Interactive Datamining and Visualization of Multi-Dimensional Datasets

Introduction

Today, chemical companies are routinely employing synthesis technologies such as High-Throughput-Screening, Combinatorial Synthesis and Parallel Synthesis that generate terabytes of data per year. These datasets are highly relevant for chemists, because they potentially deliver insights into chemical and biochemical trends and principles and could lead to faster development and higher numbers of drug candidates. However, the development of data mining and visualization tools are capable of analyzing these large amounts of multi-dimensional data appears not to have been able to keep pace with the dramatic increase in size of these datasets. Ultimately, this situation has become one of the most critical bottlenecks in chemical R&D today.

We present an applet that allows an interactive and graphical data mining and visualization of large-scale multi-dimensional datasets. Beside the applet version it also exists an application version with local file and database connection support to import own datasets easily.

Screenshot

Technique

The application is based on Java2 and the extension Java3D and uses an 3D Glyph approach to visualize the multi-dimensional datasets. Therefore, it is necessary to have a Java2 plug-in with Java3D extensions before starting the applet in the browser window. By using Java3D the applet is capable to utilize 3D hardware capabilities of modern desktop clients. This allows a graphic card dependent fast and high-quality rendering of the data. The number of datapoints depends on the capabilities of the graphic card.

Requirements

Java2 plug-in and Java3D extensions.

Windows and Solaris users can download the Java plug-in and the Java3D extension.

SGI users can download all necessary files from www.sgi.com.

Manual

An online manual is available (will open in a new window).

Demo

An online demo using an NCI anti-tumor screening dataset (applet size ~238 KB).

Example

NCI Screening Data 3D Miner is an example of InfVis in action.

Visualize Your Data

An online service for data visualization via file upload is provided.