Prediction of Linear B-cell Epitopes using Manifold Adaptive Experimental Design and Random Forest Algorithm

Identification of B-cell epitope plays an important role in the design and development of immunodiagnosis kits and vaccines. Feature preprocessing approach and machine learning method have important influence on the development of epitope prediction models. Although several epitope prediction models based on machine learning have been developed, their accuracy are still unsuitable for vaccine development. Thus, a new and suitable method is necessary to improve prediction. In this study, we developed a novel framework based on random forest algorithm (RF) and Manifold Adaptive Experimental Design (MAED) algorithm for improved linear B-cell epitope called RF-maed. For testing dataset, the sensitivity (SEN), specificity (SPE), accuracy (ACC), and Matthews correlation coefficient (MCC) the RF-maed were 0.978, 0.993, 0.985, and 0.971, respectively. These experimental results demonstrate the accuracy and efficiency of our method using random forest algorithm to promote linear B-cell epitope prediction. The results suggest that the developed RF-maed is practical for the identification of B-cell epitope and developing reliable predictive model.