We have carried out both retrospective and prospective studies of our docking program, WILMA, and binding affinity scoring function, SIE. We have participated in three SAMPL (Statistical Assessment of Modeling Proteins and Ligands) blind tests and two CSAR (Community Structure-Activity Resource) exercises. We will present some insights that we have gained into factors affecting the reliability of virtual docking and scoring predictions.  As the expression goes, the devil is in the details. For example, even methods with the same underlying theory can produce widely varying results due to differences in the details of implementation and parameterization. We will discuss the sensitivity of our results to the details of structure preparation and refinement, sampling methods and approximations used. Realism often has to be sacrificed in the interest of reducing noise in our calculations. These factors can confound the assessment of the accuracies of scoring functions per se.

Virtual docking and affinity prediction:  Insights from benchmarking and validation studies

245th National Meeting (2013)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.