Our programme builds on our world-leading expertise in causal analysis (specifically, Mendelian randomization) and population-based bioinformatics and genomic sciences. These methods facilitate robust discrimination between causal exposures that are possible targets for behavioural or therapeutic interventions to prevent cancer (primary prevention) or its progression (tertiary prevention), from non-causal biomarkers which may nevertheless act as predictive biomarkers (secondary prevention).

To address the objectives, we integrate state-of-the-art methods across four work packages (WPs):

• WP1 Mendelian randomization
• WP2 Molecular mechanisms
• WP3 Predictive biomarkers
• WP4 Designing and evaluating interventions

The programme is underpinned by strands in: Bioinformatics, AI and data mining; Evidence synthesis and triangulation; Capacity building; and Patient and public involvement and engagement (PPI/E). Data sharing is prominent.

Mendelian randomisation What can we learn from genetic information about intrinsic and extrinsic causal factors in cancer risk and progression?	Molecular mechanisms What are the molecular mechanisms that underly cancer initiation and progression, and how can we use these in chemoprevention and cancer treatment?	Predictive biomarkers How can we use the genomes, behaviours and molecular by-products of cancer cells to detect cancers early?
Designing and evaluating interventions How does evidence from in-vivo studies, feasibility trials and randomised pilot studies support potential novel interventions for cancer prediction, treatment and recovery?	Bioinformatics, artificial intelligence and data mining How can our new analytical tools and methods support research projects across the programme and more widely amongst researchers in academia and industry?	Evidence synthesis and triangulation What can we learn from analytical automation, text mining and risk-of-bias tools about the mechanistic and causal pathways that lead to cancer?