Integrating AI, big data and smart drug design against Superbugs

The World Health Organization (WHO) has identified antibiotic resistance as an urgent serious global threat. For example, Acinetobacter baumannii is among the three WHO top-priority ‘superbugs’ against which polymyxins are the last resort, and no new antibiotics will be available in the near future; however, kidney toxicity has limited the optimal use of polymyxins and resistance has emerged. This project will be the first to employ cutting-edge all-atom molecular dynamic simulations, Big Data, machine learning, and artificial intelligence to design new-generation polymyxins, and target the imperative global challenge.

Funding organisations

Project news and links

New antibiotic to combat deadly bacterial ‘superbugs’ enters clinical trials

US biopharmaceutical company licenses Monash University 'superbug' drug discovery

Jiangning Song group at Monash Biomedicine Discovery Institute

Jian Li group at Monash Biomedicine Discovery Institute

Monash Institute of Pharmaceutical Sciences - Novel treatments for infectious diseases

Project description

Problem: A. baumannii can develop resistance to all current antibiotics, including the last-line polymyxins. A major barrier in developing new-generation polymyxins is the lack of understanding on polymyxin interactions with bacterial and human kidney tubular cell membranes. For favourable in vivo disposition, better efficacy, and less kidney toxicity, it is extremely challenging to fine-tune the complex polymyxin structure using traditional experimental approaches, as synthesis and evaluation of thousands of analogues are extraordinarily expensive, time-consuming and just not feasible.

Proposal: The current polymyxin structure-activity relationship (SAR) developed by CI Li’s group is based on the interaction with a single target, lipopolysaccharide (LPS), in Gram-negative outer membrane (OM). The current polymyxin structure-toxicity relationship (STR) developed by CI Li’s group as well is empirical and very preliminary. Excitingly, our recent membrane lipidomics and all-atom molecular dynamics (MD) studies discovered that the phospholipid composition in A. baumannii (for SAR) and human kidney tubular HK-2 cells (for STR) significantly affects their interaction with polymyxins. We also discovered the very different interactions, at the atom level, of polymyxins with the major phospholipids in the OM of A. baumannii and membrane of human kidney tubular cells. This proposal is built upon our internationally leading research in antimicrobial discovery, bioinformatics and machine learning (ML); we will develop the first-ever quantitative membrane-based SAR (QSAR) and structure- toxicity relationship (QSTR) models for engineering and discovery of superior polymyxins against MDR A. baumannii.

Our Specific Aims are to: (1) Perform all-atom MD of the interactions between 550 representative polymyxin analogues (i.e. 50 modifications per position x [10 amino acids + N-terminus]) with membranes of A. baumannii and human kidney tubular HK-2 cells; (2) Develop membrane-based QSAR-QSTR with the MD data (~1,000 TB of data); and (3) Conduct virtual lead optimisation with ~6,000 analogues using cutting-edge machine learning and AI approaches. CIB Li will provide extra funding to support the subsequent experimental evaluations of the screened superior analogues. This project will lead to the identification of a superior candidate (plus a backup) for pre-clinical experimental evaluations.

Significance: Our project will address the significant challenge in discovering new-generation polymyxins against MDR A. baumannii, one of the top three high-priority Gram-negative ‘superbugs’. No action today, no cure tomorrow.” As antibiotic resistance has seriously threatened modern medicine, the WHO has urged all government sectors and society to act on it urgently. This project will develop a cutting-edge systems approach by employing all-atom molecular dynamic simulations, Big Data, machine learning, and artificial intelligence to design new-generation polymyxins. Notably, we are the first in the world to use this multi-disciplinary approach to discover new-generation polymyxin drugs. It will provide essential data for a nationally/internationally competitive grant. Importantly, our project will paradigm shift the current antibiotic discovery.

Research Platforms to be utilised: High-performance computing facilities (MASSIVE3), CAVE2 and e-Research Centre (NecTAR cloud), Bioinformatics.

Team members

Chief Investigators

A/Prof Jiangning Song, Faculty of Medicine, Nursing, Health Sciences
Prof Jian Li, Faculty of Medicine, Nursing, Health Sciences
Dr Nitin Patil, Faculty of Medicine, Nursing, Health Sciences
Dr Daniel Schmidt, Faculty of IT
Prof Philip Thompson, Faculty of Pharmacy and Pharmaceutical Sciences

Associate Investigators

Professor Geoff Webb, Faculty of Information Technology, Research Director of Monash Data Futures Institute
Dr Xukai Jiang

PhD Students

Mr Yanan Wang
Ms. Jing Xu
Ms. Yue Bi
Ms. Tong Pan
Mr. Zhikang Wang

Research assistants

Ms. Xiaoyu Wang
Mr. Tim Peng


Chen Z, Zhao P, Li C, Li F, Xiang D Chen YZ, Akutsu T, Daly RJ, Webb GI, Zhao Q, Kurgan L, Song J. iLearnPlus: a comprehensive and automated machine-learning platform for nucleic acid and protein sequence analysis, prediction and visualization. Nucleic Acids Research. 2021 June 4. 49(10):e60.

Wang Y, Coudray N, Zhao Y, Li F, Hu C, Zhang YZ, Imoto S, Tsirigos A, Webb GI, Daly RJ, Song J. HEAL: an automated deep learning framework for cancer histopathology image analysis. Bioinformatics. 2021 May 19:btab380.

Chen YZ, Wang ZZ, Wang Y, Ying G, Chen Z, Song J. nhKcr: a new bioinformatics tool for predicting crotonylation sites on human nonhistone proteins based on deep learning. Briefings in Bioinformatics. 2021 May 18:bbab146.

Fossati A, Li C, Uliana F, Wendt F, Frommelt F, Sykacek P, Heusel M, Hallal M, Bludau I, Capraz T, Xue P, Song J, Wollscheid B, Purcell AW, Gstaiger M, Aebersold R.  Comprehensive assessment of machine learning-based methods for predicting antimicrobial peptides. Nat Methods. 2021 May;18(5):520-527.

Xu J, Li F, Leier A, Xiang D, Shen HH, Marquez Lago TT, Li J, Yu DJ, Song J.  Comprehensive assessment of machine learning-based methods for predicting antimicrobial peptides. Briefings in Bioinformatics. 2021 Mar 27:bbab083.

Wang J, Li J, Hou Y, Dai W, Xie R, Marquez-Lago TT, Leier A, Zhou T, Torres V, Hay I, Stubenrauch C, Zhang Y, Song J, Lithgow T. BastionHub: a universal platform for integrating and analyzing substrates secreted by Gram-negative bacteria. Nucleic Acids Research. 2021 Jan 8;49(D1):D651-D659.

Mei S, Li F, Xiang D, Ayala R, Faridi P, Webb GI, Illing PT, Rossjohn J, Akutsu T, Croft NP, Purcell AW, Song J. Anthem: a user customised tool for fast and accurate prediction of binding between peptides and HLA class I molecules. Briefings in Bioinformatics. 2021 Jan 18:bbaa415.

Chen Z, Zhao P, Li F, Marquez-Lago TT, Leier A, Revote J, Zhu Y, Powell DR, Akutsu T, Webb GI, Chou KC, Smith AI, Daly RJ, Li J, Song J. iLearn: an integrated platform and meta-learner for feature engineering, machine-learning analysis and modeling of DNA, RNA and protein sequence data. Briefings in Bioinformatics. 2020 May 21;21(3):1047-1057. (ESI Highly-Cited Paper by Clarivate Analytics Web of Science)