Uncertainty-Aware Metabolic Stability Prediction with Dual-View Contrastive Learning
Peijin Guo, Minghui Li, Hewen Pan, Bowen Chen, Yang Wu, Zikang Guo, Leo Yu Zhang, Shengshan Hu, Shengqing Hu · 2025
Read Uncertainty-Aware Metabolic Stability Prediction with Dual-View Contrastive Learning by Peijin Guo, Minghui Li, Hewen Pan, Bowen Chen, Yang Wu, Zikang Guo, Leo Yu Zhang, Shengshan Hu, Shengqing Hu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-00936.txt.
Description
Peijin Guo and co-authors predict how stable drug molecules are in the body, with honest confidence attached, in a 2025 paper. Metabolic stability decides how long a drug lasts. Graph neural networks are the current tool of choice. They pass messages between atoms but mostly ignore bond detail. They also give no sense of their own uncertainty. The team's framework, TrustworthyMS, attacks both flaws. It remaps the molecule's graph so atoms and bonds inform each other.
A contrastive step aligns two views of each molecule. One view stresses structure, the other bond patterns. Agreement between views hardens the representation. A statistical device then prices the model's doubt. It reports prediction and confidence together. Tests show the method beating current leaders. The authors frame trust as part of accuracy in drug work. A wrong but confident answer can waste a whole program. Calibrated doubt helps chemists choose what to synthesize next.
Catalog record
| CREATOR | Peijin Guo, Minghui Li, Hewen Pan, Bowen Chen, Yang Wu, Zikang Guo, Leo Yu Zhang, Shengshan Hu, Shengqing Hu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-00936