Data-Driven Computational Methods for Cyclic Discovery Processes

The aim of this project is to investigate-in a challenging real-world drug discovery process-how data-driven computational methods can support discovery processes that operate in design-make-test-analyse cycles. This project promises to (i) enable effective, well-justified machine learning applications in discovery processes, (ii) drive forward research into theory and algorithms for machine learning in discovery processes, (iii) pioneer a sound data-driven approach to reduce attrition in drug discovery processes, and (iv) develop highly promising compounds for the treatment of idiopathic pulmonary fibrosis, a chronic lung disease. To ensure success, our consortium covers a wide range of expertise in data-driven computations techniques and drug discovery. The unique composition of challenges encountered in design-make-test-analyse cycles are partly general and partly domain-specific. Evaluation of novel methods in these settings is only meaningful in real-world applications as retrospective analyses cannot do justice to the complexity and flexibility of a true design phase coupled with real making and testing of structures. Therefore, the pharmaceutical companies GlaxoSmithKline and Evotec are partners in the project and we incorporate the synthesis capabilities of the University of Nottingham with their quality-assured in vitro testing capabilities.