Janeesh Bansal

Profile

Project Title: Bayesian inference and AI approaches to infer patterns of disease spread from the genomes of the malarial parasite

Summary: 

Understanding spatial connectivity is critical to malaria control: otherwise, human communities can be continually seeded with new infections. Traditional methods for identifying infection routes rely on patchy information about human travel. Genomic data, from the malarial parasite itself, offers far more direct and powerful information, but the analysis must account for background levels of relatedness within parasite populations. A fundamental problem is that allowing for relatedness involves partitioning the very large number of parasite genomes into different categories. The multiplicity of models makes a likelihood-based analysis prone to misinterpreting the data, by failing to find well-supported models among the profusion of less effective candidates.

To combat the challenges a strategy will be developed using MCMC chains and complex coalescent cases will be mathematically challenged. The method will be compared against General Adversarial Networks which are supervised AI learning methods. The overall aim is to analyse real Plasmodium falciparum datasets, to visualize and interpret their patterns for malarial control. 

Supervisor:

• Prof Richard Nichols

Research