The general architecture of a genetic regulatory network consists of strong connected components of its interaction graph, to which are attached three kinds of sub-structures: - a set of up-trees, rooted in the sources of the interaction graph, represented either by small RNAs like microRNAs: nuclear miRs or mitochondrial mitomiRs, i.e., translational inhibitors respectively of the messenger mRNAs and of the transfer tRNAs, or by gene repressors and/or inductors, - a set of circuits in the core (in graph sense) of the strong connected components of the interaction graph, - a set of down-trees going to the sinks of the interaction graph, i.e., to genes controlled, but not controlling any other gene. The various state configurations it is possible to observe in the above sub-structures correspond to different dynamical asymptotic behaviors. The network dynamics have in general a small number of attractors, corresponding in the Delbrück's paradigm to the functions of the tissue they represent. Examples of such dynamics will be given in embryology: cell proliferation control network in mammals and gastrulation control network in Drosophila melanogaster.