Dr Vinny Davies
- Lecturer (Statistics)
Dr Davies is a lecturer in Statistics in the School of Mathematics and Statistics specilising in computational biology and computational methods for statistics and machine learning. He completed his Ph.D. within the School of Mathematics and Statistics where he focused on variable selection models for selecting antigenic sites in virus evolution. He then completed several post-docteral research positions in both the schools of Statistics and Computing Science, as well as spending time as a Biostatistician at the University of Leeds. He recently returned to the School of Mathematics and Statistics where his research interests will focus on methods on the interface between Statistics and Machine Learning. He has a particular interest in Computational Metabolomics, but a general interest in applying statistical and machine learning methods to any biological, chemical or health problem.
If you are interested in a doing a Ph.D., please take a look at the Supervision section or email me directly.
I am looking for potential PhD students across a range of subjects and have a number of projects available below. Please contact me if you wish to discuss these or any other projects further.
Gaussian Process Emulation for Mathematical Models of the Heart
Mathematical models of the heart can help us understand how the heart functions and provide us with valuable insights into how we can treat patients or diagnose disease. Previous and ongoing work has looked at how we can use statistical and machine learning emulation strategies to speed up inference and make the mathematical models applicable within a clinical setting. The aim of this project is to further develop these methods through the application of Gaussian Processes and apply them to different mathematical problems with higher dimensional inputs. In many of the possible applications, the mathematical models will often have high dimensional and potentially correlated parameter inputs, as well as highly correlated outputs. The initial aim of this work will be to further develop the emulation methods to deal with these problems and look at how we can more effectively select the parameter inputs for the simulations we choose to generate output for. Further work will then look at how these models can potentially be combined with other techniques such as automated annotation, accelerating the construction of our emulator, or through the combination of other emulators, which would allow for the modelling of a more global system.
Metabolomics DIA Resolver
In metabolomics we take a sample (blood, urine, etc) and put it through a mass spectrometer. The mass spectrometer scans the sample in multiple ways to help us work out what metabolites can be found in the sample. Identifying these metabolites can be useful for clinical trials, disease diagnosis and progression and various other medical applications. There are various way of choosing the scans, but in one particular method (DIA) we often see multiple fragments from multiple metabolites in a single scan. In order to identify the metabolites we need to work out which fragments belong to which metabolites. There are two possible projects that could come from this which can both be tested via a virtual mass spectrometer that we have recently created (ViMMS). The first is to design a new Bayesian model to use existing data to help us link the fragments to their metabolites, existing methods are currently quite basic. This project will involve more Bayesian modelling, but will also have to link with existing work in Python. The second option would be to design new ways of collecting the data that could provide more information from a sample, we have previously done some related work. This second option would involve more programming (Python) and less Bayesian statistics most likely, but the projects would certainly be closely related. In the case of each project, some initial work would need to be done analysing existing methods using our virtual mass spectrometer.
Bayesian Modelling of Antigenic Variability in Viruses
Viruses often mutate as we have seen with Covid-19 recently. When we create vaccines we usually use parts of a previous virus strain as part of the process. What this means (roughly) is that the vaccine will protect people against virus strains that are antigenically similar to the virus strain from which the vaccine was created. If new virus strains sufficiently mutate away from the virus strain of the vaccine, then the vaccine will become ineffective. The mutations occur in the amino acids in the proteins, but not all mutations are likely to cause problems. I have previously developed a Bayesian model to identify the relevant mutations. One of the options for this project would be to further develop this model by taking into account additional sources of information in the model. The other option would be to develop a new method which was able to predict which mutations we are likely to see in the future based on previous viruses and their current prosperity.
This year I will be teaching on the online MSc Data Analytics, where I will be teaching on the following courses:
- Data programming in Python
- Large-Scale Computing for Data Analytics
I will also be supervising a number of dissertation projects.