Artificial Intelligence and Machine Learning Across TIA-supported facilities: Building Capability for Translation
Across the Therapeutic Innovation Australia (TIA) network, artificial intelligence (AI) and machine learning (ML) are increasingly embedded into existing workflows to support more efficient research translation. While these approaches are often described as emerging technologies, many TIA-supported facilities have been applying computational and data-driven methods for some time. In many workflows, AI/ML drives innovation, providing a competitive advantage to our facilities.
Through the National Collaborative Research Infrastructure Strategy (NCRIS), TIA enables facilities to develop and invest in the infrastructure, software and expertise required to integrate these capabilities in a coordinated and sustainable way.
Recent investments by TIA in AI and ML technologies indicate a broad range of approaches at Australian therapeutic development facilities and demonstrate how Australian researchers are embracing the opportunities of this paradigm shift.
From developing their own open-source software, improving therapeutic candidate selection and predicting efficacy and toxicity of novel therapeutic candidates, TIA’s facilities are routinely using these new platforms to accelerate the path from discovery to application.
Â
Responding to technological and societal changeÂ
The growth of AI/ML is creating new cross-disciplinary connections, linking biologists and biochemists with computer scientists. At the TIA-supported RNA Innovation Foundry at the University of Western Australia, the teams are not only developing in-house classical algorithms, but are also advancing research into a neural network capable of modelling mRNA dynamics and function using signals from diverse mRNA datasets and features. To support this industry-linked work, the teams are securing access to computing capabilities through collaboration with the Pawsey Supercomputing Research Centre, another NCRIS provider. This enables the computer science teams to leverage high-performance computing to train foundation models and large language models for mRNA designs.
Developing AI/ML models that are predictive of therapeutic success is a focus for other nodes of TIA’s RNA products capability and is a core function of the University of Queensland’s (UQ) BASE mRNA facility whose mRNArchitect software is now used by more than 1000 scientists globally. mRNArchitect is an open-access platform for mRNA design that optimises important attributes of mRNA, improving the strength and stability of potential therapies. Building on this user community, the platform will continue to evolve, providing vital data repositories for Australian and international research communities.
At the mRNACore facility at the Monash Institute for Pharmaceutical Sciences, sequence optimisation tools such as LinearDesign are being used to support mRNA design, complementing existing manual optimisation approaches. While the relative impact of these tools continues to be assessed, their use reflects a pragmatic approach to integrating computational methods into established workflows.
Integrating AI and ML across screening, imaging and safety
Therapeutic discovery research generates enormous amounts of data, and AI/ML approaches are proving useful in ensuring that researchers can rapidly extract the most meaningful results from these data. Computational, often referred to as in silico, drug discovery has long been used to identify molecules most suitable to proceed to in vitro studies. With a range of off-the-shelf software now available to expedite this process further, the National Biologics Facility at UQ is using in silico tools to support protein sequence design and collaborating externally to develop drug stability assessment tools that could predict shelf life of novel therapies, whilst the Australian Translational Medicinal Chemistry Facility at Monash University is generating molecule to target binding insights using Boltz-2, a structure and affinity prediction model. Â
Similarly, the Stafford Fox Drug Discovery Facility at the Murdoch Children’s Research Institute and the Cell Function & Screening Facility at the Victor Chang Cardiac Research Institute’s Innovation Centre (VCIC) integrate ML into their high-content, image-based drug screening workflows to enhance data processing, phenotypic profiling, hit candidate identification, hit-to-lead and lead assets prioritisation. This also allowed for establishment of clinically-benchmarked in vitro assessment of cardiac safety for robust early-stage derisking drug discovery and is now available at the VCIC.
The Victorian Centre for Functional Genomics at Peter MacCallum Cancer Centre uses machine learning to quantify cellular phenotypes in response to gene or drug perturbations and high throughput transcript profiling, clustering significantly similar morphologies to identify novel agents with conserved mechanisms of action. The Monash Fragment Platform (MFP) at Monash University also offers AI-based analysis of DNA-encoded library screening data, helping to identify commercially available analogues of hits emerging from screening campaigns. While this capability is still developing and delivered in collaboration with a third-party contract research organisation, it is available to clients accessing MFP services.
Accurate prediction of potential toxicity early in development enables researchers to focus resources on molecules that are more likely to be safe. Facilities like the Queensland Emory Drug Discovery Initiative at UQ and the VCIC’s Cell Function & Screening Facility are using AI and ML approaches to create predictive models for toxicity. Elsewhere in the TIA consortium, the National Biologics Facility is collaborating externally to develop drug stability assessment tools that could predict shelf life of novel therapies.
Â
Investment in infrastructure and accessÂ
It is clear that TIA’s capabilities are embracing the AI/ML revolution in therapeutic drug discovery and development to build state-of-the-art and novel models that will expedite drug development and improve safety. TIA facilities have been investing in the underlying infrastructure, such as high-performance workstations and off-the-shelf software, needed to adopt and scale these approaches.
For example, facilities are similarly exploring how emerging digital platforms may complement existing capabilities. At the Protein Expression Facility at UQ, potential collaborations with external AI-enabled service providers are being assessed to determine whether these platforms align with facility workflows and user needs. This evaluative approach reflects a broader emphasis on ensuring that new tools add clear value before being embedded into service offerings.
Assessment of how these software tools integrate into routine workflows and meet user requirements is paramount in these investments, ensuring researchers have access to the very best tools that add clear value to their research translation activities.
Across facilities, the emphasis is on access, integration, efficiency and sustainability, ensuring that digital and computational tools can be used alongside experimental capabilities rather than in isolation.
Â
Looking ahead
AI and ML will continue to play an increasing role in how therapeutics are discovered, developed and translated. Across TIA-supported facilities, future plans are focused on building on existing strengths rather than introducing entirely new directions.
This includes:
- further integrating AI and ML into translational workflows
- developing shared datasets, reference standards and training resources
- developing sovereign IP and approaches in translational and discovery AI/ML
- strengthening links between biological, computational and structural approaches
- leveraging national high-performance computing infrastructure
- addressing unmet health needs, from personalised therapeutics to rapid responses aligned with public health priorities such as pandemic preparedness
These activities align with national priorities outlined in the Australian Government’s National AI Plan, which aims to build trusted, world-class AI capability that supports productivity, resilience and the public good.
Through coordinated investment in infrastructure, software and expertise, TIA-supported facilities are well positioned to continue enabling research translation as AI and ML become standard components of therapeutic development.
