A Technique for Prediction Cytokines based On Statistical Moments and a Random Forest Classifier

Abstract

Research in the analysis of cytokine plays an important role because of the importance that cytokine has in the treatment and analysis of disease, but the current method for cytokine identification have numerous weaknesses, such as low affectability and low F-score. In this paper we purposed a new prediction method by consolidating the protein place explicit propensity into general type of pseudo amino acid sequences. Our predictor model has used CSM, PRIM, RPRIM, FMD, AAPIV, RAAPIV based on ANN or RFF algorithm to compute the Accuracy, Sensitivity, Specificity and MCC which are 96.28%,88.96%,99.94%,91.73% respectively using 10-fold cross validation. RFA shows 96.28% result. Our model has given the more accuracy other than research models using SVM.