Embedded Files
THEOCEPTORS: A new computational method for predicting protein-ligand binding geometries and energies
We have created and developed a new computational approach to predicting protein-ligand binding geometries and energies. Our aim is to use a quantum mechanical approach in order to ensure that our calculations capture the true nature of the interactions as far as possible and that our approach is as general as possible. We do not require the correct binding pose but can identify these from among an array of docked poses and our binding energies generally correlate usefully well with measured affinity. By relying on quantum mechanics, our approach provides real insights into the physical basis for observed geometries and energies and thus is a spur to human or computational creativity.
Funding
Applications of the method include:
a) Modelling cardiac drug safety
We are applying our method to the cryo-EM structures of some of the key cardiac ion channels including: hERG, Nav1.5, Cav1.2 and RyR2. We are also creating models of the human adrenergic receptors where we have found that we can predict not just binding geometry and energy but also mode of action by characterising binding strength to different protein conformations. All of these systems play an important role in regulating the heart’s regular rhythm and understanding molecules that disrupt any of them is important across all drug discovery programs.
b) Designing and optimising phosphatase inhibitors
While creating and validating computational models for some phosphatase enzymes that are associated with infectious diseases, we discovered a series of inhibitors. In this area, we are continuing to develop our computational modelling approaches but also performing synthetic chemistry, biological testing and structural biology in order to drive forward our ability to inhibit these enzymes that were once viewed as undruggable.
c) Tools for kinase inhibitor design
Kinase inhibitors have been an important advance in the treatment of cancer over the last 25 years. The kinase enzymes are key parts of a range of cellular signalling pathways and ensuring that the right kinases (and only the right kinases) are inhibited has been a longstanding challenge. Responding to mutations in kinases that arise after treatment and prevent that treatment from remaining effective are a further important challenge in this area.
d) Understanding promiscuous enzymes
The most important enzymes involved in metabolising drugs are the cytochromes P450 which have the remarkable property of processing many substrates and producing many products. We are exploring how well our approach to predicting the reversible binding of proteins with ligands translates to the binding of substrates to these promiscuous enzymes. We aim to create a tool that will allow the reliable prediction of metabolism regiochemistry and rate but more importantly to provide insights into what molecular features enable this surprising promiscuity.
Boron: An opportunity for new reactivity and pharmacology
The chemical element boron is at the heart of a 5-year collaboration between ourselves and researchers at the Universities of Edinburgh and St Andrews. We are focusing on computational aspects of the project and are seeking to understand the details of reaction mechanisms involved boron as well as a more general understanding of factors that influence the reactivity and interactions of boron. We have recently investigated the hydrolysis of boroxines and are working to understand what determines whether a Lewis base can form a complex with a boron-based Lewis acid. We are also investigating the utility of boron in medicinal chemistry by seeking to understand the stability of boron-containing molecules in physiological conditions and the formation of covalent bonds between boron and proteins.
More information available in Boron: Beyond the Reagent
Corrosion: Transferring approaches from pharmaceuticals to corrosion inhibitors
Corrosion is a major challenge for many industries and countries where there is infrastructure built of metal that is exposed to corrosive conditions. Identifying new generations of molecules that could help prevent or reduce corrosion and which do not include toxic elements is therefore of high importance. We are researching potential corrosion inhibitors for both direct application to surfaces/feedstocks or for addition to coatings. We are taking computational methods for molecular design from the pharmaceutical industry and applying them to this problem.
Funding
An electrostatic potential-led approach to understand the role of dynamic ordering in explicit solvation
Funding
Correctly accounting for the effects of the environment on physical properties and reactions has always been a big challenge for quantum chemists (and others!). We have been grappling with this problem, starting from a PhD studentship funded by Astrazeneca in which we explored a range of methods that we hoped would guide us towards the correct placement of the appropriate (small) number of solvent molecules around any system to correctly reproduce its properties. This has taken us into thinking about the ordering of charged species around one another as well as of hydrophobic molecules forming aggregates/clusters in water. We are working to secure experimental evidence to support our explicit solvation models as well as providing full benchmarking and code for the method.
Matched molecular pair analysis: Demonstrating the medicinal chemistry relevance of AI, ML and other modelling approaches.
Matched molecular pair analysis has become a well-established technique across the drug design process. Through research undertaken while at Astrazeneca and through his company, Medchemica limited, Andrew has helped popularise and develop this “explainable AI” approach to drug design. We are exploring ways that matched molecular pairs can complement ad improve Artificial Intelligence (AI) or other Machine Learning (ML) techniques in drug discovery.
Page updated
Google Sites
Report abuse