Page 276 - 2025中醫藥與天然藥物聯合學術研討會-中醫藥與天然藥物的挑戰Ｘ機遇與未來大會手冊

CM-09

Machine learning-powered discovery and MassQL-based localization of

cyclic peptides from Leonurus japonicus fructus

,2
1
Trinh-Tac-Dat Ngo, Chung-Kuang Lu*

1 School of Pharmacy, Clinical Drug Development of Herbal Medicine, Taipei Medical
University, Taipei, 11031, Taiwan
2 Department of Life Sciences and Institute of Genome Sciences, National Yang Ming Chiao

Tung University, No. 155, Sec. 2, Li-Nong Street, Peitou, Taipei, 112304, Taiwan
* Email: cklu@nricm.edu.tw

Abstract
Discovery of bioactive natural products remains a significant challenge due to compound
redundancy and complex mixtures in crude extracts. In this study, we developed a pipeline to
target the potential bioactive compounds; the pipeline combines two high-throughput
performance tools: a machine learning (ML) binary classification model and a mass
spectrometry query language (MassQL). By this approach, we efficiently identify and localize
the bioactive cyclic peptides. At the top of the pipeline, the trained ML screening model inspired
by the pharmacoepic features of the senolytic drug was used to predict the cyclic peptides
structure associated with the senolytic effect. The computationally predicted candidate
structures then guided the experimental workflow, starting with extracting and partitioning
natural material. Then, we rapidly located the target cyclic peptides within the complex
fractions by querying distinctive iminium and scrambled ions using LC-MS/MS analysis
coupled with MassQL. Chromatographic techniques further purified and localized fractions to
isolate the predicted cyclic peptides. The fragmentation patterns in MS data aid in confirming
the isolated compound, and NMR confirmed the 2D structure. Lastly, the senolytic effect of the
isolate was validated experimentally. This integrative ML-driven MASSQL-assisted workflow
provides an efficient approach for discovering interesting compounds.

Keywords: Leonurus japonicus fructus; Cyclic peptides; Machine learning; MASSQL