def test_PDistTargetSimilaritySearchlight():
# Test ability to use PDistTargetSimilarity in a searchlight
from mvpa2.testing.datasets import datasets
from mvpa2.mappers.fx import mean_group_sample
from mvpa2.mappers.shape import TransposeMapper
from mvpa2.measures.searchlight import sphere_searchlight
ds = datasets['3dsmall'][:, :3]
ds.fa['voxel_indices'] = ds.fa.myspace
# use chunks values (4 of them) for targets
ds.sa['targets'] = ds.sa.chunks
ds = mean_group_sample(['chunks'])(ds)
tdsm = np.arange(6)
# We can run on full dataset
tdcm1 = PDistTargetSimilarity(tdsm)
a1 = tdcm1(ds)
assert_array_equal(a1.fa.metrics, ['rho', 'p'])
tdcm1_rho = PDistTargetSimilarity(tdsm, corrcoef_only=True)
sl_rho = sphere_searchlight(tdcm1_rho)(ds)
assert_array_equal(sl_rho.shape, (1, ds.nfeatures))
# now with both but we need to transpose datasets
tdcm1_both = PDistTargetSimilarity(tdsm, postproc=TransposeMapper())
sl_both = sphere_searchlight(tdcm1_both)(ds)
assert_array_equal(sl_both.shape, (2, ds.nfeatures))
assert_array_equal(sl_both.sa.metrics, ['rho', 'p'])
# rho must be exactly the same
assert_array_equal(sl_both.samples[0], sl_rho.samples[0])
# just because we are here and we can
# Actually here for some reason assert_array_lequal gave me a trouble
assert_true(np.all(sl_both.samples[1] <= 1.0))
assert_true(np.all(0 <= sl_both.samples[1]))
def test_transpose():
from mvpa2.mappers.shape import TransposeMapper
ds = Dataset(np.arange(24).reshape(2, 3, 4), sa={"testsa": np.arange(2)}, fa={"testfa": np.arange(3)})
tp = TransposeMapper()
tds = tp(ds)
assert_equal(tds.shape, (3, 2, 4))
assert_true("testfa" in tds.sa)
assert_true("testsa" in tds.fa)
assert_false(tds.fa is tds.sa)
# and back
ttds = tp(tds)
assert_array_equal(ttds.samples, ds.samples)
assert_equal(ttds.sa, ds.sa)
assert_equal(ttds.fa, ds.fa)
# or this way
rds = tp.reverse(tds)
assert_array_equal(rds.samples, ds.samples)
assert_equal(rds.sa, ds.sa)
assert_equal(rds.fa, ds.fa)
assert_array_equal(rds.samples, ttds.samples)
assert_equal(rds.sa, ttds.sa)
assert_equal(rds.fa, ttds.fa)
def test_transpose():
from mvpa2.mappers.shape import TransposeMapper
ds = Dataset(np.arange(24).reshape(2, 3, 4),
sa={'testsa': np.arange(2)},
fa={'testfa': np.arange(3)})
tp = TransposeMapper()
tds = tp(ds)
assert_equal(tds.shape, (3, 2, 4))
assert_true('testfa' in tds.sa)
assert_true('testsa' in tds.fa)
assert_false(tds.fa is tds.sa)
# and back
ttds = tp(tds)
assert_array_equal(ttds.samples, ds.samples)
assert_equal(ttds.sa, ds.sa)
assert_equal(ttds.fa, ds.fa)
# or this way
rds = tp.reverse(tds)
assert_array_equal(rds.samples, ds.samples)
assert_equal(rds.sa, ds.sa)
assert_equal(rds.fa, ds.fa)
assert_array_equal(rds.samples, ttds.samples)
assert_equal(rds.sa, ttds.sa)
assert_equal(rds.fa, ttds.fa)
'Most consistent searchlight pattern correlation distances')
"""
It is all in the face!
It would be interesting to know where in the brain dissimilarity structures can be found that are similar to this one. PDistTargetSimilarity() can be used to discover this kind of information with any kind of target dissimilarity structure. We need to transpose the result for aggregation into a searchlight map, as PDistTargetSimilarity can return more features than just the correlation coefficient.
"""
# let's see where in the brain we find dissimilarity structures that are
# similar to our most stable one
tdsm = rsa.PDistTargetSimilarity(
slres.samples[:, mean_consistency.argmax()])
# using a searchlight
from mvpa2.base.learner import ChainLearner
from mvpa2.mappers.shape import TransposeMapper
sl_tdsm = sphere_searchlight(ChainLearner([tdsm, TransposeMapper()]), 2)
slres_tdsm = sl_tdsm(mtds)
"""
Lastly, we can map this result back onto the 3D voxel grid, and overlay it onto the brain anatomy.
"""
# plot the spatial distribution using NiPy
vol = ds.a.mapper.reverse1(slres_tdsm.samples[0])
import nibabel as nb
anat = nb.load(pjoin(datapath, 'sub001', 'anatomy', 'highres001.nii.gz'))
from nipy.labs.viz_tools.activation_maps import plot_map
pl.figure(figsize=(15, 4))
sp = pl.subplot(121)
pl.title('Distribution of target similarity structure correlation')
