Use MolSnapper Online
Commercially Available MolSnapper No-Code Web Server
MolSnapper: Pharmacophore-Conditioned Diffusion for Structure-Based Drug Design and De Novo Ligand Generation
MolSnapper is a tool that conditions diffusion models for structure-based drug design with 3D pharmacophores, letting chemists inject expert knowledge into molecule generation. It was developed by Yael Ziv, Fergus Imrie, Brian Marsden and Charlotte M. Deane and published in the Journal of Chemical Information and Modeling in 2025. Code is available on GitHub.
Generative models can propose new molecules, but making them bind a specific pocket remains hard. Autoregressive generators accumulate errors and impose an artificial atom ordering. Diffusion models avoid this, yet synthesizability, chemical validity and the lack of a way to include expert knowledge, such as known key interactions, still limit practical use.
Property | Detail |
|---|---|
Task | Structure-based ligand generation in a protein pocket under 3D pharmacophore constraints |
Constraint types | Hydrogen bond donor and acceptor pharmacophores with positions |
Base model | MolDiff, an unbound-molecule diffusion model |
Benchmarks | CrossDocked2020 and Binding MOAD |
Compared with | SILVR and DiffSBDD |
How MolSnapper Conditions Diffusion on 3D Pharmacophores
Define pharmacophores: choose 3D positions and types, such as hydrogen bond donors or acceptors, from a known ligand or from expert knowledge.
Guide atom placement: during the reverse diffusion process, the gradient of a loss pulls atoms of matching type toward the fixed pharmacophore positions, strongly in the final steps.
Respect the pocket: protein atom positions are included in the conditioning so molecules fit the binding site.
Filter and rank: generated molecules can be checked with PoseBusters, drug-likeness and synthetic accessibility filters.
MolSnapper Results on CrossDocked and Binding MOAD
On CrossDocked, MolSnapper was compared with MolDiff without conditioning and with SILVR, another conditioning method.
Metric (CrossDocked) | SILVR | MolSnapper |
|---|---|---|
PoseBusters pass rate | 27% | 58% |
SCRDKit similarity, Top 1 | 0.586 | 0.721 |
Interaction similarity, Top 1 | 0.493 | 0.746 |
Success rate | 7.05% | 45.00% |
Overall, the authors report about twice as many valid molecules as alternative methods, and up to a 20 percent improvement in shape and color similarity to reference ligands. MolSnapper also needs no pocket-specific training set of complexes, unlike DiffSBDD, although DiffSBDD often gives more favorable docking scores.
Case Studies
In a metallo-beta-lactamase inhibitor study, 5000 molecules were generated from biapenem-derived pharmacophores; after filtering, 1139 unique molecules remained, and 55.5 percent matched or exceeded the interactions of a published compound. A second case showed scaffold hopping, modifying and expanding existing scaffolds.
What is Tamarind Bio?
Tamarind Bio is a no-code bioinformatics platform built to give life scientists and researchers access to powerful computational tools. Many cutting-edge machine learning models are hard to deploy and use. Tamarind provides an intuitive, web-based environment that removes the complexity of high-performance computing, software dependencies and command-line interfaces.
The platform is designed for biologists, chemists and other researchers who may not have a background in programming or cloud infrastructure but want to run models on their own data. Key features include:
A user-friendly graphical interface for setting up and launching experiments
A robust API for integration into existing research pipelines
An automated system for managing and scaling computational resources
Tamarind treats information and data security as a top priority, as detailed in its Trust Center and Terms of Service.
Accelerating Drug Discovery with MolSnapper on Tamarind Bio
Pharmacophore-guided de novo design: generate ligands that keep key hydrogen bonds seen in known inhibitors.
Scaffold hopping: modify and expand existing scaffolds while preserving desired interactions.
Hit expansion: propose new molecules fitting a pocket and a defined interaction pattern.
Filtered candidate sets: generate thousands of molecules and screen them with drug-likeness and PoseBusters checks.
How to Use MolSnapper on Tamarind Bio
Log in and open the tool: sign in at tamarind.bio and select MolSnapper.
Provide the protein: upload the target structure with the binding site defined.
Specify pharmacophores: provide 3D positions and types of donor or acceptor features, for example from a reference ligand (if exposed).
Set generation options: choose the number of molecules and other parameters the tool exposes.
Run the job: submit generation.
Download and filter: retrieve the molecules and filter by PoseBusters validity, QED, synthetic accessibility and interaction similarity.
Use downstream: redock or minimize the top candidates and prioritize them for synthesis.
Parameters on Tamarind may differ from the repository, so check the tool page for what is exposed.
Things to Keep in Mind
Conditioning slightly lowers PoseBusters pass rate (58 percent) compared with unconditioned MolDiff after redocking (91 percent), and top diversity is lower than SILVR.
Vina scores are only weakly correlated with true binding affinity, so docking scores are a rough guide.
Pharmacophores in the study were hydrogen bond donors and acceptors only.
Outputs are computational proposals and need synthesis and experimental testing.
Source: Ziv, Imrie, Marsden and Deane, "MolSnapper: Conditioning Diffusion for Structure-Based Drug Design," J. Chem. Inf. Model. 2025, 65, 4263-4273, DOI 10.1021/acs.jcim.4c02008. Code: github.com/oxpig/MolSnapper.