bluebrain / bluepyopt Goto Github PK

At the moment the parameters BluePyOpt ephys optimises are range variables, section variables and global Neuron variables. We need a way to optimise point processes, and also to include these processes at the relevant locations

Needs to allow for:

• hoc code should be read from string (See #143). hoc_path is not necessary anymore. Should be first argument after name.
• we need a morph_dir argument to specify morphology directory. when this is specified morph filename in hoc can be used.
• morph_filename, overwrite morph_filename specified in hoc
• hoc code should not have full path to morpho anymore (only morph_filename)
• have an argument extra_params which is a list of (e.g. Global) parameters that have to be set outside of the hoc template, like celsius / v_init etc.

We could slowly change by calling the class with the above constructor 'HocCellModel' (that should be the name anyway). Then we can phase out slowly the HocModel.

At the moment the scores of the parameters are returned to the master process. We should also have a way to e.g. get the raw efeature values out. This would be useful for sensitivity analysis e.g.

Reported by @elisabettai.
Some features (e.g. inv_time_to_first_spike when first spike is very close to start of stimulus) can create very very high scores. This confuses the algorithm.
At the moment scores that don't generate an error are not bounded (when an error occurs the score is set to 250 though).
Option should be added to EFelFeature to allow the user to force a bound on the scores, and to allow to set the value of the bound.

@gsarma and I are working on optimization of ion channel models here for the Open Worm project, and are interested in using BluePyOpt. We reviewed your example with the HH neuron, but were wondering how we might handle an objective function constructed around channel recordings (e.g. peak and steady-state current during measurement of an IV curve). One option would be to create these objectives by hand, but are there any objective functions for this sort of purpose that are already being used with BluePyOpt, possibly for ion channel model optimizations as part of HBP? More generally, do you have any example scripts/notebooks doing ion channel model optimization?

Add a new subclass of Parameter, that would manage the value of several 'sub'Parameter objects at the same time.

Same global variable in the Neuron block in two or more mod files should have same values at the end of optimization.

At the moment variable alpha in the Neuron block of one mod file is defined as

NEURON {
SUFFIX kmdb
GLOBAL inf, tau,alpha
}

while in one other mod file alpha is defined as

NEURON {
SUFFIX kdr
GLOBAL ninf,taun,zetan,vhalfn,vhalfn2,zetan2,alpha
}

When we refer to the variable in the parameters.json for the optimization we refer to it as alpha_kmdb in the first case and alpha_kdr in the second case.

The two mod files are present in different sections in the morphology. Let's say one in the axonal and one in the somatic.

We would like to have the same value for all the alpha variables in the different mod files.

If we write in the json something like this

"optimized": {
"all": [
],
"axonal": [
["alpha_kmdb", 0.025, 0.15, "uniform" ]
],
"somatic": [
["alpha_kdr", 0.025, 0.15, "uniform" ]
]

the two values will have different values at the end of the optimization.

Hi I am comparing the results of the efel vs fitcalc.calculate_scores and it seems that there is a difference.

scores = fitcalc.calculate_scores(best_responses)


import efel
import pandas as pd
trace1 = {'T': best_responses['IDRest_7.soma.v'].response['time'], 'V':best_responses['IDRest_7.soma.v'].response['voltage']}
trace1['stim_start'] = [700]
trace1['stim_end'] = [2700]
traces_results = efel.getFeatureValues([trace1],
                                ['AP_height','AP_amplitude','Spikecount','AHP_depth_abs'])


traces_results


print('IDRest_7_AHP_depth_abs')
print('manual calculation - ' + str((-69.4687143541 - traces_results[0]['AHP_depth_abs'].mean())/0.539758874442))
print('bluepyopt calculation - ' + str(scores['IDRest_7.soma.AHP_depth_abs']))

IIDRest_7_AHP_depth_abs
manual calculation : 29.24285168089122
bluepyopt calculation : 36.269754996091436

See full example in the attached zip -Show_Bug.zip - file, open Show_error.ipynb for full details,
Thanks.

Might be related to BlueBrain/eFEL#166

Hi,

Would it be possible to please include tests and the license files in the release tars? At the moment, they are not included in the pypi release downloads.

Alternatively, if these are deemed unnecessary for the pypi tar, could the tags/releases on github please match the pypi releases so that downstream maintainers can use the sources from here?

It is currently in review for inclusion in NeuroFedora, where these issues cropped up: https://bugzilla.redhat.com/show_bug.cgi?id=1727505

After a protocol that sets cvode_minstep is run, this variable is not reset to it's original state. This can create problems if other protocols don't use cvode_minstep

Hello

Is Bluepyopt suitable for optimization of synaptic weights?
Are there any such examples?

Thank you

I have created a fork of bluepyopt, that is designed to work with dask distributed instead of scoop. Also it uses model parameter dictionaries that get mapped from GA genes. It also uses error functions objective functions that are created by interfacing with neuronunit.

Since these changes, are not necessarily helpful for general use cases of BluePyOpt, I have discussed how to best integrate these changes with a colleague, and we decided to propose making a sciunit branch of BluePyOpt.

The main places where code is substantially different are:
https://github.com/russelljjarvis/BluePyOpt/blob/scidash/bluepyopt/deapext/optimisations.py#L101-#L144

and:

https://github.com/russelljjarvis/BluePyOpt/blob/scidash/bluepyopt/deapext/optimisations.py#L218-#L231

@rgerkin

As mentioned by Johannes Hjorth, the replace axon code at:
https://github.com/BlueBrain/BluePyOpt/blob/master/bluepyopt/ephys/morphologies.py#L194
contains 'access axon...' statements. However these sections are never popped from stack using:
https://www.neuron.yale.edu/neuron/static/docs/help/neuron/neuron/secspec.html#pop_section

This has the potential of filling up the section stack in Neuron

Allow users to specify in the evaluator which protocol execution exceptions don't crash bluepyopt but return a bad fitness value instead

We should add a new mechanism to improve version tracking.
We should add also a changelog file.
It should something like this:
if travis passes in master, and it's going to push to pypi, this should happen:

• something like this should be used to create a changelog: https://github.com/vaab/gitchangelog
• the package.json should get the latest version
• these 2 files should be commit automatically into the main repo
• the current version is tagged in the repo
• the tags are pushed to the main repo
• (maybe a github 'release' is created)

There is one chicken-and-egg problem here. The commit will change the version again. An option here would be to 'predict' that version, since it will be the current_version + 1 commit.
Another issue to look at, is how versioneer reacts if all the patch versions are tagged in the repo also.

There is a before_deploy travis block we might be able to use for this.

Hello,

I'm having some trouble using the run function in BluePyOpt. I download multi-compartment models from Allen Brain Institute and use BluePyOpt to re-optimize the parameters. When running the optimization it appears to run fine and "optimized" parameter sets are generated by the end, however when I use the SequenceProtocol.run() function to plot the optimized models at the end, it generates errors along these lines (same error was generated across two tested models):

CVode-- At t = 0 and h = 2.11201e-11, the corrector
convergence failed repeatedly or with |h| = hmin.

CVode 0x7fb2c9cadb10 PutativePV[0].soma[0] advance_tn failed, err=-7.
err=-7
NEURON: variable step integrator error
 near line 0
 {dend[48] { pt3dadd(834.5686, 543.95477, 6.7634001, 0.43239999) }}
                                                                   ^
        fadvance()
      advance()
    step()
  continuerun(4000)
and others

This error is generated when I run it on my computer (i.e. with isolate = False) as well as on Neuroscience Gateway (i.e. with isolate = True). As suggested by the error message, CVode is enabled. When CVode is not enabled (i.e. on my computer), it appears to be running when viewing NEURON's RunControl, but the voltage never changes, despite square pulses being present in the protocol - also the voltage never actually plots in the NEURON Graph. As well the simulation actually keeps starting over to no end. In line with this, when CVode is not enabled when running the code on Neuroscience Gateway, the job reaches the time limit without completing.

From reading different entries on the NEURON forum, I get the impression that the error might might have to do with one of the MOD files, but removing each of the MOD mechanisms does not alter the error message. Also, if this were the case, I would imaging getting an error during the optimization itself, however the optimization appears to run to completion. Additionally, when running the models purely in NEURON/Python (i.e. outside of BluePyOpt), with CVode enabled and all the same MOD file mechanisms present, no error is generated, which tells me that it's not the MOD files, but in fact possibly something in the SequenceProtocol.run() function.

In case it helps, the two models that I tested my code with are the following:
http://celltypes.brain-map.org/experiment/electrophysiology/529807751
http://celltypes.brain-map.org/experiment/electrophysiology/528687520

Side note: I also tried using CellEvaluator.run_protocols on my computer, but nothing appears to happen when I do this (i.e. in the terminal it appears as though it's simulating, but looking at RunControl, nothing is changing - possibly because there is no option to input Isolate = False).

Any insight into this would be most helpful. Thanks,

Alex GM

At description of how the evolutionary algorithm works to the documentation

I would like to run in parallel the BluePyOpt using ipyparallel.
To instantiate an bluepyopt.optimisations.DEAPOptimisation object in ipyparallel,
do I need to set a new map_function that differ from the built-in map?

In case, what kind of parallel map object do you suggest to use?

Make the EFeatureObjective class abstract, and mention it in the documentation

When running neuron.h.topology() during the tests, one can see that some test is leaving behind a 'soma' section in the simulator (don't know yet which test).

After restoring an optimization from a checkpoint, the hall of fame object returned by DEAPoptimisation.run is empty.
The bug could be solved by returning the checkpoint loaded hall of fame in algorithms.eaAlphaMuPlusLambdaCheckpoint.

Hey,
I'm very happy you've build this toolbox but since I have set up a python 2.7 kernel on my Jupyter Notebook I can't install BluePyOpt anymore. I've explained my problem in more detail here. I hope you guys can help!

Hi!
Considering that ephys.create_hoc.create_hoc allows to specify a custom template via template_filename and template_dir, it would be really great to allow additional custom arguments for rendering. For example my custom template would be the same as cell_template.jinja2 but with custom geom_nseg that requires my_custom_number param:

// the same as `cell_template.jinja2`
...
proc geom_nseg() {
  this.geom_nseg_fixed({{my_custom_number}})
}
...
// the same as `cell_template.jinja2`

Then one can use the custom template easily with:

ephys.create_hoc.create_hoc(..., my_custom_number=13)

Also can we replace template_dir and template_filename with template_filepath? It would reduce the number of args but won't change functionality.

Reported by @pineider

When installing bluepyopt from source using
pip install -e .
one gets (after importing bluepyopt):
ValueError: unknown style 'pep440-minor'

This is due to a non-default version of versioneer.py in the root of the source dir, to control the version numbering format.
Need to figure out how to fix this.

Resuming from a checkpoint fails in Python 3 because of a missing "b" flag in pickle load.

Hi,

I get the following error after I run nrn = ephys.simulators.NrNSimulator() in simplecell.ipynb:

/Users/jyang/opt/anaconda3/lib/python3.7/site-packages/bluepyopt/ephys/simulators.py:75: UserWarning: Unable to find Neuron hoc shared library in /Applications/NEURON-7.7/nrn/lib/python/neuron, not disabling banner
'not disabling banner' % nrnpy_path)

Any ideas why I'm getting this error? Thanks in advance.

Hello,
I would like to optimize parameters of a single compartment from some experimental data. Instead of describing morphology in swf or asc files, I'd rather do it in a hoc file. How I can make it from one of the functions or classes provided by the bluepyopt.ephys Module?

thank

Sometimes certian individuals in the population take a long time to evaluate.
It's necessary to provide a mechanism to kill this individuals prematurely, and possible return a bad error score.
At least two mechanisms for this should be implemented:

• based on absolute walltime (disadvantage: walltime can e.g. depend on the compute resources)
• based on slowest in a population, e.g. kill all individuals that take longer than 80% of the population. (disadvantage: requires sending a signal from the object that knows about the population all the way down to the evaluator/simulator.)

As mentioned by @Aman-A on gitter, there is a possibility of the checkpoint disappearing if (probably) something goes wrong during the write of the checkpoint. This loses all history of the optimization. We should make this a bit safer.

One option see is doing the following:

• first write checkpoint to temporary file (just same name with extra .tmp extension?)
• issue a copy command from tmp -> main file
• leave the .tmp file there until the next generation

This way there are always two copies around, and the copy command is mostly more atomic.

I have tried to get started and to explore this issue by reproducing the examples in the example folder. I compiled the appropriate nmodl mechanisms with

/work/scidash/BluePyOpt/examples/l5pc/mechanisms$ nrnivmodl

I have both scoop and ipyparallel installed. However I just launch opt_l5pc.py without any flags first because it exposes the most fundamental error messages.

When I run the command:

jovyan@650b8092666a:~/work/scidash/BluePyOpt/examples/l5pc$ python opt_l5pc.py 

I get the output below. Is there particular branch of bluepyopt that contains a simulator module? If its an external module I wonder where I can get it?

Traceback (most recent call last):
  File "opt_l5pc.py", line 44, in <module>
    import l5pc_evaluator
  File "/home/jovyan/work/scidash/BluePyOpt/examples/l5pc/l5pc_evaluator.py", line 26, in <module>
    import l5pc_model  # NOQA
  File "/home/jovyan/work/scidash/BluePyOpt/examples/l5pc/l5pc_model.py", line 26, in <module>
    import bluepyopt.ephys as ephys
  File "/opt/conda/lib/python3.5/site-packages/bluepyopt-1.2.87-py3.5.egg/bluepyopt/ephys/__init__.py", line 24, in <module>
    import simulators  # NOQA
ImportError: No module named 'simulators'

Thanks.
Russell.

Following up on the discussion with Werner, I am creating this ticket for changes.

1. Copyright: Please update the information on copyright. (c) 2016-2020, Blue Brain Project/ EPFL. The last year (2020) should be updated each new year. Please make it consistent where it appears.

2. Authors: Currently Giuseppe, JDC, Mike, Christian, and Werner are listed as authors. Please update it to reflect the current status. (<- Tanguy!)

Thank you.

Hi,
I would like to start/restart optimization with a population rather than just specifying a random seed. This would really help if after n generations I want to change the fitness function and reinitialize the optimization with the population at the n_th generation.

Thanks

This is because of ipyparallel that have a dependency on ipython but I don't know how to solve it (other than installing ipython 5 manually)

Here is the end of the pip install bluepyopt output log:

Collecting ipython>=4 (from ipyparallel->bluepyopt)
  Downloading https://bbpteam.epfl.ch/repository/devpi/root/pypi/+f/eca/537aa61592aca/ipython-6.4.0.tar.gz (5.1MB)
    100% |████████████████████████████████| 5.1MB 69.6MB/s 
    Complete output from command python setup.py egg_info:
    
    IPython 6.0+ does not support Python 2.6, 2.7, 3.0, 3.1, or 3.2.
    When using Python 2.7, please install IPython 5.x LTS Long Term Support version.
    Beginning with IPython 6.0, Python 3.3 and above is required.
    
    See IPython `README.rst` file for more information:
    
        https://github.com/ipython/ipython/blob/master/README.rst
    
    Python sys.version_info(major=2, minor=7, micro=12, releaselevel='final', serial=0) detected.

The class DEAPOptimisation has parameters, that affect the optimization, most importantly mutpb, cxpb and eta. How do you usually chose these parameters? I see, that they default to cxpb = 1, mutpb = 1, eta = 10. In the example of the L5PC, the default values are (implicitly) used in the toy example of 2 generations. in gaupnerbrunelstdp, eta=20, mutpb=0.3, cxpb=0.7 is used. Why? Is there a rule of thumb, which parameters give good results?

If I use 1000 individuals and 1000 generations, I see, that a feasible solution is only found in ~30% of the optimizations and I am therefore wondering, if I just choose the parameters wrong. Or is this the expected success rate?

I get some warnings like:

Warning: no DISPLAY environment variable.
--No graphics will be displayed.

with python3.7, which I can reproduce by from neuron import gui, so it is related to the banner/gui of neuron. It should be disabled by https://github.com/BlueBrain/BluePyOpt/blob/master/bluepyopt/ephys/simulators.py#L65, but it seems to not be the case.

• skip intermediate files, if possible

Hi there,

It would be great if unit checking could be added to Ephys parameters when they are instantiated. This would be pretty easy to implement. Units of Hoc variables can be queried using h.units() (if they are defined), and Python has some great unit checking/conversion packages. I have hacked together some code that may be a good start:

import re
import pint # popular units package
from neuron import h
ureg = pint.UnitRegistry()

nrn_units = h.units(hoc_varname)
units_nodash = nrn_units.replace('-', '*') # some NEURON units are defined with dashes
units_exponents = re.sub(r'([a-zA-Z]+)(\d)',r'\1^\2', units_nodash) # adds '^' char for exponents
target_units = ureg(units_exponents)

param_quantity = ureg.Quantity(param_value, param_units_str)
try:
    converted_quantity = param_quantity.to(target_units)
except pint.errors.DimensionalityError:
    raise ValueError('incompatible units')

setattr(nrn_obj, hoc_varname, converted_quantity.magnitude)

Would something like this be possible?

Cheers,
Lucas.

For example here:
https://github.com/BlueBrain/BluePyOpt/blob/master/examples/thalamocortical-cell/CellEvalSetup/template.py#L85

iteritems is not longer supported in python3

There is at the moment a hardcoded internal BBP related file in the repo: 'package.json'.
This file should be automatically generated.
Related to:
#225

hello

When I execute on jupyter the following lines for the simplecell.ipynb example:

default_params = {'gnabar_hh': 0.1, 'gkbar_hh': 0.03}
responses = twostep_protocol.run(cell_model=simple_cell, param_values=default_params, sim=nrn)

the following error appears. I use Python 3.6 the Anaconda release on Windows.
Thanks for any help/suggestions.

RemoteTraceback Traceback (most recent call last)
RemoteTraceback:
"""
Traceback (most recent call last):
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\protocols.py", line 155, in _run_func
cell_model.instantiate(sim=sim)
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\models.py", line 222, in instantiate
self.morphology.instantiate(sim=sim, icell=self.icell)
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\morphologies.py", line 91, in instantiate
self.morphology_path)
OSError: Morphology not found at 'simple.swc'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "C:\Users\Alexandra\Anaconda3\lib\multiprocessing\pool.py", line 119, in worker
result = (True, func(*args, **kwds))
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\protocols.py", line 185, in _run_func
traceback.format_exception(*sys.exc_info())))
Exception: Traceback (most recent call last):
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\protocols.py", line 155, in _run_func
cell_model.instantiate(sim=sim)
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\models.py", line 222, in instantiate
self.morphology.instantiate(sim=sim, icell=self.icell)
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\morphologies.py", line 91, in instantiate
self.morphology_path)
OSError: Morphology not found at 'simple.swc'

"""

The above exception was the direct cause of the following exception:

Exception Traceback (most recent call last)
in ()
1 default_params = {'gnabar_hh': 0.1, 'gkbar_hh': 0.03}
----> 2 responses = twostep_protocol.run(cell_model=simple_cell, param_values=default_params, sim=nrn)

~\Anaconda3\lib\site-packages\bluepyopt\ephys\protocols.py in run(self, cell_model, param_values, sim, isolate)
71 param_values=param_values,
72 sim=sim,
---> 73 isolate=isolate)
74
75 key_intersect = set(

~\Anaconda3\lib\site-packages\bluepyopt\ephys\protocols.py in run(self, cell_model, param_values, sim, isolate)
208 'cell_model': cell_model,
209 'param_values': param_values,
--> 210 'sim': sim})
211
212 pool.terminate()

~\Anaconda3\lib\multiprocessing\pool.py in apply(self, func, args, kwds)
257 '''
258 assert self._state == RUN
--> 259 return self.apply_async(func, args, kwds).get()
260
261 def map(self, func, iterable, chunksize=None):

~\Anaconda3\lib\multiprocessing\pool.py in get(self, timeout)
642 return self._value
643 else:
--> 644 raise self._value
645
646 def _set(self, i, obj):

Exception: Traceback (most recent call last):
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\protocols.py", line 155, in _run_func
cell_model.instantiate(sim=sim)
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\models.py", line 222, in instantiate
self.morphology.instantiate(sim=sim, icell=self.icell)
File "C:\Users\Alexandra\Anaconda3\lib\site-packages\bluepyopt\ephys\morphologies.py", line 91, in instantiate
self.morphology_path)
OSError: Morphology not found at 'simple.swc'

Hey,
I've managed to run BluePyOpt and Neuron on a Linux Cluster and I'm very excited to work with it! Using the example you provide here, (I basically just copy pasted everything) I get the following error:

responses = twostep_protocol.run(cell_model=simple_cell, param_values=default_params, sim=nrn) 

Exception: Traceback (most recent call last):
  File "/cm/shared/uniol/software/Python/2.7.12-intel-2016b/lib/python2.7/site-packages/bluepyopt/ephys/protocols.py", line 130, in _run_func
    cell_model.instantiate(sim=sim)
  File "/cm/shared/uniol/software/Python/2.7.12-intel-2016b/lib/python2.7/site-packages/bluepyopt/ephys/models.py", line 222, in instantiate
    self.morphology.instantiate(sim=sim, icell=self.icell)
  File "/cm/shared/uniol/software/Python/2.7.12-intel-2016b/lib/python2.7/site-packages/bluepyopt/ephys/morphologies.py", line 91, in instantiate
    self.morphology_path)
IOError: Morphology not found at 'simple.swc'

Do you know how I can check where 'simple.swc' is stored and how I can tell python to grap it?

I am not completely sure but I believe this notebook simulates data to optimise a model against rather than optimizing with external experimental data. If I am right about that, it could be nice to have a markdown cell that explains that is what is happening, in order to help the audience perceive the value of the exercise.

https://github.com/BlueBrain/BluePyOpt/blob/master/examples/simplecell/simplecell.ipynb

Atm the L5PC test only creates and runs the template and imports the optimisation.
There should be a functionality test that runs a tiny optimisation (using scoop).

I am interested in this project and have worked on few such things before. I wanted this as a side project working in free time. I asked few questions on tickets, I think mostly minor inputs, do you accept contributions from people outside the organisation? Please let me know. I just felt weird w/ the lack of communication from an open source project.

Extend documentation to show how to different eFeatures can be combined into objectives

Reduce the possibility that the model will be instantiated, if created from hoc template, that runs with a different celcius or other global parameter

Hello

I use a hoc cell template with Bluepyopt, called 'class_axoaxoniccell_temp.hoc' attached below, and I run into the following issue when I run my bluepyopt script 'test.py' using my hoc cell template.
error:

  File "/usr/local/lib/python2.7/dist-packages/bluepyopt/ephys/locations.py", line 205, in instantiate
    isectionlist = getattr(icell, self.seclist_name)
AttributeError: 'hoc.HocObject' object has no attribute 'soma_list'

According to locations.py in ephys, 'hoc.HocObject namely 'icell' reflects 'my_cell[0]' in my case. 'my_cell[0]' does have a soma_list attribute which can be found in the hoc code. If for example I explicitly define "somatic_loc=my_cell.soma_list" in the python script, then the code runs just fine, it is when I define somatic_loc using the "class NrnSeclistLocation" as shown in the python script below, that I get the error mentioned above. To me both methods should ultimately call the same object: "my_cell.soma_list" but they don't. Any help with this would be much appreciated.

thank you,
Alexandra

test.py:

"""Expsyn synapse parameter fitting"""

# pylint: disable=R0914

import os

import bluepyopt as bpopt
import bluepyopt.ephys as ephys
import neuron
from neuron import h, gui
import matplotlib.pyplot as plt

h.load_file('class_axoaxoniccell_temp.hoc')
my_cell=h.axoaxoniccell_temp()

morph = ephys.morphologies.NrnFileMorphology('class_axoaxoniccell_temp.hoc') 

somatic_loc = ephys.locations.NrnSeclistLocation('somatic', seclist_name='soma_list')
basal_loc = ephys.locations.NrnSeclistLocation('basal', seclist_name='basal_list')
axonal_loc = ephys.locations.NrnSeclistLocation('axonal', seclist_name='axonal_list')

soma_loc=my_cell.soma[0]
#somatic_loc=my_cell.soma_list  #alternatively if I use this line it works 

ch_KvA_mech = ephys.mechanisms.NrnMODMechanism(
     name='ch_KvA',
     suffix='ch_KvA',
     locations=[soma_loc])#[somatic_loc])
# Non-Frozen Mechanism Parameters
gmax_ch_KvA_param = ephys.parameters.NrnSectionParameter(
    name ='gmax_ch_KvA',
    param_name ='gmax_ch_KvA',
    locations = [somatic_loc],
    value = h.axoaxoniccell_temp().gKvA,
    bounds=[0.00005, 0.00055],
    frozen = False
)

simple_cell = ephys.models.CellModel(
        name='my_cell',
        morph=morph,
        mechs=[ch_KvA_mech],
        params=[gmax_ch_KvA_param])
.....etc

class_axoaxonicell_temp.hoc

begintemplate axoaxoniccell_temp
public init, connect_sections, size_sections, append_sections, define_synapses
public mechinit, insert_mechs, set_biophys, get_root
public pre_list, connect_pre, is_art, is_connected, gid, randi
public soma, dend
public all, basal_list, apical_list, soma_list, axon_list, dendrite_list
public x, y, z, position, myroot, myrootsec, Vrest
public NumSoma, NumApical, NumBasal, NumAxon
public ExpSyn
public celsius


// strings
strdef myroot

// objects
objref syn, pre_list, templist, rootlist, myrootsec, this

// external variables
//external numCellTypes, cellType

// create the sections[segments]
NumSoma=1
NumApical=16
NumBasal=0
NumAxon=0

create soma[NumSoma], dend[NumApical]

proc init() {
        gid = 1
        randi = 2

        // morphology
        connect_sections()      // local fcn: connect soma, dendrites, axon initial segment
        size_sections()         // local fcn: set the size dimensions of each section
        define_shape()          // builtin fcn: fill in 3d info for sections defined by only L and diam, translate 3d points for consistency with their connections
        append_sections()       // local fcn: append all sections to the section list
        set_nseg()                      // local fcn: set the number of segments in each section
        get_root()                      // local fcn: perform morphology checks

        // electrophysiology
        mechinit()                      // local fcn: set values for max conductances and reversal potentials of ion channels and other ephys parameters that are subject to fitting
        insert_mechs()          // local fcn: insert ion channels and actually set values determined in the mechinit fcn
        set_chanparams()        // local fcn: after all channels have been inserted, then their other parameters can be set

        // synapses
        //pre_list = new List() // define a list for the presynaptic connections
        //define_synapses($3)   // local fcn: define all possible synaptic connections received by this cell
        NumSoma=1
        celsius = 37
}
......etc

Takes a string, replacing the current implementation if defined

I've been trying to follow these two examples:

examples/simplecell/simplecell.ipynb
examples l5pc/L5PC.ipynb

However, I cannot load the morphology with the ephys.morphologies.NrnFileMorphology method as it only supports the .asc and .swc format.

(How) can a morphology described in the .hoc format be used within BluePyOpt?

Hey again,
I wanted to ask whether you can recommend any sort of documentation, tutorial or guide for BluePyOpt? I went through your simple cell example and now I wonder where I can find ressources if I want to adapt the code to my needs. I wanted to start understanding bpop.optimisations.DEAPOptimisation() for example but the only wiki I found was pretty empty.

I was following alone the optimization set up in the jupyter notebook in Quick Start, and this happened

Either DEAPOptimization is deprecated or the example has a typo.