Skip to main content Skip to navigation


Title Description Status Research Student Supervisor
A Game of Order Parameters Predicting if, how and when ice crystals will form in clouds (and our own cells!) is important to atmospheric science (and cryopreservation!) and manufacture of pharmaceuticals. In Progress Katarina Blow (Cohort 1) David Quigley
Automatic Prediction and Characterisation of Complex Chemical Reactions Around 90% of all chemical processes use catalysis to control reactivity and selectivity, yet the design of new catalysts too often depends on informed trial-and-error to make progress. In Progress Idil Ismail (Cohort 1) Scott Habershon
Exploring the Heterogeneous Biogenesis Pathways of the Bacterial Cell Envelope The cell envelope serves as the front-line to both defence and pathogenicity in bacteria. Critical protein assemblies are required to mature, localise and assemble proteins, sugars and lipids around the cell to enable protection against antibiotics, phages and toxins, and to modulate cell structure and shape. A major component of this is the essential extracellular cell wall, which forms a mesh-like coat around the cell. Its assembly is the target for many antibiotics. Here we will use molecular simulation to study the protein machinery responsible for the formation of the cell wall and other biogenesis pathways within the cell envelope. In Progress Matyas Parrag (Cohort 3) Phillip Stansfield and David Roper
It's all in the Structure: Transforming drug design by bringing together molecular simulations and machine learning The solubility of pharmaceutical drugs determines to what extent they can be absorbed. Machine learning algorithms can predict the solubility of novel drugs without the need of actually synthetizing them - thus saving substantial time and money. However, we currently infer solubility from the structure of single molecules in vacuum - a sub-optimal approach ignoring interatomic interactions. This project, supported by AstraZeneca, will address this pitfall by generating three-dimensional molecular models of crystalline drugs polymorphs and simulate their dissolution by means of enhanced sampling simulations. These results will be used to construct a machine learning framework that will unravel the atomistic origins of drugs solubility. In Progress Steven Tseng (Cohort 2) Gabriele Sosso
Next generation sampling of organic molecules Models of simple flexible organic molecules, such as methane or carbon dioxide, have surprisingly rich phase diagrams showing numerous (and sometimes spurious) stable and metastable crystal structures. These models are of utmost interest in areas from organic electronics to pharmaceuticals. Traditional computer modelling of these systems requires laborious application of multiple techniques. We will instead adapt a novel “one shot” data-intensive algorithm (nested sampling) which makes both structural and thermodynamic predictions, to flexible organic molecules. We will use the resulting tool to automatically calculate full phase diagrams, initially for simple “toy” alkane models and ultimately realistic heterogeneous organic systems. In Progress Omar Adesida (Cohort 2) Livia Bartok-Partay
Predicting long-term materials ageing using reaction discovery and machine learning Predicting the long-term (decades or more) stability of organic polymeric materials under ambient environmental photothermal conditions is a unique challenge because experimental testing on such time-scales is often impossible or too expensive. In this project, we will merge reaction discovery tools (Habershon group) with machine-learning energy calculation methods (Maurer Group) to develop kinetic models to predict the long-term behaviour of organic polymeric materials. These predictive models will then have potential to be used to guide the choice of materials with tailored properties for long-term environmental applications. In Progress Joseph Gilkes (Cohort 2) Scott Habershon
Protein origami: New computational methods to predict protein folding ensembles How does a protein fold into its native state? This is one of the most important and challenging problems in the chemical sciences, and a key question in understanding diseases driven by protein misfolding and aggregation (such as Parkinson’s disease). In this project, we will develop and employ a new computational scheme to access long time-scale protein folding events by mapping onto a discretized connectivity-based description of protein structure. This new approach will then enable us to investigate sequence-specific folding effects and translational folding, as well as providing a new scheme for protein-structure prediction. In Progress Ziad Fakhoury (Cohort 3) Scott Habershon and Gabriele Sosso
Untangling disorder in amorphous pharmaceuticals using artificial intelligence The best pharmacologically active compound may have been found, but its function, properties and processability are heavily dependent on the solid form. Solid-state Nuclear Magnetic Resonance and diffraction experiments, coupled with quantum mechanical calculations are powerful tools to elucidate the atomic structure, which is needed to understand and control macroscopic properties. The challenge in amorphous systems is that experiments only provide globally averaged measurements, while large length scales are needed to capture the disorder, which render quantum mechanical calculations unfeasible. In this project, machine learning models will be used to bypass the need for these costly computations, allowing interpretation of experimental data and structure determination with unprecedented accuracy. In Progress Jeremy Thorn (Cohort 3) Albert Bartok-Partay and Steven Brown