Given a sample space with many random variables defined on it, it makes lots of scene to think of some kind of association among them. Generally, sample space due to biological experiment is thought to create many random variable with biologically motivated inter or intra-relationship. In this study, our sample space contained mRNA profile of 6 immune system candidate genes: Interfron Gamma (IFN-γ), Tomur Necrotic Factor alpha (TNF-α), Granolocyt-Macrofage Colony Stimulating Factor (GM-CSF), Interleukine-2 (IL-2), Interleukine-6 (IL-6) and Interleukine-8 (IL-8) in which their transcriptome were obtained. These genes were due to dairy cattle in different biological states e.g. the early, middle, late of lactation and cows suffering from mastitis. Using Bayesian Network (BN), we concluded that first, many genes showed independent behaviour in terms of regulation (there were not any wired regulation among them over different biological states) second, the mode of regulation changed across different biological states. In mastitis state, IL-8 shown to be regulator of TNF-a, IL-8 was regulator of GM-CSF in early lactation, in middle of lactation, IL-2 shown to be regulator of IFN- γ and in late of lactation IL-8 turned out to be regulator of GM-CSF. This research revealed that the mode of regulation of candidate genes was not identical over different biological states. Overall, we showed that aforementioned immune system candidate genes should be seen in the biological context which they are functioning. Therefore, if the objective is to tackle with mastitis using drug targeting studies or in genetic selection, it is more relevant to pay close attention to IL-8 as it is predicted to show the mode of regulator in mastitis state in dairy cattle. However, this gene was also the regulator of other different genes in across different biological states.