Connecting Biological Themes Using a Single Human Network of Gene Associations

The accumulations of biological data in various knowledge domains (biological themes) provide rich resources for extracting deeper biological insights into biosystems. However, relations among these biological themes remain uncharacterized. Here, we present a systematic approach to the discovery of these relationships using a single human network of gene associations. We first constructed the network by incorporating the compiled human interactome and the predicted human associatome resulting from the Bayesian supervised integration of functional annotation data, model organism functional linkage data, and human functional linkage data. Using the network, we then performed the binomial distribution-based enrichment assessment to examine the high-level relationships among biological themes. Applications of the approach in biological processes from the Gene Ontology indicated that, although intra-process connections are of highly modular, distinct processes differ considerably in their abilities to interconnect with others. We also showed that such interconnectedness of processes cannot be explained by the modular constraints, but is largely due to the connectivity of their individual members to other processes. Moreover, we extended applications in other biological themes to find connections among regulatory profiles of transcription factors and microRNAs. These results demonstrated the feasibility of the approach, combining network biology with systematic information, to characterize high-level connections of any new biological themes.