13 March 2024 | Peng Cheng, Cong Mao, Jin Tang, Sen Yang, Yu Cheng, Wuke Wang, Qixiu Gu, Wei Han, Hao Chen, Sihan Li, Yaofeng Chen, Jianglin Zhou, Wuju Li, Aimin Pan, Suwen Zhao, Xingxu Huang, Shiqiang Zhu, Jun Zhang, Wenjie Shu, and Shengqi Wang
Protein Mutational Effect Predictor (ProMEP) is a novel, zero-shot method for predicting mutation effects on proteins, leveraging multimodal deep representation learning. The model integrates both sequence and structure contexts from approximately 160 million proteins, achieving state-of-the-art performance in mutational effect prediction. ProMEP significantly improves speed compared to existing methods, making it efficient for protein engineering. It accurately forecasts the effects of mutations on gene-editing enzymes TnPB and TadA, leading to the development of high-performance gene-editing tools. Specifically, the 5-site mutant of TnPB achieves a gene-editing efficiency of 74.04%, and the base editing tool based on a 15-site mutant of TadA exhibits an A-to-G conversion frequency of 77.27%, with reduced bystander and off-target effects. ProMEP's superior performance and ability to guide protein engineering make it a powerful tool for exploring the vast protein space and advancing biomedicine and synthetic biology.Protein Mutational Effect Predictor (ProMEP) is a novel, zero-shot method for predicting mutation effects on proteins, leveraging multimodal deep representation learning. The model integrates both sequence and structure contexts from approximately 160 million proteins, achieving state-of-the-art performance in mutational effect prediction. ProMEP significantly improves speed compared to existing methods, making it efficient for protein engineering. It accurately forecasts the effects of mutations on gene-editing enzymes TnPB and TadA, leading to the development of high-performance gene-editing tools. Specifically, the 5-site mutant of TnPB achieves a gene-editing efficiency of 74.04%, and the base editing tool based on a 15-site mutant of TadA exhibits an A-to-G conversion frequency of 77.27%, with reduced bystander and off-target effects. ProMEP's superior performance and ability to guide protein engineering make it a powerful tool for exploring the vast protein space and advancing biomedicine and synthetic biology.