ABC Parameterization of the Hodgkin-Huxley Model
On this page
This tutorial describes how to download and execute the Approximate Bayesian Computation code for inference on the Hodgkin-Huxley model as described in “Hodgkin-Huxley revisited: reparameterization and identifiability analysis of the classic action potential model with approximate Bayesian methods”, J. R. Soc. Interface 2:150499, 2015.
Installation
This project requires the Functional Curation add-on to Chaste in order to run, which in turn requires the Chaste source tree to be installed. To install these dependencies, follow the installation instructions in the Functional Curation paper tutorial. In addition, you will need the dependencies for the new Python implementation detailed at FunctionalCuration/PythonImplementation.
Afterwards, obtain the latest version of the project code from the Chaste repository using:
#!sh
cd [path_to_Chaste]/projects
svn checkout --username anonymous https://chaste.cs.ox.ac.uk/svn/chaste/projects/HodgkinHuxleyABC
Use your email address as the password for the ‘anonymous’ account.
Usage
Source code for the ABC-SMC parameter fitting algorithm is contained in the src
folder.
Python scripts for performing inference on both the simplified and full Hodgkin-Huxley
model can be found in tests
.
All functional curation protocol files can be found in tests/protocols
, while the
annotated CellML model file can be found in the top-level project directory as
hodgkin_huxley.cellml
.
A description of important files and their contents follows below:
src/fitting.py
contains the implementation of ABC-SMC with adaptive error shrinking.src/HodgkinHuxley.py
contains accessors for voltage clamp experimental data digitized
from the original 1952 publication. Used for all fitting experiments.
tests/HodgkinHuxleyFittingABC_simple.py
contains the python script for performing
inference on the simplified models of potassium (2 parameters) and sodium (4 parameters) conductance.
tests/HodgkinHuxleyFittingABC.py
contains the python script for performing
inference on the full models of potassium (5 parameters) and sodium (9 parameters) conductance.
tests/ProbabalisticAP
generates graphs of differential response of models parameterized
according to a sample ABC posterior estimate of sodium/potassium conductance when subjected to more complex voltage protocols.
To generate ABC posterior estimates for the six-parameter simplified Hodgkin-Huxley model as described in the paper, use:
scons projects/HodgkinHuxleyABC/test/HodgkinHuxleyFittingABC_simple.py
To generate ABC posterior estimates for the 14-parameter full Hodgkin-Huxley model as described in the paper, use:
scons projects/HodgkinHuxleyABC/test/HodgkinHuxleyFittingABC.py
To see verbose output on the progress of the ABC algorithm, add the flag no_store_results=1
to the scons
commands above.
Note however that this will prevent storing a copy of the output on disk.
TODO: Check that the above commands run on a fresh install.
TODO: Information on producing each graph in the paper.