Loading…
A Beijing research institute pairing first-principles physics with machine learning to build AI for Science infrastructure for chemistry and biology.
The AI for Science Institute (北京科学智能研究院, AISI) is an independent research institute in Beijing, founded in 2021 by mathematician Weinan E. Its mission is to advance the frontiers of science by integrating first-principles physical modeling with state-of-the-art machine learning, with research spanning molecular simulation, materials, life sciences, and proteomics.
Contrastive dual-encoder model for DIA proteomics, embedding peptides and spectra in a shared space for zero-shot peptide-spectrum matching.
DP Technology / AI for Science Institute / Shanghai Jiao Tong University / Fudan University
Released August 4, 2025
Molecular reasoning language model for molecule captioning and text-to-molecule generation, trained by chain-of-thought distillation then reward RL.
Organic reaction foundation model that tokenizes 3D molecular structure to predict products, retrosynthetic routes, conditions, and yields.
Institute of Computing Technology, Chinese Academy of Sciences / DP Technology / University of Chinese Academy of Sciences / AI for Science Institute / State Key Laboratory of Medical Proteomics / Peking University
Released June 30, 2025
Proteomics foundation model for peptide-spectrum scoring and open de novo sequencing, reading over 1,300 modifications from tandem mass spectra.
Autoregressive 3D structure model built on an octree tokenizer, spanning molecule generation, molecular docking, and protein pocket prediction.
Westlake University / Westlake Omics / DP Technology / AI for Science Institute / Peking University
Released February 12, 2025
Neural ODE model of protein network dynamics, pretrained on 38 million perturbed protein measurements for drug efficacy and synergy prediction.
DNA methylation foundation model that encodes 5mC as a fifth base, pretrained on 568 million BS-seq reads for tissue-of-origin and expression tasks.