An Efficient Approach for Prediction of Nuclear Receptor and Their Subfamilies Based on Fuzzy k-Nearest Neighbor with Maximum Relevance Minimum Redundancy

The efficient classification of nuclear receptors and their subfamilies plays an important role in the detection of various diseases such as diabetes, cancer, and inflammatory diseases and their related drug design and discovery. As of now, few methods have been reported in literature for the same but the performance and efficacy of these methods are not up to the desired level. To address the issue of efficient classification of nuclear receptor and their subfamilies, here in this paper we propose to use a fuzzy k-nearest neighbor classifier with minimum redundancy maximum relevance for the classification of nuclear receptor and their eight subfamilies. The minimum redundancy maximum relevance algorithm is used to select the optimal feature subset and observed that highest accuracy and Matthew's correlation coefficient is obtained with 150 features among 753 features through fuzzy kNN classifier. The performance of fuzzy kNN classifier depends on two parameter number of nearest neighbor (k) and fuzzy coefficient (m) and it is observed that the highest accuracy and MCC is obtained at k = 7 and m = 1.25. The overall accuracies of tenfold cross validation with optimal number of features, k and m are 100 and 91.7% and the MCC values of 1.00 and 0.89 for the prediction of nuclear receptor families and subfamilies respectively. From the obtained results and analysis it is observed that the performance of the proposed approach for the classification of nuclear receptor and their eight subfamilies is very competitive with some other standard methods available in literature.