Efficient calculation of binding free-energies and druggability in challenging drug targets
This four year PhD research project, due to start in October 2019, is a collaboration between University College London and UCB pharma, a global biopharmaceutical company (www.ucb.com). The doctoral student will be formally based at UCL in the group of Francesco Gervasio but will also have a placement of three months or more at UCB, Slough, UK. The successful candidate will be awarded a doctorate from UCL.
Recent estimates put the development cost of a new medicine at over 2 billion dollars. Of the enormous number of compounds entering the drug discovery pipeline, only a tiny fraction reach the market. Failures in the later (clinical) stages are extremely costly. It is therefore of great importance to identify early on lead compounds that interact with biologically validated target and exert their mechanism without inducing adverse effects (toxicity). In this respect, computational approaches are expected to play an increasing role. Fast docking algorithms can now reliably predict binding modes. But the ranking of compounds based on affinity still requires a more in-depth understanding of the association mechanisms.3 In this respect, fully atomistic molecular dynamics based approaches are increasingly useful. Over the last 10 years, thanks to better protein and ligand force-fields, exponential increase in computer performance and better sampling and free-energy algorithms, these methods made enormous progress and can now accurately predict binding mechanisms, relative BFEs and even absolute BFEs and binding kinetics. However, to obtain absolute free energies and binding kinetics, they require very significant sampling and are still too computationally demanding for most de-novo design projects. For this reason, triaging from a virtual screening is often carried out based on knowledge about the interactions of known active compounds and visual inspection.
FLG’s group contributed to understand binding mechanisms and developed methods for accurate binding free energy calculations. His “coarse Metadynamics” approach inspired the now widely-used “binding pose Metadynamics”scoring method by a software company. Together with UCB, his group has recently thoroughly tested a number of fast free energy estimation approaches (BP Metadynamics, FEP WaterSwap) in a number of relevant biological targets. However, they concluded that, albeit they are fast they are often not able to distinguish active from inactive, so can only have limited impact. Here we plan to build on our extensive experience in method development and drug discovery and develop a computational platform based on methods such as SWISH and multiple-walker Metadynamics with funnel shaped restraints to quickly achieve the required accuracy for semi-quantitative and quantitative scoring on cheap graphic processing units (GPUs). We are confident that this is achievable based on extensive preliminary data on several targets (including GPCRs). The level of accuracy required should be tuneable: active/inactive (where active is a kd of < 200 µm) for relatively quick screenings and up to ~2kcal/mol or better for hit optimization. The proposed techniques will be experimentally validated by UCB by biophysical (microcalorimetry, SPR) and structural approaches (crystallography).
This project is fully funded by the EPSRC with an additional stipend provided by UCB. To be eligible, applicants must satisfy 3 years UK residency criteria. Please submit a full CV and covering letter to Dr. Ben Cossins ([Email Address Removed]) and Prof. Francesco Gervasio ([Email Address Removed]). The deadline for applications is June 2019.