Assistant Professor Fairfield University Fairfield, Connecticut, United States
Introduction: : Pulmonary hypertension is a condition in which blood pressure rises in the pulmonary arteries, which is often caused by low oxygen levels or hypoxia. Hypoxia causes arteries to constrict, making it difficult for blood to flow to the lungs, which also puts strain on the heart. Computational modeling of the cell-signaling network is a useful tool in investigating and understanding the effects of hypoxia. The investigation of the sensitivity of the computational model is also important in analyzing the model’s accuracy.
Materials and
Methods: : For this research all computational work was done using Netflux, a user-friendly computational modeling tool used to create computational models of cell-signaling networks. These models can simulate cell behaviors and allow for accessible experimentation of parameters within the cell-signaling networks, such as changing the weight of a reaction to examine different outputs. The main objective of this research was to test the sensitivity of a pulmonary artery smooth muscle cell model. More specifically, we tested the sensitivity of the model by knocking down characteristic values as well as upregulating them. We used Netflux to export a system of ordinary differential equations into MATLAB and then wrote custom scripts to simulate knock-down and upregulation of each node in the model. We changed the weight of reactions (w) between the values of 0-1 in intervals of 0.25 and then examined how simulations varied after each trial. The variables of initial activation y0 and maximum activation ymax were also tested. We tested knockdowns by decreasing the values of ymax to values of 0 and 0.25. A difference plot was then simulated to compare the results with the results of the variable’s original value of 1. We tested upregulation by increasing the value of ymax to 2 and 5, and the same process was used in which a difference plot was simulated to compare results and analyze sensitivity.
Results, Conclusions, and Discussions:: We found that changing ymax had the greatest impact on the model, so we focused on the results of these simulations. When setting the ymax variable to 0 the plots indicate that some of the species start to react and by the colormap we can see that the activity is on the negative side but still different from the reactions when ymax is equal to 1. Once ymax is increased to 0.5 which is much closer to 1 than 0, the plot indicates that most of the species remain unchanged while only a handful of species have some reaction. When the variables are upregulated so that ymax is equal to 2, similar species that reacted negatively when ymax equaled 0, now react positively. Finally, when ymax was set to 5 a larger number of species had reactions and most notably ET1 and ETAR displayed the most amount of activity across all the plots (Figure 1). The computational model reacts sensibly to changes in variable value. The next step with this research is to quantify the uncertainty of the computational model and how accurately they can simulate real cell behaviors. This will be done by simulating perturbations to the parameters using Monte Carlo simulations. An expanded version of the model will also be tested with Netflux, including additional hypoxia signaling pathways and cell signaling models for fibroblasts and macrophages. Overall, the goal is to quantify the uncertainty that these models have when simulating cell behavior compared to how cells behave and to compare this simulated cell behavior to experimental results of cells from the pulmonary artery exposed to hypoxia. When ymax is changed to 0 and 0.25 from 1, species such as Wss show more activity. When ymax is changed to 2 and 5 from 1, species such as Stress, ET1 and ETAR display much more activity. This activity shows that computational models do react to changes in parameters which closely simulate the behavior of the actual cell-signaling network. Therefore, computational models such as those run in Netflux are useful in investigating cell-signaling networks.
Acknowledgements and/or References (Optional): : Stenmark et al., Circulation Research, 2006 Clark et al. PLOS Computational Biology, 2025 Irons et al., Annals of Biomedical Engineering, 2021
