发表机构
National University of Singapore; Tencent AI for Life Science Lab（新加坡国立大学; 腾讯生命科学人工智能实验室）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对小分子、晶体和蛋白质跨域生成管道分散问题，提出单架构流匹配自动编码器SinAE，用普通Transformer编码器和解码器，将重建负担转移到迭代流匹配解码器，实现跨域近无损重建，在生成基准测试中表现出色。
AI 中文摘要
小分子、晶体和蛋白质在三维空间中都可简化为原子，但其生成管道在各领域仍分散，各有自己的架构。跨域训练可缓解数据稀缺，但在三维坐标空间直接生成难以处理所有三个领域的异构结构先验，且之前没有潜在自动编码器能同时无损且通用。我们引入SinAE，一种用于分子、晶体和蛋白质的单架构流匹配自动编码器，用普通Transformer编码器和解码器，无特定领域算子。SinAE将重建负担转移到迭代流匹配解码器，实现跨域近无损重建，相对于之前的潜在基线，重建误差降低几个数量级。相同的 per-token 潜在支持标准扩散Transformer先验，在分子、晶体和蛋白质生成基准测试中表现出色。联合分子 - 晶体训练严格改善了两个领域，通过共享原子潜在提供了跨域转移的直接证据。代码可在指定网址获取。
英文摘要
Small molecules, crystals, and proteins all reduce to atoms in 3D space, yet their generative pipelines remain fragmented across domains, each with its Small molecules, crystals, and proteins all reduce to atoms in 3D space, yet their generative pipelines remain fragmented across domains, each with its own graph, equivariant, or frame-based architecture. Cross-domain training would mitigate per-domain data scarcity, but direct generation in 3D coordinate space cannot easily handle the heterogeneous structural priors of all three domains, and no prior latent autoencoder is simultaneously lossless and architecturally general across all three. We introduce SinAE, a single-architecture flow-matching autoencoder for molecules, crystals, and proteins, with vanilla Transformer encoder and decoder and no equivariant, graph, or domain-specific operators. Rather than requiring the encoder to capture fine-grained geometry, SinAE shifts the reconstruction burden into an iterative flow-matching decoder, achieving near-lossless reconstruction across domains and reducing reconstruction errors by orders of magnitude relative to prior latent baselines. The same per-token latent supports a standard Diffusion Transformer prior that reaches strong performance on molecular, crystal, and protein generation benchmarks. Joint molecule--crystal training strictly improves both domains, providing direct evidence of cross-domain transfer through a shared atomic latent. Code is available at https://github.com/BlueWhaleLab/SinAE .
Commentsconference
