Assessment of MHC class II-restricted epitope prediction in silico

Antigen processing and presentation by HLA class II is a complex process and current in silico predictions are not able to account for all the factors involved in immunodominance/immunogenicity as they focus only the binding component of the process. Generation of large training datasets from single HLA class II allele cell-based or cell-free MAPPS-like platforms is enabling development of algorithms for identification of naturally processed ligands. The combination of binding and processing is a more reliable approach to the identification of viable CD4+ T cell epitopes, and to rank molecules. In silico prediction still needs to be part of a more complex immunognicity risk asessment strategy.