Online Platform Predicts Artificial Enzyme Activity

A tool that can predict catalytic activity of nanozymes, an artificial form of enzymes widely used in the life sciences and environmental industry sectors, has been developed by researchers at ITMO University. Using AI algorythms, the DiZyme platform can rapidly identify the main reaction features of different nanoparticles such as its velocity and the affinity of its compounds, as well as suggesting its optimum run conditions.

More stable and easier to store than natural enzymes, nanozymes are also cheaper and simpler for use in manufacturing; learning which systems will ensure the ultimate effect of a given reaction; could significantly accelerate the process of producing new materials and biochemical systems.

“Our service is tailored, first of all, to the study of nanoparticles with peroxidase activity. It’s a class of enzymes that catalyze oxidation of substrates with hydrogen peroxide,” said Julia Razlivina, the article’s first author and a Master’s student at ITMO. “We have accumulated quite a lot of data on such systems. Any user can open our website, enter a chemical formula and nanoparticle parameters and then let the algorithm predict its activity in seconds.” 

The new service will come in handy both for those looking to develop a specific material – the algorithm will produce activity limits for the material’s chemical formula – and fundamental researchers in the field. 

“Moreover, the platform is not limited to peroxidase activity and can potentially be used for other reactions, which we are planning to do when there is enough digitised experimental data,” added Nikita Serov, a co-author of the article and a PhD student at ITMO. “We are aiming to significantly accelerate experimental studies and minimise their costs. The more researchers use our platform, the better it will become.”

 “We wanted to demonstrate the model’s precision and to this end we’ve selected 16 different samples and measured their peroxidase activity and compared the results to the ones produced by the platform. It turned out that the service matched our results almost ideally for 70% of samples, while for the other 30% the results fit into the accepted accuracy limit. This demonstrates the high precision level of our platform in predicting enzymatic activity,” commented Vladimir Vinogradov, head of ITMO’s SCAMT Institute.

More information online