BACKGROUND: Walking along disease-associated protein sequences in the search for specific segments able to induce cellular immune response may direct clinical research towards effective peptide-based vaccines. To this aim, we are studying the targets of the immune response in autoimmune diseases by applying the principle of non-self-discrimination as a driving concept in the identification of the autoimmunogenic peptide sequences. METHODS: Computer-assisted proteomic analysis of the autoantigen protein sequence and dot-blot/NMR immunoassays are applied to the prediction and subsequent validation of the epitopic sequences. RESULTS: Using the experimental model Pemphigus vulgaris/desmoglein 3, we have identified the antigenic linear determinant recognized by MAb 5H10, a monoclonal antibody raised against the extracellular domain of human desmoglein-3. The computer-assisted search for the Dsg3 epitope was conducted by analyzing the similarity level to the mouse proteome of the human desmoglein protein sequence. Dot-blot immunoassay analyses mapped the epitope within the sequence Dsg349-60 REWVKFAKPCRE, which shows low similarity to the mouse proteome. NMR spectroscopy analyses confirmed the specificity of MAb 5H10 for the predicted epitope. CONCLUSIONS: This report promotes the concept that low level of sequence similarity to the host's proteome may modulate peptide epitopicity.