Predicting Membrane Proteins Type Using Inter-domain Linker Knowledge

Predicting membrane type is a crucial problem in computational biology. It is closely related to the biological function of the protein and its interaction process with other molecules in a biological system. The function of a membrane protein is closely correlated with the type it belongs to. In this paper we introduce DomMat, a novel method to predict membrane protein types. The method extracts functional domains by removing all the corresponding inter-domain linkers from the membrane protein sequence. A novel matching algorithm is then introduced to measure the sensitivity of the functional domains information to the membrane protein sequences of interest. Two protein sequences are expected to be related if they contain similar functional domain information. DomMat was tested in a highquality benchmark dataset. The dataset consists of eight different membrane protein types collected from the SwissProt database. The results obtained suggested that DomMat is comparable to the state-of-the–art methods and indeed a very useful method in identifying membrane protein types.