MazeBio turns protein dynamics into functional understanding — what a protein actually does, and how a mutation, a binding partner, or a candidate changes it. Not another end-to-end discovery engine: the functional layer your programs are missing.
We're focusing on protein dynamics
A molecule can bind and still do nothing useful — or bind elsewhere and change everything. Function lives in how a protein moves between states, something a single structure can't show. We read that motion with AI and turn it into functional insight development teams can act on.
Adenylate kinase — our model mapping the conformational landscape, apo → active.
We quantify how a molecule changes what a protein does — its mechanism of action — not only whether it sticks.
Thousands of distinct conformations in minutes, without running molecular dynamics — fast enough to inform a whole program, not one target.
Functional read-outs to choose targets, understand selectivity and resistance, and explain why a candidate works — before the bench and the clinic.
Where it matters most
Dynamics-driven functional insight matters most for proteins defined by the states they move through.
Function is the motion between open and closed states. One snapshot tells you little; the cycle tells you everything.
Druggable sites that only appear in certain conformations — invisible to a method that sees one shape.
How a mutation or compound reshapes the landscape — and how to select against off-targets, not just for the target.
Across the program
The same engine informs decisions across a development program — target selection, mechanism of action, selectivity against off-targets, and resistance — across GPCRs and CNS receptors, antibacterial membrane targets, and enzymes with moving active sites.
Team
MazeBio is an early-stage team building protein-dynamics AI out of Lausanne, Switzerland.
Contact
Working on a program where function — not just binding — decides the outcome? We'd like to hear about it.
matthieu.marfoglia@maze-bio.com