Gero unveils AI model for small molecule design without structure

ProtoBind-Diff generates novel compounds from protein sequence alone – a potential accelerant for early-stage discovery in aging biology.

AI in drug discovery is evolving fast. Singapore’s Gero has launched ProtoBind-Diff – a generative model that designs small molecules without using protein structures.

Trained on over a million protein–ligand pairs, it relies only on sequence data, enabling it to target proteins with no structural info. According to its creators, it works even for “orphan, flexible, or rapidly emerging targets.”

For aging research, where many targets lack clear structures, this could open faster paths into uncharted biology.

My take on this: AI in drug discovery usually focuses on speed and optimization. ProtoBind-Diff flips the script – targeting the unknowns in the proteome by generating molecules from protein sequences alone. That means even disordered or obscure targets are now in play. In aging research – often full of theories but light on clear targets – this shifts the game: the less we know, the more interesting it gets.

What’s bold isn’t just the model – it’s Gero’s plan to open-source it. In a field of closed systems, that’s rare. If it works, ProtoBind-Diff could jumpstart early discovery, even without structural data. It’s not about ditching structure – it’s about embracing uncertainty. And that might be exactly what geroscience needs.

“Designing small molecules that hit protein targets is one of the hardest problems in drug discovery,” said Peter Fedichev, CEO and Co-founder at Gero. “Classical modeling struggles because the energy scales, polarization effects, and the complexity of protein dynamics make high-resolution predictions nearly impossible. But maybe we’ve been asking the wrong question.”

He continued: “Nature had to solve this puzzle already – evolution optimized a biochemical language that encodes how proteins and molecules interact. With ProtoBind-Diff, we’re tapping into that. It’s a language model that learns from sequences, not structures. It doesn’t simulate physics – it learns the grammar of bioactivity from a million real examples.”

Dive deeper into Gero's new AI model, with more insights from Peter Fedichev right HERE. Plus discover how Epigenica is transforming how we detect and scale the measurement of epigenetic shifts and finally, read how Oxford Biohacking Society Founder Marie B. Kruth (Mary) captured a night of bold ideas and community spirit at the Oxford & Cambridge Club in London.

Thanks for reading! If you’ve enjoyed this article, it would mean a lot to me if you could subscribe and share it on LinkedIn!