Empirical quantification of predictive uncertainty due to model discrepancy by training with an ensemble of experimental designs: an application to ion channel kinetics

This repository contains the code necessary to reproduce the figures presented in our publication: "Empirical quantification of predictive uncertainty due to model discrepancy by training with an ensemble of experimental designs: an application to ion channel kinetics", Bulletin of Mathematical Biology (2023) Joseph G. Shuttleworth, Chon Lok Lei, Dominic G. Whittaker, Monique J. Windley, Adam P. Hill, Simon P. Preston and Gary R. Mirams.

This repository consists of package for the processing and analysis of patch-clamp electrophysiology data. Some of this functionality is used in the paper. The code to produce our figures is in the scripts directory, and the corresponding output is provided in the output directory.

Installation

These scripts have been tested with Python version 3.9 (see Dockerfile). It is recommended to install libraries and run scripts in a virtual environment to avoid version conflicts between different projects. To do this:

• Clone this repository git clone https://github.com/CardiacModelling/empricial_quantification_of_model_discrepancy
• Create a python virtual environment python -m venv .venv or (python -m virtualenv folder_name). If that doesn't work you may need to install virtualenv first pip install virtualenv.
• Activate the virtual environment using source folder_name/bin/activate. Simply type deactivate to exit the virtual environment at the end of a session.
• Install graphviz. On Ubuntu, this is done by running sudo apt install graphviz graphviz-dev.
• Install gcc and build essential: sudo apt-get install gcc build-essential
• Install cmake: sudo apt-get install cmake
• Install LaTeX (for plots): sudo apt install texlive texlive-latex-extra texlive-fonts-recommended cm-super dvipng
• Install scikit-build: pip install scikit-build
• Install the MarkovModels package by running pip install -e ..

Alternatively, you can create a docker image using Dockerfile.

Scripts

Figure 1 was produced using scripts/fix_wrong_param_study/simple_example.py.

For Case I, the computations are performed using scripts/fix_wrong_param_study/fix_wrong_params. Then, Figure 4 is produced using scripts/fix_wrong_param_study/big_multi with these results.

The synthetic dataset used for Case II was produced using scripts/fix_param_study/generate_synthetic_data.py and this dataset was used to fit both the Wang and Beattie models using scripts/fix_wrong_param_study/fit_all_wells_and_protocols.py. These results are summarised using scripts/fix_wrong_parma_study/CaseII_figure.py and error_compare_plot.py.

The scripts used to produce each figure are shown in the following table:

Figure	script
Fig1.pdf	simple_example.py
Fig3.pdf	plot_protocols.py
Fig4.pdf	CaseI_prediction_plots.py
Fig5.pdf	CaseI_main.py
Fig6.pdf	CaseII_prediction_plots.py
Fig7.pdf	CaseII_figure.py
Fig8.pdf	error_compare_plot.py
Fig10.pdf	CaseI_main.py

Running

To run a script execute it using Python. For example, python3 scripts/fix_wrong_param_study/fix_wrong_parameter.py --protocols sis staircase

Protocols

A list of voltage-clamp protocols are provided in /MarkovModels/protocols. These are .csv files which describe time-series data. The filenames which correspond to the protocols used in the data are shown in the table below.

protocol	filename
d0	longap
d1	hhbrute3gstep'
d2	sis
d3	spacefill19
d4	staircaseramp1
d5	wangbrute3gstep

Results

All of the computational results mentioned in the paper are provided in the paper_output directory. In paper_output, each subdirectory includes an info.txt file which lists the command run to produce the output.