Common Intermediate Format (CIF)¶
Yimeng Zhang
Feb. 29, 2016
To use cdttable, the user must present their spiking neural data in a particular format called
Common Intermediate Format (CIF). To see some examples of CIF, simply run tests.generate_one_preprocess_result(), and the first argument returned by the function is a valid CIF.
Formal Specification¶
The CIF for some spiking neural data for N trials and M spikes
must be a MATLAB struct with the following fields.
event_codes- a N x 1 cell array, i-th element being a column vector storing the event codes for i-th trial. The order of
event codes must reflect the actual temporal order they appear, which is natural for real data.
The order of N elements should follow the actual temporal order trials appear. In any case, the i-th element in the
cell array will be considered as the i-th trial by the program, and passed in
iwhen computing condition number. event_times- a N x 1 cell array, i-th element being a column vector of same length as i-th element in
event_codes, storing the timestamps of event codes. The timestamps with in any trial must be non-decreasing and match the order of corresponding codes, which is natural for real data. Similarly, the order of N elements should follow the actual temporal order trials appear. spike_times- a M x 1 column vector, specifying the timestamps of all spikes in the data. The order of spikes can be shuffled without affecting the correctness of the program.
spike_locations- a M x 2 matrix, i-th row corresponding to the i-th spike in
spike_times. For each row, first element denotes the electrode number of the spike, and second element the unit number. For our lab’s settting (NEV files), electrode number has a range of 1-96, and unit number has a range of 0-255. The combination of electrode number and unit number is the spike’s location, which is the neuron that emits the spike in this context. Similarly, the rows can be shuffled, but should match that ofspike_times.
Notice that all timestamps (for codes and spikes) must be w.r.t. to some common origin, such as start of recording. Such timestamps are the most natural ones when extracted from the raw recording file, but may not be the case if they are extracted from the experiment control program, which may use the start of each trial as time origin.