Scientists in Australia are employing artificial intelligence and biotechnology to identify enzymes capable of breaking down plastic and other persistent pollutants. Researchers at Murdoch University are leveraging machine learning alongside biochemistry to analyze millions of enzymes catalogued in global biological databases, aiming to discover those that could effectively degrade harmful substances.

Joseph Boctor, a PhD candidate at Murdoch University, outlined this approach in a recent review published in *Nature Reviews: Earth & Environment*. He explained that the solution to finding suitable enzymes to combat specific pollutants likely resides within already existing biological data. By using machine-learning algorithms, researchers can sift through vast quantities of unexamined biological information to pinpoint enzyme candidates that have the potential to interact with and dismantle target pollutants.

The machine-learning models draw on previous knowledge of characterized enzymes to forecast how various enzyme structures might engage with contaminants, thereby predicting their pollutant-degrading capabilities. This computational method, Boctor noted, could expedite the remediation of pollutants currently affecting the environment while research continues on developing bioplastic alternatives.

Among the pollutants of concern are persistent substances such as per- and polyfluoroalkyl substances (PFAS) and microplastics, which pose significant health risks. Boctor highlighted that agricultural soils now contain approximately 23 times more microplastics than the oceans, underscoring the widespread environmental impact of these contaminants. The integration of AI-driven enzyme discovery represents a promising avenue for addressing plastic pollution and mitigating its effects on ecosystems and human health.