Figure 2.

The structural equation model (SEM). A simplified cartoon of the principal behind structural equation modeling (SEM). Here a cell is influenced by several “environmental” factors e.g., insulin (Ins) which induce a response in the cellular regulatory network cascade. The effect of this signal transduction is to induce or inhibit transcription of genes or genetic regulatory structures (Genomics). In addition, the signaling may directly influence the functionality of effector networks e.g., by promoting phosphorylation of effector proteins in all resulting in regulation of the blood pressure. In real life the pathway from the external factors to the ultimate response is much more complex as several cell types are involved in the local process (the vessel) and several tissues or organs will be involved (e.g., fatty acids generated in the liver) in a complex interactive network. The entire system is formulated by a complex set of matrix formulated regression algorithms, which may be proximal to the processes we actually can obtain information above or just by summary expressions. The subpopulations are then defined by simultaneous optimizing the SEM and the latent class (or rather latent profile) defining the population structure. A detailed account can be found in Muthen and Muthen (2004); Fenger (2008, 2012); Fenger et al. (2011).