A new in silico method is introduced for the dereplication of natural metabolite mixtures based on HMBC and HSQC spectra that inform about short-range and long-range H–C correlations occurring in the carbon skeleton of individual chemical entities. Starting from the HMBC spectrum of a metabolite mixture, an algorithm was developed in order to recover individualized HMBC footprints of the mixture constituents. The collected H–C correlations are represented by a network of NMR peaks connected to each other when sharing either a ¹H or ¹³C chemical shift value. The network obtained is then divided into clusters using a community detection algorithm, and finally each cluster is tentatively assigned to a molecular structure by means of a NMR chemical shift database containing the theoretical HMBC and HSQC correlation data of a range of natural metabolites. The proof of principle of this method is demonstrated on a model mixture of 3 known natural compounds and then on a real-life bark extract obtained from the common spruce (Picea abies L.).

中文翻译：

---

HMBC相关网络的重建：基于新的基于NMR的代谢物混合物分析贡献

引入了一种新的计算机模拟方法，用于基于HMBC和HSQC光谱图谱的自然代谢物混合物的重复计算，该图谱揭示了各个化学实体碳骨架中发生的短程和长程H–C相关性。从代谢物混合物的HMBC光谱开始，开发了一种算法，以恢复混合物成分的个性化HMBC足迹。当共享¹ H或¹³ H时，所收集的H–C相关性由相互连接的NMR峰网络表示C化学位移值。然后使用社区检测算法将获得的网络划分为多个簇，最后，通过一个NMR化学位移数据库，将每个簇暂时分配给分子结构，该数据库包含一系列天然代谢物的理论HMBC和HSQC相关数据。在3种已知天然化合物的模型混合物上，然后在从常见云杉（Picea abies L.）获得的真实树皮提取物中，证明了该方法的原理。