It is believed that allostery (a change in enzyme behaviour in one place caused by some other change at a distant location) plays a key role in enzyme mechanism, activation and inhibition, and is of paramount importance in enzyme functioning. In sharp contrast with its significance, the mechanism of allostery is still poorly understood in most enzymes. Current proposal will try to tackle this problem in a innovative way, by combining several very different methodologies that have seldom been used together in this context. As the 3D structure coming from X-ray crystallography (XC) alone is not capable to explain a fundamentally dynamic basis of allostery, it will serve as a basis for molecular dynamics (MD) simulations. Resulting static (XC) and dynamic (MD) data will be unified in a form of relational database. This database will be ideally suited for identifying hidden connections (pathways) between different enzyme regions, which is the basis of allostery. To address the complexity and the amount of data contained in the database, this proposal will leverage machine learning (ML) algorithms that will operate on this highly structured data. Therefore, the overall scientific objective of the project is to uncover the mechanism of allostery in oligomeric enzymes using a novel approach that combines the information obtained by different techniques: XC, MD and ML. Along with the mechanism of allostery, the project will try to understand the role of oligomerization in connection with allostery. Finally, the proposed allosteric mechanisms will be experimentally validated by mutating the most important residues in allosteric pathways found by ML algorithms and observing the change in enzyme behaviour upon mutations. This will then represent an ultimate proof of the validity of the concept and algorithms found in this project and show great prospect for their applicability in much broader domain of other enzymes.