Combining NMR and artificial intelligence to characterise large enzymes involved in diseases

Today, powerful deep learning and artificial intelligence approaches are capable of resolving high resolution static structures of proteins. However, critical aspects of protein function such as their dynamics, allostery and conformational heterogeneity are more challenging to elucidate. Obtaining this information is often the barrier between understanding regulatory and disease mechanisms of challenging therapeutic targets. To this end, Nuclear Magnetic Resonance (NMR) spectroscopy is a powerful tool and has the potential to provide experimental atomistic insight into these aspects of protein function. Furthermore, access to experiments in solution means NMR provides physiologically relevant insights. However, characterising large complexes using NMR spectroscopy is very time-consuming and often limited to methyl-bearing side chains. Combining NMR spectroscopy with tools of artificial intelligence, e.g. deep learning, has the potential the open up new avenues for characterising large protein complexes and enzymes involved in diseases.

This project aims to explore the promising synergy between NMR and deep learning. Specifically, this project builds off a recently developed strategy being developed in the Hansen lab for analysing and transforming complex NMR data. The specific aims are:
(1) Unlock new types of NMR experiments for new insights into large proteins.
(2) Demonstrate applications to challenging therapeutic targets beyond the reach of conventional NMR.
(3) Improve the accessibility of NMR to other scientists by
(3a) removing expensive sample preparation requirements and
(3b) streamlining highly technical acquisition and analysis workflow