Practical application of CellML 1.1: the integration of new mechanisms into a human ventricular myocyte model

David P Nickerson & Martin L Buist.

Computational Bioengineering Laboratory, Division of Bioengineering, National University of Singapore.

David Nickerson & Martin Buist. 'Practical application of CellML 1.1: The integration of new mechanisms into a human ventricular myocyte model. Prog Biophys Mol Biol, 98:38—51, 2008. [doi: 10.1016/j.pbiomolbio.2008.05.006]

In the above article we present concepts aiding the application of CellML standards with the goal of easing integration of new mechanisms into cardiac cell models. In order to further aid the presentation of the ideas discussed in the article, here we provide access to all the mathematical models and numerical experiments covered in the article. As mentioned in the paper, all numerical simulations were performed using CellMLSimulator, which is freely available from http://cellml.SourceForge.net.

In the following, we list all the simulation result graphs presented in the article and provide links to the original source used to generate the graphs. These source descriptions are all defined using CellML and associated metadata standards. All graphing metadata which is used to create the graphs presented in the article can be found in the experiments directory and the mathematical models and other metadata can be found in our local CellML model repository.

• Figure 6
• Figure 7
• Figure 8
• Figure 9
  • Figure 9(a)
  • Figure 9(b)
  • Figure 9(c)
• Figure 10
• Figure 11
• Figure 12
• Figure 13
• Figure 14
• Figure 15

Figure 6

This is a simple validation simulation presenting results for comparison to the original ten Tusscher et al. (2004) article. The graphing metadata used to generate these graphs is TNNP-periodic-graphs.xml, which uses the CellML model TNNP-periodic.xml which imports the base TNNP model and defines a periodic electrical stimulus protocol.

If you have CellMLSimulator available, you can recreate the graphs with a command similar to:

Figure 7

The APD restitution curve. In order to obtain the APD restitution curve, the S1-S2 protocol needs to be applied to the TNNP model multiple times with varying diastolic interval. We achieve this be defining a template model (TNNP-S1S2-template.xml) which imports the TNNP model with the S1-S2 pacing protocol applied (TNNP-S1S2.xml) and exposes only the diastolic interval as a model parameter for external control. We then use a perl script (runAPDR.pl) and the template model to generate models (and simulation metadata) for each diastolic interval spanning the range of interest.

The perl script also generates graphing metadata which defines a trace for each membrane potential variable from each of the generated models. In addition, the script defines analysis metadata from the template TNNP-S1S2-APDs-template.xml — analysis metadata is used by CellMLSimulator to analyze simulation results and in this instance is used to calculate the APD for each trace in the defined graphing metadata. The same perl script will then extract the appropriate APD values from the analysis results to assemble the data for the graph given in Figure 7.

Figure 8

Changes in intracellular Ca²⁺, Na⁺, and K⁺ under increasing pacing frequencies (cf. ten Tusscher et al. 2004, Fig. 9). Pacing frequency is varied from 0.25 to 3.0 Hz and each frequency is maintained for 10 minutes. The graphing metadata used to generate these graphs is TNNP-periodic-increasing-frequency-graphs.xml, which uses the CellML model TNNP-periodic-increaseing-frequency.xml.

If you have CellMLSimulator available, you can recreate the graphs with a command similar to:

Figure 9

Alterations in AP morphology due to the inclusion of the I_K(ATP) current.

Figure 9(a)

The unmodified TNNP model compared to those with either the Shaw & Rudy (1997) or Matsuoka et al. (2003) formulations of I_K(ATP) under control conditions. The graphing metadata used to generate these graphs is IKATP-graphs.xml, which uses the CellML models TNNP-periodic.xml, IKATP-Shaw-control.xml, and IKATP-Matsuoka-control.xml.

If you have CellMLSimulator available, you can recreate the graph with a command similar to:

Figure 9(b)

The base TNNP model with the Shaw & Rudy (1997) I_K(ATP) formulation. In generating this graph, we first created a template for the graphing metadata (IKATP-Shaw-graph-template.xml) and the modified model (IKATP-Shaw-template.xml). The model template imports the modified TNNP model and provides access only to the I_K(ATP) parameters. The templates are then used with a perl script (runModelRange.pl) to create the complete graphing metadata document and each of the required models with the required changes in parameter values.

As the runModelRange.pl script currently produces metadata and models using absolute file paths from the local file system, we currently do not provide the generated documents via this website. However, the following steps should allow Figure 9(b) to be recreated locally using CellMLSimulator.

1. Download the runModelRange.pl script. You'll need to update line 12 to point to your local CellMLSimulator.
2. Download the two template files (IKATP-Shaw-graph-template.xml and IKATP-Shaw-template.xml) to the same directory.
3. Run the script with the command:
   ~/> perl runModelRange.pl IKATP-Shaw-template.xml IKATP-Shaw-graph-template.xml . png landscape ATP_VALUE=6.8,6.0,6.0,5,4,3,2 K_HALF_VALUE=0.042,0.042,0.25
   which will generate the series of model files and the graphing metadata and then run the simulations which produce the graph presented in Figure 9(b).

Figure 9(c)

Same as Figure 9(b) but for the Matsuoka et al. (2003) formulation of I_K(ATP). The following steps should allow Figure 9(c) to be recreated locally using CellMLSimulator.

1. Download the runModelRange.pl script. You'll need to update line 12 to point to your local CellMLSimulator.
2. Download the two template files (IKATP-Matsuoka-graph-template.xml and IKATP-Matsuoka-template.xml) to the same directory.
3. Run the script with the command:
   ~/> perl runModelRange.pl IKATP-Matsuoka-template.xml IKATP-Matsuoka-graph-template.xml . png landscape ATP_VALUE=6.8,6.0,2.0,1.0,0.5,0.36
   which will generate the series of model files and the graphing metadata and then run the simulations which produce the graph presented in Figure 9(c).

Figure 10

The effects of changing pH_i on the inward rectifier K⁺ current, I_K1. The graphing metadata used to generate these graphs is IK1-pH_clamp-graphs.rdf, which uses the CellML model IK1-pH_clamp.xml. Figure 10(b) also illustrates the use of experimental data with graphing metadata. The experimental data is described in the document IK1-pH-fit-unit_amp.rdf with the actual data given in IK1-pH-fit-unit_amp.csv.

If you have CellMLSimulator available, you can recreate the graphs with a command similar to:

Figure 11

The effect of adding the Ito et al. (1992) formulation of pH_i dependence of I_K1 to the model. The graphing metadata used to generate these graphs is IK1-Ito-graphs.xml, which uses the CellML models TNNP-periodic.xml, IK1-Ito-control.xml, and IK1-Ito-ischemic.xml.

If you have CellMLSimulator available, you can recreate the graphs with a command similar to:

Figure 12

Examining the changes in action potentials and [Ca]_i due to modification of I_CaL. The graphing metadata used to generate these graphs is ICaL-modified-graphs.xml, which uses the CellML models TNNP-periodic.xml, ICaL-modified-control.xml, ICaL-modified-ischemic-pHo.xml, ICaL-modified-ischemic-pHi.xml, ICaL-modified-ischemic-ATPi.xml, and ICaL-modified-ischemic.xml.

If you have CellMLSimulator available, you can recreate the graphs with a command similar to:

Figure 13

Examining the changes in action potentials due to modification of I_NaCa. The graphing metadata used to generate these graphs is INaCa-modified-graphs.xml, which uses the CellML models TNNP-periodic.xml, INaCa-modified-control.xml, INaCa-modified-ischemic-pHi.xml, INaCa-modified-ischemic-ATPi.xml, and INaCa-modified-ischemic.xml.

If you have CellMLSimulator available, you can recreate the graphs with a command similar to:

Figure 14

Examining the changes in intracellular calcium dynamics due to modification of J_rel. The graphing metadata used to generate these graphs is Jrel-modified-graphs.xml, which uses the CellML models TNNP-periodic.xml, Jrel-modified-control.xml, and Jrel-modified-ischemic-pHi.xml.

If you have CellMLSimulator available, you can recreate the graphs with a command similar to:

Figure 15

Examining the changes in intracellular calcium dynamics due to modification of J_up. The graphing metadata used to generate these graphs is Jup-modified-graphs.xml, which uses the CellML models TNNP-periodic.xml, Jup-modified-control.xml, and Jup-modified-ischemic-pHi.xml.

If you have CellMLSimulator available, you can recreate the graphs with a command similar to: