Identification of Anticancer Peptides Using Optimal Feature Space of Chou's Split Amino Acid Composition and Support Vector Machine

Abstract

Cancer is a serious disease and occurs the cause of death around the world. Various traditional methods i.e. targeted therapy chemotherapy and radiation based therapies have been extensively used by the investigators but still it is considered ineffective due to its high cost, side effects and Vulnerability towards finding errors. Therefore; an automatic and efficient model highly desirable to identify anticancer peptides. In this paper, the peptides sequences are formulated using three numerical descriptors namely; Split amino acid composition, dipeptide composition and Pseudo amino acid composition. The predicted outcomes of the proposed method is evaluated using two different nature classification learners, i.e., instance based k-nearest neighbor and Support vector machine. Our proposed model achieved the an accuracy of 93.31% sensitivity of 86.23% and specificity of 98.06% and MCC of 0.86, the success rate shows the remarkable improvement in performance matrices in comparison with existing techniques in the literature. It is observed that our proposed method will be useful for the investigators in the area of drugs design and proteomics.