The concept behind the Systems Biology Iterative Refinement Cycle is to iterate and refine the modeling results with specific model features validated by supplemental in vitro or in vivo experiments. The resulting models accrue knowledge at each iterative step.
The first step in the Iterative Refinement Cycle process is the design of time-course in vivo experiments which must include systematic perturbations (such as a virulent and non-virulent pathogen challenge). The second step collects multi-dimensional "omic" data from both the pathogen and the host (if possible for the pathogen) from in vivo experiments. The third step completes comprehensive systems biology analysis to identify the important pathways, genes, and biological functions for both the pathogen and host related to invasion, evasion and defensive response over the course of time. The fourth step creates the predictive interactome model, a DBN which is used to identify candidate points of interaction between host and pathogen or to learn just the host response. The model is evaluated through computational means to determine confidence of the interconnections and predicting points of interaction which becomes part of the model knowledgebase that accrues at each step of the cycle. The fifth step is to interrogate and visualize the model by biologist to determine features which warrant further validation. The sixth step is in vitro validation of these selected features, with the objective to achieve higher confidence in host-pathogen interaction points. Validated features are then fed back into the model for refinement and database for knowledge accrual and future reuse. The final step determines if additional perturbations are necessary, i.e., perturbations could be either in the form of alternate mutant pathogen challenge or different strains (or knockouts) of the host species, etc.. The iterative refinement cycle is repeated and the new "omic" data employed to improve and validate the interactome model for the given experimental conditions. Models are iteratively updated and validated until the biologists are satisfied with the model's predictive capability.return to pipeline