B03_cleanData.m

% Load the data
%load('processedAnimalData.mat');

%%%
%WARNING: only run step 1 if just post trials are being analyzed
%%%

% Step 1: Remove entries where nansum(x.data.X_mm) == 0
indicesToDelete = arrayfun(@(x) nansum(x.data.X_mm) == 0, animal);
animal(indicesToDelete) = [];

% Step 2: Ensure each day/group/ROI combination has both 'pre' and 'post' phases
% Convert animal to a table for easier manipulation
animalTable = struct2table(animal);

% Get unique combinations of day, group, and roi
[unique_combinations, ~, ic] = unique(animalTable(:, {'day', 'group', 'roi'}), 'rows');

indicesToKeep = false(height(animalTable), 1);

for i = 1:height(unique_combinations)
    % Get the indices for the current combination
    combo_indices = find(ic == i);
    
    % Check for 'pre' and 'post' phases
    phases = animalTable.phase(combo_indices);
    if any(strcmp(phases, 'pre')) && any(strcmp(phases, 'post'))
        indicesToKeep(combo_indices) = true;
    end
end

% Keep only the valid entries
animal = table2struct(animalTable(indicesToKeep, :))';

clearvars -except animal

% Save the cleaned data
save('cleanedAnimalData.mat', 'animal');

% % quickly look at perf of all rois - for priors
% 
% for ij=1:numel(animal)
%     prior(ij)=nanmedian(animal(ij).data.performance_index_binned)
% end