AI 中文总结
本研究开发了基于GROMACS的混合蒙特卡罗/分子动力学代码MolDStruct，经基准测试可模拟蛋白质尺度生物分子的辐射损伤与库仑爆炸动力学，为蛋白质爆炸成像提供实用工具。
AI 中文摘要
利用强X射线自由电子激光脉冲进行单粒子成像时，需要对生物分子产生的电离及库仑爆炸动力学进行建模，以优化实验参数并实现正确的结构重建，但在蛋白质尺度上模拟完整动力学超出了量子力学方法的能力范围。为解决该问题，我们开发了MolDStruct，这是一个基于GROMACS构建的混合蒙特卡罗/分子动力学代码，其通过蒙特卡罗模块模拟强X射线电离动力学，结合经典分子动力学实现原子传播。针对丙氨酸二肽的基准测试显示，当原子平均电荷$\bar{z} \approx 1.35$以上时，MolDStruct的碎裂模式与量子力学计算结果一致；与2-碘吡啶的库仑爆炸成像实验数据相比，模拟的动量分布在碎片方向上重现了实验的牛顿图，且落在实验绝对动量范围内，仅存在适度高估。我们进一步将MolDStruct应用于蛋白质爆炸成像（一种根据探测器记录的爆炸离子图对分子结构进行分类的方法），通过降维证明16残基肽$\text{Ala}_{16}$的构象体与5个生色团标记的泛素突变体可被区分。这些结果确立了MolDStruct作为蛋白质尺度生物分子辐射损伤与库仑爆炸动力学模拟实用工具的地位，而量子力学方法在该尺度下计算上不可行。
英文摘要
Single Particle Imaging with intense X-ray free-electron laser pulses requires modelling of the resulting ionisation and Coulomb explosion dynamics of biomolecules to optimise experimental parameters and enable correct structural reconstruction, yet simulating the complete dynamics at protein scale is beyond the reach of quantum-mechanical methods. To address this, we developed \moldstruct, a hybrid Monte Carlo/Molecular Dynamics code built on GROMACS that couples high intense X-ray ionisation dynamics modelled through a Monte Carlo module with classical Molecular Dynamics for atomic propagation. Benchmarked against quantum mechanical calculations for di-alanine, MolDStruct agrees in fragmentation patterns above a mean charge per atom of $\bar{z} \approx 1.35$. Compared with Coulomb explosion imaging experimental data for 2-iodopyridine, simulated momentum distributions reproduce the experimental Newton plots in fragment direction and fall within the range of experimental absolute momenta, with a moderate overestimation. We further apply MolDStruct to Protein Explosion Imaging, a method that classifies molecular structures from explosion ion maps recorded on a detector, demonstrating via dimensionality reduction that conformers of the 16-residue peptide $\mathrm{Ala}_16$ and five chromophore-labelled ubiquitin mutants are distinguishable. These results establish MolDStruct as a practical tool for simulating radiation damage and Coulomb explosion dynamics of biomolecules at protein scale, where quantum-mechanical methods are computationally infeasible.
