Information Loss and Noise Inclusion Risk in Mimotope Based Epitope Mapping

Mimotopes are peptides mimicking an epitope. They can readily be selected out from phage-displayed random peptides libraries and have been widely used, especially in epitope mapping. There are now several computational tools available for the mimotope based epitope mapping. Most of these tools implicitly take peptides acquired through phage display as mimotopes in an extended conformation and map the sequential neighbor residue pair of the mimotopes to the spatial neighbor residue pair on the antigen structure. In this study, we evaluated the implicit assumptions above. The results show that mimotopes have many conformations and the indirect sequential neighbors separated by one amino acid should also map to spatial neighbors, indicating there is information loss with these tools. Furthermore, all the existing tools do not have a noise sifting procedure, though peptides acquired through phage display can be target unrelated peptides also. Thus, the risk of noise inclusion may exist. Our results may be helpful for developing new tools with improved performance on mimotope mapping.