Published October 26, 2026 | Version v1

基于混合计算流水线的极速多肽构象集合生成:MDM2-p53 界面的基准验证

Authors/Creators

  • 1. 独立研究者

Description

多肽药物发现通常受限于分子动力学(MD)模拟在构象采样时高昂的算力成本。本文提出了一种未经过深度算法优化的混合计算流水线。通过融合拓扑状态空间中的极速启发式搜索策略与高保真全原子三维几何还原算法,该引擎仅使用 CPU 便在数秒内生成了高亲和力的多肽构象。我们在经典的 MDM2-p53 靶点(PDB: 1YCR)上验证了其基准性能。从完全随机的初始状态出发,在经过标准的全柔性对接优化后,其生成的最佳构象对接打分为 -20.15 kcal/mol,极度逼近真实晶体结构的金标准 -22.92 kcal/mol(差距仅 2.77 kcal/mol)。这一基准验证凸显了该流水线作为多肽药物研发超高速前置过滤器的巨大潜力。

Peptide drug discovery is often bottlenecked by the immense computational cost of Molecular Dynamics (MD) simulations required to sample flexible conformations. Here, we present an unoptimized hybrid computational pipeline. By integrating a rapid heuristic search strategy in topological state space with a high-fidelity all-atom 3D geometric reconstruction algorithm, our engine generates high-affinity peptide conformers within seconds using only CPU. We demonstrate its baseline performance on the classic MDM2-p53 target (PDB: 1YCR). From a completely random initial state, the best generated conformer achieved a docking score of -20.15 kcal/mol, closely approaching the crystallographic ground truth of -22.92 kcal/mol (a gap of 2.77 kcal/mol). This baseline validation highlights the viability of this pipeline as an ultra-fast pre-filter for peptide drug discovery.

Files

paper_zh_safe.pdf

Files (78.2 kB)

Name Size
md5:faa79e3aaa5e4f6a422320acf7d8cdf9
78.2 kB Preview Download

Additional details

Additional titles

Translated title (English)
Fast Peptide Ensemble Generation via a Hybrid Computational Pipeline: A Baseline Validation on the MDM2-p53 Interface