Polaris, the N8 High Performance Computing facility, is being used to identify new opportunities for drug design by enabling changes in protein structures to be studied.
Protein structures shown by X-ray crystallography provide a static picture of protein topology. Computational approaches have the potential to reveal concavities in this topology. However, the energetic barriers between proteins can be large, which hinders the use of conventional computational approaches.
The project team, working with N8 HPC, have developed an approach which improves the conformational sampling of molecular dynamics simulation by using a swarm of coupled replicate simulations. In this method, which the team call swarm-enhanced sampling molecular dynamics (sesMD), the replicas are able to sense each other’s presence.
Accurate prediction of protein plasticity has the potential to reveal ‘pockets’ on the surface of protein structures that may not have been apparent from experimental structure determination methods and which may be capable of transporting drugs more effectively. Targeting these pockets with structure-based design can provide new directions in the design of pharmaceuticals and other therapeutics.
The sesMD method involves multiple interacting simulation systems and therefore is inherently a parallel algorithm, requiring good interconnect linking multiple fast cores. N8 HPC provided this capability and also the capacity to run multiple simulations to allow the exploration of different initial structures and simulation conditions.
The research team applied the sesMD approach to predict the plasticity of p38α mitogen-activated protein kinase, an anti-cancer enzyme target linked to key cellular processes. From a single protein structure they were able to generate a range of different conformations. While some of these conformations had been observed in other crystal structures, some conformations had never before been revealed. These hidden pockets in the protein structure have formed the basis of further investigation of their ‘druggability’ at Johnson and Johnson.
- “The simulations were aimed at understanding protein kinase conformational changes and were conducted together with the team from the University of Manchester. The expertise and flexibility of the team from the University of Manchester, as well as the computational resources from N8 HPC were key in defining new opportunities for drug design.”
Dr Berthold Wroblowski, Johnson & Johnson R&D