Machine learning methods for data-driven discovery of antibiotic resistance plasmid dissemination and evolution
The capacity of bacteria to develop resistance to antibiotics is driven by their ability to exchange small non-chromosomal circular pieces of DNA, termed plasmids. These may encode antimicrobial resistance genes and virulence factors with major impact on human, animal, and plant health.
Experimental mapping of plasmids is possible but labour intensive and expensive, preventing large scale discovery and monitoring of the plasmid repertoire of bacterial communities or individual isolated strains, e.g., from infected patients.
Our project will address this challenge by applying deep learning to a large data set of curated plasmids to develop new computational models. The models will be used to discover and track plasmids in the massive amounts of sequencing information which already exists in the public domain. We will also study to which degree plasmids are exchanged between bacterial communities in humans, e.g., gut microbiome, and other environments, such as soil or wastewater. This could provide new insights on how bacteria acquire new capabilities, like resistance to antibiotic treatment, and thus impact human, animal, and plant health.