Paul Hawkins Abstract

Accelerating problem solving and decision making in medicinal chemistry through visualisation

PAUL HAWKINS1, Krisztina Boda1

1OpenEye Scientific
Modern ligand discovery and optimization projects rely heavily on complex three-dimensional data for success. Whether this data is obtained from experiment (structural data from crystallography), or computation (active site pose and interaction predictions, molecular simulations, quantum mechanics) it is valuable and frequently expensive to obtain. Medicinal chemists’ efforts can be greatly accelerated if they can utilize this data effectively, but that utilization has been impeded by a lack of suitable analysis tools. Efficient analysis of 3D data first requires it to be made into comprehensible information and then, more importantly, transformed into actionable knowledge. This last part is vital to the rapid progress of a project, but is a long neglected area of cheminformatics. Taking action based on complex 3D data requires it to be understood clearly and the correct context, but this process is often impeded by the substantial language barrier that exists between chemists and their highest value data; the natural language of chemists is 2D, while the native form of their data is 3D. Proper rendition of complex 3D information into easily accessible 2D form can accelerate the understanding of that information not only for chemists, but all those involved in a project.
Here we present an approach to the effective and efficient visualization of 3D data in 2D in order to accelerate the process of decision making in molecular design. Examples of this approach will be provided for 3D data from a variety of sources:
1. Crystallography. Easy identification of problems with model fit and interpretation of B-factors to aid selection of the appropriate protein model for structure-based design.
2. Conformational analysis. At-a-glance overview of accessible conformational space. Informative comparison of conformations generated under different regimes or by different methods.
3. Docking/pose prediction. Highlighting important interactions that are and ARE NOT made between receptor and ligand to accelerate design decisions.
4. Molecular dynamics. Monitoring receptor-ligand interactions over time (transforming 4D information into 2D). Intuitive interpretation of pose stability of lead optimization candidates within the binding site to aid prioritization of chemistry effort.