Mechanistic Analysis and Modeling
Following the data integration and preprocessing steps in the computational workflow, the system executes the Mechanistic Analysis and Modeling steps (Dynamic Bayesian Gene Group Activation and Comparative Analysis) to analyze, discern, and model the unique differences and similarities over the consolidated multi-condition observations. Central to the analysis and modeling is the application of Dynamic Bayesian Networks (DBNs). For this computational flow, a technique was developed for identifying groups of genes that as a whole represent the activation/inactivation of metabolic and signaling pathways, pathway subnets, biological processes, and genes over time. This technique is extremely important due to lack of sample measurement repetitions, the variable span of time between measurements, and the resulting expression level variability commonly observed between experiments. This technique is called Dynamic Bayesian Gene Group Activation (DBGGA). DBGGA allows the examination, scoring, and ranking of groups of genes across all time points in lieu of just individual genes in a single time slice to determine differences within components (subsystems) of a biological system as a function of the experimental conditions. These components are defined by the prior biological knowledge such as the best known gene/protein relationships involved, for example, in the Toll-like Receptor signaling pathway or a group of genes associated with the Gene Ontology term "dorsal/ventral pattern formation". Hundreds of pathways and thousands of biological processes are analyzed in this fashion to systematically identify the biological system behavior as a function of the different experimental conditions.