def test_plot_surf():
s1 = to_data(vtk.vtkSphereSource())
s2 = to_data(vtk.vtkSphereSource())
surfs = {'s1': s1, 's2': s2}
layout = np.array([['s1', 's2'], ['s2', 's2']])
plot_surf(surfs, layout, offscreen=True)
def test_build_plotter():
s1 = to_data(vtk.vtkSphereSource())
s2 = to_data(vtk.vtkSphereSource())
surfs = {'s1': s1, 's2': s2}
layout = np.array([['s1', 's2'], ['s2', 's2']])
p = build_plotter(surfs, layout, offscreen=True)
assert isinstance(p, Plotter)
def plotter_single_renderer():
s = to_data(vtk.vtkSphereSource())
p = Plotter(offscreen=True)
ren0 = p.AddRenderer(row=0, col=0)
ac0 = ren0.AddActor()
ac0.SetMapper(inputdata=s)
return p
def test_moran():
# Sphere with points as locations to build spatial matrix
sphere = wrap_vtk(vtk.vtkSphereSource,
radius=20,
thetaResolution=10,
phiResolution=5)
sphere = to_data(sphere)
n_pts = sphere.n_points
# Features to randomize
rs = np.random.RandomState(0)
feats = rs.randn(n_pts, 2)
# build spatial weight matrix
a = me.get_immediate_distance(sphere)
a.data **= -1
# test default
v, w = compute_mem(a, tol=1e-7)
assert w.shape[0] <= (n_pts - 1)
assert v.shape == (n_pts, w.shape[0])
r1 = moran_randomization(feats[:, 0], v, n_rep=10, random_state=0)
assert r1.shape == (10, n_pts)
r2 = moran_randomization(feats, v, n_rep=10, random_state=0)
assert r2.shape == (10, n_pts, 2)
# test default dense
mem, ev = compute_mem(a.toarray(), tol=1e-7)
assert np.allclose(w, ev)
assert np.allclose(v, mem)
r1 = moran_randomization(feats[:, 0], mem, n_rep=10, random_state=0)
assert r1.shape == (10, n_pts)
r2 = moran_randomization(feats, mem, n_rep=10, random_state=0)
assert r2.shape == (10, n_pts, 2)
# test object api
msr = MoranRandomization(n_rep=10, random_state=0, tol=1e-7)
msr.fit(a)
assert np.allclose(msr.mev_, ev)
assert np.allclose(msr.mem_, mem)
assert np.allclose(r1, msr.randomize(feats[:, 0]))
assert np.allclose(r2, msr.randomize(feats))
# test object api with PolyData
msr = MoranRandomization(n_rep=10, random_state=0, tol=1e-7)
msr.fit(sphere)
assert np.allclose(msr.mev_, ev)
assert np.allclose(msr.mem_, mem)
assert np.allclose(r1, msr.randomize(feats[:, 0]))
assert np.allclose(r2, msr.randomize(feats))
def plotter_multiple_renderers():
s = to_data(vtk.vtkSphereSource())
p = Plotter(nrow=1, ncol=2, offscreen=True)
ren0 = p.AddRenderer(row=0, col=0)
ac0 = ren0.AddActor()
ac0.SetMapper(inputdata=s)
ren1 = p.AddRenderer(row=0, col=1)
ac1 = ren1.AddActor()
ac1.SetMapper(inputdata=s)
return p
def test_plotter():
s = to_data(vtk.vtkSphereSource())
p = Plotter(offscreen=True)
ren0 = p.AddRenderer(row=0, col=0)
ac0 = ren0.AddActor()
m0 = ac0.SetMapper(inputdata=s)
p.show(embed_nb=False)
img = p.screenshot()
assert isinstance(img, PIL.Image.Image)
p.close()
def test_variogram_sampled():
# Sampled class
# Sphere with points as locations to build distance matrix
sphere = wrap_vtk(vtk.vtkSphereSource,
radius=20,
thetaResolution=50,
phiResolution=25)
sphere = to_data(sphere)
points = sphere.GetPoints()
npoints = sphere.n_points
distmat = squareform(pdist(points)) # pairwise distance matrix
rs = np.random.RandomState(0)
brainmap = rs.randn(npoints, 1).flatten()
brainmap[0] = np.nan
# Create a mask file
mask = np.zeros(npoints, dtype=int)
mask[[3, 4, 5]] = 1
# Save distmat and mask file
temp = gettempdir()
dist_file = join(temp, 'distmat.txt')
mask_file = join(temp, 'mask.txt')
np.savetxt(dist_file, distmat)
np.savetxt(mask_file, mask)
assert exists(dist_file) and exists(mask_file)
# Convert to memmap
files = txt2memmap(dist_file=dist_file,
output_dir=temp,
maskfile=mask_file)
assert exists(files['index'])
assert exists(files['distmat'])
masked_map = brainmap[np.where(mask == 0)]
# Generate surrogates
gen = Sampled(masked_map,
files['distmat'],
files['index'],
knn=100,
ns=100)
surrs = gen(n=5)
assert gen.D.shape == (npoints - 3, 100)
assert surrs.shape == (5, npoints - 3)
assert np.allclose(gen.x.data[1:], masked_map[1:])
assert np.isnan(gen.x.data[0])
assert not np.isnan(surrs).any()
def test_variogram_base():
# Base class
# Sphere with points as locations to build distance matrix
sphere = wrap_vtk(vtk.vtkSphereSource,
radius=20,
thetaResolution=20,
phiResolution=10)
sphere = to_data(sphere)
points = sphere.GetPoints()
npoints = sphere.n_points
distmat = squareform(pdist(points)) # pairwise distance matrix
rs = np.random.RandomState(0)
brainmap = rs.randn(npoints, 1).flatten()
brainmap[0] = np.nan
# Generate surrogates
gen = Base(brainmap, distmat)
surrs = gen(n=10)
assert surrs.shape == (10, npoints)
assert np.allclose(gen.D, distmat)
assert np.allclose(gen.x.data[1:], brainmap[1:])
assert np.isnan(gen.x.data[0])
assert not np.isnan(surrs).any()
def test_spin():
# Create dummy spheres or left and right hemispheres
sphere_lh = wrap_vtk(vtk.vtkSphereSource,
radius=20,
thetaResolution=10,
phiResolution=5)
sphere_lh = to_data(sphere_lh)
pts_lh = sphere_lh.Points
n_pts_lh = sphere_lh.n_points
# Right with more points
sphere_rh = wrap_vtk(vtk.vtkSphereSource,
radius=20,
thetaResolution=10,
phiResolution=6)
sphere_rh = to_data(sphere_rh)
pts_rh = sphere_rh.Points
n_pts_rh = sphere_rh.n_points
# Features to randomize
rs = np.random.RandomState(0)
feats_lh = rs.randn(n_pts_lh, 2)
feats_rh = rs.randn(n_pts_rh, 2)
# generate spin indices
ridx1 = _generate_spins(pts_lh, n_rep=10, random_state=0)
assert ridx1['lh'].shape == (10, n_pts_lh)
assert 'rh' not in ridx1
ridx2 = _generate_spins(pts_lh, points_rh=pts_rh, n_rep=10, random_state=0)
assert ridx2['lh'].shape == (10, n_pts_lh)
assert ridx2['rh'].shape == (10, n_pts_rh)
# test api lh
sp = SpinPermutations(n_rep=10, random_state=0)
sp.fit(pts_lh)
assert np.all(sp.spin_lh_ == ridx1['lh'])
assert sp.spin_rh_ is None
assert sp.randomize(feats_lh[:, 0]).shape == (10, n_pts_lh)
assert sp.randomize(feats_lh).shape == (10, n_pts_lh, 2)
# test api lh and rh
sp = SpinPermutations(n_rep=10, random_state=0)
sp.fit(pts_lh, points_rh=pts_rh)
assert np.all(sp.spin_lh_ == ridx2['lh'])
assert np.all(sp.spin_rh_ == ridx2['rh'])
r1 = sp.randomize(feats_lh[:, 0])
assert r1[0].shape == (10, n_pts_lh)
assert r1[1] is None
r1bis = spin_permutations(pts_lh, feats_lh[:, 0], n_rep=10, random_state=0)
assert np.all(r1[0] == r1bis)
r2 = sp.randomize(feats_lh)
assert r2[0].shape == (10, n_pts_lh, 2)
assert r2[1] is None
r2bis = spin_permutations(pts_lh, feats_lh, n_rep=10, random_state=0)
assert np.all(r2[0] == r2bis)
r1 = sp.randomize(feats_lh[:, 0], x_rh=feats_rh[:, 0])
assert r1[0].shape == (10, n_pts_lh)
assert r1[1].shape == (10, n_pts_rh)
r1bis = spin_permutations({
'lh': pts_lh,
'rh': pts_rh
}, {
'lh': feats_lh[:, 0],
'rh': feats_rh[:, 0]
},
n_rep=10,
random_state=0)
assert np.all(r1[0] == r1bis[0])
assert np.all(r1[1] == r1bis[1])
r2 = sp.randomize(feats_lh, x_rh=feats_rh)
assert r2[0].shape == (10, n_pts_lh, 2)
assert r2[1].shape == (10, n_pts_rh, 2)
r2bis = spin_permutations({
'lh': pts_lh,
'rh': pts_rh
}, {
'lh': feats_lh,
'rh': feats_rh
},
n_rep=10,
random_state=0)
assert np.all(r2[0] == r2bis[0])
assert np.all(r2[1] == r2bis[1])
def test_plot_hemispheres():
s1 = to_data(vtk.vtkSphereSource())
s2 = to_data(vtk.vtkSphereSource())
plot_hemispheres(s1, s2, offscreen=True)
