Algorae Pharmaceuticals (ASX:1AI) has taken a step forward in proving the real-world capability of its AI-driven drug-combination engine, following the release of independently generated validation data from the Peter MacCallum Cancer Centre.
The findings mark the first external confirmation that the company’s proprietary platform, AlgoraeOS, can successfully predict and prioritise drug combinations with true biological synergy.
The study, completed at the Victorian Centre for Functional Genomics (VCFG) and led by Professor Kaylene Simpson and Algorae’s Chief Scientific Officer, Dr James McKenna, tested 21 drugs predicted by AlgoraeOS v1 (AOS1) for synergy with cannabidiol (CBD) across four cancer cell models. In all, more than 10,000 individual data points were generated from 96 preclinical assessments, covering 84 CBD–drug–cell combinations.
Using four internationally recognised models of drug interaction, the researchers confirmed that AOS1 could meaningfully differentiate synergistic from non-synergistic combinations, demonstrating predictive strength across three of the four recognised synergy frameworks.
“These validation experiments mark an important milestone in the real-world testing of the AOS1 in silico predictions,” Dr McKenna said. “By testing both positive and negative predictions of synergy, we have gained important insights into the strengths of the baseline version of AlgoraeOS. Completing this program in collaboration with the VCFG has allowed us to successfully benchmark the capabilities of AOS1 and build capacity to rapidly test future in silico predictions. The promising combinations identified from this initial study can now be considered for further assessment.”
Algorae described the results as a “technical de-risking milestone”, with the study confirming that AOS1 can both identify combinations likely to show real synergy and filter out those unlikely to yield clinical value. The company says these insights will directly shape future R&D decisions, ensuring that laboratory resources are directed toward the most promising combinations.
The results also validate which analytical synergy models are most informative for future work and will guide refinements to the platform’s internal scoring, reference thresholds, and preclinical study design.
In parallel, Algorae confirmed the launch of AlgoraeOS v2 (AOS2), a significant upgrade developed in collaboration with UNSW and CSIRO’s Data61. The company said the new version, trained on more than 5.5 million inhibition records, has already outperformed state-of-the-art models, including frameworks from Google DeepMind, on published benchmarks.
The first in silico predictions using AOS2 are expected before the end of December 2025.
The validation strengthens Algorae’s broader strategy, which combines AI-powered drug-combination discovery with a commercial pharmaceutical business supplying generic and specialty medicines in Australia and New Zealand. The company says its platform will now be applied across oncology and other therapeutic areas through partnerships with leading research centres.