ComiRNet is a database of miRNA target predictions and predicted miRNA regulatory networks. ComiRNet stores approximately 5 million predicted interactions between 934 human miRNAs and 30,875 gene transcripts (mRNAs) which are exploited in the construction of the hierarchies of overlapping biclusters representing potential miRNA regulatory networks.
All target predictions were scored by a learning algorithm that learns to combine the score returned by several prediction algorithms (i.e., 10 stablished miRNA targets prediction programs) by exploiting information conveyed by both labeled (validated interactions, from miRTarBase) and unlabeled (predicted interactions, from mirDIP) instances.
miRNA regulatory networks are discovered by analyzing scored miRNAs target interactions with the algorithm HOCCLUS2 (see  in Publications).
ComiRNet database consists of two main modules, each one equipped with a web query interface for the retrieval and visualization of data. A series of filters can be used to refine the query on the basis of different criteria and parameters.
The Search Interaction module allows the extraction of miRNA-target predicted interactions on the 3' -UTR of all known Human genes. Results are presented with the score (i.e. probability) assigned by the ComiRNet learning algorithm.
The Search Bicluster module hosts predicted interaction networks, identified by HOCCLUS2 from the set of predicted interactions.
In addition, the system provides a graph-based visualization of interaction networks, which summarizes the predicted associations for a particular bicluster. Nodes represent genes and interacting miRNAs. Edges represent the predicted functional associations. Hovering over an edge will display ComiRNet predicted interaction scores. Network view can be changed by filtering interactions according to the score.