Through type Ⅲ, Ⅳ, Ⅵ secretion systems, Gram-negative bacterial effector proteins can be directly injected into host cells, which causes the hosts infected with various diseases. Because both of type Ⅲ, Ⅳ secreted effector proteins belong to the non-classically secreted proteins (NCSPs), and may have similar sequence motifs or evolutionary conservation profiles, it is hard to distinguish them from each other. Based on support vector machine (SVM) and pseudo position specific scoring matrix (PsePSSM), a computational approach is proposed to fastly and accurately classify the type Ⅲ and Ⅳ effector proteins of Gram-negative bacteria. The test results show that this approach has a good effect on the classification of type Ⅲ and Ⅳ effector proteins, and could be used as a supplementary tool for further studies of secreted effector proteins.
