Dr. Joseph Sang-II Kwon, associate professor in the Artie McFerrin Department of Chemical Engineering at Texas A&M University, alongside Dr. Arul Jayaraman, executive associate dean of the College of Engineering, received the 2021 Premium Award for Best Paper for The Institution of Engineering and Technology (IET) in Systems Biology.
Their study was published in vol. 13, issue 4 of IET Systems Biology.
For any biological system, there are signaling pathways — a series of chemical reactions that allow the cell to perform functions such as cell divisions or cell death. Within these reactions, several proteins and catalysts are involved that contribute to the difficulty of developing mathematical models that accurately describe these pathways.
In addition, the cost of obtaining an accurate model can be high and labor-intensive, and there is often limited information about the pathway.
The researchers took a systems biology approach to combat this issue, which uses data clustering to combine different iterations of pathways, creating a time-varying model based on nominal models available in literature.
“If you were to use this one simple model for the entirety of systems that change frequently, the model’s accuracy could be damaged,” said Kwon. “If you make small adjustments to a model depending on the time domain, you can expect much better accuracy.”
Using the researcher’s intracellular signaling model starts with a global sensitivity analysis that helps give value to the importance of the model and the way inputs will impact outputs, as well as the most critical parameters. Second, measurement data is clustered to determine temporal subdomains where the parameters take different values. Finally, a least-squares problem is solved, which helps identify the best options for a data set.
The proposed methodology is a semi-data-driven approach, where the model construction is guided by both the available experimental data and the mechanistic model. Specifically, based on the experimental data, the temporal profiles of the model parameters are inferred to complement the model mismatch due to the use of a nominal model. The resultant model can provide relatively accurate predictions in spite of the incomplete knowledge of the underlying system. At the same time, the use of the mechanistic model allows the resultant model to be used in the detailed analysis of the underlying mechanisms, which is difficult to perform through a data-driven model.
A better understanding of the nuclear factor kappa B (NF-κB) pathway would enable the design of more specific and effective therapeutic approaches for treating inflammatory diseases.
The researchers hope to use this pathway in broader capacities, as it shows potential to be used when mathematical models are not sufficient for the design of experiments.