Algorae Pharmaceuticals (ASX: 1AI) has reported independent preclinical validation of its AlgoraeOS v1 drug-combination prediction platform following a major study at the Victorian Centre for Functional Genomics at the Peter MacCallum Cancer Centre.
The program tested 21 AI-predicted drug combinations across four cancer cell models, generating about 10,000 data points to assess predicted synergy.
Algorae believes the findings confirm that AlgoraeOS can prioritise combinations more likely to demonstrate real biological synergy, marking a critical technical de-risking milestone.
Chief scientific officer James McKenna said the results “mark an important milestone in the real-world testing of the AOS1 in silico predictions” and provide the basis for future preclinical programs.
Validation Program Outcomes
The study assessed 84 cannabidiol–drug–cell combinations, including predicted synergistic and non-synergistic pairs, using four established synergy models: Highest Single Agent, Bliss independence, Zero Interaction Potency and Loewe additivity.
Algorae reported that AlgoraeOS v1 showed strong predictive alignment with three of the four independent models and correctly excluded most combinations identified as non-synergistic.
The validation demonstrated that AlgoraeOS v1 can discriminate between combinations more likely to show synergy and those that are not, enabling laboratory resources to be focused on the most promising candidates.
Several drug combinations met predefined thresholds for synergy during testing and will now be evaluated for further in vitro or in vivo assessment.
Algorae said the exercise also clarified which analytical models provide the most reliable insights for future platform development and optimisation.
Integration into Platform Development
Key learnings from the study are being integrated into the continued evolution of AlgoraeOS through refinements to internal ranking functions, reference models, and threshold settings.
The data will inform the design of future preclinical programs aimed at efficiently identifying high-value synergistic combinations.
Algorae said the validation supports its broader strategy to use the platform across oncology and additional therapeutic areas in partnership with leading research centres.
The company expects to release the first in silico predictions from AlgoraeOS v2 in December 2025, which will guide the next phase of preclinical evaluation.
This will include identification of new synergistic drug combinations and prioritisation of repurposed candidates.
AlgoraeOS v2 Launch
In parallel with the validation program, Algorae launched AlgoraeOS v2—a major upgrade built in collaboration with the University of New South Wales and CSIRO Data61.
The new version has trained on more than 5.5 million inhibition records, models full dose–response surfaces across all four synergy formalisms, and provides confidence-weighted predictions.
Benchmarking shows v2 outperforming representative state-of-the-art models—including those developed by Google DeepMind—with lower error and stronger correlation across key prediction metrics.
Algorae said the combination of independent validation for v1 and the launch of v2 positions the platform as a decision-grade engine for drug-combination discovery and pharmaceutical development.
