Skolkovo Institute of Science and Technology Novaya St., 100, Skolkovo 143025, Russian Federation.
Department of Analytical and Forensic Medical Toxicology, Sechenov University, 8-2 Trubetskaya St., Moscow 119048, Russian Federation.
Anal Chem. 2020 May 19;92(10):6877-6885. doi: 10.1021/acs.analchem.9b05379. Epub 2020 May 6.
Accurate and reliable identification of chemical compounds is the ultimate goal of mass spectrometry analyses. Currently, identification of compounds is usually based on the measurement of the accurate mass and fragmentation spectrum, chromatographic elution time, and collisional cross section. Unfortunately, despite the growth of databases of experimentally measured MS/MS spectra (such as MzCloud and Metlin) and developing software for predicting MS/MS fragments in silico from SMILES patterns (such as MetFrag, CFM-ID, and Ms-Finder), the problem of identification is still unsolved. The major issue is that the elution time and fragmentation spectra depend considerably on the equipment used and are not the same for different LC-MS systems. It means that any additional descriptors depending only on the structure of the chemical compound will be of big help for LC-MS/MS-based omics. Our approach is based on the characterization of compounds by the number of labile hydrogen and oxygen atoms in the molecule, which can be measured using hydrogen/deuterium and O/O-exchange approaches. The number of labile atoms (those from -OH, -NH, ═O, and -COOH groups) can be predicted from SMILES patterns and serves as an additional structural descriptor when performing a database search. In addition, distribution of isotope labels among MS/MS fragments can be roughly predicted by software such as MetFrag or CFM-ID. Here, we present an approach utilizing the selection of structural candidates from a database on the basis of the number of functional groups and analysis of isotope labels distribution among fragments. It was found that our approach allows reduction of the search space by a factor of 10 and considerably increases the reliability of the compound identification.
准确可靠地识别化合物是质谱分析的最终目标。目前，化合物的鉴定通常基于精确质量和碎片谱、色谱洗脱时间和碰撞截面的测量。不幸的是，尽管实验测量的 MS/MS 谱数据库（如 MzCloud 和 Metlin）不断增长，并且开发了用于从 SMILES 模式预测 MS/MS 碎片的软件（如 MetFrag、CFM-ID 和 Ms-Finder），但鉴定问题仍然没有解决。主要问题是洗脱时间和碎片谱在很大程度上取决于所使用的设备，并且在不同的 LC-MS 系统中并不相同。这意味着任何仅依赖于化合物结构的附加描述符对于基于 LC-MS/MS 的组学都将有很大帮助。我们的方法基于通过分子中不稳定的氢和氧原子的数量来表征化合物，这些原子可以使用氢/氘和 O/O 交换方法来测量。不稳定原子（来自-OH、-NH、═O 和-COOH 基团的原子）的数量可以从 SMILES 模式预测，并在进行数据库搜索时作为附加结构描述符。此外，同位素标签在 MS/MS 碎片中的分布可以通过 MetFrag 或 CFM-ID 等软件进行大致预测。在这里，我们提出了一种利用数据库中基于官能团数量的结构候选物选择和碎片中同位素标签分布分析的方法。结果表明，我们的方法可以将搜索空间缩小 10 倍，并大大提高化合物鉴定的可靠性。
