Prediction of transmembrane protein topology based on power average operator

The prediction of the topology of transmembrane proteins is an important foundation in the research of bioinformatics. In this paper, information fusion theory has been introduced to locate the transmembrane regions and find their orientation. The proposed method, which is on the basis of the power average operator and a kind of similarity degree of interval numbers, considers a transmembrane region as an interval number. By constructing some base predictor, several independent prediction results have been generated. And then, the real transmembrane regions will be predicted by fusing those prediction results using power average operator. The algorithm is tested on a data set extracted from SWISS-PROT. The results shown are satisfactory.