Predicting enzyme class with Rough Sets
Lihua Tang, Jun Wang · 2016
Protein structure based enzyme class prediction is a useful and challenging task in protein analysis. Here we describe a new method for the prediction of enzyme class based on Rough Sets theory, which is a supervised and rule-based learning method. The method can assign protein function from structure with simple attributes calculated from the primary sequence, such as amino acid compositions and physicochemical properties. Regarded as conditional attributes, these attributes are used in constructing the decision system and generating associate rules, which could be applied in classifying new objects. The results showed that compared with other approaches, the Rough Sets based method can achieve acceptable accuracy in enzyme class prediction and it may become a promising high-throughput tool for proteomics and bioinformatics.