2024. 08.28 (수) ~ 2024. 08.30 (금)
군산새만금컨벤션센터(GSCO)
| 2024 한국질량분석학회 여름학술대회 및 총회 Brief Oral Presentaionof Selected Posters | |
제목 | Development of the screening software using machine learning models, hybrid similarity searches, and molecular networking for identifying unknown narcotic analogues. |
---|---|
작성자 | 이소연 (서강대학교) |
발표구분 | 포스터발표 |
발표분야 | 5. Life & Informatics |
발표자 |
이소연 (서강대학교) |
주저자 | 이소연 (서강대학교) |
교신저자 |
오한빈 (서강대학교) |
저자 |
이소연 (서강대학교) 오한빈 (서강대학교) |
A software integrated machine learning, a hybrid similarity search (HSS) algorithm, and molecular networking was developed. Although comparing liquid chromatography tandem mass spectrometry (LC-MS/MS) data with a pre-built LC MS/MS database is common strategy for screening unknown narcotic drugs, it faces limitations due to the recent increase in the number of synthesized structural analogues that does not affect the nature of drugs to avoid regulation. To effectively identify synthesized narcotic analogues, we generated AI-SNPS (AI-Screener of narcotic drugs and psychotropic substance) version 2, which consists of four layers: LC/MS Viewer, AI Classifier, Identifier, and MN Constructor. The model to be used in AI Classifier layer is currently under development. The Identifier layer uses pick-count scoring, simple and hybrid similarity searches to identify unknown narcotics. The MN Constructor aids in identifying structurally similar compounds within complex mixtures, thereby facilitating structural predictions. Future enhancements will focus on expanding the database to refine the identification of narcotic analogues and applying the system to a broader range of synthesized structural analogues. We believe that the developed AI-SNPS version 2 will serve as a powerful platform for effectively screening unknown narcotic drugs and their analogues.
Acknowledgements This research was supported by R&D funding from the Ministry of Food and Drug Safety, Republic of Korea (RS2024-0033209). |