I am testing a new scoring function for docking and I am looking for simple protein-ligand data sets from DUD-E (only!) in which the binding cavity is well defined and not very flexible. As such, rigid receptor docking is more likely to produce a near native pose than in cases where the protein is very flexible and induced fit ligand binding is prominent. Also, if possible, cases with not too many ambiguous crystal waters that mediate the binding of a subset of the ligands but are replaced by the rest. Essentially, I am looking for data sets where correct docking pose prediction is easy, since I am interested only in the performance of the scoring function. So far the HSP90 set yielded very good results. Has anybody favorable experience with docking using DUD-E data sets?
I think there are no best set.
There are a number of available database can be used, such as PBDbind.
Stefan M Noha
I am aware that DEKOIS 2.0 is superior, but you are out of topic. I explicitly state that I need a dataset from DUD-E.- Badges
- Science topic