Skip to main content
Research_

Computational modelling

Using powerful computation and artificial intelligence to help us understand how cannabinoids work

With computational modelling, we can simulate the molecular interactions between cannabinoids and receptors to gain insight into their physiological activity and expand their therapeutic applications.

Our research

Cannabinoid-receptormolecular simulations

Cannabinoids have a wide range of effects on the brain and body by interacting with receptors and proteins that exist in the body to regulate normal biological functions.

Cannabinoids interact with these existing biological scaffolds by mimicking the shape of natural chemicals, like hormones, and plugging-in to nano-scale pockets on receptors (see figure).

As these receptors and proteins are smaller than a microscope can observe, we use computer simulations to predict how cannabinoids are interaction with receptors and proteins to help explain their effects.

Machine learning

We also use machine learning models to predict novel and unknown protein targets for cannabinoids. While cannabis has wide-ranging effects, the mechanism of action for some of the substituent chemicals is not fully understood.

Our models use state-of-the-art statistical techniques to learn from large pharmacological databases of protein-ligand assay data, inferring new rules that can be used to predict which proteins are undiscovered targets for cannabinoids.

These predictions are then validated pharmacologically using cellular screening, opening up exciting new avenues for therapeutic treatments.

 

Current research projects

Many cannabinoid-protein interactions have been discovered using in vitro pharmacology. However, the precise location of binding on the protein surface is often unknown. Molecular docking and all-atom molecular dynamics simulations are being used to model potential binding sites, helping us understand possible therapeutic utility and drug-development opportunities.

Many effects of cannabinoids are reported through anecdotal reports or phenotypic assays but the pharmacological targets for these actions are not always known. Using machine learning on databases of pharmacological data (also known as virtual screening) allows us to screen cannabinoids against protein targets quickly and affordably to determine the likely mechanism of action for any therapeutic effects.

The therapeutic benefits of cannabidiol (CBD) are claimed for a wide range of diseases, perhaps more than any other FDA-approved drug. Bioinformatics analyses of reported interactions for CBD and drug targets are being used to predict which potential indications arise from genuine drug-target interactions, and which from non-specific effects of in vitro assay conditions.