用于生成式分子设计的 SAFE 设置
SAFE setup for generative molecular design
AI总结：
本文系统研究 SAFE 分子生成模型的训练设置，发现大规模多样化数据、随机化增强、LLaMA 架构及 BRICS 分解可带来最佳且稳健的片段约束设计性能。
摘要：
基于 SMILES 的分子生成模型在药物设计中一直发挥着关键作用，但在受片段约束的任务中面临挑战。为解决这一问题，基于顺序连接的片段嵌入（Sequential Attachment-based Fragment Embedding，SAFE）表示最近被提出，作为一种可简化这些任务的替代方案。在本研究中，我们考察了训练 SAFE 生成模型的最佳设置，重点关注数据集规模、通过随机化进行的数据增强、模型架构以及键断裂算法。我们发现，规模更大、多样性更高的数据集能够提升性能，其中采用 Rotary Positional Embedding 的 LLaMA 架构被证明最为稳健。基于 SAFE 的模型在骨架修饰和连接子设计中也持续优于基于 SMILES 的方法，尤其是在使用 BRICS 分解时取得了最佳结果。这些见解突出了显著影响基于 SAFE 的生成模型效能的关键因素。
英文摘要：
SMILES-based molecular generative models have been pivotal in drug design but face challenges in fragment-constrained tasks. To address this, the Sequential Attachment-based Fragment Embedding (SAFE) representation was recently introduced as an alternative that streamlines those tasks. In this study, we investigate the optimal setups for training SAFE generative models, focusing on dataset size, data augmentation through randomization, model architecture, and bond disconnection algorithms. We found that larger, more diverse datasets improve performance, with the LLaMA architecture using Rotary Positional Embedding proving most robust. SAFE-based models also consistently outperform SMILES-based approaches in scaffold decoration and linker design, particularly with BRICS decomposition yielding the best results. These insights highlight key factors that significantly impact the efficacy of SAFE-based generative models.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
