www.neuron.yale.edu

Hi,

I am trying to optimize a model of an oscillating single-compartment neuron to get it to oscillate (and "rest") at a specific membrane potential. I've written python code to create a class for my neuron (using Neuron) and then by creating hundreds of thousands of instances of the model with parameters randomized around a set point, I have been to able to find a few oscillating neurons. However, all of these models rest at a membrane potential that is more positive than I would like. Does anyone have any suggestions? I know I need to use optimization but I'm a newbie when it comes to this sort of optimization. What would be the easiest way to do optimization? Are there pre-existing tools, especially ones that work with well with python, or do I need to create my own pipeline?

Thanks,
Dan

I don't know what it means to say that an oscillating model has a resting potential (unless the model's phase space has a feature that separates trajectories that collapse to a stable node from trajectories that don't).

There are many ways to shift the resting potential of a model that has a resting potential, without altering the model's dynamics. A trivial way is to shift all ionic equilibrium potentials and the voltage dependencies of all gating variables by the desired amount.

Okay, good point. I mean average membrane potential (say over one cycle of oscillation) by rest. The trouble is I am trying to constrain at least some of my channel kinetics to the experimentally measured values (at the soma in a pyramidal-type cell with terrible space clamp). The real cells that I'm trying to model have an average membrane potential about 10 mV more negative than the model. My idea was to simplify the cells by considering them to have only the conductances experimentally shown to be involved in the oscillation and by considering them to be a single compartment. I'm starting to wonder if my simplifications are the problem?

If the only aspect of the oscillation that you want to change is average membrane potential, do what I suggested in my previous post.