Predicting routes of Phase I and II metabolism based on quantum mechanics and machine learning

Predicting routes of Phase I and II metabolism

This peer-reviewed paper in Xenobiotica describes a new method to determine the most likely experimentally-observed routes of metabolism and metabolites based on our WhichP450, regioselectivity and new WhichEnzyme model.

The paper builds on previous metabolism prediction research, developing the capabilities available in our StarDrop Metabolism module.

Authors

Mario Öeren, Peter A. Hunt, Charlotte E. Wharrick, Hamed Tabatabaei Ghomi and Matt D. Segall

Summary

Unexpected metabolism can lead to late-stage drug candidate failure. Early in silico prediction of the dominant routes of metabolism is therefore vital to improve a drug’s chance of success.

In this paper, we describe the development of our new WhichEnzyme model. This model tells the user which enzymes are most likely to metabolise their compound.

By training heuristics on a combination of the outputs of the new WhichEnzyme model, our WhichP450 and regioselectivity models, we developed a new method to determine the most likely routes of metabolism and metabolites to be observed experimentally. This method delivers high sensitivity, with great success in identifying experimentally reported metabolites. It also demonstrates higher precision than other methods for predicting in vivo metabolite profiles.

Find out more

Read the full paper on the journal webpage, and discover our new method for predicting routes of Phase I and II metabolism, or get in touch with us if you’re interested in seeing a copy of the pre-print.

If you would like to discover more of our metabolism research, take a look at our recent J. Med. Chem. paper, which describes how to predict the regioselectivity of AO, CYP, FMO and UGT metabolism, or watch our webinar on the integrated prediction of phase I and II metabolism.