A central challenge of modern biology is not merely to correlate observable data on an organism’s genotype and environment with its observable characteristics (i.e., phenotype), but to develop predictive models of phenotypic trajectories—such as disease emergence, progression, and remission—from high-dimensional, heterogeneous data. Increasingly, biological data span multiple molecular layers, tissues, and time points, yet predictive performance often plateaus because many models implicitly assume that phenotypes are direct functions of individual features rather than emergent properties of interacting systems. Crucially, many clinically relevant phenotypes arise through progressive network reconfiguration rather than isolated molecular failures. For example, two genes may exhibit strong correlation across patient cohorts, yet this association may reflect shared regulation by an upstream signaling module or feedback loop rather than a direct causal relationship. Conversely, perturbations to sparsely connected but strategically positioned network components can induce large-scale phenotypic shifts while remaining weakly correlated with disease labels. Such phenomena illustrate how biological complexity can mask causation when analyses rely solely on feature-level correlations. Predicting phenotypic trajectories—particularly for chronic, immune-mediated conditions—therefore requires modeling frameworks that can represent conditional dependencies, feedback, and context specificity. Here, I propose adopting a network-science perspective, in which phenotypes are understood as system-level outcomes arising from structured interactions among genes, cells, signaling molecules, and environmental inputs. This perspective enables explicit modeling of how perturbations propagate through biological systems and how disease states emerge as stable or metastable network configurations. This thesis will center on finding governing subnetworks within the molecular networks governing immune response to predict phenotypic outcomes. I contribute to the applications and methodological developments of this process. The subnetworks and driver pathways discovered in human and fish immune networks leads to identification of genes, and molecules essential for therapeutic intervention for immune system dysfunctions.