In the NAU-CHINA 2023 project, we conducted our modeling from both molecular and mathematical aspects. By establishing molecular models, we predicted protein structures, which helped us better understand the structures and properties of the proteins used, as well as optimize experimental designs. Through building mathematical models, we simulated the concentration changes of OTα during the reaction process using ordinary differential equations, providing guidance for experiments. We also applied machine learning methods to protein secondary structure prediction, demonstrating the rationality of protein structures through data and enhancing the feasibility of experimental plans.

1、Structure prediction provides reliable models for the proteins and structures encountered during the experimental process.

2、Molecular dynamics simulations are used to test the stability of molecules with the aid of GROMACS while leveraging the advantages of various structure prediction servers, thereby demonstrating the reliability of the docking structures.

3、The establishment of ordinary differential equations monitors the concentration changes of OTα during the reaction process. Stability analysis is conducted to reflect the stability of our reaction and provide directional guidance for experiments.

4、The use of machine learning methods, which are applied to the prediction of protein secondary structure by selecting machine learning methods with better training effect to enhance the feasibility of experimental operation schemes.

In constructing the ordinary differential equations to simulate the concentration changes during the OTα reaction, we used the Mie function and logistic model, and adopted the stability analysis to prove the stability of the reaction in different environments, as well as to provide guidance for the experimental design, saving the time and cost of the experiment.

In the part of predicting the secondary structure of proteins based on machine learning, we selected three machine learning methods to train the model and applied the most effective random forest model to actual prediction, providing a reliable model and theoretical basis for the protein and structure encountered in the experimental process, which corresponds to the prediction of the tertiary structure.

read more

In molecular modeling, we utilized Swiss-Model, I-TASSER, and AlphaFold2 to predict the structures of SpyCatcher, SpyTag, C3, and T3. Docking results of SpyCatcher and SpyTag, C3 and T3 were obtained using ClusPro 2.0, ZDOCK, and GRAMM. Subsequently, we referred to existing literature to select reliable docking results for the SpyCatcher and SpyTag complex and assessed the stability of the selected C3 and T3 docking structures using GROMACS. Finally, with the assistance of GROMACS, we also explored the lattice structure.

read more