Preparing the Next Generation of Students for AI-Driven Biomedical Research: Opportunities in Higher Education
Joshua Hardy, Dylan Silke, Ashley Weir, Daniel Brown, Kerry Ko, Brendan Ansell, Jessica Borger and Megan Taylor
Walter and Eliza Hall Institute of Medical Research and Department of Medical Biology, University of Melbourne

The rapid advancement of artificial intelligence (AI) in biomedical research has accelerated scientific discovery, particularly in fields like de novo protein design. A recent University of Melbourne, Bachelor of Science Honours project, jointly supervised in the Lucet and Papenfuss labs, at the Walter and Eliza Hall Institute of Medical Research (WEHI) exemplified this potential, showcasing the power of interdisciplinary collaboration between experimental and computational scientists in de novo protein design. Recognising the need to prepare future biomedical researchers for an AI-embedded research landscape, we have conceptualised a new subject that incorporates coding and AI literacy in a biomedical research context.

ProteinDJ: where bench science meets AI

Protein binders can be engineered to recognise and specifically bind to a defined target protein to support its visualisation or enact post-translational modifications, relocalisation or degradation.

ProteinDJ¹, a new tool for de novo protein binder design, was developed using funding from the NHMRC. It was part of a novel research project that brought together a supervisor with extensive structural biology knowledge and a student with strong computational skills and AI expertise.

ProteinDJ integrates deep learning AI tools AlphaFold2 (1) and ProteinMPNN (2) with the de novo protein design tool RFDiffusion (3), to generate novel functional protein binders. Using AI and coding, the student developed this user-friendly package that can leverage high-performance computing systems, dramatically speeding up binder generation (Fig. 1). ProteinDJ implements new metrics to evaluate binder quality, including their biochemical attributes such as hydrophobicity, size and charge, with visual plotting of results to help with design filtering and selection. The designs can then be validated by wet lab researchers, using in vitro binding assays and assessing functional impact.

Fig. 1. Overview of the ProteinDJ de novo protein binder design pipeline. The target protein structure is supplied to RFDiffusion, which generates thousands of potential binder structures through iterative refinement. The output structures are suppli
Fig. 1. Overview of the ProteinDJ de novo protein binder design pipeline. The target protein structure is supplied to RFDiffusion, which generates thousands of potential binder structures through iterative refinement. The output structures are supplied to ProteinMPNN, which generates a feasible sequence for each putative binder. Finally, AlphaFold2 is used to validate each binder–target structure. The putative protein must then be validated experimentally to check for specificity of binding to its target and functional impact. Created in biorender.com

The success of the ProteinDJ project highlights the importance of collaboration between experimental and computational scientists to integrate traditional wet lab techniques with cutting-edge AI technologies. To address the rapidly evolving landscape of biomedical research and the growing role of AI, the Scientific Education team at WEHI have developed a new Master of Biomedical Science elective subject, Coding and Data Analysis in Biomedicine, at the University of Melbourne.


Preparing biomedical graduates for AI-integrated biomedical research

The Coding and Data Analysis in Biomedicine graduate subject combines theoretical lectures and tutorials presented by leading experts in biomedical research with hands-on coding workshops. The subject aims to equip students with the core skills needed to succeed in an AI-embedded research environment.

The lecture series exposes students to a broad range of computational approaches used in biomedical research. For example, one lecture introduces students to core principles of machine learning and how these tools are evaluated and applied in biomedical datasets (4). In another, Professor Jodie McVernon explores the use of modelling techniques in interdisciplinary settings, such as COVID-19, demonstrating how computational tools can inform public health decision-making (5).

In parallel, the subject offers a series of coding workshops in which students learn to program using R. These workshops build foundational coding skills and expose biomedical students to one of the languages commonly used by computational biologists. This shared understanding enhances communication between biomedical and computational researchers, fostering more effective interdisciplinary collaboration. For some of the biomedical science students, these workshops may serve as an entry point into computational biology and may inspire a transition into AI and machine learning careers.

Ultimately, this subject aims to build students’ confidence in using AI tools and navigating the intersection of computational and biological approaches.

By integrating these skills, students will be better prepared to undertake research projects that combine AI and biomedical science, accelerating discoveries and contributing to the next generation of AI-embedded biomedical research.

References
1. Jumper J, Evans R, Pritzel A, et al. Nature 2021. 596(7873):583–589.
2. Dauparas J, Anishchenko I, Bennett N, et al. Science 2022;378(6615):49–56.
3. Watson JL, Juergens D, Bennett NR, et al. Nature 2023;620(7976):1089–1100.
4. Zhu JY, Zheng DW, Zhang MK, et al. Nano Lett 2016;16(9):5895–5901.
5. Marcato AJ, Black AJ, Walker CR, et al. Lancet Reg Health West Pac 2022;28:100573.

About the authors
Author photoDr Josh Hardy is a Senior Research Officer in the Department of Medical Biology (WEHI), University of Melbourne. josh.hardy@unimelb.edu.au

Author photoDylan Silke is an Honours student in the Department of Medical Biology (WEHI), University of Melbourne dsilke@student.unimelb.edu.au

Author photoAshley Weir is a PhD candidate in the Department of Medical Biology (WEHI), University of Melbourne. ashley.weir@student.unimelb.edu.au

Author photoDr Daniel Brown is a Senior Research Officer in the Department of Medical Biology (WEHI), University of Melbourne dbrown1@unimelb.edu.au

Author photoKerry Ko is the Senior Scientific Education Officer in the Department of Medical Biology (WEHI), University of Melbourne. kerry.ko@unimelb.edu.au

Author photoDr Brendan Ansell is a Senior Research Officer in the Department of Medical Biology (WEHI), University of Melbourne brendan.ansell@unimelb.edu.au

Author photoAssociate Professor Jessica Borger is the Director of Education in the School of Translational Medicine, Monash University. jessica.borger@monash.edu

Author photoDr Meg Taylor is the Scientific Education Program Leader in the Department of Medical Biology (WEHI), University of Melbourne. meg.taylor2@unimelb.edu.au Corresponding author

This article first appeared in the ASBMB Education Feature of Australian Biochemist Vol 56 No 2, August 2025, pages 17–18.


ASBMB Membership Office
T +61 (03) 9907 8648 
E asbmb@wsm.com.au

© 2026 ASBMB. All rights reserved.