Deep Learning for CAR-T cell therapy optimisation

点击翻译按钮获取中文摘要

Adoptive cell therapy is a treatment of patients using their own, modified immune cells. It involves harvesting T cells - the immune cells that normally recognise and eliminate pathogens - from the patients blood, reprogramming them to recognise cancer cells and re-introducing the newly modified cells back into the patient to destroy tumours. The modifying factor used to arm these T cells against cancer is called Chimeric Antigen Receptor or CAR, in short. Using the lock and key analogy, just as every lock can only be opened with the appropriate key, each type of cancer, with its unique antigens, can only be recognised by a specific engineered CAR.Currently, the NHS is providing CAR-T cell therapy against malignant B cells that express antigen called CD19 to children and young adults with relapsed or difficult- to-treat leukaemia and lymphoma. The treatment is particularly effective in patients who did not respond to chemotherapy, offering real hope to those who otherwise would have no other options. However, this CAR therapy can only treat a very specific type of cancer, leaving much room for improvement.The concerted efforts into development of specific CAR therapies against other cancers are hindered by the tedious process of target discovery and receptor optimisation. Coding.bio streamlined the process of building better receptors, making it faster, cheaper, and more diverse. This project will develop a deep learning method to select lead candidates from big datasets of these receptors, making candidates more likely to succeed in the clinicOur objective is to help bring this transformative therapy to a wider patient population.