De novo generation of superior polymyxins using artificial intelligence and big data

Project supervisors

A/Prof Jiangning Song, Faculty of Medicine, Nursing and Health Sciences (Main Supervisor)
Dr Daniel Schmidt, Faculty of IT
Prof Jian Li, Faculty of Medicine, Nursing and Health Sciences

PhD project abstract

The World Health Organization (WHO) has identified antibiotic resistance as an urgent serious global threat. Acinetobacter baumannii, Klebsiella pneumoniae and Pseudomonas aeruginosa are the three WHO top-priority ‘superbugs’ against which polymyxins are the last-resort and no new antibiotics will be available in the near future. Unfortunately, kidney toxicity has limited the optimal use of polymyxins and resistance has emerged. We recently developed structure-activity-toxicity relationship models for polymyxins, and synthesised more than 2,000 different analogues for mechanistic investigations. This project will be the first to employ cutting-edge generative adversarial networks, big data, deep learning, and artificial intelligence to design and generate de novo polymyxins against bacterial 'superbugs', and target the imperative global challenge.

Areas of research

AI and Data Science in Health Sciences

Project description

Machine learning employs automated data-driven strategies to correlate complex chemical features of drug candidates with systems-level biological data for the prediction of ideal pharmacological properties. Machine learning algorithms are intelligent and adaptive as they continuously improve through each iterative cycle to increase their prediction accuracy. Notably, machine learning has emerged as a powerful computational approach in modern drug discovery; its high-throughput and excellent predictability have facilitated the discovery of several approved drugs and small molecule antibacterials. Particularly, cutting-edge machine learning techniques (e.g. deep learning) are vastly powerful in processing large datasets, which is simply impossible with traditional computational approaches. This multi-disciplinary project will combine cutting-edge deep lerning and currently available polymyxin data to develop a quantitative membrane-based structure-activity relationship model for the discovery of novel polymyxins against polymyxin-resistant Gram-negative bacterial pathogens. The developed model will allow rapid, high-precision in silico evaluations of enormous chemical and pharmacological datasets for more than 50,000 analogues, and dramatically minimise labour-intensive and time-consuming experimental work required by traditional drug discovery.

PhD student role description

In this project, the PhD student will be supervised by the supervisory team to design, develop and deploy cutting-edge data-driven computational approaches by combining big data, generative adversarial networks, deep learning, molecular dynamics, chemoinformatics and AI to design new-generation polymyxins. The PhD student should ideally have data-driven bioinformatics/machine learning/deep learning background and skillsets. Applicants with related first-authored papers in CS/IT/SE or Bioinformatics & Computational Biology are preferable.

Required skills and experience

Algorithms; Bioinformatics; Data Mining; Deep Learning; Machine Learning; Software Engineering