Jcim ｜ AI based protein structure prediction in drug discovery: impacts and challenges

2022 year 6 month 21 Japan , From small molecule allosteric drug discovery company HotSpot Therapeutics Of Michael Schauperl Et al. J Chem Inf Model Magazine articles , Based on AI The key contribution of protein structure prediction method to the field of drug discovery , And the limitations and challenges .

Thesis link ：

https://pubs.acs.org/doi/abs/10.1021/acs.jcim.2c00026

Abstract

Protein is the body's molecular machine , Its dysfunction often leads to diseases . therefore , Protein is the key target of drug discovery . The three-dimensional structure of protein determines its biological function , Its conformational state determines the substrate 、 The binding of cofactors to proteins . Rational drug discovery uses engineering small molecules to selectively interact with proteins to regulate their functions . To selectively target proteins and design small molecules , It is important to understand the structure of proteins and all their specific conformations . Unfortunately , For a large number of proteins related to drug discovery , Its three-dimensional structure has not been solved by experiments .

lately ,AlphaFold2, A machine learning application based on deep neural network , It can predict the unknown structure of protein with unprecedented accuracy . Even though AlphaFold2 Impressive progress has been made , But nature still challenges the field of structure prediction . In this paper , We discussed AlphaFold2 And how related methods can help improve the efficiency of drug design . We emphasize the aspects in which advanced machine learning methods need to be further improved , In order to successfully 、 Fully applied to the pharmaceutical industry .