Overview of the bioPIXIE system. Diverse data sets are integrated with a Bayesian network, which weighs each evidence type probabilistically based on its accuracy (a). This Bayesian integration produces a graph with confidence-weighted relationships between each gene pair (characterized in supplemental Figure S1 in ). Based on this integrated network graph and a user-defined query set of proteins of interest (b), the network prediction algorithm identifies novel network components by finding proteins with the maximum expected number of direct and indirect relationships with the query set (c). The resulting network is then displayed to the user using a spring model layout, such that the geometric proximity of genes reflects how related they are to each other, and the edge color reflects the confidence of pair-wise connections (d). Details of each component are presented in Materials and methods.