生成式 AI 如何助力蛋白质科学研究.pdf

1、Large-Scale Generative AI for Protein Modeling&DesignZaixiang ZhengByteDance Rhttps:/%zhengzx-nlp.github.ioYSSNLP2024YSSNLP2024YSSNLP2024 YSSNLP2024YSSNLP2024YSSNLP2024Were doing Generative AI for Science at ByteDance ResearchProteinLearning Harmonic Molecular Representations on Riemannian Manifold.

2、In ICLR 2023On Pre-training Language Model for Antibody.In ICLR 2023Structure-informed Language Models Are Protein Designers.In ICML 2023(oral)Diffusion Language Models Are Versatile Protein Learners.In ICML 2024.Protein Conformation Generation via Force-Guided SE(3)Diffusion Models.In ICML 2024.Ant

3、igen-Specific Antibody Design via Direct Energy-based Preference Optimization.preprint.2024Small MoleculeRegularized Molecular Conformation Fields.In NeurIPS 2022Zero-Shot 3D Drug Design by Sketching and Generating.In NeurIPS 2022Diffusion Models with Decomposed Priors for Structure-Based Drug Desig

4、n.In ICML 2023DecompOpt:Controllable and Decomposed Diffusion Models for Structure-based Molecular Optimization.In ICLR 2024Cryo-EMCryoSTAR:Leveraging Structural Prior and Constraints for Cryo-EM Heterogeneous Reconstruction.preprint.2023YSSNLP2024YSSNLP2024YSSNLP2024 YSSNLP2024YSSNLP2024YSSNLP2024_

5、Structure-informed Language Models Are Protein Designers.In ICML 2023(oral)LM-DESIGN:steering large protein LMs to design protein sequences as structure-conditioned sequence generative modelsiteratively refine T?clsYKTVRAGRLGSISRSLEReosclsMKTVRQERLKSIVRILEReosstructural adapter N?structure encoder(G

6、NNs,ProteinMPNN,GVP,IPA,etc.)Multihead ATTNFFNTransformer layerMultihead ATTN+FFNsequence decoder:pLM(ESM series,etc)Fstructure-based sequence design models(GNNs,ProteinMPNN,GVP,PiFold,IPA,etc.)C N?Multihead ATTN+FFNTransformer layerprotein language models(pLMs)(ESM-1b,ESM-2 series)DEUniRef-50 seque