MetaMHCIIpan:a consensus approach for pan-specific HLA-DR binding predictions
【摘要】：正Background:Binding of antigen peptides to Major Histocompatibility Complex ClassⅡ(MHC-Ⅱ,mainly HLA-DR for human) molecules is a core step in the adaptive immune responses.Accurate identification of MHC-restricted peptides is of great importance for elucidating the underlying mechanism of immune recognition,as well as for developing effective cpitope-based vaccines and promising immunotherapies for many severe diseases. Due to extreme polymorphism of MHC-Ⅱalleles and the high cost of biochemical experiments,the development of computational methods for accurate prediction of binding peptides of MHC-Ⅱmolecules,particularly for the ones with few or no experimental data,has become a topic of increasing interest.Recently a few so called pan-specific methods have been developed to meet this need,including our newly proposed method,TEPITOPEpan,which can make prediction with decent accuracy for over 700 HLA-DR Molecules.The performance of these methods is however not good enough and should be improved. Methods:MetaMHCIIpan is a consensus method which integrates by AvgTanh the outputs of multiple pan-specific methods for predicting binding peptides.Four state-of-the-art pan-specific predictors and two leading normal allele-specific predictors are involved in this study.After checking the performance of various combinations,an optimal one is selected as the default setting of MetaMHCIIpan. Results:Experimental results over three benchmark datasets showed MetaMHCIIpan with the default predictor combination,which outperformed all individual predictors in prediction accuracy,being statistically significant.MetaMHCIIpan is freely available at http:// www.biokdd.fudan.edu.cn/Service/MetaMHCIIpan. Conclusions:MetaMHCIIpan utilizes diverse base predictors to achieve the current best performance in predict peptides binding to MHCⅡ,keeping a nice feature of pan-specific methods.That is,MetaMHCIIpan is applicable to any HLA-DR alleles.The consensus method thus shows great potential for effective boosting the discovery of T-cell epitopes.