Data mining application in biomedical informatics for probing into protein stability upon double mutation

To explore the mechanism of protein stability change is one of the important topics in protein design. The accurate prediction of protein stability change upon mutation is very useful for enhancing the experimental efficiency in many biological and medical studies. In this work, we aimed at effectively introducing data mining technologies for investigating the understanding of protein stability change upon double mutation. We constructed a non-redundant dataset of protein mutants with various attributes and applied systematically analyses on the dataset. Therefore, we developed general knowledge from the dataset by several data mining techniques, including decision tree, decision table and association rule algorithms. Furthermore, we interpreted, evaluated, and compared those knowledge outcomes obtained from different techniques. The observations on the experimental results demonstrated that the present method may serve as an effective tool in biomedical informatics to understand the prediction of protein stability change upon double mutation.