Construction of Chinese baijiu compound database using text mining and its application in assisting compound identification of liquid chromatography-high-resolution mass spectrometry data

Comprehensive understanding of the molecular composition in Chinese baijiu is of a great value for understanding the flavor and improving the quality. Liquid chromatography-mass spectrometry (LC-MS) is a good tool for the analysis of Chinese baijiu. However, lack of the related database makes corresponding LC-MS data difficult to be annotated. Therefore, a Chinese baijiu compound database (CBDB) containing 5709 compounds was constructed based on literatures and text mining, and further a Chinese baijiu LC-MS database (LCMS-CBDB) was built by endowing each compound with retention time, exact mass, and MS/MS spectrum. In total 468 compounds were identified in six Chinese baijiu samples by searching LCMS-CBDB. Further, database enhancement strategies were proposed based on biological transformations and theoretical homologue expansion, which allows the identification of new compounds. Four monoglycerides were successfully identified and validated using the extended CBDB, three of them were reported in baijiu for the first time.全面解析中国白酒的分子组成，对于阐释其风味特质与提升产品品质具有重要研究价值。液相色谱-质谱联用法（LC-MS）是分析中国白酒的常用技术手段。然而，相关专用数据库的缺失使得对应LC-MS检测数据难以实现有效注释。为此，本研究基于文献调研与文本挖掘，构建了包含5709种化合物的中国白酒化合物数据库（CBDB）；进一步通过为每种化合物赋予保留时间、精确质量数与二级质谱（MS/MS）谱图信息，搭建了中国白酒LC-MS数据库（LCMS-CBDB）。通过检索LCMS-CBDB，研究人员在6份中国白酒样品中共计鉴定出468种化合物。此外，本研究基于生物转化与理论同系物拓展提出了数据库增强策略，可实现新化合物的高效鉴定。利用拓展后的CBDB，研究人员成功鉴定并验证了4种单甘油酯，其中3种为首次在白酒中被报道。