In recent years, protein kinases have gained significant attention in the treatment of human diseases. Kinase mutations and the over expression of kinases can cause abnormal frequency of protein phosphorylation and thus their activation, and this plays a significant role in diseases such as cancer and Parkinson’s disease. Such pathways can be blocked by designing kinase inhibitors. However, one of the main obstacles in designing an inhibitor is its specificity in binding to the desired kinase. Inhibitors end up having kinase “off-targets” that were not meant to be inhibited due to the structural similarities of kinases. By looking deeper into kinase structural patterns, we can better differentiate protein kinases and raise the specificity of inhibitors. Conserved, stable water clusters found in the kinome could possibly influence kinase functionality and inhibitor specificity, but so far, only little research has been done on these water clusters. This research project investigates the interaction between kinases and the water solvent and how water clusters stabilise kinase amino acid residues and works as a 'bridge' between residues and an inhibitor. This research was focused on the said stable water clusters: creation of a database of all stable clusters throughout the whole human kinome; creation of a new kinome tree composed according to the similarity of stable water networks; their interactions with kinase inhibitors. One of the cases studied is that of the kinase Aurora A.Through MD simulations of Aurora A with one of its inhibitors and with ATP, we investigated how stable water clusters could affect the specificity of inhibitors and ATP binding. Furthermore, cross comparison studies of inhibitors of different kinases within or outside of the newly generated families are being done with a purpose of understanding more about the possible ways of inhibitor off-targets, that are of a big importance in pharmaceutical research. Suggestions about possible inhibitor modifications that account for stable water clusters and their similarity across some of the kinases of interest are made. The study is conducted using in-silico methods such as MD-simulations, as well as Boltz-2 for binding affinity predictions and various softwares that allow us to view the structures more closely and manipulate them. To receive reliable data about stable water clusters from the simulation results, coding with DBSCAN was used.