energy minimization would work in this but have you converted your ligand into 3d because sometimes 2d ligand will result in positive value.!!!

please change ligand

Getting a high affinity on dockings of very small molecules like the one you are using is "expected". Most of these molecules rely on water molecules to get to / interact with the active site of the enzyme. Removing water molecules from a structure is one of the first steps when preparing it for docking with AutoDock and AutoDock Vina. In addition you also have to consider that these small molecules are often only used as a product when they reach a "high" concentration. I have recently tried to dock ethanol to alcohol dehydrogenase (EC 1.1.1.1) and, it's affinity was very high (-2.4kcal/mol) despite the fact that this is its natural substrate.

Something else you should also consider is that AutoDock and AutoDock Vina exhibit size-related bias on the scoring and as such, smaller molecules will generally get worse results. (You can read about this in the article that I have attached)

Like the previous answers have already mentioned you should always try to minimize the energy of your ligand and check if it has been converted to 3D.

Vinod Kumar also raises an important issue, you should change your ligand. You have to consider the fact that this enzyme is usually responsible for reducing alcohols to aldehydes in animals (usually ethanol to ethanal (first link) but they can also work with secondary alcohols (second link)). In this scenario, your cyclopentanone is a product and not a substrate, meaning it will have a worse affinity. Try to perform the dock again but using cyclopentanol and see if your results differ greatly. If you are using AutoDock Vina you should also improve the "exhaustiveness" value.

Finally I would just like to add that there are two things that you should keep in mind the next time that you need help on docking / enzyme:

http://www.genome.jp/dbget-bin/www_bget?rn:R00623

http://www.genome.jp/dbget-bin/www_bget?rn:R00624

Article Virtual Screening for HIV Protease Inhibitors: A Comparison ...