Tomsett et al. (2014)


The parameter files for the model described by Tomsett et al. (2014) are included in the main VERTEX download (in the bsfModel folder), and are also available on figshare and ModelDB. The parameters are set to run the full model as described in the paper, using VERTEX in parallel mode with 12 labs, to run the simulation for 2400 ms. To run the model, first make sure you have added the VERTEX simulator to your Matlab path as described in Tutorial 0. You should also compile the MEX file as described in that tutorial to allow more than one connection from a presynaptic to a postsynaptic neuron. Then run the script bsf_model_run.m – this will load the parameters from the other .m files, initialise the network, and run the simulation. You can modify parameters either in the other parameter files, or in the bsf_model_run.m file (for example, if you want to change the number of labs or the location to save results). Results can then be loaded and analysed as described in the tutorials.

NB: the results will be qualitatively, but not quantitatively, the same as in the paper, because the random number generator control code was not implemented at that time.

The model contains approximately 175,000 neurons, each with several compartments, so the memory requirements are quite large. Including the overheads for running on 12 Matlab workers, the simulation uses about 30 GB RAM on our Intel Xeon based Linux server (Matlab 2012b).

We have also uploaded the raw experimental data used for comparison in figure 9 of the paper, along with a script to generate a similar gamma power map to that shown in figure 9a. The data can be downloaded from figshare.

If you have any trouble running the simulation or want further details on either the simulation or the experimental data, please contact us using the contact form.

EDIT 1: The version of VERTEX used to generate the results in the paper contained a bug that caused the synaptic release time to be added to the total axonal delay time twice, resulting in a synaptic release time of 1 ms, rather than 0.5 ms as reported in the paper. This bug was fixed prior to the initial VERTEX release. The parameter files for the neocortical slice model available from ModelDB, figshare, and in the VERTEX simulator download therefore set a synaptic release time of 1 ms so that the model behaviour reported in the paper can be reproduced.

EDIT 2: VERTEX version 0.91 corrected a bug in the original VERTEX release that caused LFPs to be calculated with the wrong sign. This bug did not affect the results reported by Tomsett et al. (2014) as this error was introduced during edits just prior to the initial VERTEX release (compare simulated and experimental LFPs in figure 9 – simulated results have the correct sign).

 

Reference

Tomsett RJ, Ainsworth M, Thiele A, Sanayei M, Chen X, Gieselmann MA, Whittington MA, Cunningham MO, Kaiser M. (2014) Virtual Electrode Recording Tool for EXtracellular potentials (VERTEX): comparing multi-electrode recordings from simulated and biological mammalian cortical tissue, Brain Structure and Function. doi:10.1007/s00429-014-0793-x



Leave a Reply

Your email address will not be published. Required fields are marked *