def _main(argv=None):
"""Entry point for aroma CLI."""
options = _get_parser().parse_args(argv)
kwargs = vars(options)
aroma.aroma_workflow(**kwargs)
def test_integration(skip_integration, nilearn_data):
if skip_integration:
pytest.skip("Skipping integration test")
resources_path = get_tests_resource_path()
# Obtain test path
test_path, _ = split(nilearn_data.func[0])
# Create output path
out_path = join(test_path, "out")
# Read confounds
confounds = pd.read_csv(nilearn_data.confounds[0], sep="\t")
# Extract motion parameters from confounds
mc = confounds[[
"rot_x", "rot_y", "rot_z", "trans_x", "trans_y", "trans_z"
]]
mc_path = join(test_path, "mc.txt")
mc.to_csv(mc_path, sep="\t", index=False, header=None)
mixing = join(resources_path, "melodic_mix")
component_maps = join(resources_path, "melodic_IC_thr_MNI2mm.nii.gz")
aroma_workflow(
in_file=nilearn_data.func[0],
mixing=mixing,
component_maps=component_maps,
mc=mc_path,
out_dir=out_path,
TR=2,
den_type="nonaggr",
generate_plots=False,
overwrite=True,
)
# Make sure files are generated
assert isfile(join(out_path, "classification_overview.txt"))
assert isfile(join(out_path, "classified_motion_ICs.txt"))
assert isfile(join(out_path, "denoised_func_data_nonaggr.nii.gz"))
assert isfile(join(out_path, "feature_scores.tsv"))
# Check classification overview file
true_classification_overview = pd.read_csv(
join(resources_path, "classification_overview.txt"),
sep="\t",
index_col="IC",
nrows=4,
)
classification_overview = pd.read_csv(join(out_path,
"classification_overview.txt"),
sep="\t",
index_col="IC",
nrows=4)
assert np.allclose(true_classification_overview.iloc[:, :-1].values,
classification_overview.iloc[:, :-1].values,
atol=0.9)
# Check feature scores
f_scores = pd.read_table(join(out_path, "feature_scores.tsv"))
f_true = pd.read_table(join(resources_path, "feature_scores.txt"))
assert np.allclose(f_true.values, f_scores.values, atol=0.9)
# Check motion ICs
mot_ics = np.loadtxt(join(out_path, "classified_motion_ICs.txt"),
delimiter=",")
true_mot_ics = np.loadtxt(join(resources_path,
"classified_motion_ICs.txt"),
delimiter=",")
assert np.allclose(true_mot_ics[:4], mot_ics[:4])
def test_integration(skip_integration, nilearn_data, mel_mix, mel_IC,
classification_overview, classified_motion_ICs):
"""Perform integration test."""
if skip_integration:
pytest.skip("Skipping integration test")
# Obtain test path
test_path, _ = op.split(nilearn_data.func[0])
# Create output path
out_path = op.join(test_path, "out")
# Read confounds
confounds = pd.read_table(nilearn_data.confounds[0])
# Extract motion parameters from confounds
mc = confounds[[
"rot_x", "rot_y", "rot_z", "trans_x", "trans_y", "trans_z"
]]
mc_path = op.join(test_path, "mc.txt")
mc.to_csv(mc_path, sep="\t", index=False, header=None)
# Add seed for reproducibility
np.random.seed(42)
aroma_workflow(
in_file=nilearn_data.func[0],
mixing=mel_mix,
component_maps=mel_IC,
mc=mc_path,
out_dir=out_path,
TR=2,
den_type="nonaggr",
generate_plots=False,
overwrite=True,
)
# Make sure files are generated
assert op.isfile(op.join(out_path, "desc-AROMA_metrics.tsv"))
assert op.isfile(op.join(out_path, "AROMAnoiseICs.csv"))
assert op.isfile(op.join(out_path, "desc-smoothAROMAnonaggr_bold.nii.gz"))
# Load classification overview file
true_classification_overview = pd.read_table(
classification_overview,
index_col="IC",
)
test_classification_overview = pd.read_table(
op.join(out_path, "desc-AROMA_metrics.tsv"),
index_col="IC",
)
# Check feature scores
f_scores = test_classification_overview[[
"edge_fract", "csf_fract", "max_RP_corr", "HFC"
]]
f_true = true_classification_overview[[
"edge_fract", "csf_fract", "max_RP_corr", "HFC"
]]
assert np.allclose(f_true.values, f_scores.values,
atol=0.01), f_true.values - f_scores.values
# Check classifications
assert (true_classification_overview["classification"].tolist() ==
test_classification_overview["classification"].tolist())
# Check motion ICs
test_mot_ics = np.loadtxt(op.join(out_path, "AROMAnoiseICs.csv"),
delimiter=",")
true_mot_ics = np.loadtxt(classified_motion_ICs, delimiter=",")
assert np.allclose(true_mot_ics, test_mot_ics)