Evaluating GPCR modeling and docking strategies in the era of deep learning-based protein structure prediction
Problem
It was unclear whether deep learning-based protein structure predictions could actually improve docking and virtual screening for GPCR drug discovery compared to pre-deep-learning tools.
Why it matters
GPCRs are key drug targets, and better computational docking could speed up the discovery of new medicines.
What the AI did
The study benchmarked modeling and docking strategies on 70 diverse GPCR complexes, showing that deep learning-based predicted structures substantially improve docking and virtual screening outcomes.
Results
Docking success on DL-based model structures approaches cross-docking on experimental structures, with over 30% improvement over the best pre-DL protocols.
Impact
Evidence supports that DL structure predictions materially improve GPCR docking and virtual screening performance.
Record
| Venue | Computational and Structural Biotechnology Journal |
|---|---|
| Published | 2022 |
| Work type | article |
| Authors | Sumin Lee, Seeun Kim, Gyu Rie Lee, Sohee Kwon, Hyeonuk Woo, Chaok Seok, Hahnbeom Park |
| Citations | 42 (OpenAlex, at import time) |
| References | 48 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1016/j.csbj.2022.11.057 |
| Open access | gold |
| Keywords | GPCR, docking, virtual screening, deep learning, protein structure prediction |
Importance rating
significant — thousands to millions affected
GPCRs are key drug targets; DL-based structure predictions substantially improve docking and virtual screening, potentially speeding drug discovery.