Related Experiment Video
Updated: May 28, 2026

06:26
Nano-Differential Scanning Fluorimetry for Screening in Fragment-based Lead Discovery
Published on: May 16, 2021
A Molecular-Protein Fusion Framework for Rapid Virtual Screening: Accelerating Lead Discovery for "Undruggable''
Chenxi Zhou1, Yanni Zhu2, Chenrui Yang3
1College of Biotechnology and Pharmaceutical Engineering, Nanjing Tech University, 30 South Puzhu Road, Jiangbei New District, Nanjing 211816, China.
Pharmaceuticals (Basel, Switzerland)
|May 27, 2026
Summary
A new machine learning framework, MPFF-IS, efficiently identifies potential KRAS G12D inhibitors for pancreatic cancer. This computational approach accelerates drug discovery for challenging oncogenic targets.
Area of Science:
- Computational chemistry
- Oncology
- Drug discovery
Background:
- KRAS G12D mutations are prevalent in pancreatic ductal adenocarcinoma (PDAC).
- Targeting KRAS G12D is difficult due to limited druggable pockets.
- Novel therapeutic strategies are urgently needed for PDAC.
Purpose of the Study:
- To develop a machine learning (ML) framework for rapid virtual screening of KRAS G12D inhibitors.
- To identify novel drug candidates targeting KRAS G12D mutations.
- To establish a computational tool for precision drug discovery.
Main Methods:
- Developed MPFF-IS, a molecular-protein fusion prediction framework integrating ESM2 and MPNN-GNN.
- Trained the model on a KRAS G12D inhibitor dataset and screened over 134,000 compounds.
- Validated top candidates using AutoDock Vina docking and GROMACS molecular dynamics simulations.
Main Results:
- MPFF-IS identified 2663 candidate KRAS G12D inhibitors with high predictive performance.
- Several identified compounds showed favorable binding affinity and stable interactions.
- Molecular dynamics confirmed the stability of predicted complexes and key binding residue interactions.
Conclusions:
- MPFF-IS effectively identifies potential KRAS G12D inhibitors.
- The framework offers a valuable computational tool for targeting difficult oncogenic proteins.
- This approach advances precision medicine for KRAS G12D-driven cancers.

