Update optimization.py to run with neurotune instead of optimalneuron

[VahidGh]

Based on the changes in the neurotune, I guess optimalneuron should be replaced with neurotune in the optimization.py.

[pgleeson]

Thanks for checking through these @VahidGh. Yes, neurotune (and pyelectro) have replaced optimalneuron. Unfortunately these files haven't been updated since that change.

Also in that time, libNeuroML has become more stable, pyramidal hasn't been updated much & jNeuroML is the most well tested way for running NeuroML2 models. Now the preferred pipeline is:

libNeuroML (create/edit model) -> NeuroML2 XML files -> jNeuroML (load) -> simulate -> check & optimise

or

libNeuroML (create/edit model) -> NeuroML2 XML files -> jNeuroML (load & export to NEURON) -> simulate -> check & optimise

as opposed to:

libNeuroML -> Pyramidal -> NEURON -> simulate -> check & optimise

My suggestions for updating this code are:

• create a new LEMS file (e.g. based on this to run a simulation with just the current NeuroML2 muscle & an input current
• get main.py running with the NeuroML2 muscle model
  • load in ../NeuroML2/SingleCompMuscle.cell.nml & *.channel.nml using loaders.NeuroMLLoader.load in init() into one NeuroMLDocument
  • adjust the values in the nml files using libNeuroML
  • save this new NeuroML file (include cell & channels in one file) in a new file with unique name
  • create a copy of the LEMS file above which uses this new cell
  • run the new lems file with jNeuroML when run() is called in main.py and reload the saved trace for v in a local array
• update run.py to use the updated main.py and use neurotune/pyelectro instead

Try my forks of neurotune and pyelectro, as I've retested & updated these recently.

[VahidGh]

@pgleeson, Could you please explain what is the meaning of these gating parameters, and why do we choose these values?

[pgleeson]

I wouldn't take any meaning from those particular values, they look like they are based on the HH model and probably were used just for testing. It's a long time since that code worked...

[VahidGh]

@pgleeson, Thanks for your reply. I was going to look if I can do something with this code, regarding to the optimization case study issue. Is this still your suggested pipeline? or it could be implemented easier via PyNeuroML, now?

[VahidGh]

@pgleeson, another question related to the gating states of an ion channel: is there a way that I can represent a Markov model (gating dynamics) via NeuroML2/LEMS?

[pgleeson]

For your 2nd question, see here an example of a kinetic scheme based channel specification in NeuroML2: https://github.com/NeuroML/NeuroML2/blob/development/LEMSexamples/LEMS_NML2_Ex4_KS.xml. This runs with jNeuroML (but not jNeuroML->NEURON), but I don't believe this channel type has been tested/used beyond this example...

Re optimisation, I started this updated version of the optimiser code to work with c302, but it's not yet finished. Have a look if you like. Requires my forks of neurotune and pyelectro. Hope to get this more stable in the next week.

[VahidGh]

Actually I needed that type of gating kinetics for other purposes (representing channel kinetics with Markov models I saw in some studies). Although here was not a right place to ask the question, but thank you for your explanation :)

[pgleeson]

This issue is redundant, due to the work ongoing to use neurotune in c302, see https://github.com/openworm/CElegansNeuroML/blob/master/CElegans/pythonScripts/c302/tune/c302tuner.py, and ChannelWorm https://github.com/VahidGh/ChannelWorm/tree/cwFitter/cwFitter.

The code in this repo using neurotune/optimalneuron probably won't be revived (sorry @vellamike)...

[VahidGh]

@pgleeson, Now I'm much more familiar with all these stuff. I can also work on this if we need to.