Prediction can be done by using the caret package in R.

To read more about Caret package, you may like to check the URL attached.

R can read .CSV or .txt files, so you don't need a database here as your prediction problem is analytical and not interactive.

Good Luck

http://machinelearningmastery.com/caret-r-package-for-applied-predictive-modeling/

Thanks a lot Noha for the valuable information...i will check the attached link.

The following are the databases that would  give relevant or related information on interaction partners based on the sequence info, protein name, host, disease, etc.

http://www.phi-base.org/

http://www.uniprot.org/

Thanks a lot Dr. Amalraj. These links are helpful for me.