Every natural enzyme is an answer to a question evolution encountered. Digest this nutrient. Copy this molecule. Defend against that threat. The answers can be astonishingly precise, but they are bounded by history.
Some of the chemistry humanity now needs sits outside that history. Synthetic pollutants, new industrial feedstocks, and reactions developed only in the last century gave evolution no time and often no reason to build a biological solution.
That is where DeNovoLabs begins.
The missing enzyme is the point
Traditional enzyme discovery asks a sensible question: what has nature already made that could be adapted for our purpose? For many valuable reactions, that is the fastest route. But it becomes limiting when the underlying chemistry is nearly absent from biology.
Our first program focuses on PFAS, a broad class of persistent synthetic chemicals built around unusually strong carbon–fluorine bonds. These bonds barely appear in natural metabolism. There is no deep catalogue of natural enzymes waiting to be screened for the exact job we need.
The fact that the enzyme has no ancestor is not a defect in the search. It is the reason to design.
Start with the reaction
We work reaction-first. Instead of beginning with an existing protein and asking what it might do, we begin with the bond transformation that must happen. We map the physical requirements of that event: how the target must be held, where charge and spin may need to move, what geometry could support the reaction, and which competing outcomes must be avoided.
Then we design the molecular environment around those requirements. This is not a claim that computation makes the chemistry true. It is a way to turn an enormous search problem into testable candidates and explicit failure modes.
Design the protein and the metal together
Many difficult reactions demand chemistry that amino-acid side chains alone cannot provide. Nature often solves this by using metals. We adopt the same physical principle without being constrained to a naturally evolved active site.
The protein has one job: bind the target, control its orientation, shape the electrostatic environment, and exclude unproductive arrangements. The metal cofactor has another: make electronic states available that a plain protein cannot easily reach. Neither part is designed in isolation.
This protein–metal co-design is central to our platform. It is what allows us to pursue metalloenzymes for reactions with no useful natural template while keeping every design accountable to a specific mechanism.
Our PFAS program has computational candidates and is advancing toward experimental validation. We are not claiming a working PFAS-degrading enzyme today. The decisive evidence must come from expression, cofactor loading, binding, product analysis, fluoride release, controls, and reproducibility in the wet lab.
Why begin with PFAS?
PFAS is not an easy demonstration. That is precisely why it is useful. A platform intended for hard chemistry should be tested against a problem where binding alone is not enough, electronic activation matters, and false positives are easy to produce.
Starting here forces discipline. A computational candidate has to survive multiple layers of scrutiny before it deserves experimental attention. A failed design is useful only when the failure teaches us which part of the mechanism, geometry, or molecular environment must change.
PFAS is the first proof, not the ceiling
The broader ambition is to make biology programmable for chemistry without a natural enzyme. The same reaction-first approach can be directed toward plastic breakdown, carbon conversion, and pharmaceutical transformations where selectivity and low-energy operation matter.
That future will not arrive from a single model or a single heroic experiment. It comes from a repeatable loop: define the chemistry, design candidates, test them, learn from what fails, and carry the evidence forward.
We are building that loop in public, carefully enough that progress means something.