def __init__(self, **kwargs):
self.nodes = [
Detrender(chunks_attr='file'),
Detrender(),
FeatureZNormalizer(),
SampleSlicer(selection_dictionary={'events_number': range(1, 13)})
]
PreprocessingPipeline.__init__(self, nodes=self.nodes)
def loading_data(img_pattern):
loader = DataLoader(configuration_file=conf_file,
loader='mat',
task='fcmri',
event_file=img_pattern[:-4] + ".txt",
img_pattern=img_pattern,
atlas='findlab')
prepro = PreprocessingPipeline(nodes=[
Transformer(),
#Detrender(),
#SampleZNormalizer(),
#FeatureZNormalizer()
])
#prepro = PreprocessingPipeline()
ds = loader.fetch(prepro=prepro)
return ds
configuration_file="/home/carlos/fmri/carlo_ofp/ofp_new.conf",
task='OFP_NORES')
ds = loader.fetch()
decoding = RoiDecoding(n_jobs=20, scoring=['accuracy'])
results = dict()
for subject in np.unique(ds.sa.subject):
results[subject] = []
for evidence in [1, 2, 3]:
pipeline = PreprocessingPipeline(nodes=[
TargetTransformer('decision'),
SampleSlicer(**{
'subject': [subject],
'evidence': [evidence]
}),
Balancer(balancer=RandomUnderSampler(return_indices=True),
attr='chunks'),
])
ds_ = pipeline.transform(ds)
decoding.fit(ds_, roi=['lateral_ips'])
results[subject].append(decoding.scores)
with open(
os.path.join(loader._data_path,
'0_results/lateral_ips_decoding.pickle'), 'wb') as output:
pickle.dump(results, output)
matrix_list = glob.glob("/media/robbis/DATA/fmri/monks/061102chrwoo/fcmri/*.mat")
matrix_list = [m.split("/")[-1] for m in matrix_list]
for m in matrix_list:
m = '20151103_132009_connectivity_filtered_first_filtered_after_each_run_no_gsr_findlab_fmri.mat'
loader = DataLoader(configuration_file=conf_file,
loader='mat',
task='fcmri',
atlas='findlab',
event_file=m[:-4]+".txt",
img_pattern=m)
prepro = PreprocessingPipeline(nodes=[
#Transformer(),
#Detrender(),
SampleZNormalizer(),
#FeatureZNormalizer()
])
#prepro = PreprocessingPipeline()
ds = loader.fetch(prepro=prepro)
_default_options = [
{
'prepro':['sample_slicer', 'target_transformer'],
'target_transformer__attr': 'expertise',
'sample_slicer__targets': ['Samatha']
},
{
'prepro':['sample_slicer', 'target_transformer'],
nodes = ds.fa.nodes_1
matrix = np.zeros_like(ds.samples[0])
nanmask = np.logical_not(np.isnan(ds.samples).sum(0))
ds = ds[:, nanmask]
# 1. Transform dataset to have mean 0 and std 1
prepro = [
FeatureZNormalizer(),
SampleAttributeTransformer(attr='dexterity1', fx=('zscore', zscore)),
SampleAttributeTransformer(attr='dexterity2', fx=('zscore', zscore)),
]
ds = PreprocessingPipeline(nodes=prepro).transform(ds)
formulas = [
'task + dexterity1 - 1',
'task + dexterity2 - 1',
'task - 1',
'dexterity1 - 1',
'dexterity2 - 1',
]
contrasts = [
{'f+restvstask': [1, -1/6, -1/6, -1/6, -1/6, -1/6, -1/6, 0],
'f+taskvsdext': [1/7, 1/7, 1/7, 1/7, 1/7, 1/7, 1/7, -1],
'f+task': [[1, -1, 0, 0, 0, 0, 0, 0],
}
configuration_file = "/home/carlos/fmri/carlo_mdm/memory.conf"
#configuration_file = "/media/robbis/DATA/fmri/carlo_mdm/memory.conf"
loader = DataLoader(
configuration_file=configuration_file,
#data_path="/home/carlos/mount/meg_workstation/Carlo_MDM/",
task='BETA_MVPA',
roi_labels=roi_labels,
event_file="beta_attributes_full",
brain_mask="mask_intersection")
prepro = PreprocessingPipeline(nodes=[
#Transformer(),
Detrender(),
SampleZNormalizer(),
FeatureZNormalizer(),
])
#prepro = PreprocessingPipeline()
ds = loader.fetch(prepro=prepro)
ds = MemoryReducer(dtype=np.float16).transform(ds)
_default_options = [
{
'target_transformer__attr':
"image_type",
'sample_slicer__attr': {
'image_type': ["I", "O"],
'evidence': [1]
},
from joblib import Parallel, delayed
path = "/media/robbis/Seagate_Pt1/data/working_memory/"
conf_file = "%s/data/working_memory.conf" % (path)
task = 'PSI'
task = 'PSICORR'
loader = DataLoader(configuration_file=conf_file,
loader='mat',
task=task,
data_path="%s/data/" % (path),
subjects="%s/data/participants.csv" % (path))
prepro = PreprocessingPipeline(nodes=[
Transformer(),
#SampleZNormalizer()
])
ds = loader.fetch(prepro=prepro)
_default_options = {
'sample_slicer__targets': [['0back', '2back']],
'sample_slicer__band': [[c] for c in np.unique(ds.sa.band)],
'estimator__fsel__k': np.arange(1, 1200, 50),
}
_default_config = {
'prepro': ['sample_slicer'],
#'ds_normalizer__ds_fx': np.std,
'sample_slicer__band': ['gamma'],
'sample_slicer__targets': ['0back', '2back'],
root = enable_logging()
#####################################
#conf_file = "/home/carlos/mount/megmri03/working_memory/working_memory_remote.conf"
conf_file = "/media/robbis/DATA/fmri/working_memory/working_memory.conf"
conf_file = "/media/robbis/Seagate_Pt1/data/working_memory/working_memory.conf"
loader = DataLoader(configuration_file=conf_file,
loader=load_mat_ds,
task='CONN')
prepro = PreprocessingPipeline(nodes=[
Transformer(),
#SignTransformer(),
Detrender(),
#AbsoluteValueTransformer(),
SignTransformer(),
#SampleSigmaNormalizer(),
#FeatureSigmaNormalizer(),
])
#prepro = PreprocessingPipeline()
ds = loader.fetch(prepro=prepro)
_default_options = {
'sample_slicer__targets' : [['0back', '2back'], ['0back', 'rest'], ['rest', '2back']],
#'sample_slicer__targets' : [['0back', '2back']],
'sample_slicer__band': [[c] for c in np.unique(ds.sa.band)],
#'estimator__clf__C': [0.1, 1, 10],
#'cv__n_splits': [75, 150, 200, 250],
# Only when running on permut1
from pyitab.utils import enable_logging
root = enable_logging()
#####################################
conf_file = ""
conf_file = '/media/guidotr1/Seagate_Pt1/data/Viviana2018/meg/movie.conf'
loader = DataLoader(configuration_file=conf_file,
loader='mat',
task='conn')
prepro = PreprocessingPipeline(nodes=[
#SampleZNormalizer(),
#FeatureZNormalizer(),
Resampler(down=5)
])
#prepro = PreprocessingPipeline()
ds = loader.fetch(prepro=prepro)
_default_options = {
'sample_slicer__targets' : [['movie', 'rest'],
['movie', 'scramble']],
'sample_slicer__band' : [['alpha'], ['beta']],
}
_default_config = {
# Only when running on permut1
from mvpa_itab.utils import enable_logging
root = enable_logging()
#####################################
conf_file = "/media/robbis/DATA/fmri/carlo_mdm/memory.conf"
loader = DataLoader(
configuration_file=conf_file,
#loader=load_mat_ds,
task='BETA_MVPA')
prepro = PreprocessingPipeline(nodes=[
#Transformer(),
Detrender(),
SampleZNormalizer(),
FeatureZNormalizer()
])
#prepro = PreprocessingPipeline()
ds = loader.fetch(prepro=prepro)
_default_options = {
#'target_trans__target': ["decision"],
'sample_slicer__accuracy': [[1], [0]],
}
_default_config = {
'prepro': ['sample_slicer', 'target_transformer', 'balancer'],
'sample_slicer__decision': ['NEW', 'OLD'],
'sample_slicer__evidence': [1],
SampleAttributeTransformer(attr='dexterity2', fx=('zscore', zscore)),
]
networks = ['AN', 'CON', 'DAN', 'DMN', 'FPN', 'LN', 'MN', 'VAN', 'VFN', 'VPN']
tasks = ['task1']
betas = list()
intercepts = list()
for net in networks:
for subj in np.unique(ds.sa.subject):
prepro_X = [
SampleSlicer(subject=[subj], band=['alpha'], task=['rest']),
FeatureSlicer(nodes_1=['MN'], nodes_2=['MN']),
SampleZNormalizer()
]
X = PreprocessingPipeline(nodes=prepro_X).transform(ds).samples.T
prepro_y = [
SampleSlicer(subject=[subj], band=['alpha'], task=['task1']),
FeatureSlicer(nodes_1=['MN'], nodes_2=['MN']),
SampleZNormalizer()
]
y = PreprocessingPipeline(nodes=prepro_y).transform(ds).samples.T
linear = LinearRegression()
linear.fit(X, y)
betas.append(linear.coef_.squeeze())
intercepts.append(linear.intercept_.squeeze())