Implement trial selection option

Consistent implementation of option for trial selection in all relevant functions (such as plotting functions, functions that handle raw data, etc.).

Objectives

• implement configuration option for doing trial selection in all relevant functions
• consistent implementation and documentation

Step 1: get an overview of all functions for which this is relevant

• functions that use raw data
• functions that (could) use data with an ‘rpt’ dimension
• functions where this has already been implemented

Functions that use raw data:

• preprocessing.m (when called with preprocessed data) -done-
• combineplanar.m -done-
• megplanar.m -done-
• megrealign.m -done-
• megrepair.m -done-
• scalpcurrentdensity.m -done-
• rejectvisual.m -done-
• redefinetrial.m -done-
• resampledata.m -done-
• timelockanalysis.m -done-
• freqanalysis.m -done-

Functions that (could) use data with an ‘rpt’ dimension

• freqdescriptives.m -done-
• singleplotER.m -done-
• singleplotTFR.m -done-
• topoplotER.m -done-
• topoplotTFR.m -done-
• multiplotER.m -done-
• multiplotTFR.m -done-

Functions where this has already been implemented

• componentanalysis.m -done-
• nonlinearassociation.m -done-

How it’s currently implemented in these function

% cfg.trials = ‘all’ or a selection like 1:10 (default = ‘all’)

% set the defaults
if ~isfield(cfg, 'trials'),   cfg.trials = 'all';  end

% select trials of interest
if ~strcmp(cfg.trials, 'all')
fprintf('selecting %d trials\n', length(cfg.trials));
data.trial  = data.trial(cfg.trials);
data.time   = data.time(cfg.trials);
end

Step 2: develop and implement solution

• ensure consistency in solution
• relevant fields (trl) should be adjusted accordingly
• documentation

Ensure consistency in solution

For raw data input functions (implement before ‘Ntrials=…’ or equivalent):

% set the defaults
if ~isfield(cfg, 'trials'),   cfg.trials = 'all';  end


% select trials of interest
if ~strcmp(cfg.trials, 'all')
  fprintf('selecting %d trials\n', length(cfg.trials));
  data.trial  = data.trial(cfg.trials);
  data.time   = data.time(cfg.trials);
end

For rpt data input functions:

% set the defaults
if ~isfield(cfg, 'trials'),   cfg.trials = 'all';  end

• for topoplotE

  elseif strcmp(data.dimord, ‘rpt_chan_time’) tmpcfg = []; tmpcfg.trials = cfg.trials; data = timelockanalysis(tmpcfg, data); if ~isfield(cfg, ‘xparam’), cfg.xparam=’time’; end if ~isfield(cfg, ‘yparam’), cfg.yparam=’’; end if ~isfield(cfg, ‘zparam’), cfg.zparam=’avg’; end

• for singleplotER, multiplotER (varargin

  elseif strcmp(varargin{1}.dimord, ‘rpt_chan_time’) tmpcfg = []; tmpcfg.trials = cfg.trials; for i=1:(nargin-1) varargin{i} = timelockanalysis(tmpcfg, varargin{i}); end if ~isfield(cfg, ‘xparam’), cfg.xparam=’time’; end if ~isfield(cfg, ‘zparam’), cfg.zparam=’avg’; end

• for topoplotER, singleplotTFR, multiplotTF

  elseif strcmp(data.dimord, ‘rpt_chan_freq_time’) tmpcfg = []; tmpcfg.trials = cfg.trials; tmpcfg.jackknife= ‘no’; data = freqdescriptives(tmpcfg, data); if ~isfield(cfg, ‘xparam’), cfg.xparam=’time’; end if ~isfield(cfg, ‘yparam’), cfg.yparam=’freq’; end if ~isfield(cfg, ‘zparam’), cfg.zparam=’powspctrm’; end

Relevant fields (trl) should be adjusted accordingly

(note: when cfg.trials=’all’ this doesn’t apply)

the code for adjusting the trl should look something like this:

finding the trl (see e.g., appenddata.m

% adjust the trial definition (trl) in case of trial selection
if ~strcmp(cfg.trials, 'all')
  % try to locate the trial definition (trl) in the nested configuration
  if isfield(data, 'cfg')
    trl = findcfg(data.cfg, 'trl');
  else
    trl = [];
  end
  if isempty(trl)
    % a trial definition is expected in each continuous data set
    warning('could not locate the trial definition ''trl'' in the data structure');
  else
    cfg.trl=trl(cfg.trials,:);
  end
end

adjusting the tr

% adjust the trial definition (trl)
if ~isempty(trl) && ~strcmp(cfg.trials, 'all')
cfg.trl=trl(cfg.trials,:);
end

Documentatio

%   cfg.trials       = 'all' or a selection given as a 1xN vector (default = 'all')

Appendix: useful Linux commands

Find functions that use raw data:

grep -n datatype.*raw *.m

Find functions that already have cfg.trials option:

grep -n cfg.trials *.m