The UK Biobank study recruited 500,000 participants aged 40-69 years (2006-2010) with extensive phenotypic and genetic data. Participants were genotyped using two arrays and imputed centrally based on the 1000 Genomes Project phase 3. Adiposity measures, including VAT, WFM, and WHR, were used, with MRI-derived PDFF and liver iron corrected T1 provided by the UK Biobank. Phenotypic prediction models were developed using penalized linear regression and tenfold nested cross-validation. Genome-wide association analysis was conducted using REGENIE, and independent variants were identified through LD clumping and conditional joint SNP analysis. Heritability and genetic correlations were estimated using LD score regression. Functionally informed fine-mapping was performed with PolyFun and SuSiE. Colocalization analysis was conducted between genetic loci and eQTLs from GTEx. Gene mapping and functional enrichment analysis were performed using FUMA and Enrichr. Partitioned polygenic risk scores were defined based on concordant or discordant associations with PDFF and circulating triglycerides. Replication cohorts included NEO, Liver BIBLE, MAFALDA, and Dallas Heart Study. Meta-analysis was performed using inverse-variance weighted fixed-effect models. RNA-seq analysis was conducted on paired liver and VAT samples from MAFALDA. Follow-up analysis tested the longitudinal association of PRS with outcomes using Cox proportional hazard regression. Gene-adiposity interaction analysis was performed in REGENIE. Mediation analysis examined the impact of loci on PDFF or liver iron corrected T1 mediated via adiposity measures. Association analysis with adiposity measures was conducted using REGENIE. bNMF clustering was applied to define clusters of independent loci. Comparison between bNMF and PDFF-TGs pPRS was performed using Wald test and Akaike information criterion.