We analyzed the sera of patients with melanoma to define the human humoral autoantibody profile towards HMW-MAA. Computational proteome scanning using the non-selfdiscrimination principle as a guide led to the individuation of the low-similarity HMW-MAA 781 The vast majority of melanomas develop from cutaneous intraepidermal melanocytes either in normal skin or within melanocytic nevi and progress through radial and vertical growth phases. Cutaneous melanoma accounts for about 10% of skin cancer cases but 75% of skin cancer deaths. The incidence of cutaneous melanoma is rising steadily worldwide, and the median survival of patients with distant metastases remains Ͻ1 year. 1,2 Adolescent cases are on the rise too: the incidence of melanoma increased by 85% among 15-to 19-year-olds from 1973 to 1996. 3 This situation demands efforts to develop new biologic therapies. 4 A promising approach is the use of specific immunotherapy with peptide-based vaccine strategies. Conditio sine qua non to this goal is the identification of immunogenic tumor antigens, including antigenic peptides able to evoke humoral/cellular responses. [5][6][7] Melanoma presents a heterogeneous pattern of various tumor markers, including melanocytic antigens and cancer testis antigens. 8 Among them, HMW-MAA 9 is a melanoma marker of particular interest since (i) it is highly expressed at the surface of tumor cells, (ii) it has restricted distribution in normal tissues 10 -12 and (iii) the induction of specific humoral response to anti-idiotypic anti-HMW-MAA MAb increases survival in patients with advanced melanoma. 13,14 Given these premises, identification of the epitopic sequences of this proteoglycan MAA is of great interest for obtaining specific antibodies to activate the destruction of melanoma cells. However, the dimensions of the HMW-MAA molecule do not allow the definition of HMW-MAA conformational determinants and hamper the exact individuation of the immunogenic linear peptide portions of this antigen through the usual epitope mapping methodology.We have used computational biology to exploring TAA epitopes recognized by MAbs or polyclonal antibodies. [15][16][17][18] By measuring the antigenic peptide similarity to the host's proteome, we demonstrated that a low level of similarity is a necessary condition for a peptide to be immunogenic. Consequently, based on the assumption that antigenic motifs that are scarcely represented in human proteins may offer epitopic determinants unknown to the host, we searched for immunogenic epitopes of the human melanomaassociated chondroitin sulfate proteoglycan by, first, selecting for portions of the HMW-MAA not shared with the human proteome and, then, analyzing the low-similarity peptide reactivity pattern using sera from 26 melanoma patients.