AI for GPCR Drug Discovery – Advantages and Limitations
Speaker
Time
Venue
Online
Tencent
Abstract
Despite the tremendous success of crystallography, Cryo-EM and NMR methods, G protein-coupled receptor (GPCR) structures are still extremely difficult to be experimentally determined owning to the challenges in low expression level and conformation flexibilities. Recently, the artificial intelligence and deep learning methods led breakthrough in protein structure predictions, such as AlphaFold2 and RoseTTAFold, dramatically advanced the field of structural biology. In this study, via the comparative analysis between the predicted and experimentally determined structures, the structure prediction methods are evaluated for their application and limitations in GPCR research. Our results show that the overall folding prediction on GPCR structure is reliable, even for structureless GPCR sub-families, while the sidechains are less accurate, which will limit its application in elucidating function and structure-based drug discovery of GPCRs. This study is intended to probe the boundary of structure prediction methods in GPCR research and provide guidance for its future development.
Bio
刘志杰教授现任上海科技大学iHuman研究所执行所长，大道书院院长，国际欧亚科学院院士。曾任中国科学院生物物理研究所“百人计划”研究员，主要研究方向为人类重大疾病相关GPCR结构与功能研究和药物设计。已在Cell, Nature, Science, Immunity, Nat Stru Mole Biol, Cell Research, PNAS等SCI期刊发表研究论文160余篇。于2013年调入上海科技大学与Ray Stevens教授共同创建iHuman研究所。获得上海市自然科学奖一等奖、上海市领军人才、药明康德生命化学研究奖、上海市劳动模范、上海市科技精英提名奖、三次获中国科学院优秀导师奖。现任中国生物物理学会副理事长，中国晶体学会常务理事，上海生物物理学会副理事长，国际生物结晶组织（IOBCr）理事、亚洲晶体学会理事，国际专业杂志JMB及IUCr J 编委。
