Mango Tree Software Logo
     
 

 

  

RELNET

Relnet is a software tool written by Atul Butte that allows genomics and bioinformatics researchers to construct relevance networks from their gene expression data.

There are three main advantages to using relevance networks:
  • Negative associations are shown. An example of a negative association is that of tumor suppressor genes. The more tumor suppressor genes are expressed, the less other genes are expressed. This type of negative interaction is commonly missed with other "clustering" techniques.
  • Disparate data types can be included in the same analysis (i.e. clinical, expression, and phenotypic)
  • Multiple connections are allowed for each gene (e.g. a transcription factor may be responsible for regulating the expression of more than one other gene
There are two steps to making relevance networks. First, a comprehensive comparison of all pairs of genes (also known as features) must be performed using a dissimilarity measure (also known as a metric). For example, all genes can be compared against each other by calculating pairwise correlation coefficients. Second, the complete set of comparisons must be filtered using threshold values leaving behind only the strongest connections (these may be either positive or negative associations), which are represented in a graph-network form.
The algorithm was first described in Butte, AJ, et al. Proceedings of the American Medical Informatics Association Symposium. 1999, 711-715 and was then later applied to expression data in Butte AJ, et al. Proceedings of the National Academy of Science USA. 2000, 97(22):12182-12186. We encourage users to cite this latter publication in their own publications.
Features of this software include:
  • The software is written in Java, so it runs under essentially any operating system, including Macintosh, Windows and Linux
  • It handles gene expression data in any tab-delimited format, and intelligently determines the appropriate columns holding gene symbols and expression amounts
  • It dynamically determines the latest names, symbols, functions, and genome position for each gene and includes these in the relevance networks output
  • The software outputs text and GMF (Tom Sawyer Software) formats, and also outputs publication-quality figures in PDF (Adobe) and PostScript
The software has the following requirements:
  • A Java Runtime Environment (JRE) must first be installed. Though this is automatically provided by many modern operating systems, one can download and install this for free from Sun Microsystems.
  • Because gene names and information for each accession number is determined dynamically, an Internet connection is required for actually generating the networks.
  • Construction of the networks may transiently require up to 400 MB memory (either actual memory or virtual memory).
This version of the software is for academic and non-profit use only; commercial use is allowed with a signed license.
Funding for this work was provided in part by:
 

Updated March 8, 2004