MACHINE LEARNING TECHNIQUES TO UNVEIL AND UNDERSTAND PSEUDOMONAS AERUGINOSA SURVIVAL MECHANISM IN NUTRIENT DEPLETED WATER

Abstract

Pseudomonas aeruginosa is a Gram-negative organism that is ubiquitous in the ecosystem and antibiotic resistant. Capable of long-term survival, it is a common cause of hospital-acquired infections. The focus of this thesis is to unveil P. ae-ruginosa genes interactions and identify those that are pivotal to its mechanisms of survival. With unlabeled data collected from P. aeruginosa gene expression in response to low nutrient water, a Bayesian Networks Machine Learning methodology was implemented, and a static regulatory network of its survival was modeled. Subsequently, node influence techniques were used to infer a dozen genes as key orchestrators of the survival phenotype. Among these genes, PA0272 was identified to be the root node in the learned network model. Water survival experiments were conducted in the lab on PA0272 mutants, and it was interestingly found that their survival declined by 10-fold compared to the wild type PA01; 10-fold or higher being significant.