Enhancing Accuracy and Correlation Coefficients of Cancer epitopes through Machine Learning Approaches

Determining the epitope's binding regions is incredibly difficult. Scientific research on epitopes in folded proteins is still challenging and time-consuming. Various services have been created in the past to forecast B-cell epitopes. Due to the diversity of datasets and the fact that the same technique does not produce acceptable accuracy for all window lengths, the server's accuracy has only made moderate improvement despite tremendous effort. In this study, two models were subjected to machine learning approaches. Using several models, such as IBK and Nave Bayes, both models are trained to distinguish between epitopes and non-epitopes. For Model 2 we achieved an improvement in Mathew's correlation coefficient in the range of 0.9-1.0 and for Model 1 in the range of 0.5-0.6.