多肽稳定性的AI预测和修饰策略
Artificial Intelligence for Peptide Stability Prediction and Peptide Modifi-cation Strategies
蒋迅 1马天玥 1郑珩1
作者信息
- 1. 中国药科大学生命科学与技术学院,江苏南京 211198
- 折叠
摘要
多肽是分子大小介于小分子和蛋白质之间,具有独特类药空间的一类分子,具有亲和力强、选择性高、免疫原性低等优点.然而,多肽的一级结构和高级结构易被破坏,导致多肽在体内易降解、半衰期短、生物利用度低.多肽的不稳定性制约了多肽的成药性,对多肽进行骨架改造和基团修饰可延长半衰期.基于人工智能的多肽稳定性预测能够加速多肽的开发进程,具有颠覆多肽类药物研发范式的潜力.鉴于影响多肽稳定性的因素众多,多肽的结构稳定性、酶稳定性、代谢稳定性等概念首先被疏理,接着总结了可作为机器学习标签的湿实验指标.近期的多肽稳定性预测研究涉及到多肽的体外稳定性和体内稳定性,据此介绍了多肽制剂稳定性、多肽体外稳定性、多肽胃肠道稳定性、胰岛素类似物PK参数预测的进展,并总结了提升多肽稳定性的常见方法.
Abstract
Peptides are a class of molecules whose molecular size is between small molecules and proteins,and which have a unique drug-like space.Peptides have the advantages of high selectivity and low immunogenicity.However,the primary and advanced struc-tures of peptides are easily destroyed,leading to easy degradation of peptides in vivo with short half-life and low bioavailability.Pep-tide instability restricts its druggability.Backbone modification and motif modification of peptides prolong half-life.Peptide stability prediction based on artificial intelligence can accelerate the development process of peptide formulations,which has the potential to shake the paradigm of peptide drug development.Because of the richness of the influencing factors of peptide stability,the concepts of structural stability,metabolic stability,and thermodynamic stability of peptides were introduced,and the wet experimental metrics that could be used as machine learning labels was summarized.Recent peptide stability prediction studies cover both in vitro and in vivo stability of peptides.Accordingly,the advances in the prediction of peptide formulation stability,peptide in vitro stability,peptide gastrointestinal stability,and PK parameters of insulin analogs were presented.Common methods to improve the stability of peptides were summarized.
关键词
多肽稳定性/计算机辅助药物分子设计/多肽修饰/胰岛素类似物/机器学习/深度学习Key words
peptide stability/computer aided drug design/peptide modification/insulin analogue/machine learning/deep learning引用本文复制引用
出版年
2024
NETL
NSTL
万方数据