AI Protein Engineering: Universal Design Strategy
AI-Powered Protein Engineering Strategy Promises Rapid Advances in Medicine adn Biotechnology
A new artificial intelligence-informed strategy is poised to revolutionize protein engineering, offering a faster, more efficient, and broadly applicable method for designing proteins with tailored functions. Developed by researchers at the Institute of Genetics and Developmental Biology (IGDB) of the Chinese Academy of Sciences, lead by Professor GAO Caixia, the approach – dubbed AI-informed Constraints for protein Engineering (aice) – bypasses the need for extensive, specialized AI training, making advanced protein design accessible to a wider range of researchers.
The findings, published July 7th in Cell, address longstanding challenges in protein engineering. Traditionally, achieving optimal protein performance has been a costly, time-consuming, and often limited process. While recent AI-based methods have shown promise, they frequently demand significant computational resources, hindering widespread adoption. AiCE offers a compelling alternative, maintaining predictive accuracy while dramatically improving accessibility.
How AiCE Works: Leveraging Structure and Evolution
AiCE’s core innovation lies in its integration of structural and evolutionary constraints into existing “inverse folding” models – AI tools that predict amino acid sequences compatible with a desired protein 3D structure. The team first developed AiCEsingle, a module focused on predicting high-fitness (HF) single amino acid substitutions. By extensively sampling inverse folding models and incorporating structural constraints, AiCEsingle significantly enhances prediction accuracy.
Rigorous benchmarking against 60 deep mutational scanning (DMS) datasets revealed AiCEsingle outperformed other AI-based methods by a remarkable 36-90%. The module’s effectiveness extended to complex proteins and even protein-nucleic acid complexes, with structural constraints alone contributing to a 37% improvement in accuracy.
Recognizing that multiple mutations can interact in complex ways – sometimes negatively – the researchers then created AiCEmulti.This module incorporates evolutionary coupling constraints to accurately predict multiple high-fitness mutations with minimal computational demand, further expanding the tool’s versatility.
From Lab to Submission: Next-Generation Base Editors
to demonstrate AiCE’s practical power, the researchers successfully evolved eight proteins with diverse structures and functions, including deaminases, nuclear localization sequences, nucleases, and reverse transcriptases. This engineering feat has directly led to the creation of several next-generation base editors with significant improvements for precision medicine and molecular breeding.
Specifically, the team developed:
enABE8e: A cytosine base editor with a ~50% narrower editing window, increasing precision and reducing off-target effects.
enSdd6-CBE: An adenine base editor exhibiting 1.3× higher fidelity, minimizing unwanted genomic alterations.
* enDdd1-DdCBE: A mitochondrial base editor demonstrating a 13× increase in activity, opening new avenues for treating mitochondrial diseases.
A simpler, More Powerful Future for Protein engineering
AiCE represents a significant leap forward in protein engineering. By unlocking the potential of existing AI models, it provides a simple, efficient, and broadly applicable strategy for protein design. Beyond its immediate applications, AiCE enhances the interpretability of AI-driven protein redesign, fostering a deeper understanding of the relationship between protein structure, sequence, and function. This breakthrough promises to accelerate research and progress across a wide range of fields, from drug finding to materials science.
https://www.miragenews.com/ai-powered-universal-strategy-for-protein-1491888/