Funded Grants

Modeling biocomplexity: From molecular interactions to population genetics

Living processes are governed by complex networks of interactions that are precisely regulated to produce highly specific biological effects (cellular growth, differentiation, development, etc), yet flexible enough to adapt to environmental changes (eg, the immune response to infection or the triggering of apoptosis following DNA damage) and robust enough to tolerate extrinsic and intrinsic variability (such as genetic polymorphisms). While components of these networks have been identified, a complete description of cellular regulation remains the central goal of modern systems biology research. Advances in high–throughput (HT) assays, which enable simultaneous measurement of millions of molecular markers per sample on a genome–wide scale, now provide an unprecedented opportunity to investigate these systems in extremely fine detail.

The enormous wealth of HT data brings with it significant analytical challenges. Examining each assayed gene independently (as is often done in HT studies) will fail to capture crucial systems –level effects, yet the temptation to construct models that are as comprehensively detailed as our measurements will lead to de-scriptions that—like Borges’ fictional map∗—are so elaborate as to be useless. To fully realize the potential of genome–wide studies, it is necessary to develop minimal models that articulate the crucial determinants of biological function.

My group’s research, broadly described as computational systems biology, includes the development of novel statistical methods for the analysis of HT data as well as in silico simulation studies of biophysical dynamics, with the goal of modeling the processes that lead to the emergence of complex phenotypes. The proposed work is organized around three broad aims: to develop novel computational techniques to infer and analyze regulatory networks from high-dimensional HT data; to make testable predictions about the network dynamics based on their structural properties; and to investigate the crucial interactions within those networks through dynamical simulation. In collaboration with experimentalists and clinical investigators, we will apply these models to investigate the mechanisms underlying malignant transformation, myelodysplastic syndromes, and the antiviral response.

A key innovation of our approach is the development of a novel graph–theoretic framework to summarize HT data in the context of biological network topology. Because our methods do not rely upon single–gene association statistics, they are able to articulate bulk pathway–level differences even when samples are molecularly diverse or exhibit nonlinear patterns of gene expression. More importantly, our approach characterizes the connectivity of signaling networks in a manner that is directly related to the network’s dynamic properties. This enables us to make testable predictions about the responsiveness and robustness of regulatory networks using based on static “snapshot” data obtained in HT assays. In contrast to existing systems analysis techniques, which only yield lists of significant pathways, our approach has the potential to not only enumerate which systems differ, but also indicate how those systems will behave when perturbed. This feature enables us to go beyond “stamp collections” of significant genes/pathways and make mechanistic predictions about pathway behavior that can be validated using simulation studies or experimental modulation.