• University of Cambridge Featured PhD Programmes
  • University of Glasgow Featured PhD Programmes
  • University of Birmingham Featured PhD Programmes
  • Aberdeen University Featured PhD Programmes
  • University of East Anglia Featured PhD Programmes
  • Castelldefels School of Social Sciences Featured PhD Programmes
University of Liverpool Featured PhD Programmes
John Innes Centre Featured PhD Programmes
Coventry University Featured PhD Programmes
Imperial College London Featured PhD Programmes
Coventry University Featured PhD Programmes

Development of Semi-Supervised Ensemble Learning Models for Serotype O and A FMDV Vaccine Strain Selection using Heterogeneous Data


Project Description

This studentship is open to science graduates (with, or who anticipate obtaining, at least 2.1 or equivalent in relevant biological subject in their undergraduate degree, or a Masters degree - subject to university regulations). Other first degrees, e.g. veterinary science, mathematics or statistics will be considered. Knowledge in statistics/mathematics or machine learning is desired. Programming skills in one of the major computer programming languages such as Matlab, R, Python, Java, or C/C++ are expected. You should be looking for a challenging, interdisciplinary research training environment and have an active interest in the control of infectious diseases.

The current method of Foot-and-Mouth Disease (FMD) vaccine strain selection is based on the serological tests that determine the antigenic relationships (r1 value) of circulating viruses with vaccine strains using antisera raised in bovines. This method is time consuming, expensive and the selected vaccines may not be able to provide broad-spectrum protection against emerging antigenically novel strains. In addition, the in-vivo cross-protection tests do not always agree with the in-vitro serological test results.

As the antigenic determinants of FMD viruses (FMDV) lie in the viral capsid proteins, viral capsid sequences provide an alternative dataset for vaccine strain selection. A powerful computational model that incorporates viral immunogenic protein or RNA level changes can effectively predict antigenically variant strains (1-Rahman et al., 2015; 2-Reeve et al., 2010), and minimise the amount of resources spent on serological testing of vaccines.

Using evolutionary algorithms, we have developed anon-linear (1) regression model for FMDV vaccine strains selection that revealed better performance than the -linear model (2). Although the non-linear model optimized by an evolutionary algorithm has shown to be very promising with 80% accuracy (1), the availability of limited serological data has seriously affected further improvement of the prediction accuracy so far.

This project aims at developing the ability to predict the antigenic sites of serotype O and A FMD viruses from in-vitro vaccine matching and capsid sequence data using computational and statistical models. This will involve improving the level of accuracy achieved by non-linear model by using a huge data set (already available in the laboratory), including antigenic distances (antigenic cartography) and more powerful learning techniques, in particular semi-supervised ensemble classification models. This would be a significant break-through as it would allow rapid selection of FMD vaccine strain in the event of an outbreak leading to better control of the disease. Furthermore, analysis of the data will result in identification of antigenically important motifs that can be tested using an existing cDNA clone.

Funding Notes

This is a fully-funded studentship open to UK students and eligible EU students who qualify for home-rated fees, in line with BBSRC criteria:
View Website.
International students are welcome to apply, however tuition fees will not be funded.
Eligible students will receive a minimum annual stipend of £14,553 and university registration fees will be paid.
Students without English as a first language must provide evidence that they meet the English language requirement - eg with an IELTS score of 7.0, with no less than 6.5 in any of the subsections.

References

1. Rahman T., Mahapatra M., Laing E., and Jin Y. (2015). Evolutionary non-linear modelling for selecting vaccines against antigenically variable viruses, Bioinformatics, 31(6):834-840.
2. Reeve R, Blignaut B, Esterhuysen JJ, Opperman P, Matthews L, Fry EE, de Beer TA, Theron J, Rieder E, Vosloo W, O'Neill HG, Haydon DT, Maree FF. Sequence-based prediction for vaccine strain selection and identification of antigenic variability in foot-and-mouth disease virus. PLoS Comput Biol. 9; 6 (12):e1001027.
3. Bari FD, Parida S, Tekleghiorghis T, Dekker A, Sangula A, Reeve R, Haydon DT, Paton DJ, Mahapatra M (2014). Genetic and antigenic characterisation of serotype A FMD viruses from East Africa to select new vaccine strains. Vaccine. 2014 Oct 7; 32 (44):5794-800.
4. Asfor AS, Upadhyaya S, Knowles NJ, King DP, Paton DJ, Mahapatra M (2014). Novel antibody binding determinants on the capsid surface of serotype O foot-and-mouth disease virus. J Gen Virol., 95 (Pt 5):1104-16.
5. Bari FD, Parida S, Asfor AS, Haydon DT, Reeve R, Paton DJ, Mahapatra M (2015). Prediction and characterisation of novel epitopes of serotype A FMD viruses circulating in East Africa using site-directed mutagenesis. J Gen Virol., 96 (Pt 5):1033-41.

Email Now

Insert previous message below for editing? 
You haven’t included a message. Providing a specific message means universities will take your enquiry more seriously and helps them provide the information you need.
Why not add a message here
* required field
Send a copy to me for my own records.

Your enquiry has been emailed successfully




Cookie Policy    X