Current Version: Gene Diver 1.05 is now avaliable for download

Gene Density Interactive Visual ExplorER

About Gene DIVER: Gene DIVER is a scalable implementation of a recently developed non-parameteric hierarchical cluster mining algorithm that also incorporates a powerful visualization and browsing of clusters. It is a great tool for users interested in finding pure, small, dense clusters, especially in scenarios where a large fraction of the data may not cluster well. Gene DIVER provides a new level of robustness in cluster mining with it's ability to automatically identify: (1) if there are any clusters in any part of the data, (2) (if present) where they are, (3) what they are (their member data points), and (4) how they are related to each other. Gene DIVER produces a compact hierarchy of all the discovered clusters, which can then be visualized and browsed within Gene DIVER. Areas where Gene DIVER holds great promise is Bioinformatics, and some features of Gene DIVER are specifically tailored for clustering genes. Other potential non-biological applications include market-basket data, biometrics, and web data.

Free Download (please register, see below): Click here to download a free copy. To install, simply unzip all files into a single directory. Gene diver executable file is genediverw.bat. See the lower part of this page for two demo datasets for Gene DIVER.Note: Gene DIVER needs JRE 1.5 (same as JRE 5.0) or higher that can be downloaded and installed from this page. Currently tested on Windows and AMD64 bit linux, but should work with any platform supported by Java.

Support and Registration:To better understand our users, we encourage you to register your copy of Gene DIVER by simply sending an email to gunjan@iname.com, with subject "register", and your name in the email body. We will only contact you if further updates or bug fixes are available, and your email will not be shared with anyone else.

Related papers and publications: To find out more about the clustering algorithm behind Gene DIVER, please see Auto-HDS website and related papers. The last chapter of this Ph. D. dissertation describes Gene DIVER in some detail, especially the scalable architecture of Gene DIVER and some of its unique user interface features that make it well-suited for finding small, pure clusters on large, high-dimensional, noisy datasets.

Test/Demo datasets for Gene DIVER: Two demo datasets are available for demostrating clustering using Gene DIVER:

• Gene DIVER for non-bio applications: Click here to download the Sim-2 demo dataset for Gene Diver: A 2-d Gaussian dataset described and used in the related Auto-HDS paper. The data file to load is Sim2.txt but make sure to unzip all files in one directory.
• Gene DIVER for bio applications: Click here to download the Lee gene dataset used in the Auto-HDS paper. The data file that can be loaded into Gene DIVER under data tab is Insuk1.txt. Make sure to unzip all files in Insuk1.zip into the same directory. Recommended configuration for Lee dataset: Data tab: skip first data line, Gene names available, Vector Space, Delimiter: comma, Class Label Column: classlabel (Name Specified), Cluster tab: n_eps=4, fshave=0.3, rshave=0.05, npart=1. Browse Clustering tab: Access to internet required for browsing resulting genes, System tab: default should work fine on most machines.
• Click here to download pre-clustered Sim-2 data (17MB). Parameters for Gene DIVER will be automatically set for you when you load this data. Make sure to unzip all files in ClusteredSim2.zip into one directory. The data file to load is Sim2.txt.
• Click here to download pre-clustered Lee data using Cosine measure (326 MB). Parameters for Gene DIVER will be automatically set for you when you load this data. Make sure to unzip all files in InsukCosine.zip into one directory. The data file to load is Insuk1.txt.

Click here to visit the IDEAL LAB at University of Texas at Austin, where Gene DIVER was developed.