||Barrett JH, Taylor JC, Bright C, Harland M, Dunning AM, Akslen LA, Andresen PA, Avril MF, Azizi E, Bianchi Scarrà G, Brossard M, Brown KM, Dębniak T, Elder DE, Friedman E, Ghiorzo P, Gillanders EM, Gruis NA, Hansson J, Helsing P, Hočevar M, Höiom V, Ingvar C, Landi MT, Lang J, Lathrop GM, Lubiński J, Mackie RM, Molven A, Novaković S, Olsson H, Puig S, Puig-Butille JA, van der Stoep N, van Doorn R, van Workum W, Goldstein AM, Kanetsky PA, Pharoah PD, Demenais F, Hayward NK, Newton Bishop JA, Bishop DT, Iles MM, GenoMEL Consortium
||At least 17 genomic regions are established as harboring melanoma susceptibility variants, in most instances with genome-wide levels of significance and replication in independent samples. Based on genome-wide single nucleotide polymorphism (SNP) data augmented by imputation to the 1,000 Genomes reference panel, we have fine mapped these regions in over 5,000 individuals with melanoma (mainly from the GenoMEL consortium) and over 7,000 ethnically matched controls. A penalized regression approach was used to discover those SNP markers that most parsimoniously explain the observed association in each genomic region. For the majority of the regions, the signal is best explained by a single SNP, which sometimes, as in the tyrosinase region, is a known functional variant. However in five regions the explanation is more complex. At the CDKN2A locus, for example, there is strong evidence that not only multiple SNPs but also multiple genes are involved. Our results illustrate the variability in the biology underlying genome-wide susceptibility loci and make steps toward accounting for some of the "missing heritability."