Evaluation of Multi-site Lambda Dynamics Simulations Using Factor Xa, Tankyrase and Kinase Inhibitors
Optimization of protein-ligand binding affinities belongs to one of the central goals in early-stage drug discovery. Tremendous progresses in molecular dynamics based alchemical free energy calculations allows computational chemists to predict changes in binding free energies between ligands in response to variation of substituents in an accurate and robust manner. Multi-site lambda dynamics (MSλD) improves scalability by evaluating relative free energies of multiple ligands in a single simulation setup, mimicking a competitive binding assay. Recent implementation of the MSλD method as an automated workflow in BIOVIA Discovery Studio makes it possible to assess free-energy changes of large compound sets with different substituents on multiple sites within reasonable computing time.
In this talk, I will present three challenging test cases for evaluation of MSλD method on real medicinal chemistry projects. In the first two cases, we predicted relative binding free-energies of factor Xa-inhibitor and tankyrase-inhibitor complexes, where for each system several sets of inhibitors with substantially different scaffolds and binding modes were investigated. In the third test case, we predicted the binding free-energy differences of the same compound sets on a kinase and its gate-keeper mutant. I will present the accuracy of the prediction and the underlying problem for each test cases. Possible solutions to improve the prediction will be provided and discussed.