Enzymology for Environment
Enzymology for Environment
Generative Enzyme Design for Targeted Bioremediation
Pollutants like aromatic hydrocarbons, halogenated solvents, plastic monomers, dyes, and complex agrochemical residues persist in the environment, posing long-term risks to ecosystems and human health. Enzymes, a specialised class of proteins, possess exceptional catalytic efficiency, enabling the rapid synthesis and transformation of complex organic molecules. When an enzyme is able to recognise a pollutant as its substrate, adopt a productive catalytic geometry, and maintain stability and selectivity under environmental conditions, it becomes a powerful tool for safe and effective bioremediation of water, soil, and industrial waste streams.
Intractābilis has developed a generative design platform specifically for engineering enzymes targeted for bioremediation. For any specified chemical reaction, the system predicts the minimal catalytic arrangement of residues and cofactors required to stabilise the key transition state. It then proposes residue identities, side-chain conformations, and local structural geometries, along with predicted protonation states and metal coordination where relevant.
Our solution also ranks candidate enzymes according to predicted activation barriers, binding free energies, hydration patterns within the active site, structural stability, and the likelihood of expression in common microbial hosts. It further assesses potential off-target activities and the toxicity of reaction products to ensure safe application. This approach provides a scalable and adaptable framework for reaction-conditioned enzyme generation that can be rapidly extended to new classes of pollutants. It offers a practical route to cleaner water and soil while reducing dependence on hazardous chemical treatments.
Transformer-based Cleavage Site Prediction Framework
In enzyme-catalysed protein hydrolysis, proteins are split at specific cleavage sites through the action of proteolytic enzymes. Accurate prediction of enzyme-catalysed cleavage sites in substrate proteins enables the identification of potential therapeutic targets and supports drug design. Cleavage sites of proteins acted upon by proteolytic enzymes can be identified through peptide specificity assays or high-throughput mass spectrometry. While effective, these experimental methods are resource-intensive, technically challenging, and often costly, making scalable applications difficult.
Intractābilis has developed an advanced Transformer-based encoder framework for predicting protein cleavage sites, designed to generalise across a wide range of proteolytic enzymes. This approach incorporates enzyme active-site knowledge to improve the precision of cleavage site prediction in enzyme–protein interactions.
The framework utilises a biochemically informed enzyme encoder in conjunction with active-site-aware pooling to generate high-quality enzyme representations. The enzyme encoder is further enhanced through pretraining on an expanded enzyme set specifically for active-site prediction, ensuring robust and accurate outputs across diverse enzymatic profiles.
Together, these platforms redefine what is possible in enzymology from predictive mutation screening and rational catalyst generation to fine-grained functional modelling.
Our solutions eliminate the guesswork from protein engineering, reduce reliance on wet-lab iteration, and empower researchers and developers with tools that work at the speed of thought—ushering in a new era of intelligent, mechanistically grounded enzyme design.
Ready to Work Together?
© 2025 Intractābilis