Requirements: - JVM 1.6 or above - Metis 5.0 or above METIS must be installed in the system. You can download it from http://dl.dropbox.com/u/8633823/coclustering/metis-5.0.tar.gz To install METIS, just run this commands in a terminal: sudo make config sudo make sudo make install Please note: - cmake>=2.8 and g++ must be installed in the system in order to compile METIS. - METIS source path must not contain any blank character, in order to avoid cmake compilation errors. How to run HOCCLUS: HOCCLUS can be run with the following syntax: java -jar hocclus.jar [options] -alpha Options: -q Specify the quality function -d Specify the format of the input dataset mirnamap: miRNAMap format sparsematrix: sparse matrix format (used for synthetic datasets). Each row consists of: "row object id" "column object id" "value" -stats Calculate results info (precision, recall, etc). is a file containing the ground truth. -alpha (mandatory) Specifies the quality threshold, that must be between 0 and 1. We also provide the real dataset (dataset.zip) and the synthetic datasets (SyntheticDataset.zip) we used for the experiments reported in the paper. Please note that any METIS version >= 5.0 can be used, whose the latest version can be downloaded at: http://glaros.dtc.umn.edu/gkhome/metis/metis/download However, experiments have been performed with METIS version 5.0.0, which is the version we suggest to use for replicating experiments.