Enzyme vs. non-enzyme classification based on principal component analysis and AdaBoost classifier

Classifying Enzymes and Non-Enzymes sequences are important and it saves efforts and time. In this paper, machine learning model is proposed to classify the unknown sequences into Enzymes or Non-Enzymes. The proposed model extracts the features from all sequences using the online tool Pepstate from repository “SWISS-PORT” database. Due to high dimensionality, Principal Component Analysis (PCA) technique is used to reduce the dimension of the extracted features. AdaBoost classifier is then used to match the features of unknown sequence with the features of training or labelled sequences. The number of the eigenvectors and the size of the AdaBoost ensemble are investigated to show the influence of each on the accuracy and the CPU time of the proposed model. The proposed model achieved promising results compared with many state-of-the-art methods.