LDD-score: A new index for topology prediction of transmembrane proteins based on data fusion

The topology prediction of transmembrane proteins has been a hot topic in bioinformatics for the last decades, and many prediction methods have been developed. Method based on the Dempster-Shafer theory of evidence is one of them. The basic probability assignment (BPA) is an essential element in the Dempster-Shafer theory of evidence, which is widely used in many fields. In this paper, a new index called LDDscore has been proposed, which can be regarded as BPA in the topology prediction of transmembrane proteins based on data fusion. We use five individual prediction methods as basic predictors. After fusing the above methods' prediction results, the final prediction result can be constructed. During the process, the LDD-scores are regarded as the BPAs, so that a better result can be obtained. An illustrative example is given to demonstrate the effectiveness of our proposed method.