Model Overview | SCU-China 2024

The development of synthetic biology relies heavily on various biological modeling techniques and methods. Models enable researchers to design experiments and analyze results more efficiently. This year, regarding the novel Nicotiana benthamiana synthetic biology chassis, VersaTobacco, created by SCU-China, we conducted dry lab explorations in four areas: metabolic flux analysis, market promising analysis, promoter optimization, and protein modeling. Through simulation and prediction using models in these areas, we provided guidance for wet lab experiments, saving substantial manpower and time, while also offering crucial support for the advancement and feasibility of our project.

Metabolic Flow Analysis

Using constraint-based reconstruction and analysis (COBRA) techniques, we constructed a genome-scale metabolic network model (GSMM) for Nicotiana benthamiana to investigate the key genes involved in the accumulation of chlorogenic acid, as well as the synthesis of target metabolites in mutant N. benthamiana under various knockout strategies. Through flux balance analysis (FBA), we identified HQT as our knockout target, leading to a reduction in chlorogenic acid accumulation while enhancing the synthesis of target metabolites in related pathways, thereby guiding the wet lab experiments.
Metabolic Flow Analysis

Market Promising Analysis

Based on the results of metabolic flux analysis, we conducted a market promising analysis on our novel Nicotiana benthamiana chassis to assess the profitability of VersaTobacco as a universal plant chassis for producing resveratrol. Through return on investment (ROI) and risk-return analysis, we identified the optimal time frame for the highest profit margins in compound production and the expected returns, indicating that our plant chassis has significant market application potential.
Market Promising Analysis

Promoter Optimization

To further leverage the advantages of VersaTobacco as a synthetic biology chassis, we optimized its promoters to enhance product yield. Utilizing deep learning techniques, we trained a Conformer-based neural network architecture to predict promoter strength based on their sequences. Subsequently, we employed a global random mutation approach based on a greedy algorithm for promoter optimization. The promoter optimizer we designed provides a novel toolkit for tobacco-related synthetic biology applications.
Promoter Optimization

Protein Modeling

In order to optimize enzymes more efficiently, it is necessary to build structure model of enzyme, perform molecular docking and dynamic simulations. Tools such as Alphafold3 can provide reliable three-dimensional enzyme structures. Based on this, docking of enzyme and substrate molecules can let us understand the mechanism of enzyme substrate interaction and identify key residues which affect catalysis. Molecular dynamics simulation can further simulate the interaction between enzymes and substrates, and perform virtual screening on the constructed mutants, thereby reducing the workload of wet experiments.
Protein Modeling