Project AURORA

AURORA is a pioneering project exploring how agentic AI could help scientists accelerate progress against some of the toughest challenges in cancer.

Funded by Cancer Research UK through Cancer Grand Challenges, the AURORA AI co-scientists are designed to support researchers, helping them navigate complex scientific data, generate and evaluate new ideas, and identify the most promising experiments to pursue. By combining the power of AI with human expertise, AURORA aims to accelerate discovery whilst keeping scientists firmly at the centre of every decision.

Led by Marinka Zitnik at Harvard University, the $10m programme is testing how the AURORA AI co-scientists can support researchers' work on real-world cancer challenges. Three Cancer Grand Challenges pioneer teams, NexTGen, PROMINENT and MATCHMAKERS, will use the technology to build on their existing research and explore whether AI can accelerate scientific discovery.

Cancer research generates enormous amounts of information, from the molecular changes within individual cells to findings from laboratory studies and clinical trials. Bringing all this evidence together and deciding which ideas are most worth pursuing is a major challenge for researchers today.

AURORA is exploring whether agentic AI can help scientists navigate this complexity. By working alongside researchers, AI co-scientists will help generate and evaluate hypotheses, identify knowledge gaps and prioritise the experiments most likely to produce new insights.

Building on ToolUniverse, a platform developed by Marinka Zitnik’s group at Harvard University, the AI co-scientists draw on scientific literature, biological databases, experimental datasets and computational tools. This enables the AI co-scientists to bring together evidence from multiple sources, carry out analyses and explore different explanations for complex biological questions.

Cancer Grand Challenge researchers remain firmly in control throughout the process. They define the scientific questions, set the experimental constraints, evaluate the AI's recommendations and decide which ideas should be tested. Experimental results are then fed back into the system, helping to refine both the hypotheses and the AI's future recommendations.

By testing AI co-scientists within active cancer research programmes, AURORA will help determine whether AI co-scientists can help researchers make better-informed decisions and accelerate discovery against some of cancer's toughest challenges.

Accelerating the solid tumours in children challenge

Led by Nicky McGranahan at University College London, NexTGen’s AURORA project will help prioritise designs for CAR T-cell therapies for children with solid tumours. The co-scientist will combine evidence about tumour targets, the surrounding tumour microenvironment and engineered T-cell behaviour to propose combinations for experimental testing. The aim is to identify designs that warrant further development and could inform the team’s clinical programme.

Learn more about team NexTGen

AURORA could help scientists make faster progress against some of the toughest challenges in cancer research.

For NexTGen, AI co-scientists could help guide the design of CAR T cells better equipped to recognise and attack childhood solid tumours, supporting the development of more effective treatments for children. For PROMINENT, they could help researchers understand why some cells carrying cancer-driving mutations remain normal while others progress towards cancer, revealing opportunities to intervene before tumours develop. For MATCHMAKERS, they could help scientists decipher how T cells recognise cancer and identify promising tumour antigen and T-cell receptor combinations, informing the development of more effective, personalised immunotherapies.

Together, these projects will test how AI co-scientists can help researchers turn complex biological data into new insights and discoveries. The lessons learned through AURORA could shape how AI is applied across cancer research in the future, providing new tools and approaches to help researchers understand, prevent and treat cancer.

Cancer researchers have more information than ever before, but working out what it means and which experiments to try next can be difficult.

AURORA is developing AI systems that will work alongside scientists to help with this process. They will bring together information from research papers and experiments, suggest possible explanations and help scientists choose how to test them. Scientists will assess the suggestions and carry out experiments, using the results to improve the AI systems and guide further research.

Three Cancer Grand Challenges teams will help develop and test this approach. Their projects focus on designing immune-cell treatments for children with solid tumours, understanding how genetic changes contribute to cancer development, and discovering how immune cells recognise cancer. By testing the systems in these different settings, AURORA aims to establish how AI can help researchers make discoveries more quickly and effectively.