Wieners Luc, Garcia Martin E
Institute of Physics, University of Kassel, Heinrich-Plett-Straße 40, 34132 Kassel, Germany.
Res Sq. 2025 Sep 16:rs.3.rs-7327472. doi: 10.21203/rs.3.rs-7327472/v1.
The behaviour of proteins and other biomolecules is mainly governed by the quantum-mechanical character of their electrons. Accurately capturing the resulting interactions is essential for predicting molecular properties, obtaining spectroscopic data, and advancing drug design. However, the extreme computational cost of quantum calculations has historically limited their application to small systems of just a few hundred atoms. Here, we present a quantum-mechanical method that enables electronic structure calculations on biological systems at unprecedented scales, up to millions of atoms, while drastically reducing computational costs. We apply this approach to entire proteins and large biomolecular assemblies, including a complete bacteriophage in a solution containing over 150 million electrons. Additionally, we show that atomic energies computed for AlphaFold-predicted protein structures strongly correlate with AlphaFold's confidence scores, providing a new quantum-based validation metric. The method's efficiency also allows the accurate prediction of spectroscopic properties for biomolecules previously out of reach for first-principles techniques. We present computed spectra for DNA and the anticancer drug Actinomycin, involving hundreds to thousands of atoms, in close agreement with experimental measurements. This advance bridges quantum mechanics and biology at a previously inaccessible scale, enabling large-scale, first-principles simulations with broad applications in quantum biology, structural biology, medicine, and materials science.
蛋白质和其他生物分子的行为主要由其电子的量子力学特性决定。准确捕捉由此产生的相互作用对于预测分子性质、获取光谱数据以及推进药物设计至关重要。然而，量子计算极高的计算成本在历史上一直限制了其应用范围，仅适用于由几百个原子组成的小系统。在此，我们提出一种量子力学方法，该方法能够以前所未有的规模对生物系统进行电子结构计算，规模可达数百万个原子，同时大幅降低计算成本。我们将这种方法应用于完整的蛋白质和大型生物分子组装体，包括在含有超过1.5亿个电子的溶液中的一个完整噬菌体。此外，我们表明，为AlphaFold预测的蛋白质结构计算出的原子能量与AlphaFold的置信度得分高度相关，提供了一种基于量子的新验证指标。该方法的效率还使得能够准确预测此前第一性原理技术无法企及的生物分子的光谱性质。我们展示了DNA和抗癌药物放线菌素的计算光谱，涉及数百到数千个原子，与实验测量结果高度吻合。这一进展在以前无法达到的规模上搭建了量子力学与生物学之间的桥梁，实现了具有广泛应用的大规模第一性原理模拟，涵盖量子生物学、结构生物学、医学和材料科学等领域。
