一种融合酶促反应多模态特征的酶周转数预测方法及系统
By constructing a multimodal feature prediction method and integrating information from enzymes, substrates, and products using deep learning technology, the high cost and low accuracy of enzyme turnover prediction are solved, achieving more efficient prediction of enzyme catalytic constants and construction of metabolic network models.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- JIANGNAN UNIV
- Filing Date
- 2025-09-22
- Publication Date
- 2026-07-17
AI Technical Summary
Existing methods for predicting enzyme turnover are costly, time-consuming, and lack accuracy and reliability, failing to effectively integrate multimodal information of enzyme-catalyzed reactions.
A prediction method integrating multimodal features of enzyme-catalyzed reactions is adopted. By constructing an optimized pre-trained model of protein and chemical reaction, the feature matrices of enzyme, substrate and product are extracted by combining graph attention network and graph isomorphic network, and the feature is enhanced by cross attention mechanism. Finally, the data are input into a multilayer perceptron to predict enzyme turnover.
It significantly improves the accuracy and reliability of predicting enzyme catalytic constants, solves the problems of high cost and low precision of traditional methods, and provides more efficient support for enzyme screening and metabolic network model construction.
Smart Images
Figure CN121483405B_ABST