In the past several years, the field of biology has been undergoing a revolution.  The rapid advent of new technology in the areas of molecular and cellular biology has resulted in the ability of scientists and engineers to analyze biological systems on a scale never before possible.  The result has been a deluge of new data at the genetic, protein, and metabolic levels just to name a few.  To further complicate matters, each of these domains interact with each other resulting in a level of complexity which traditional means of biological analysis are simply unable to deal with.  Reconciling the interactions of these biological components with each other to elucidate the functioning of a given system as a whole has resulted in the interdisciplinary field of “systems biology.”  

A hallmark of systems biology is the use of quantitative frameworks to describe the non-linear interactions occurring at the molecular, cellular, and higher biological levels.  As a result, modeling these dynamics mathematically draws upon the skills of mathematicians, physicists, computer scientists, and engineers.  However, a fundamental understanding of the underlying biology is critical, lest poor modeling assumptions be made resulting in an unrealistic model.  Thus there is a critical need for quantitatively trained scientists and engineers to work closely with biologists.  Even more critical is the need to train a new generation of quantitative biologists or systems biologists who possess the requisite skills to work more effectively in this growing field.

The focus of my research group, which does both theoretical and experimental work, has been to use systems biology to integrate genomic, proteomic, and metabolomic data to develop models of various pathogens, ranging from viruses to bacteria.  We are motivated by two over-arching goals.  The first is to use the models we develop to identify potential drug targets for treating these pathogens.  Along those lines, we are experimentally developing new types of inhibitory RNA to target the sites we identify and treat these pathogens.  The second goal is to use models of pathogen-drug interaction to develop optimal therapeutic treatment strategies.  For example, HIV treatment using highly active anti-retroviral therapy (HAART) often results in severe health side effects.  If a strategy to minimize patient exposure to the HAART drugs while maintaining the white blood cell count and viral titers within pre-determined bounds could be determined, the quality of life of the patient could potentially be dramatically increased.

An example of how we are approaching these two over-arching goals is through the development and application of model discrimination theory to identify the most probable mathematical model of a biological system.  To identify a drug target or develop an optimal treatment strategy for a patient, it is necessary to have a high quality model that accurately describes the biology of the underlying system.  Model discrimination is a Bayesian based method that allows one to determine the most probable of any number of postulated models relative to each other given a set of relevant experimental data.  We recently used this technique to evaluate over 20 published HIV-1 viral dynamic models to identify the most probable one.  We have further adapted this approach for use in genome-scale flux balance analysis (FBA).  The goal of FBA is to identify the distribution of metabolites across an organism’s metabolic reaction network.  The metabolic reaction network is generally constructed using publicly available genomic information for the organism.  Since there generally are far more reactions than there are experimentally known values of metabolites, the system is underdetermined.  It is possible to calculate the optimal metabolic distribution of such a system using linear programming if an appropriate objective function is known.  However, identifying the appropriate objective function is not a trivial matter.  To overcome this problem, we have been able to adapt the model discrimination approach to evaluate any number of different objective functions to identify the most probable one.

The field of systems biology is a growing one and holds much promise for the future.  It holds great potential for benefiting society, as well as posing great intellectual challenges for the curious.  Personally, I cannot imagine a more rewarding field to be a part of.

|