The Evozyne platform is distinctive to traditional methods of protein engineering in several key attributes.

Starting point: We use the evolutionary history of a protein

Rather than varying a single natural sequence (directed evolution) or manipulating a single atomic structure (physics-based design), we capture information from genome databases representing the evolutionary record of a target protein. This enables us to learn the non-intuitive hidden rules of protein function.

Sequence exploration: We direct our search rationally

Protein sequence space is impossibly large – much larger than the number of atoms in the known universe – and cannot be searched comprehensively by any existing approach. Using principles of molecular evolution, our models uniquely find the vanishingly small subset of this space that encodes functional, natural-like proteins. Though the model space is a tiny fraction of the total sequence space, it is enormous in absolute terms and vastly larger than what can be explored through directed evolution and physics-based design. This opens opportunities for designing novel proteins with non-natural functions.

Sequence generation: We generate novel sequences

Traditional approaches use random mutagensis and local perturbations from known sequences to generate new designs. The Evozyne process uses a deep and guided search within the sequence space predicted by evolution-based computational models, an approach that is neither random nor subject to physics-based approximations. This enables large sequence diversity of solutions and unlocks the natural potential of evolved proteins for functional innovation.

Learning mechanism: We learn from every sequence generated

Rather than single mutation trial-and-error, our models quickly evolve through rounds of high-throughput learning from the thousands to millions of measurements conducted in each iteration. Genotype-phenotype mapping for every sequence generated provides unbiased insights as feedback to our models, enabling novel sequence solutions for complex design challenges.

Performance optimization: We tune multiple parameters simultaneously

Successful protein engineering often involves a difficult multidimensional optimization over potentially correlated biochemical properties – catalytic power, substrate specificity, stability, environmental sensitivity, etc. Traditional methods are prone to becoming trapped in local performance minima, where improving one property may degrade performance of other properties. The Evozyne approach uncovers adaptive regions of a protein to enable simultaneous optimization over complex objective functions. Rather than modifying one parameter at a time, we can improve all desired parameters at once, thus dramatically accelerating protein engineering campaigns.

Delivered output: We provide a library of solutions

We do not just produce a single molecule. Just like natural evolution, our models produce a great diversity of sequences, all of which are solutions to a given design problem. This can facilitate downstream optimization for specialized environments, proprietary host strains, manufacturability, etc. This enables our solutions to scale up from the lab to commercial operation.

Discover next

Our impact