Bioinformatics approaches to vaccine design for bacterial pathogens

Abstract

This thesis focused on bacterial vaccinology and employed a newly emergent branch of vaccinology; reverse vaccinology (RV). RV is an in silico process that predicts vaccine candidates from an entire bacterial proteome, thus enabling the realisation of a greater number of putative vaccine candidates when compared to conventional vaccinology approaches. A previous RV classifier that utilised the computational field of machine learning (ML) was used to predict bacterial protective antigens (BPAs) (i.e. vaccine candidates) for Mycobacterium tuberculosis (Mtb). Mtb was chosen as the initial focus for RV approaches in this thesis because one third of the world’s population are infected with Mtb and in 2015 Mtb infection killed 1.8 million people. It is also being recognised that the only clinically licensed vaccine against Mtb infection, Bacille Calmette-Guérin (BCG), has varying rates of protection. Predicted BPAs by a published RV classifier were synthesised as DNA vaccines and tested in a mouse model of Mtb infection. However, the predicted BPAs were shown not to generate protection in repeat animal trials. To address the negative result obtained when testing BPAs predicted by a previous RV classifier, enhancements were made to the previously published RV classifier (i.e. nested leave-tenth-out cross-validation, subcellular localisation bias removal, increased size of training dataset and increased type of protein annotation tools used to generate features). Finally, the enhanced RV classifier, developed in this thesis, was assessed using a more biologically revealing metric termed recall in the proteomes of Mtb and Neisseria meningitidis serogroup B (MenB). MenB was chosen to assess the recall metric as it enabled comparisons to the BEXSERO vaccine, which was the first clinically licensed vaccine developed using RV. In summary, this thesis has developed a biologically relevant RV classifier that can now be used to predict BPAs for any bacterial pathogen with a sequenced genome. It is envisaged that these predicted BPAs could then be used to facilitate the rapid formulation of novel subunit vaccines.

More information

Identifiers

ORCID for Mahesan Niranjan: orcid.org/0000-0001-7021-140X

Catalogue record

Export record