update demo

Author: Remi-Gau

demos/transformers/.gitignore

@@ -0,0 +1,3 @@
+*.mat
+new_events.tsv
+transformers.json

demos/transformers/split_by_trials.m

@@ -1,15 +1,110 @@
-% Example of how to use transformers to:
+% In this example we use a matlab script to generate a series of transformations
+% and apply them to the content of a TSV file.
+%
+% This also shows how to edit the bids stats model directly,
+% so you can directly "paste" the series of transformations into the JSON
+% of your BIDS stats model.
+%
+% ---
+%
+% The transformations include:
 %
 % - "merge" certain trial type by renaming them using the Replace transformer
 % - "split" the trials of certain conditions
 %
-% For MVPA analyses, this can be used to have 1 beta per trial (and not 1 per run per condition).
+% For MVPA analyses, the latter can be used to have 1 beta per trial
+% (and not 1 per run per condition).
+%
+%
 %
 % (C) Copyright 2021 Remi Gau
 
+clear;
+
+% conditions_to_split = {'CONG_LEFT'
+%                        'CONG_RIGHT'
+%                        'INCONG_VL_PR'
+%                        'INCONG_VR_PL'
+%                        'P_LEFT'
+%                        'P_RIGHT'
+%                        'V_LEFT'
+%                        'V_RIGHT'};
+
+% same but using as regular expressions
+conditions_to_split = {'^.*LEFT$'
+                       '^.*RIGHT$'
+                       '^INCONG.*$'};
+
+% columns headers where to store the new conditions
+headers = {'LEFT'
+           'RIGHT'
+           'INCONG'};
+
+tsv_file = fullfile(pwd, 'data', 'sub-03_task-VisuoTact_run-02_events.tsv');
+
+data = bids.util.tsvread(tsv_file);
+
+%% merge responses
+
+transformers{1}.Name = 'Replace';
+transformers{1}.Input = 'trial_type';
+transformers{1}.Replace = struct('key', '^RESPONSE.*', 'value', 'RESPONSE');
+transformers{1}.Attribute = 'value';
+
+%% split by trial
+
+for i = 1:numel(conditions_to_split)
+
+  % create a new column where each event of a condition is labelled
+  % creates a "tmp" and "label" columns that are deleted after.
+  transformers{end + 1}.Name = 'Filter'; %#ok<*SAGROW>
+  transformers{end}.Input = 'trial_type';
+  transformers{end}.Query = ['trial_type==' conditions_to_split{i}];
+  transformers{end}.Output = 'tmp';
+
+  transformers{end + 1}.Name = 'LabelIdenticalRows';
+  transformers{end}.Cumulative = true;
+  transformers{end}.Input = {'tmp'};
+  transformers{end}.Output = {'label'};
+
+  transformers{end + 1}.Name = 'Concatenate';
+  transformers{end}.Input = {'tmp', 'label'};
+  transformers{end}.Output = headers(i);
+
+  % clean up
+  % insert actual NaN
+  transformers{end + 1}.Name = 'Replace';
+  transformers{end}.Input = headers(i);
+  transformers{end}.Replace = struct('key', '^NaN.*', 'value', 'n/a');
+  transformers{end}.Attribute = 'value';
+
+  % remove temporary columns
+  transformers{end + 1}.Name = 'Delete';
+  transformers{end}.Input = {'tmp', 'label'};
+
+end
+
+[new_content, json] = bids.transformers(transformers, data);
+
+% save the new TSV for inspection to make sure it looks like what we expect
+bids.util.tsvwrite(fullfile(pwd, 'new_events.tsv'), new_content);
+
+% generate the transformation section that can be added to the bids stats model
+bids.util.jsonencode(fullfile(pwd, 'transformers.json'), json);
+
+%% update the BIDS stats model
+
+model_file = fullfile(pwd, 'models', 'model-VisuoTact_smdl.json');
+
+bm = BidsModel('file', model_file);
+bm.Nodes{1}.Transformations.Instructions = transformers;
+bm.write(model_file);
+
+%% generate the mat file that SPM will use to specify the model sor the run 2 of subject 3
+
 tsvFile = fullfile(pwd, 'data', 'sub-03_task-VisuoTact_run-02_events.tsv');
 
-opt.model.file = fullfile(pwd, 'models', 'model-VisuoTact_smdl.json');
+opt.model.file = model_file;
 opt.verbosity = 2;
 opt.glm.useDummyRegressor = false;