Every year, thousands of new synthetic chemicals enter our environment, yet we have detailed safety data for only a small fraction of them. Some of these chemicals are particularly concerning because they do not break down easily and can move rapidly through water systems, potentially reaching drinking water sources and accumulating in ecosystems. Identifying which chemicals pose this kind of risk is a major challenge, because running laboratory experiments for every substance is too time-consuming and expensive.
This project develops computer-based tools that use the molecular structure of a chemical to predict two key environmental risk properties: how easily it moves through water systems, and how quickly it breaks down.
The first tool predicts chemical mobility using data from thousands of laboratory liquid chromatography experiments. Rather than relying on scarce experimental measurements, it uses patterns in how chemicals interact with water during analysis to classify them as very mobile, mobile, or non-mobile. Applied to over 64,000 registered industrial chemicals, the model found that roughly one in five would be considered very mobile; a scale of assessment that would be impossible through traditional testing alone.
The second tool predicts how readily a chemical biodegrades. That is, how quickly microorganisms can break it down. This model was trained on data from nearly 5,000 chemicals and outperforms existing tools. Most importantly it also works for chemicals that have never been identified: using only the fragmentation pattern a chemical produces in a mass spectrometer, the model can estimate whether an unknown substance is likely to persist in the environment. This is particularly valuable because many chemicals detected in environmental samples remain completely unidentified.
Together, these tools provide a scalable, data-driven approach to prioritising chemicals of concern.
Conference Abstracts
Hulleman, T., Samanipour, S., Turkina, V., Rauert, C., Li, J., Okoffo, E., Thomas, K.V. & O’Brien, J.W. Prioritisation of persistent known and unknown chemicals using synthetic accessibility and tandem mass spectrometry, Queensland Mass Spectrometry Symposium, Brisbane, Australia, 28-29 January 2026.
Hulleman, T., Samanipour, S., Haddad, P.R., Rauert, C., Okoffo, E., Thomas, K. & O’Brien, J. Machine learning for predicting environmental mobility based on retention behaviour, SETAC Australasia, New Zealand, 25-28 August 2025.
Hulleman, T., Turkina, V., O’Brien, J.W., Chojnacka, A., Thomas, K.V. & Samanipour, S. Critical assessment of the chemical space covered by LC-HRMS non-targeted analysis, International Mass Spectrometry Conference 2024, Melbourne, Australia, 17-23 August 2024.
Hulleman, T., Samanipour, S., Haddad, P.R., Rauert, C., Okoffo, E.D., Thomas, K.V. & O’Brien, J.W. Machine learning for predicting environmental mobility based on retention behaviour, 20th Annual Workshop On Emerging High-Resolution Mass Spectrometry (HRMS) And LC-MS/MS Applications In Environmental Analysis And Food Safety, Spain, 7-8 October 2024.
Research Outputs
Industry Placement
- 2025-2026 SCIEX Targeted high resolution mass spectrometry to build a method to distinguish algae species using lipodomics and multivariate data analysis
Awards
- 2025 People's Choice winner, Three Minute Thesis (3MT) Competition at UQ's Faculty of Health, Medicine and Behavioural Sciences Faculty Final.
- 2025 First Place & People's Choice winner, Three Minute Thesis (3MT) Competition QAEHS/School of Pharmacy.
- 2025 HyTECH Biannual Meeting, Best Student Presentation for Non-Targeted High-Resolution Mass Spectrometric Characterisation of Highly Mobile and Highly Persistent Chemicals.
- 2024 Second Place for Best Poster at the 20th Annual Workshop On Emerging High-Resolution Mass Spectrometry (HRMS) And LC-MS/MS Applications In Environmental Analysis And Food Safety
