DEKP: a deep learning model for enzyme kinetic parameter prediction based on pretrained models and graph neural...

Area of Science:

• Biotechnology and Biochemistry
• Computational Biology and Cheminformatics
• Enzyme Engineering and Directed Evolution

Background:

• Accurate prediction of enzyme kinetic parameters is vital for enzyme screening and engineering.
• Current machine learning models face performance limitations due to sequence similarity variations in datasets.
• Experimental methods for determining enzyme kinetics are costly and time-consuming.

Purpose of the Study:

• To develop a novel deep learning approach for predicting enzyme kinetic parameters.
• To address the performance degradation issues caused by sequence similarity variance in existing models.
• To enhance the enzyme screening process and support enzyme-directed evolution research.

Main Methods:

• Introduction of DEKP, an integrated deep learning approach for enzyme kinetic parameter prediction.
• Leveraging pretrained models for protein sequence representation.
• Incorporating enhanced graph neural networks for comprehensive protein structural feature analysis.

Main Results:

• DEKP effectively alleviates performance degradation caused by sequence similarity variations.
• The approach demonstrates sensitive detection of catalytic efficiency changes due to enzyme mutations.
• Experimental validation shows DEKP outperforms existing models in predicting enzyme kinetic parameters.

Conclusions:

• DEKP offers a robust and accurate method for predicting enzyme kinetic parameters.
• This approach significantly improves enzyme screening efficiency and aids in enzyme design.
• DEKP is a valuable tool for advancing enzyme-directed evolution research.