, 'wb'))