Non-alcoholic fatty liver disease (NAFLD) is one of the major health problems in the western world. Despite this, there is currently no approved pharmacological treatment for NAFLD. Difficulties in diagnosing and accessing liver tissue samples and the lack of good preclinical animal models mimicking human disease progression are major challenges in developing treatment for NAFLD. Both obesity and high fasting insulin and high liver triglycerides are strong risk factors for NAFLD. Here, we utilize the Diversity Outbred Australia (DOz) mice, a unique cohort of outbred mice with genetic diversity resembling that of the human population, together with the wealth of data from human genetics to find novel drivers of liver fat accumulation. We detect ~100-fold difference in liver triglycerides levels across 243 male DOz mice fed a healthy chow or a high-fat diet concomitant with a 40-fold difference in fasting blood insulin levels, highlighting the DOz population as an excellent model for studying NAFLD. We quantified ~4000 proteins across all livers by total proteomics. By combining the mixed linear modelling and network analysis, overlaid with genes that increase risk of developing NAFLD in the human population, two modules of co-regulated proteins were identified, which were highly associated with liver triglycerides in the mouse livers. Proteins involved in DNL, such as SCD1, ACC1, where its inhibitor is currently in clinical trials, were among the proteins in these two modules and interestingly were significantly affect by diet through linear model. By filtering proteins based on features from the linear modelling, such as effect size and significant affected by triglycerides, we identified 30 proteins, put forward for validation for their role in liver fat accumulation. In conclusion, we have identified a number of potential drivers for liver fat accumulation by combining mouse systems biology and human genetics.