Direct discovery · Biology · Medicine

Machine-learning-designed receptors become artificial protein switches

Artificial allosteric protein switches and biosensors built from machine-learning-designed ligand-binding domains.

Summary

The Nature Biotechnology paper shows that machine-learning-designed ligand-binding proteins can serve as receptors in artificial allosteric switches. The authors demonstrate colorimetric, luminescent and electrochemical biosensors, logic gates, steroid-dependent bacterial antibiotic resistance and bioelectronic steroid-hormone sensing.

AI role

Machine learning supplied artificial receptor proteins that researchers converted into switchable enzymatic, luminescent and electrochemical biosensors.

Narrative role

This fills a direct synthetic-biology output gap: AI-designed protein components were incorporated into experimentally tested molecular devices with specified sensing behavior.

Caveat

The work demonstrates platform potential in engineered constructs; clinical, environmental or industrial deployment is not established by this paper.