Algorae Pharmaceuticals (ASX:1AI) has taken a significant step forward in AI-enabled drug discovery, announcing the completion of a sweeping set of in silico synergy predictions generated by its upgraded platform, AlgoraeOS v2 (AOS2).
The new dataset, covering more than 500,000 potential CBD–drug–cell line combinations, marks a significant expansion of the company’s capability to identify and prioritise promising therapeutic candidates at unprecedented scale and depth.
AOS2 analysed combinations of cannabidiol (CBD) with more than 3,000 approved and investigational medicines across 170 cell lines, producing one of the most comprehensive AI-driven landscapes yet reported by an Australian biotechnology company.
The company said early performance results suggest that AOS2 outperforms other global models, including those developed by Google DeepMind, across all four primary metrics. They are ZIP, Bliss, HSA and Loewe. The model also demonstrated stronger calibration across biologically diverse and clinically relevant interaction patterns.
Importantly, each prediction is accompanied by confidence-weighted uncertainty scores that quantify both data and model-driven uncertainty. This provides Algorae with a sophisticated framework for risk-aware prioritisation, sharpening decision-making as candidates move toward laboratory testing. A preliminary review of the dataset has already identified 90 high-quality combination candidates for potential progression into preclinical validation.
Chief Scientific Officer Dr James McKenna said the enhanced granularity of AOS2 is critical for improving the precision and efficiency of early-stage drug discovery. “While the improved predictive capabilities of AOS2 are obviously advantageous, the enhanced granularity around synergy predictions, coupled with embedded risk-aware uncertainty metrics, cannot be overstated. Together, these tools provide a powerful approach for prioritising candidates for further interrogation and progression to in vitro validation assays.”
Associate Professor Fatemeh Vafaee of UNSW Sydney and the UNSW AI Institute, who collaborated on the development of AOS2, said the platform demonstrates the potential for sovereign Australian AI to reshape drug discovery pipelines. “AOS2 demonstrates the kind of high-performance AI needed to expand the therapeutic landscape. What makes this platform compelling is that it outperforms global state-of-the-art models while remaining lightweight, deployable and biologically grounded. Its ability to generalise across diverse contexts, combined with confidence-weighted outputs, gives researchers higher clarity in prioritising combinations for downstream validation.”
Algorae said it will now begin a deeper assessment of the 90 shortlisted combinations, with commercial and intellectual property considerations guiding the final selection. The company has also commenced discussions with the Peter MacCallum Cancer Centre regarding a second program and independent validation of AOS2 predictions—building on PMCC’s earlier validation work with AlgoraeOS v1.
Alongside its AI discovery activities, Algorae continues to expand its pharmaceutical commercialisation business, AlgoraeRx, which sources and supplies generic and specialty medicines across Australia and New Zealand. The company says it will keep shareholders updated as further milestones are met.