This repository contains the code necessary to reproduce the results and figures in Evaluating the predictive accuracy of ion-channel models using data from multiple experimental designs.
Running this code requires an installation of python 3 and the markovmodels package. The code in this repository has been tested on MacOS where the requisite packages have been installed using micromamba.
It is recommended to install libraries and run scripts in a virtual environment to avoid version conflicts between different projects. First, clone the repository.
git clone https://github.com/CardiacModelling/multiprotocol_data_fitting
Then, ensure that your pip installation is up to date.
python3 -m pip install --upgrade pip
Then, install the required packages.
python3 -m pip install -r requirements.txt
Data can be downloaded through FigShare by running
wget https://figshare.com/ndownloader/files/48628150 -O 25112022_MW1_FF.tar.xz
wget https://figshare.com/ndownloader/files/48632359 -O 25112022_MW_FF_processed.tar.xz
The data tarballs can then be extracted by running
tar xvf 25112022_MW1_FF.tar.xz -C data
tar xvf 25112022_MW_FF_processed.tar.xz -C data
Similarly, the HPC model fits can be downloaded and extracted by running
wget https://figshare.com/ndownloader/files/48634114 -O 25112022MW_fitting.tar.xz
tar xvf 25112022MW_fitting.tar.xz -c data
Fitting was performed using scripts/fit_all_wells_and_protocols.py which can be provided with the -w, --sweeps --protocols command to select specific data traces to fit. The --experiment_name`` flag should be set to 25112022_MW. This fitting is performed on data postprocessed by the pcpostprocess package. The exact command line parameters used for each instance are found in the info.txt folders in the subdirectories of 25112022MW_fitting. As are the SLURM scripts used, and command-line output.
The list of figures in the paper and the scripts that are run to produce them are shown in the table below. Some of these figures were produced using pcpostprocess.
| figure | script | filename |
|---|---|---|
| 1 | NA | NA |
| 2 | NA | NA |
| 3 | plot_protocols.py | protocols_figure.pdf |
| 4 | NA | NA |
| 5 | optimisation_results.py | B20_staircaseramp1_sweep0.pdf |
| 6 | prediction_comparison.py | prediction_comparison.pdf |
| 7 | t_test_plots.py | average_sweep_0_t_scores_model3_0c_fitting.pdf |
| 8 | t_test_plots.py | average_sweep_0_t_scores_model3_0c_prediction.pdf |
| 9 | heatmaps.py | best_worst_0c_model3_heatmap.pdf |
| 10 | heatmaps.py | Case0c_heatmap_comparison.pdf |
| 11 | scatterplots.py | per_well_p1_p2_d_1.pdf |
| A.1 | pcpostprocess.scripts.run_herg_qc | B20_staircaseramp1_before0.pdf |
| A.2 | pcpostprocess.scripts.run_herg_qc | 25112022_MW-staircaseramp1-B20-sweep1-subtraction.pdf |
| D.1 | pcpostprocess.scripts.summarise_herg_export | E_rev.pdf |
| D.2 | heatmaps.py | averaged_well_heatmaps.pdf |
| E.1 | t_tests_plots.py | average_sweep0_t_scores_model2_0c_fitting.pdf |
| E.2 | t_test_plots.py | average_sweep0_t_scores_model10_0c_fitting.pdf |
| E.3 | t_test_plots.py | average_sweep0_t_scores_Wang_0c_fitting.pdf |
| E.4 | t_test_plots.py | average_sweep0_t_scores_model2_0c_prediction.pdf |
| E.5 | t_test_plots.py | average_sweep0_t_scores_model10_0c_prediction.pdf |
| E.6 | t_test_plots.py | average_sweep0_t_scores_Wang_0c_prediction.pdf |