Division of Mathematical Sciences, School of Physical and Mathematical Sciences, Nanyang Technological University, 637371, Singapore.
Phys Chem Chem Phys. 2019 Feb 20;21(8):4359-4366. doi: 10.1039/c8cp07442a.
Understanding the molecular flexibility and dynamics is central to the analysis of biomolecular functions. In this work, complex multiscale virtual particle model based elastic network models (CMVP-ENMs) have been proposed for the normal mode analysis of biomolecular complexes or biomolecular assemblies. The term complex used in our CMVP-ENMs refers to the multi-material or multi-constituent. Different "materials" or constituents contribute differently to the general flexibility and dynamics of complex biomolecular structures. In our CMVP-ENMs, the key idea is to incorporate relative density or weight information of different components into the spring parameter of elastic network models. Two different models, including the CMVP based Gaussian network model (CMVP-GNM) and the CMVP based anisotropic network model (CMVP-ANM), have been proposed. With the consideration of complex component information, our CMVP-GNM, compared with the traditional GNM, can deliver a better accuracy in the B-factor prediction of protein-nucleic acid complexes. Moreover, our CMVP-ANM can be used to remove the "tip effect" by systematically suppressing the extremely-large vectors, in the highly flexible regions, of the normal modes generated by the ANM. In this way, our CMVP-ANM can be used to handle biomolecular structures with large hanging loops or extruding ends, which usually cause an irrationally-large-vector problem in ANM predictions. Finally, we explore the potential applications of our method by the cryo-EM data analysis. We find that by tuning the relative density ratio, we can systematically enhance or suppress the modes in different components, so that it can reveal the dynamics of the special regions that we are interested in.
理解分子的柔韧性和动态性是分析生物分子功能的核心。在这项工作中，提出了基于复杂多尺度虚拟粒子的弹性网络模型（CMVP-ENMs），用于生物分子复合物或生物分子组装的正常模式分析。我们的 CMVP-ENMs 中的“复杂”一词是指多材料或多成分。不同的“材料”或成分对复杂生物分子结构的整体柔韧性和动态性有不同的贡献。在我们的 CMVP-ENMs 中，关键思想是将不同成分的相对密度或权重信息纳入弹性网络模型的弹簧参数中。提出了两种不同的模型，包括基于 CMVP 的高斯网络模型（CMVP-GNM）和基于 CMVP 的各向异性网络模型（CMVP-ANM）。考虑到复杂的成分信息，与传统的 GNM 相比，我们的 CMVP-GNM 可以在蛋白质-核酸复合物的 B 因子预测中提供更高的准确性。此外，我们的 CMVP-ANM 可以通过系统地抑制各向异性网络模型生成的正常模式中高度灵活区域的极大数据向量，来消除“尖端效应”。通过这种方式，我们的 CMVP-ANM 可以用于处理具有大悬垂环或伸出末端的生物分子结构，这些结构通常会导致各向异性网络模型预测中的不合理大向量问题。最后，我们通过冷冻电镜数据分析探索了我们方法的潜在应用。我们发现，通过调整相对密度比，我们可以系统地增强或抑制不同成分中的模式，从而可以揭示我们感兴趣的特殊区域的动力学。
