DBMol: Design of High-Affinity, Target-Specific Small Molecules through Structure Prediction Model

Yiming QIN, Kai Yi, Miruna Cretu, Sjors HW Scheres, Pietro Lio, Pascal Frossard

Keywords: Small-molecule design, structure prediction, flow matching, protein–ligand binding

TL;DR: DBMol is a docking-dataset-free framework that uses structure prediction models to guide small-molecule design, achieving strong affinity and pocket specificity without reinforcement learning or docking supervision.

Abstract:

Designing small molecule ligands that bind with high affinity to specific protein pockets is a fundamental goal in drug discovery. Recent breakthroughs in structure prediction, such as AlphaFold-3 and Boltz-2, enable accurate biomolecular interaction prediction and show promise as foundation models for downstream tasks, including binding-affinity prediction. We propose to leverage these models and introduce DBMol, a new structure-predictor-guided framework for de novo small molecule design.
By leveraging broad protein knowledge from foundation structure models, DBMol targets a challenging setting where the protein pocket is given but no known ligand is available.
DBMol formulates small molecule generation as an alternating optimization and projection process. In the optimization stage, DBMol starts from an initial molecule and uses gradient-based optimization to improve pocket-specific interactions and predicted binding affinity using a structure prediction model. In the projection stage, a flow-matching model maps the optimized molecular graph to discrete and chemically valid molecules.
Optionally, a post-processing step produces synthetically accessible molecules.
Experiments show that DBMol effectively optimizes the Boltz-2 proxy and generates molecules with strong predicted affinity and specificity under Boltz-2 evaluation.
To reduce self-confirmation bias, we further evaluate generated molecules using held-out metrics, including AF3-based evaluation.
DBMol substantially improves over unconditional generation and achieves competitive performance against stronger baselines under these held-out metrics.
These results suggest that structure prediction models can provide useful optimization signals for de novo molecular design, especially in low-supervision settings where reference ligands are unavailable.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 47