Skip to content

Latest commit

 

History

History

AP_simulation

README.md

Propagation of uncertainty in drug effects

This code performs uncertainty propagation for the CiPA in silico model. Model inputs are the pharmacological parameters for the hERG/IKr Markov model (README.md) and Hill equation parameters for drug block of other ionic currents (README.md). These inputs are used to simulate action potentials (APs) with the optimized IKr-dynamic ORd model (Dutta et al. 2017). The primary model output considered here is the qNet metric for proarrhythmia risk described by Dutta et al. 2017.

Running the code

This code uses the following R packages: optparse (version 1.4.4), deSolve (version 1.14), ggplot2 (version 2.2.0), and rms (version 4.5-0).

The IKr-dynamic ORd model C code is provided in models/ and must be compiled:

cd models
R CMD SHLIB newordherg_qNet.c

To run simulations, either fixed-input (optimal-fit) parameters or uncertainty-input parameters must be supplied to the model. By default, drug-hERG parameters are located here and Hill equation parameters are located here.

Results and figures are automatically saved to results/ and figs/, respectively.

This bash script provides a short example of how to run the simulations. The full process is explained below.

Fixed-input simulations

The control (no drug) AP simulation can be run by calling the script:

Rscript AP_simulation.R

AP simulations with drug can be run by specifying the drug name (case sensitive) and dose (interpreted as multiples of the therapeutic concentration (nM) listed in data/CiPA_training_drugs.csv):

Rscript AP_simulation.R -d dofetilide -x "1-10,15,20,25"

New drugs can be simulated if therapeutic (Cmax) values and input parameters are provided at the specified locations:

Rscript AP_simulation.R -d drug1 -x "1-10,15,20,25" --cmaxfile="my_cmax_table.csv" --hergpath="path/to/herg/results/" --hillpath="path/to/hill/results/"

If the therapeutic concentration is not available in the CSV file specified by "--cmaxfile", doses are interpreted as nM concentrations.

Uncertainty-input simulations

Samples from uncertainty-input probability distributions can be simulated by specifying sample indices:

Rscript AP_simulation.R -d dofetilide -x "1-10,15,20,25" -i "1-2000"

The appropriate parameters will be looked up in the default directories (here and here).

Note that running many AP simulations is computationally intensive, and it is recommended to run them in parallel on a high-performance computing resource. (See this script for an example of how to split up the simulations.

Postprocessing

Results for all fixed-input simulations can be combined with:

Rscript combine_results.R

Results for all uncertainty-input simulations can be combined by specifying the number of samples that were run:

Rscript combine_results.R -n 2000

The scripts compute_qNet_CI.R and compute_TdP_error.R perform analysis on the combined results and generate figures. With these scripts, TdP risk categories for each drug (0, 1, or 2) are read from a CSV file specified by the "--tdpfile" option (data/CiPA_training_drugs.csv by default):

Rscript compute_qNet_CI.R
Rscript compute_TdP_error.R
Rscript compute_TdP_error.R --uncertainty

References

  • Dutta, S., Chang, K.C., Beattie, K.A., Sheng, J., Tran, P.N., Wu, W.W., et al. (2017). Optimization of an In silico Cardiac Cell Model for Proarrhythmia Risk Assessment. Frontiers in Physiology 8(616). doi: 10.3389/fphys.2017.00616.
  • O'Hara, T., Virag, L., Varro, A., and Rudy, Y. (2011). Simulation of the undiseased human cardiac ventricular action potential: model formulation and experimental validation. PLoS Comput Biol 7(5), e1002061. doi: 10.1371/journal.pcbi.1002061.