Knowing Where to Look

Human metabolism is a good way to illustrate our research process. It is a highly interconnected network of thousands of ongoing biochemical reactions which influence each other in subtle and unforeseen ways.

The reactions come with an exponentially greater number of possible relationships between them - homeostasis, feedback loops, and other non-obvious effects. Biochemical panels alone are not sufficient to gain a full understanding of this network, and the challenge is to locate which relationships among this immense number could represent a viable target for treatment.

A method of simplifying these relationships is needed to identify the most relevant ones. We can use a variety of mathematical techniques, such as principal component analysis, LLE, isomap, diffusion maps, and kernel PCA to develop an accurate lower-dimensional approximation of this high-dimensional network of metabolic relationships. This helps to identify the most meaningful patterns in the patient’s metabolism, such as relationships between metabolic imbalances and disease, presenting potential avenues for intervention.

Research in dimensional reduction of biochemical network has shown a great deal of promise in enhancing our understanding of metabolism. We apply these methods to identify the physical mechanisms of disease, elucidating relationships between the patient’s symptoms and predicting what additional symptoms they may experience as a result of their metabolic abnormalities. The data we acquire assists us - and you - in developing these hypotheses.