Over the last decades, several features of obesity have been identified at behavioral, physiological, endocrine and genomic levels, and they have revealed the complexity of the disease; obesity results from a combination of genetic predisposition, endocrine disorders, and dysregulation of both food intake and energy expenditure. This complexity makes the development of new therapeutic regimens challenging and bariatric surgery is still the treatment of choice for many obese patients. Given the need for noninvasive therapeutic intervention strategies, we sought to systematically study the biological manifestations of obesity in peripheral organs. We analyzed publicly available datasets of genes, genomic determinants, and levels of obesity-related hormones in the blood, using a combination of methodologies, including graph theory and dynamical modeling, that allow for the integration of different types of datasets. The analysis revealed tissue-and organ-specific metabolic impairments and potential new drug targets. All the data are organized into a tissue/organ-based subcellular-function atlas for human obesity. The data show that the complexity of the obesity arises due to the multiplicity of subcellular processes in different peripheral organs. AMY, CCK, GIP). Plasma levels of some hormones in post-bariatric surgery patients lie between the lean-curves and the obese-curves (GHRL, GCG, LEP, AMY, GIP), while others appear higher than in either lean or obese individuals (INS, CCK, GLP-1, PYY, and OXM) ( Figure 3B). We ran simulated time courses for lean, obese and post-bariatric surgery subjects in a fasting and a postprandial state to estimate reliable initial values for the kinetic parameters such that the simulated curves closely resembled published experimental data [8, 25, 27, 28,[30][31][32][33] ( Figure 3B and Supplementary File 3).The magnitude of the endocrine response to nutrients can be assessed by the difference in plasma levels of the hormones before and after the meal. To quantify it, we chose to calculate the ratio between the 2-hour area under the curve (AUC) of the postprandial plasma concentration and the 2-hour AUC of the fasting plasma concentration for each hormone (Figure 3C and Supplementary Figure S8). To identify which kinetic parameters are more likely to affect the endocrine response represented by the AUC, we performed a multivariable regression analysis (MRA) [34]. The logic of this analysis was to randomly vary the initial values of the set of kinetic parameters used in the model a number of times (Supplementary Figure S9), utilizing each new set of values to run a simulation that by varying each parameter affects the output and expresses it in terms of a numerical Partial Least Square Regression (PLSR) coefficient: the higher the coefficient, the more sensitive is the response to the parameter; the lower the coefficient, the more robust is the response to changes in the parameter. The kinetic parameters for lean, obese, and post-surgery populations were then ranked according to the...