opensourcebrain/multiscaleisn

Inhibition Stabilized Networks at multiple scales based on Sadeh et al. 2017

Multiscale ISN

Inhibition Stabilized Networks at multiple scales based on Sadeh et al. 2017.

To generate network/run model

The main script to generate the model is ISN.py and changing the parameters to the main generate() function here will create different configurations of the network:

def generate(scale_populations = 1,
             percentage_exc_detailed=0,
             exc2_cell = 'SmithEtAl2013/L23_NoHotSpot',
             percentage_inh_detailed=0,
             scalex=1,
             scaley=1,
             scalez=1,
             exc_exc_conn_prob = 0.25,
             exc_inh_conn_prob = 0.25,
             inh_exc_conn_prob = 0.75,
             inh_inh_conn_prob = 0.75,
             ee2_conn_prob = 0,
             ie2_conn_prob = 0,
             Bee = .1,
             Bei = .1,
             Bie = -.2,
             Bii = -.2,
             Bee2 = 1,
             Bie2 = -2,
             Be_bkg = .1,
             Be_stim = .1,
             r_bkg = 0,
             r_bkg_ExtExc=0,
             r_bkg_ExtInh=0,
             r_bkg_ExtExc2=0,
             r_stim = 0,
             fraction_inh_pert=0.75,
             fraction_inh_offset=0,
             inh_offset_amp=0,  # hyperpolarising/depolarising current to inh fraction_inh_offset of cells 
             Ttrans = 500, # transitent time to discard the data (ms)
             Tblank= 1500, # simulation time before perturbation (ms)
             Tstim = 1500, # simulation time of perturbation (ms)
             Tpost = 500, # simulation time after perturbation (ms)
             connections=True,
             connections2=False,
             exc_target_dendrites=False,
             inh_target_dendrites=False,
             duration = 1000,
             dt = 0.025,
             global_delay = .1,
             max_in_pop_to_plot_and_save = 10,
             format='xml',
             suffix='',
             run_in_simulator = None,
             num_processors = 1,
             target_dir='./temp/',
             v_clamp=False,
             simulation_seed=11111):       

Generally the defaults work well to generate a spiking network showing ISN properties.

To generate the 2 main configurations of the network (point neurons only, point neurons + 10 detailed neurons) and save as NeuroML, run:

./regenerate_neuroml.sh

To run the 40 network simulations in NetPyNE for the point neuron network, run:

./runall.sh

To run the 40 network simulations in NetPyNE for the point neuron network with 10 detailed cells, run:

./runall_detailed.sh

Reusing this model

The code in this repository is provided under the terms of the software license included with it. If you use this model in your research, we respectfully ask you to cite the references outlined in the CITATION file.

multiscaleisn's Issues

Loading errors when `optimized=True` in `read_neuroml_file`

f = read_neuroml2_file("./ISN_net.net.nml.h5", optimized=True)
net = f.networks[0]
for p in net.populations:
    for i in p.instances:
        print(i)

gives:

....
    <transientPoissonFiringSynapse id="tpfsPertInh_after" averageRate="3600 Hz" delay="1500.0ms" duration="500.0ms" synapse="ampaBkg" spikeTarget="./ampaBkg"/>
</neuroml>
Traceback (most recent call last):
  File "/home/asinha/Documents/02_Code/00_mine/models/MultiscaleISN/NeuroML2/test.py", line 19, in <module>
    for i in p.instances:
  File "/home/asinha/.local/share/virtualenvs/neuroml-311-dev/lib/python3.11/site-packages/neuroml/hdf5/NetworkContainer.py", line 57, in __next__
    instance = self.__getitem__(self.cursor)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/asinha/.local/share/virtualenvs/neuroml-311-dev/lib/python3.11/site-packages/neuroml/hdf5/NetworkContainer.py", line 96, in __getitem__
    z=self.array[index][self._get_index_or_add("z", 3)],
      ~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
IndexError: index 3 is out of bounds for axis 0 with size 3

Something to do with the old vs new HDF5 export API? We should document this change somewhere so users know when to use optimized and when not.