Structural input
X-ray, NMR, cryo-EM, and public fragment data provide the starting points for target-aware design.
Target-based drug design
AI for molecular design
FragmentScreen connects fragment-based screening with AI-supported fragment-to-lead optimisation. This repository is the open computational side: train target-specific screeners, generate and decorate molecular scaffolds, evaluate properties, and plan synthesis routes.
FragmentScreen
FragmentScreen develops instrumentation, workflows, and experimental and computational methods for fragment-based drug discovery. Work package 6 focuses on AI-supported fragment-to-lead optimisation, bridging structural biology, medicinal chemistry, and generative AI.
AI design
Chemical language models combine motifs, grow scaffolds, and use Regression Transformer models to decorate promising analogues.
DMTA loop
Designed molecules move through synthesis planning, experimental validation, and model refinement in a design-make-test-analyse cycle.
Project links
The FragmentScreen WP6 blog describes the FFUL and IBM Research work on AI-supported design for SARS-CoV-2 targets, including motif-guided generation, Regression Transformer optimisation, RXN-assisted synthesis planning, and wet-lab validation.