LIBGS

A MATLAB Package for Gene Selection

School of Computing and Information Sciences
Florida International University

Last Modified: 08/22/2007

Overview

LIBGS is a Matlab package for gene selection. It includes many popular gene selection methods widely used for expression data sets and provides a platform to perform performance comparisons.

NOTE: Currently, this software is under review for publication and it will take a couple of weeks. Thank you for your interest and the software will be open for download upon the completion of the review process. If you have any problems, please contact me before you download any file in this site.

Gene Selection Algorithms

To summarize, the main features selection algorithms are listed as follows:

ReliefF: Description | Matlab Code Download;
F-test: Description | Matlab Code Download;
mRMR: Description | Original Version Link | Matlab Code Download;
ReliefF-mRMR: Description | Matlab Code Download;
GSNR: Description | Matlab Code Download;
Clustering Based Method: Description | Matlab Code Download;
D-optimality and A-optimality: Description | Matlab Code Download;
Rankgene 1.1: Original Version Link | Matlab Interface Download ;
- Information gain: Description;
- Twoing rule: Description;
- Gini index: Description;
- Sum minority: Description;
- Max minority: Description;
- Sum of variances: Description;
- T-statistic: Description;
- One dimensional SVM: Description.

Datasets

Six gene datasets are provided as following, all of them are fomated by .mat:

The ALL dataset covers six subtypes of acute lymphoblastic leukemia: BCR, E2A, Hyperdip, MLL, T and TEL Download;
The GCM dataset consists of 198 human tumor samples of fifteen types Download;
The HBC dataset consists of 22 herediray breast cancer samples with three classes Download;
The LYM dataset is of the three most prevalent adult lymphoid malignancies Download;
The MLL dataset consists three classes Download;
The NCI60 dataset Download.

Data Format

Convert .csv file to .mat file: Download;
Convert .mat to .csv file: Download;
Gene data discretization: Download.

Assistant Tools for Classification

To evaluate the performance of gene selection algorithms by classification accuracy, the Matlab interfaces for two classification tools are provided as following:

LIBSVM is an integrated software for support vector classifiction, (C-SVC, nu-SVC), regression (spsilon-SVR, nu-SVR) and distribution estimation(one-class SVM). It supports multi-class classification. LIBSVM Link
Weka is a collection of machine learning algorithms for data mining tasks. The algorithms can either be applied directly to a dataset or called from your own Java code. Weka contains tools for data pre-processing, classification, regression, clustering, association rules, and visualization. It is also well-suited for developing new machine learning schemes. WEKA Link Matlab interface for WEKA classification.