Google's Isomorphic Labs unveiled IsoDDE, dubbed AlphaFold 4, which dramatically outperforms AlphaFold 3 on hard protein‑structure benchmarks and antibody‑binding predictions, yet the model is fully closed‑source, sparking a debate about the future of open scientific AI.
This article reviews the evolution from AlphaFold 2 to AlphaFold 3, introduces Baidu's HelixFold 3 as the first domestic model matching AlphaFold 3, compares their benchmark results on small‑molecule ligands, nucleic acids and protein complexes, and explains the cloud‑based service and confidence scoring that make high‑throughput structure prediction accessible.
HelixFold‑Multimer, a new Baidu PaddleHelix model, outperforms AlphaFold 3 on antigen‑antibody and peptide‑protein complex predictions, achieving mean DockQ scores of 0.41 and 0.38 respectively and success rates up to 77 % when epitope data are used, and is already deployed in large‑molecule drug pipelines.
This talk presents how large language models are leveraged for predicting protein, antibody, and RNA structures, covering background, model stability, generative approaches, antibody-specific models, RNA modeling, and protein‑RNA interaction prediction, along with experimental results and future research directions.
The article introduces the MEGA-Protein tool, which integrates an AI‑driven MSA engine and the MindSpore framework to overcome AlphaFold 2’s limitations and achieve high‑accuracy protein structure prediction, and also announces the MSG Enterprise Tour and Hangzhou Developer Day event on April 27.
