Dr. Huttenhower's research is concerned with the discovery of useful biological knowledge in large collections of genomic data. Modern biological experiments each represent a detailed snapshot of a cell or organism's internal state, and public repositories already contain many thousands of experimental results and are constantly growing in size and diversity. Taken together, these data can be used to reconstruct detailed models of cellular behavior in response to changing environmental conditions, regulatory and metabolic regimes, and disease states. The goal of this research is to allow any new biomedical question to be answered by extracting information from the entire body of existing and novel experimental data, using data integration to allow results from basic research to be applied to genomic and personalized medicine (and vice versa).
In practice, this requires the development of computational methodology that is efficient enough to deal with billions of data points while remaining biologically rich enough to capture the complexities of molecular biology. This can include techniques from statistical machine learning, graphical models, and information retrieval. The lab is currently applying these methods to biological systems including both human populations (e.g. genomic data from large prospective cohort studies) and microbial populations (e.g. pathogen populations and metagenomic communities such as the gut microflora). Many results also rely on the construction and analysis of biological networks, including physical protein-protein interactions, regulatory networks, or functional associations among genes and gene products. The challenge is not only to develop useful computational models, but also to apply them collaboratively to drive novel experimental biology and to better understand biomedical results.