Quantitative Systems Biology

Our focus in the area of quantitative systems biology spans all aspect cellular and sub-cellular. Projects are widely varied. However, they all come back to a common underlying theme of using integrated computational and experimental approaches to understand and engineer cellular and sub-cellular systems, including viruses. Areas of research include microbiome modeling, inferring multi-omic networks, and developing new approaches for antisense RNA design. We also work on creation and curation of genome-scale metabolic models, as well as kinetics models of bacteria and viruses. We also carry out work in cancer biology. Most of those efforts are targeted at oncoviruses and microbiomes in a cancer biology context.

Computational Materials Design

Our efforts in computational materials design have primarily focused on nanoporous materials. Our research in this area has two thrusts. The first thrust is the development of algorithms to design/discover more effective materials for whatever purpose we are interested in. Those interests, to date, have been in creating materials that are better for adsorption or separations processes. Our second thrust has been in using computational methods to acquire a deeper insight into the fundamental chemistry and physics of the materials we are dealing with. We believe that such insight will advance our basic understanding of these materials, as well as allow the development of more effective, accurate, and efficient algorithms. While we have worked with a few different types of materials, we are currently focusing mainly on metal-organic frameworks.

Applied Machine Learning

While we have a strong emphasis on mechanistic modeling, we also have tremendous interest in artificial intelligence. Machine learning is one aspect of artificial intelligence that has garnered a great deal of interest and for good reason. For many of the systems that we study, whether biological or materials-focused, there may not be sufficient understanding to generate a mechanistic model. However, if there is enough data available, a machine learning model can be used for classification, prediction and possibly understanding. While our interest is primarily in understanding, the other facets of machine learning are extremely useful. To that end, we mainly use various types of neural networks, random forests, and gradient boosting for our work, although we do use other approaches if they are more appropriate.

Applied Evolutionary Dynamics

As mentioned on the about us page, the underlying driver for our group is our interest in evolutionary dynamics. Evolutionary dynamics represent powerful stochastic optimization approach that can be implemented either biologically or algorithmically. We are interested in both areas. Furthermore, we are interested in the ramifications and applications. On the experimental side, our focus is primarily on molecular evolution. A great example of applied molecular evolution is the directed evoluion of proteins, which won the Nobel Prize in 2018. On the algorithmic side, we study and use a variety of evolutionary algorithms, as well as other nature-inspired algorithms. A key class of problem that frequently arises in engineering and which we are interested in is the inverse function problem. Inverse function problems are those where you wish to know what type of input to the function gives you a desired output. Evolution is an example of an inverse function problem. Given a desired phenotype, what is the genotype that will get you there. We believe that by better understanding the principles of evolutionary dynamics, we can use it to solve a much broader range of problems.