In this paper we take an extended reaction scheme framework to model general genetic regulatory pathways. Extended reaction schemes are an extension of normal chemical reaction schemes. Extended reaction schemes allow molecules to be structured not only by type (or species) and their copy numbers, but in addition allow to attach to any type of molecule discrete internal states. This makes it ideal for applications in genetics, where some of the copy numbers of molecules stay constant in system time (such as genes), whereas other types of molecules vary greatly, such as transcription factors. Moreover, the additional degree of freedom allowing internal discrete states enables to write models where genes are switched on and off, proteins can change their conformation, and mRNA can differ in their binding properties, among other possibilities. Because some of the molecules will only be available in low copy numbers, stochastic effects have to be taken into account. This is done by generalising the chemical master equation to an extended chemical master equation, where most of the original Gillespie approach can be recovered. The framework is able to resolve many of the modelling mysteries of genetic regulatory networks in a very natural way, and is leading to a consistent extension of static regulatory pathway analysis to truly dynamic genetical network theory. Moreover, together with novel methods of model selection, such as deep learning, the framework allows for a data driven selection of model variants.