PRISM：通过结构-序列多模态表示的细粒度检索增强蛋白质反向折叠
PRISM: Enhancing Protein Inverse Folding through Fine-Grained Retrieval on Structure-Sequence Multimodal Representations
AI总结：
PRISM通过结构-序列多模态表示的细粒度检索提升蛋白质反向折叠性能，结合混合自交叉注意力解码器，实现更高效的序列生成与折叠预测。
摘要：
设计能够折叠成目标三维结构的蛋白质序列，即反向折叠问题，在蛋白质工程中至关重要。然而，由于序列空间广阔且局部结构约束重要，现有深度学习方法在恢复率方面表现强劲，但缺乏显式机制重用自然蛋白质中保守的细粒度结构-序列模式。为缓解这一问题，我们提出了PRISM，一种用于反向折叠的多模态检索增强生成框架。PRISM从已知蛋白质中检索潜在motif的细粒度表示，并将其与混合自交叉注意力解码器相结合。PRISM被建模为一个潜在变量概率模型，并通过高效的近似方法实现，结合了理论基础与实际可扩展性。在多个基准测试中，包括CATH-4.2、TS50、TS500、CAMEO 2022以及PDB数据集分割，实验表明PRISM在生成最先进 perplexity 和氨基酸恢复率的同时，也提高了折叠性度量（RMSD、TM-score、pLDDT）。
英文摘要：
Designing protein sequences that fold into a target 3-D structure, termed as the inverse folding problem, is central to protein engineering. However, it remains challenging due to the vast sequence space and the importance of local structural constraints. Existing deep learning approaches achieve strong recovery rates, however, lack explicit mechanisms to reuse fine-grained structure-sequence patterns conserved across natural proteins. To mitigate this, we present PRISM a multimodal retrieval-augmented generation framework for inverse folding. PRISM retrieves fine-grained representations of potential motifs from known proteins and integrates them with a hybrid self-cross attention decoder. PRISM is formulated as a latent-variable probabilistic model and implemented with an efficient approximation, combining theoretical grounding with practical scalability. Experiments across multiple benchmarks, including CATH-4.2, TS50, TS500, CAMEO 2022, and the PDB date split, demonstrate the fine-grained multimodal retrieval efficacy of PRISM in yielding SoTA perplexity and amino acid recovery, while also improving the foldability metrics (RMSD, TM-score, pLDDT).
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。
