Differential expression among tissues in morbidly obese individuals using a finite mixture model under BLUP approach

Morbid obesity, the excessive accumulation of body fat, has major consequences for human health by its association with several severe diseases, like Type 2 Diabetes. The biological mechanisms behind this association are mostly unclear, however, the biological complexity of morbid obesity indicates an important role for pathogenesis arising from different tissues/organs and interactions among them. We hypothesized that differentially expressed (DE) genes between different organs in morbidly obese individuals result in a better insight in the biological mechanisms explaining the link between obesity and obesity-related diseases. We used whole-transcriptome expression levels of subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), liver, and muscle of 93 morbidly obese individuals (26 males, 67 females) who were all phenotyped for different metabolic parameters. We estimated genetic random effects of the interactions between tissues and probes using BLUP (Best Linear Unbiased Prediction) linear models correcting for gender, which were subsequently used in a finite mixture model to detect DE genes in each tissue. This approach evades the multiple-testing problem and is able to detect variation among different biological states. Preliminary results show evidence of DE genes operating within tissues, but variation in these effect sizes was similar between the four tissues. We detected 17% of the transcripts to be DE in liver, 16% in muscle, 13% in SAT and 23% in VAT (Probability (DE) > 0.9). Results will be used to gain insight into the underlying genetic and biological mechanisms by functional annotation of DE genes, detection of pathways and construction of gene networks.