The basic tenet of systems biology is to (a) perturb a system; (b) observe and measure systems-wide consequences of the perturbation; and, finally, (c) integrate the disparate kinds of measurements into a quantitative model to explain experimental observations and predict responses to novel perturbations. A key aspect of this approach is that it is iterative – predictions made using a previous version of the model drive new experimentation. The iterative cycles refine our understanding of the biological system and bring the models into close apposition with real biology. This approach begins with integration of existing data into a preliminary model to drive the first set of experiments and set up the iterative cycles. An important outcome of this first step is that it helps to align the varied expertise such as is represented in this multi-institutional, interdisciplinary team. We have already constructed such models for M. tuberculosis (MTB) and activated macrophages (MΦs) . These preliminary models will drive the subsequent iterations of a hypothesis-driven systems approach to both reveal novel insights into the biology of both host and pathogen and also refine the models. In his leadership role in the Modeling Core, Dr. Baliga, who is trained in both microbiology and computational biology, will ensure seamless integration of experimentation and computation.
The goal of this U19 proposal is to use a systems approach to construct a predictive model for TB-host interactions, and to use this model to test a key hypothesis that behaviors of host and pathogen are coordinated by interwoven regulatory networks, and that the outcome of infection (bacterial containment or active disease) is the product of many network-network interactions that vary both spatially and temporally. It is possible to construct such predictive models because the functional capability of every organism is constrained by the limits of environmental factor (EF) variations it has experienced during its evolution. Furthermore, environmental changes (e.g. temperature, pH, O2 etc.) occur in a temporally coupled and non-random manner for well-characterized physicochemical reasons. Even pathogens have taken advantage of such repetitive environmental changes within the host by evolving coupled and anticipatory behaviorTagkopoulos, et al.. The fitness gain associated with internalizing coupled changes is evident in the fusion of multiple environmental sensing domains in numerous response regulatory proteinsMascher, et al.. These linkages are also manifested in the molecular architecture of gene and protein networks and the anticipatory behavior they encode have been known for some timeDodd, et al.,Nikaido and Johnson. Predictive modeling of such complex systems-level phenomena has become tractable at a molecular level because of high throughput technologies to measure molecular changes in system-wide genetic information processing (RNA and protein levels, protein-protein (P-P) and protein-DNA (P-D) interactions, protein modifications and so forth). This collection of coupled EF changes also aids in predictive modeling of regulatory networks from a relatively modest number of systematic perturbation experiments. This is because even a simple perturbation propagates through the highly structured network of physicochemical relationships to give complex and reproducible multi-factor changes. Consequently, these perturbations trigger linked and anticipatory networks within the organism generating a structured response that can be learned and modeledBonneau, et al..
The goal of the Modeling Core is to integrate data from Projects 1 and 2 to generate a multi-scale predictive gene regulatory network (GRN) model that will link quantitative phenotypes in the host to causative mechanisms and their influences in the pathogen, and vice-versa. This model will predict host and pathogen factors that regulate or influence the outcome of TB infection, specifically susceptibility, resistance, infection progression, persistence, and clearance. We will describe existing and novel computational strategies (algorithms, approaches and principles) in the following three aims to drive iterative modeling and experimentation: