Improved machine learning using adaptive boosting algorithm in membrane protein prediction

Abstract

Membrane protein are very important and play significantly in the field of biology and medicine. The main purpose is to find suitable features of a membrane protein. Various features extraction methods are use to find membrane protein and their types. PseAAC (Pseudo Amino Acid Composition) is a one of the feature extraction method which was used to find the localization of the protein, which helps in the detection of membrane types. Therefore, in this study, a novel feature extraction method which is an integration of the pseudo amino acid composition integer values mapped in discrete sequence numbers in a matrix. The proposed scheme avoids biasing among the different membrane proteins and their types. Decision making for predicting the identification of membrane protein types was performed using an algorithm framework to improve the learning accuracy, by putting the training samples weights in the learning process of AdaBoost. The performance of different ensemble classifiers such as Random Forest, AdaBoost, is analyzed. The best accuracy achieved is 91.50% for with the Matthews correlation coefficient is 83.0%, and Cohen’s Kappa value is 82.7%.