Protein sub-nuclear location by fusing AAC and PSSM features based on sequence information

To achieve good performance on Protein sub-nuclear location, one needs to extract a powerful representation containing rich information for identification. Various favorable techniques have been proposed, but it is believed that the single representations, containing one-sided information of protein sequence, are insufficient for discrimination. To this end, we in this paper propose the fused representations by integrating two single representations, the amino acid composition (AAC) and the position specific scoring matrix (PSSM). Due to two forms of PSSM, PsePSSM and GreyPSSM, two integrated representations, called briefly AACPsePSSM and AACGreyPSSM, are given. To evaluate the proposed representations, a benchmark data set is employed and the classical K nearest neighbor (KNN) is adopted classifier. And the experimental results show our proposed fusion representations outperform AAC and PSSM.