Presentation description

Ion transport through channels and transporters is found in nearly all cell types and organisms and controls a wide range of essential biological functions. Given their widespread role in disease states, channels and transporters are important targets for further therapeutic studies. Kinetic modeling of ion transport can be categorized into two major approaches: top-down based on experimental data and bottom-up based on simulation data. While the former often lacks the resolution to narrow the solution space sufficiently, the latter characteristically struggles with compounding numerical errors. Our multiscale responsive kinetic modeling (MsRKM) method combines the strengths of both approaches, fine-tuning computed transition rates with experimental data, such as current-voltage assays. However, this is a computationally expensive task, requiring iterative optimizations on a network structure where the nodes of the network represent metastable states, and the edges define transition rates. Our recent feature-rich updates, including the parameterization of ion-ion coulombin interactions and pH-dependent rates, significantly amplify the computational cost of optimization. To reduce the new higher-dimension model runtime and thereby computational accessibility, we use automated code profiling to identify bottlenecks and high-throughput functions as candidates for refactoring. In addition to runtime improvements, we revisit analysis scripts to simplify code logic, increase modularity, and raise code testing coverage. The updated model runs more than six times faster than the initial implementation, and we are considering future optimizations with threading.