Predicting The Single And Multiple Subcellular Location Of Apoptosis Proteins Based On Multi-Features Fusion

Proteins are key ingredients in organisms, However, proteins have different functions in various parts of the cell. Therefore, predicting the subcellular location of proteins to better understand their functionwe constructed a new single-location dataset, further enrich the protein sequence number and subcellular location categories. Several biological features, protein blocks composition, Amino acid n-peptide composition infonnation, average chemical shifts composition, amino acid hydropathy infomation and PSSM, were effectively applied to predict the subcellular location of apoptosis protein by using support vector machine (SVM) algorithm and K-Nearest Neighbors algorithm,The overall prediction accuracies of combining with above information parameters by the jackknife tests are 81%,77.9%.The single-location apoptotic protein dataset as the standard dataset, we also constructed a multiple-location data set as an independent test set. The approach for predicting the subcellular location of multiple-location apoptosis protein is proposed by combining the dipeptide composition and protein blocks composition, The overall success rate is 60.9% by using K-Nearest Neighbors algorithm.