Application of graph modularity inference to represent protein structures.

Proteins are large molecules that are implicated in the majority of the work done in living cells. We have the ability to solve the shape of the protein encoded by a particular gene. However, their three-dimensional structures are large and complex. This research program aims to provide new ways to conceptualize the inner mechanics of these large molecules. We are first interested in understanding how these long chains arrange themselves into a smaller number of rigid components, or modules. To do this, we are using cutting edge analysis of molecular simulations, and also will develop new analyses. We ascribe that attempts to engineer proteins to achieve a particular function should consider this modular structure. We are also interested in inferring modules that are stable over evolutionary time. This is done by comparing the shape of many proteins sharing the same ancestor. Modules that are stable over evolutionary time are likely to be the result of networks of complementary mutations. Since protein engineering is commonly using mutations to alter the function of a protein, understanding better the relation of one position in the protein with respect to its neighbors increases the chance of anticipating the indirect consequences of mutations. We achieve this by applying and extending work in computer science on network analyses. Going forward, we are interested to explore how evolutionary and stability modules are related. This kind of knowledge is important to further our understanding of protein stability, evolution at the molecular level, and to assist in the rational engineering of protein to design proteins capable of novel functions.

What makes proteins semi-rigid and capable of performing important functions is their ability to adopt and maintain a stable 3D structure. Let's conceptualise stable conformations as low energy structures. To transition from one stable conformation to another, the structure must cross a peak of high energy. We coin the term “threshold event” when a structure crosses a high energy peak. To study threshold events, we use the simulation of a small protein called Amyloid-beta. The transition between two stable structures for this protein is believed to be one of the factors in the pathology of Alzheimer's disease. We use computational and numerical techniques to discover the key components of the threshold events between the normal and pathological form of this protein. Understanding threshold events in general is important in protein science, and in the design of new protein structural elements. Particularly, studying the causative agent of Alzheimer's disease is important because the mechanism is common to many neurodegenerative diseases: discovering a mean to disrupt protein aggregation could be the basic science seed of new treatments for these terrible conditions.