47993Personalised therapies based on simultaneous targeting of complex oncogenic networks identified by WGS.ClosedCollaborative R&DISCFIn theory, to find an efficient treatment for cancer we have only to identify an oncogenic mutation and target it with a well characterized drug. In practice, however, every cancer is driven by multiple mutations occurring in unique combinations, which generate a complex tumour network in each patient. When oncologists have tried to practice single-target precision medicine, promising initial results fade as the full genetic complexity of the tumour is not captured. Although whole-genome sequencing (WGS) offers the opportunity to identify tumour-associated gene mutations, its present utility in personalised cancer therapy is rather limited. To give WGS an effective clinical significance it is essential to use the genomic information to build models that can match cancer patients to the optimum drug mixtures. My Personal Therapeutics (MPT) use fruit flies (_Drosophila)_ and its arsenal of sophisticated genetic tools to model individual patient's tumours features. We reconstruct tumour genetic complexity identified by WGS, in the intestine of flies. These fly "avatars" will develop the patient's tumour and will die by the same cancer, unless they take the right combination of drugs. Then we use these cancer avatar flies to test thousands of approved compounds to identify the best drug combinations that rescue cancerous avatars from lethality. In this way, we produce a precise picture of the interaction between a growing tumour within an entire animal and the drug combinations tested. Via this method, called Personal Discovery Process (PDP), MPT delivers ultra-personalised drug treatment recommendations for patients with gastro-intestinal cancers (GIC), a hard-to-treat malignancy diagnosed for 180 people every day in the UK. In this project, MPT will partner with the London IVD Co-operative, a UK NIHR centre that conglomerate UK top GIC specialists and the best infrastructure to accrue GIC patients. Our process currently takes four to six months and is expensive, making widespread uptake unlikely. In this project, we will create patient avatar models and produce drug screening data that can be used to train a machine learning tool that will match incoming patient's tumour profile with previous patient's data. In this way, using artificial intelligence, we will achieve a reduction in time taken and cost. This tool will predict the best possible combination therapies and could be widely adopted as part of cancer diagnosis and treatment.