蛋白质是生命活动的物质基础。蛋白质的功能取决于其空间结构,因此解析蛋白质的空间结构对于理解其功能至关重要。人工智能(artificial intelligence, AI)模型的利用极大地推进了蛋白质空间结构预测算法的开发,AlphaFold2是该领域里程碑式的成果,使得快速、准确且大规模的蛋白质空间结构预测成为可能。此外,蛋白质语言模型、蛋白质相互作用预测以及蛋白质设计等领域均在AlphaFold时代迎来快速发展,代表性的模型包括ESM2、ScanNet、RFdiffusion和RoseTTAFold-All Atom 等。这些基于人工智能的新算法的开发极大地促进了蛋白质功能、诱发疾病的机制和药物设计等领域的研究。

Proteins serve as the fundamental building blocks of life processes. Their functions are highly determined by the spatial structure, making analyzing these spatial structures essential for fully comprehending their roles. The application of artificial intelligence (AI) models has significantly advanced the development of algorithms for predicting protein spatial structure. AlphaFold2 represents a groundbreaking advancement in this field, enabling rapid, accurate, and large-scale predictions of protein spatial structures. In the AlphaFold era, substantial progress has been made in fields such as protein language model development, protein-protein interaction prediction, and protein design, with notable models including ESM2, ScanNet, RFdiffusion, and RoseTTAFold-All Atom, among others. These novel AI-based algorithms have profoundly facilitated research into protein function, disease mechanisms, and drug design.