Related Experiment Video
Updated: Jun 16, 2026
Drug Repurposing Hypothesis Generation Using the "RE:fine Drugs" System
Published on: December 11, 2016
Machine Learning-Driven Drug Repurposing for KRAS G12C and KRAS G12D Inhibition
Gianluca Fuschi1, Julia St Germain1, David Bebensee1
1Department of Sciences, University College Groningen, University of Groningen, Groningen 9718 BG, The Netherlands.
Abstract:
KRAS is a predominant oncogenic driver across multiple cancers and was long considered undruggable due to its high nucleotide affinity and lack of classical binding pockets. Although recent advances have led to covalent inhibitors such as Sotorasib and Adagrasib for the KRAS G12C mutation, effective therapies for other common variants, particularly KRAS G12D, which is highly prevalent in aggressive pancreatic cancers, remain limited. In this study, we employed machine learning approaches to identify potential inhibitors of KRAS G12D and G12C by screening FDA-approved compounds curated from the ChEMBL database. Random Forest and Neural Network models were trained on bioactivity data from three BindingDB data sets: wild-type KRAS GTPase, KRAS G12C, and KRAS G12D. The trained models demonstrated strong predictive performance, achieving high correlation coefficients on independent test sets. To further validate the predictive capability of the models, two compounds identified as high-confidence candidates, Cobimetinib and Etrasimod, were selected for experimental evaluation. In vitro testing revealed measurable IC50 and E max values, with both compounds exhibiting preferential activity in KRAS G12D cellular backgrounds. While additional biochemical and pathway-level studies are required to confirm direct target engagement, these results support the model's utility in prioritizing candidate compounds with allele-specific activity profiles. Overall, this study provides a data-driven framework for identifying potential KRAS-targeted therapies and highlights the value of integrating machine learning predictions with experimental validation.
Insights
Machine learning identified potential KRAS G12D inhibitors among FDA-approved drugs. Experimental validation showed promising allele-specific activity, offering a new framework for developing KRAS-targeted cancer therapies.
Area of Science:
- Oncology
- Computational Biology
- Pharmacology
Background:
- KRAS mutations drive numerous cancers and have historically been difficult to target therapeutically.
- While KRAS G12C inhibitors exist, effective treatments for other variants like KRAS G12D, common in pancreatic cancer, are limited.
Purpose of the Study:
- To leverage machine learning to screen FDA-approved compounds for potential KRAS G12D and KRAS G12C inhibitors.
- To establish a data-driven framework for identifying novel KRAS-targeted therapies with allele-specific activity.
Main Methods:
- Machine learning models (Random Forest, Neural Network) were trained on KRAS bioactivity data from BindingDB.
- FDA-approved compounds from the ChEMBL database were screened using the trained predictive models.
- High-confidence candidate compounds were selected for in vitro experimental validation.
Main Results:
- The machine learning models demonstrated strong predictive performance on independent test sets.
- In vitro testing of Cobimetinib and Etrasimod showed measurable activity, with preferential efficacy in KRAS G12D cellular backgrounds.
- The study identified a data-driven approach for prioritizing compounds with allele-specific KRAS activity.
Conclusions:
- Machine learning, combined with experimental validation, provides a viable strategy for discovering targeted KRAS inhibitors.
- The identified compounds warrant further investigation for their potential as KRAS G12D-targeted therapies.
- This approach advances the development of precision oncology treatments for KRAS-mutated cancers.
