CReM-opt
Docking-guided evolutionary molecular optimization using chemically constrained CReM fragment replacements.
Scientific problem
Structure-based optimization must efficiently explore a large chemical space while keeping generated molecules chemically reasonable and improving a chosen optimization objective.
Motivation
CReM-opt was developed for focused exploration around promising molecules, combining CReM transformations with docking and user-defined structural or physicochemical constraints.
My role
- Developed and applied the molecular generation, evolutionary optimization, docking, and prioritization workflow.
- Added atom-wise docking-score contribution to support interpretation and guidance of optimization decisions.
- Applied CReM-opt in the prospective CACHE Challenge #1 workflow.
Methodology
- CReM fragment replacement coupled to an evolutionary population-based search.
- Docking through EasyDock, using Vina-family methods including Gnina.
- Optional physicochemical filters, scaffold protection, pose-RMSD constraints, and protein–ligand interaction similarity.
Results
- Used prospectively as part of the team's CACHE Challenge #1 hit-finding workflow.
- Presented at RDKit UGM 2025
- Manuscript in preparation.
Software
- Python
- RDKit
- CReM
- EasyDock
- AutoDock Vina
- Gnina
- pandas
- NumPy