run dataset basics

run_up:

run_dataset_basics

%% Dataset basics
%
% #   For CoSMoMVPA's copyright information and license terms,   #
% #   see the COPYING file distributed with CoSMoMVPA.           #

% Set data path, load dataset, set targets and chunks, and add labels as
% sample attributes

% Set the data path
config = cosmo_config();
data_path = fullfile(config.tutorial_data_path, 'ak6', 's01');

% Set the filename to the glm_T_stats_perrun NIFTI file; assign
% the filename to a variable named 'fn'
fn = fullfile(data_path, 'glm_T_stats_perrun.nii');

% Load data using cosmo_fmri_dataset, and assign the result to a variable
% named 'ds'
ds = cosmo_fmri_dataset(fn);

% set ds.sa.targets (trial conditions) to the 60x1 column vector:
% [ 1 2 3 4 5 6 1 2 3 ... 5 6 ]'
targets = repmat((1:6)', 10, 1);
ds.sa.targets = targets;

% sanity check
cosmo_check_dataset(ds);

% set ds.sa.chunks (acquisition run number) to the 60x1 column vector:
% [ 1 1 1 1 1 1 2 2 2 ... 10 10 ]'
chunks = zeros(60, 1);
for i = 1:10
    idxs = (i - 1) * 6 + (1:6);
    chunks(idxs) = i;
end
ds.sa.chunks = chunks;

% ( alternative:   ds.sa.chunks = ceil((1:60)'/6);     )

% sanity check
cosmo_check_dataset(ds);

% Add ds.sa.labels as sample attributes as a 60x1 cell with strings
% The order is {monkey, lemur, mallard, warbler, ladybug, and lunamoth},
% repeated 10 times
labels = repmat({   'monkey'
                 'lemur'
                 'mallard'
                 'warbler'
                 'ladybug'
                 'lunamoth'}, 10, 1);
ds.sa.labels = labels;

% sanity check
cosmo_check_dataset(ds);

%% Overview of the dataset
fprintf('\nOverview of dataset:\n');
cosmo_disp(ds);

%% Apply mask to dataset
fn_mask = fullfile(data_path, 'vt_mask.nii');

% Load mask using cosmo_fmri_dataset and store in ds_mask
ds_mask = cosmo_fmri_dataset(fn_mask);

% sanity check: ensure the same feature attributes
assert(isequal(ds_mask.fa, ds.fa));

% Now slice the dataset using cosmo_slice, where the third argument must be
% 2 to indicate the slicing of features (not samples).
% Assign the result to a variable 'ds_masked'.
mask_indices = find(ds_mask.samples);
ds_masked = cosmo_slice(ds, mask_indices, 2);
cosmo_disp(ds_masked);

%% Now use cosmo_fmri_dataset with all input parameters

% We do exactly the same as above, but shorter:
% Load the dataset again using cosmo_fmri_dataset, but now use the
% 'mask','targets','chunks' parameters.
% Assign the result to ds_masked_alt
ds_masked_alt = cosmo_fmri_dataset(fn, 'mask', fn_mask, ...
                                   'targets', targets, ...
                                   'chunks', chunks);

% set labels again
ds_masked_alt.sa.labels = labels;

% check ds_masked and ds_masked_alt are the same
assert(isequal(ds_masked_alt, ds_masked));