Biological networks can be broken into functional modules, which are groups of interacting molecules. We have developed a new strategy to identify repeating interacting patterns in molecular networks by integrating multiple sources of information, including sequence, RNA expression and protein-protein interactions. This information is represented as a labeled graph, and scanned to find significantly frequent interacting patterns, “network motifs”. Each instance of a network motif is a subgraph and is predicted to be a functional module, with biological function(s) that has been adapted for multiple purposes in the organism. This approach has been successfully applied to several large datasets. Here we give an overview of results from the analysis of yeast protein-protein interaction data, malaria life cycle gene expression data, mouse protein-protein interaction data and mouse tissue gene expression data. Also we show that different instances of the same network motif can have similar functions, but may function at different times or tissue locations, presumably due to different transcriptional regulation. Our approach presents a novel strategy to study the modularity of biological systems through integrating different types of data. The implementation of this approach is through a new software package (BLUNT) for mining complex biological data which will be made publicly available. Advantages of our approach include the ability to identify small clusters of genes or proteins for more detailed experimental study, and reducing the number of false positives due to the use of multiple data types to validate the motifs identified.