We highlight key advances and remaining challenges in this emerging structure-aware paradigm, and discuss how integrating protein information can enable AI to design molecules with stronger binding potential and real-world drug-like properties.

Check out the paper: pubs.rsc.org/en/content/a...

We cover the evolution from early shape-based strategies to modern co-folding models, detailing how different representations of protein pockets (voxel and graph) can encode structural information (shape, interactions, all-atom detail) and guide molecular generation.

Great work by DPhil student Kate Fieseler. Thanks also to co-authors Max Winokan, Joseph Morrone, Charlotte Deane, Frank von Delft, and Warren Thompson

Syndirella proposes congeneric series you can actually make (multi-step, digitized routes) and explores the pocket more broadly. Additionally, by buying reactants (not products), it allows you to test far more designs for the same budget

If this sounds interesting, we’d love to hear from you (email [email protected] or [email protected])!

Positions available to start immediately.

Postdocs will contribute to:

- Developing and applying AI/ML methods for small molecule design and selection
- Running blind community challenges
- Assessing the value of large-scale structural biology datasets

Congratulations to the authors: Henriette Capel, Isaac Ellmen, Chris Murray, Giulia Mignone, Megan Black, Brendan Clarke, Conor Breen, Sean Tierney, Patrick Dougan, Richard Buick, Alex Greenshields-Watson, and Charlotte Deane, for their contributions and support on the project.

See more in the
Paper: www.biorxiv.org/content/10.1...
GitHub: github.com/oxpig/LICHEN
Web tool: opig.stats.ox.ac.uk/webapps/lich...

LICHEN outputs are customisable and tuneable to experimental needs or desired content, enabling a collaborative light sequence design by leveraging computational capabilities alongside experimental expertise.