RATIONALE Genome-wide associat10n studies (GW AS) are an efficient approach to dissect the genetic architecture of complex traits. Lipids, including the diterpenes cafestol and kahweol, are key compounds that contribute to the quality of coffee beverages. A genome-wide association study was performed to identify genomic regions associated with lipid, cafestol and kahweol. METHODS Using GBS, we genotyped 107 Cojfea arabica accessions, including wild genotypes from the historical F AO collection from Ethiopia and cultivars. Lipids, cafestol and kahweol contents in green beans were determined by NIRS. For full description of genetic analysis and the four association methods: mrMLM, F ASTmrEMMA, ISIS EM-BLAS SO, pLARmEB see Sant' Ana et al.2018. RESULTS Using the diploid Cojfea canephora genome as a reference for GBS data, we identified 6,696 SNPs. Population structure analyses suggested the presence of two to three groups (K = 2 and K = 3) corresponding to the east and west sides of the Great Rift Valley and an additional group formed by wild accessions collected in western forests. We identified 5 SNPs associated with lipid content, 4 with cafestol, 3 with kahweol and 9 with cafestol/kahweol ratio. Most of these SNPs are near by genes related to metabolic pathways of those compounds. Among all trait-associated SNPs detected by GWAS, three showed strong signals of directional selection between genetic groups identified using STRUCTURE (K = 3). The group with most wild accessions presented very low frequencies of the reference alleles compared with the other two groups. This indicates that domestication and/or breeding process of C. arabica may have changed allelic frequencies of these loci in order to modulate lipids and diterpenes content. CONCLUSIONS & PERSPECTIVES The genetic anaysis helped to define which accessions are more important to preserve in order to have a good genetic representation of the F AO collection and provided insights on C. arabica domestication.