Glif Models Implementation in NEST Simulator

Build and install modules dynamically

bash $ mkdir build $ cd build $ cmake --Dwith-nest=nest-config -Dwith-ltdl=ON [-Dwith-mpi=ON] ../GlifModel $ make $ make install

Issues

• Use NEST Simulator 2.14.0. The v2.10.0 isn't working.
• Make sure ltdl-dev libraries are available (before compiling nest). On CentOS run sudo yum install libtool-ltdl-devel, on ubuntu libltdl-dev.
• When compiling nest make sure to use absolute paths. When completed run nest-config --libs to make sure a full path to the nest libraries are used.

Instantiate Modules in pynest

python import nest nest.Install('glifmodule') neuron = nest.Create('glif_lif') # or glif_lif_r, glif_lif_asc, glif_lif_r_asc

Issues

• If you get a 'File not found' message when trying to install the module:
  • Try using nest.Install('glifmodule.so') instead (On CentOS 6 lt_dlopenext() isn't working properly).
  • Check LDLIBRARYPATH, if needed set export LD_LIBRARY_PATH="/full/path/to/nest/module:$LD_LIBRARY_PATH

Running and Testing

Download Cell-Types-DB models to local machine

In scripts/ folder, run the following command to install 10 specific modules (AllenSDK is required) bash $ python allensdk_helper.py Or to get a specific set of models for a given cell-id bash $ python allensdk_helper.py CELL-ID1 [CELL-ID2 CELL-ID3 ...]

Test all downloaded models

bash $ python test_glif2nest.py 1> /dev/null

Run and qualitativly compare NEST and AllenSDK implementation

First determine the type in injection schemes are available bash $ python run_model.py --list-stimuli The following will run both NEST and AllenSDK implementation of a model and plot voltage-traces and spike-trains. Model download is not required. bash $ python run_model.py --cells cell-id[,cell_id,...] --model LIF[-R|-ASC|-R-ASC|-R-ASC-A] --stimulus ramp-1[,long-square-1,ramp-2,...]

Run NEST implementation of Glif models with current-based synaptic ports

First determine the type in injection schemes are available bash $ python run_model_psc.py --list-stimuli The following will run NEST implementation of a 4 neurons network as described below and plot voltage-traces and spike-trains. Model download is not required. * One neuron is without synaptic port, the other three are with 2 syaptic ports (one port is 2.0ms and one port is 1.0ms); * The first neuron is connected the first port of the second neuron; * The first neuron is connected the second port of the third neuron; * The first neuron is also connected both ports of the fourth neuron; * The weights between first neuron and other neurons are all 1000.0. bash $ python run_model_psc.py --cells cell-id[,cell_id,...] --model LIF[-R|-ASC|-R-ASC|-R-ASC-A] --stimulus ramp-1[,long-square-1,ramp-2,...]

Run NEST implementation of Glif models with conductance-based synaptic ports

First determine the type in injection schemes are available bash $ python run_model_cond.py --list-stimuli The following will run NEST implementation of a 4 neurons network as described below and plot voltage-traces and spike-trains. Model download is not required. * One neuron is without synaptic port, the other three are with 2 syaptic ports (one port is 2.0ms and one port is 1.0ms); * The first neuron is connected the first port of the second neuron; * The first neuron is connected the second port of the third neuron; * The first neuron is also connected both ports of the fourth neuron; * The weights between first neuron and other neurons are all 30.0. bash $ python run_model_cond.py --cells cell-id[,cell_id,...] --model LIF[-R|-ASC|-R-ASC|-R-ASC-A] --stimulus ramp-1[,long-square-1,ramp-2,...]

Notes

• Has only been tested with Python 2.7 and NEST 2.14.0

Update

• The most updated implementation of GLIF models have been integrated into NEST Simulator master repo as glif_psc (current-based model) and glif_cond (conductant-based model) and is now included in NEST Simulator 2.20.0.