arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2410.16735q-bio.BM
收藏

用于膜蛋白设计的 Token 级引导离散扩散

Token-Level Guided Discrete Diffusion for Membrane Protein Design

University of Pennsylvania(宾夕法尼亚大学);Duke University(杜克大学);University of North Carolina at Chapel Hill(北卡罗来纳大学教堂山分校);Duke-NUS Medical School(杜克-新加坡国立大学医学院)

Shrey Goel, Peregrine M. Schray, Yinuo Zhang, Sophia Vincoff, Huong T. Kratochvil, Pranam Chatterjee

AI总结:

本文提出基于 RDM 的 MemDLM 蛋白质语言模型及 PET 逐 token 引导采样,实现可控膜蛋白从头设计、基序支架与性质优化,并经 TOXCAT 实验验证跨膜插入。

摘要:

重参数化扩散模型(RDM)最近在蛋白质生成方面已媲美自回归方法,这促使其被用于设计膜蛋白等具有挑战性的任务;膜蛋白具有可溶性区域与跨膜(TM)区域交错排列的特点。我们提出 Membrane Diffusion Language Model(MemDLM),这是一种经过微调、基于 RDM 的蛋白质语言模型,可实现可控的膜蛋白序列设计。MemDLM 生成的序列再现了天然膜蛋白的 TM 残基密度和结构特征,达到相当的生物合理性,并在 motif scaffolding(基序支架)任务中优于最先进的扩散基线,表现为更低的困惑度、更高的 BLOSUM-62 分数以及提升的 pLDDT 置信度。为增强可控性,我们开发了 Per-Token Guidance(PET),这是一种新颖的分类器引导采样策略,可在保留保守 TM 结构域的同时选择性地使残基可溶,从而产生 TM 密度降低但功能核心完整的序列。重要的是,在 TOXCAT β-内酰胺酶生长实验中验证的 MemDLM 设计展示出成功的 TM 插入,可区分高质量与低质量生成序列。总体而言,我们的框架建立了首个经实验验证、用于理性膜蛋白生成的扩散模型,整合了从头设计、基序支架和定向性质优化。

英文摘要:

Reparameterized diffusion models (RDMs) have recently matched autoregressive methods in protein generation, motivating their use for challenging tasks such as designing membrane proteins, which possess interleaved soluble and transmembrane (TM) regions. We introduce the Membrane Diffusion Language Model (MemDLM), a fine-tuned RDM-based protein language model that enables controllable membrane protein sequence design. MemDLM-generated sequences recapitulate the TM residue density and structural features of natural membrane proteins, achieving comparable biological plausibility and outperforming state-of-the-art diffusion baselines in motif scaffolding tasks by producing lower perplexity, higher BLOSUM-62 scores, and improved pLDDT confidence. To enhance controllability, we develop Per-Token Guidance (PET), a novel classifier-guided sampling strategy that selectively solubilizes residues while preserving conserved TM domains, yielding sequences with reduced TM density but intact functional cores. Importantly, MemDLM designs validated in TOXCAT beta-lactamase growth assays demonstrate successful TM insertion, distinguishing high-quality generated sequences from poor ones. Together, our framework establishes the first experimentally-validated diffusion-based model for rational membrane protein generation, integrating de novo design, motif scaffolding, and targeted property optimization.

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。

↑