Unlike some Neuron models, the intent of this model is primarily to facilitate batch simulations involving the analysis of thousands of individual data traces across multiple cells. Data traces from single synaptic simulations can be obtained as described below, but no user-friendly pushbutton interface is provided for doing so. Batch simulations create summary results for each synaptic activation but do not save individual data trace results.

In executing the model it may be useful to understand the role of the different HOC files provided. These files fall into the following groups.

1. HOC files that define the geometry of individual cells. For example, ama-c30573.CNG.hoc describes cell C30573 from the Amaral lab. File axon-common.hoc describes a common axon segment that is attached to each of the individual cells during the course of the simulation. Note that these geometries were generated from NeuroMorpho.org and hence may differ in detail from similar descriptions of the same cells found elsewhere.
   
   
2. HOC files that provide parameters specific to individual cells. These parameters account for spatial orientation of the apical direction of the cell and the division between layers stratum oriens (SO), stratum lucidium (SL), stratum radiatum (SR), and  stratum lacunosum-moleculare (LM). The output file name for simulations of synaptic responses throughout the cell is also supplied.
    
3. HOC file synresp.hoc contains both the common parameters of the simulation and the logic for actually running the simulation. File synresp.hoc is loaded by the HOC file containing parameters for an individual cell, ensuring that cell-specific parameters have already been set when synresp.hoc is loaded. Only one cell at a time is simulated per execution of the model.

Files from this model can be copied to a directory of the user's choice. The model can then be executed using a recent release of Neuron. Versions 6 and 7 were used during the preparation of the article, but there should be only small differences between results using different execution platforms and different Neuron releases. Of course, MOD files must be compiled before the model can be executed. This is done using the usual commands. For MS Windows use the mknrndll command and for Unix use nrnivmodl. File exp2nmdar.mod is used to simulate NMDA receptors using a dual-exponent model. Other MOD files are associated with the various active models considered in the article.

Files demo.hoc, file demo.ses, and files beginning with demo-fig2a are included to permit a simple verification of the installation of the model and to illustrate a method for obtaining the results of a single synaptic activation by invoking functions from the Neuron command line. This demonstration reproduces the synaptic activation shown in Figure 2a of the article. It can be launched either directly from demo.hoc or else by loading demo.hoc into Neuron using the load hoc menu item, provided that the current directory has previously been set to the directory containing the files of this model. demo.hoc automatically loads the other files as needed in turn. Results of executing the demo should, after some rearranging of windows, look like the following.

Simulation parameters are defined in the beginning of file synresp.hoc and must be set as appropriate for the type of simulation being performed. See synresp.hoc for a description of each of these parameters. For parameter settings used in generating files supporting figures in the article, see file params-by-fig.csv. Note that synresp.hoc is loaded by one of the cell-specific HOC files and will not function correctly without the necessary cell-specific parameter values being set before synresp.hoc is loaded.

Simulations are started by invoking Neuron and loading one of the cell-specific HOC files. This can be done in the usual way, but depends on the type of system being used. For example,  in MS Windows, Neuron is typically associated with files containing the suffix .hoc and it is sufficient to double-click on a cell-specific HOC file (for example synresp-cell1zr.hoc) to launch the simulation. Before loading the cell-specific HOC file using the Neuron menu item File->load hoc, set the current directory to be the directory containing synresp.hoc and other model files using the menu item File->working dir.

Control parameters of particular interest are isInteractive, which controls whether interactive components of Neuron are loaded, and runStim, which controls whether or not stimulations for synaptic activations are initiated automatically when synresp.hoc is loaded. For batch executions, normal settings would be isInteractive=0 and runStim=1. For running simulations under manual control, the settings would be isInteractive=1 and runStim=0, which are the values set initially in the copy of synresp.hoc supplied here.

When executing manually controlled simulations, the following functions defined in synresp.hoc can be invoked from the Neuron command line:

• getSynResp(x) simulates a synaptic activation at location x within the current section. Summary results are written to the Neuron console log. Normal Neuron plotting mechanisms can be used to display state variables of relevant portions of the cell, but the required plotting settings would need to be put into place prior to invoking getSynResp. Note that if the current section and location lie at the soma or in layer SL, there is no corresponding synapse.  In this case, the function returns without conducting a simulated activation.
• saveSynResp(x) simulates a synaptic activation at location x within the current section. Results are written to the file named in string variable savePath (default is savedsynresp.csv).

Output files are in comma-separated-values (.csv) format and contain the following columns:

Sample data traces used in the article are included here to permit testing of any new installations of the model. Correspondence between figures and the traces files is:

Note that the synaptic stimulation event occurs at t=2000 ms. Data values appearing prior to that time are associated with settling of the simulated cell from its initial rest state under the influence of either a current or voltage clamp at the soma as appropriate.  After loading the model into Neuron, data traces were written to external files using HOC commands of the form:

By convention, values for the receptor type, clamp mode, cell name, section number, and position were encoded into the output file name, though this is not enforced by the model software. The output file name should be placed in string variable savePath before using the above command. 

Data in the generated files can be plotted using a number of tools. Most statistical packages are capable of reading .csv files such as these. For example, when using the statistical package R, the following procedure can be used to generate a simple EPSC plot:

1. In an R workspace, set the current directory to one containing the trace files.
2. Execute the following commands in the workspace:
   
   SR<-read.csv("out-vc-ampar-c31162-ad67-022.csv")
   SR<-SR[SR$time>=1990 & SR$time<=2060,]
   plot(SR$time-2000,     # offset by settling time
     SR$soma-SR$soma[1],  # offset by resting current
     type='l',xlab="time",ylab="EPSC")
   rm(SR)
   

When runStim=1 is specified, the function sampleSynResps() is automatically invoked to sample synaptic responses throughout the current cell. A comma-separated-values (.csv) file is written containing the following columns:

The path name of the output file must be placed in the variable outPath. This is done in the cell-specific HOC files before synresp.hoc is loaded. Note that for batch execution, individual synaptic activation data traces are not saved.

Procedures for automating simulations involving multiple cells can be implemented in Unix systems through shell scripts and in MS Windows through batch command files. In this case, it may be useful to redirect sysout to a file so that the simulation results can be scanned for any error messages. Similarly, scripts can be developed to improve load balancing when simulating multiple cells on a multi-core system. These, however are left as an exercise for the reader because of differences among execution platforms.

For questions regarding this model, please contact the article's corresponding author, Dr. Giorgio Ascoli. His email address is (replace -at- with the usual @ symbol): ascoli-at-gmu.edu