Geometric deep learning methods for bacterial taxonomic classification

The task of bacterial taxonomic classification plays a pivotal role in the public health management of infectious disease outbreaks. Accurate identification of bacterial strains is essential for effective treatment protocols and subsequent genomic analyses.

With the growing prevalence of culture-independent methodologies, such as metagenomics, there is a pressing need for more rapid and precise approaches to bacterial classification. Current leading methods, while proficient, often require significant computational resources and extended processing times.

Considering recent advancements in artificial intelligence, particularly in processing DNA sequences, this project proposes the application of geometric deep learning techniques. These techniques are anticipated to surpass existing sequence homology-based methods in efficiency and accuracy for bacterial classification.

Our goal is to optimise these methods for computational resource management and processing speed. We will work with the MDAP team to apply expertise in deep learning and artificial intelligence.

The outcome of this project will be the development of open-source tools, designed for seamless integration into existing public health genomic workflows.