Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix the two sided test on IBMA Fishers and Stouffers estimators #903

Closed
wants to merge 7 commits into from
Closed

Fix the two sided test on IBMA Fishers and Stouffers estimators #903

Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
f1c7c4b
Fix the two sided test on IBMA Fishers and Stouffers estimators
JulioAPeraza Sep 10, 2024
f8a8e87
Upgrade upload-artifact to v4
JulioAPeraza Sep 10, 2024
43e5c31
Update testing.yml
JulioAPeraza Sep 10, 2024
69cc4d2
Update 12_plot_ibma_workflow.py
JulioAPeraza Sep 10, 2024
68c4425
Update testing.yml
JulioAPeraza Sep 10, 2024
e072d4a
Update ibma.py
JulioAPeraza Sep 10, 2024
23865c5
Update ibma.py
JulioAPeraza Sep 10, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
65 changes: 50 additions & 15 deletions nimare/meta/ibma.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -228,7 +228,6 @@ class Fishers(IBMAEstimator):
def __init__(self, two_sided=True, **kwargs):
super().__init__(**kwargs)
self.two_sided = two_sided
self._mode = "concordant" if self.two_sided else "directed"

def _generate_description(self):
description = (
Expand All @@ -243,13 +242,32 @@ def _fit_model(self, stat_maps):
"""Fit the model to the data."""
n_studies, n_voxels = stat_maps.shape

pymare_dset = pymare.Dataset(y=stat_maps)
est = pymare.estimators.FisherCombinationTest(mode=self._mode)
est.fit_dataset(pymare_dset)
est_summary = est.summary()
# Run Stouffers for right tail
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

suggestion (performance): Consider performance implications of running two separate tests for two-sided cases

The new implementation runs two separate tests when two_sided is True, which could potentially double the computation time. Consider if there's a way to optimize this, especially for large datasets.

pos_est = pymare.estimators.FisherCombinationTest(mode="directed")
pos_pymare_dset = pymare.Dataset(y=stat_maps)
pos_est.fit_dataset(pos_pymare_dset)
pos_est_summary = pos_est.summary()

if self.two_sided:
# Run Stouffers for left tail
neg_est = pymare.estimators.FisherCombinationTest(mode="directed")
neg_pymare_dset = pymare.Dataset(y=-stat_maps)
neg_est.fit_dataset(neg_pymare_dset)
neg_est_summary = neg_est.summary()

# Find the minimum p-values
p_map = np.minimum(pos_est_summary.p.squeeze(), neg_est_summary.p.squeeze())

# Create z_map based on which p-value is smaller
z_map = np.where(
pos_est_summary.p.squeeze() <= neg_est_summary.p.squeeze(),
pos_est_summary.z.squeeze(),
-neg_est_summary.z.squeeze(),
)
else:
z_map = pos_est_summary.z.squeeze()
p_map = pos_est_summary.p.squeeze()

z_map = est_summary.z.squeeze()
p_map = est_summary.p.squeeze()
dof_map = np.tile(n_studies - 1, n_voxels).astype(np.int32)

return z_map, p_map, dof_map
Expand Down Expand Up @@ -381,7 +399,6 @@ def __init__(
self.normalize_contrast_weights = normalize_contrast_weights

self.two_sided = two_sided
self._mode = "concordant" if self.two_sided else "directed"

def _preprocess_input(self, dataset):
"""Preprocess additional inputs to the Estimator from the Dataset as needed."""
Expand Down Expand Up @@ -440,8 +457,6 @@ def _fit_model(self, stat_maps, study_mask=None, corr=None):
# when using the aggressive mask.
study_mask = np.arange(n_studies)

est = pymare.estimators.StoufferCombinationTest(mode=self._mode)

contrast_maps, sub_corr = None, None
if corr is not None:
contrast_maps = np.tile(self.inputs_["contrast_names"][study_mask], (n_voxels, 1)).T
Expand All @@ -460,12 +475,32 @@ def _fit_model(self, stat_maps, study_mask=None, corr=None):

weight_maps = np.tile(weights, (n_voxels, 1)).T

pymare_dset = pymare.Dataset(y=stat_maps, n=weight_maps, v=contrast_maps)
est.fit_dataset(pymare_dset, corr=sub_corr)
est_summary = est.summary()
# Run Stouffers for right tail
pos_est = pymare.estimators.StoufferCombinationTest(mode="directed")
pos_pymare_dset = pymare.Dataset(y=stat_maps, n=weight_maps, v=contrast_maps)
pos_est.fit_dataset(pos_pymare_dset, corr=sub_corr)
pos_est_summary = pos_est.summary()

if self.two_sided:
# Run Stouffers for left tail
neg_est = pymare.estimators.StoufferCombinationTest(mode="directed")
neg_pymare_dset = pymare.Dataset(y=-stat_maps, n=weight_maps, v=contrast_maps)
neg_est.fit_dataset(neg_pymare_dset, corr=sub_corr)
neg_est_summary = neg_est.summary()

# Find the minimum p-values
p_map = np.minimum(pos_est_summary.p.squeeze(), neg_est_summary.p.squeeze())

# Create z_map based on which p-value is smaller
z_map = np.where(
pos_est_summary.p.squeeze() <= neg_est_summary.p.squeeze(),
pos_est_summary.z.squeeze(),
-neg_est_summary.z.squeeze(),
)
else:
z_map = pos_est_summary.z.squeeze()
p_map = pos_est_summary.p.squeeze()

z_map = est_summary.z.squeeze()
p_map = est_summary.p.squeeze()
dof_map = np.tile(n_studies - 1, n_voxels).astype(np.int32)

return z_map, p_map, dof_map
Expand Down
Loading