Selection of Causal Gene Sets from Gene Expression Profiles Using GeneFis , New Software Based on FNN

Microarray data were useful in disease diagnosis and prognosis. Most approaches to the computational analysis of gene expression data are functionally significant classification of genes [1, 6, 7]. Fuzzy Neural Network (FNN) is one of the advanced ANN models. FNN system can automatically select the causal gene set consisting of several genes for prediction of the disease diagnosis and prognosis and the constructed prediction models showed more than 90% accuracy from cDNA microarray [2] or oligonucleotide microarrays [3] for diffuse large B cell lymphoma (DLBCL) patients. In the present paper, we introduce the customized software, GeneFIS (Fuzzy Inference System for Gene expression analysis), which is incorporated in FNN modeling for prognostic prediction from gene expression data. The majoritarian decision using multiple noninferior models can be also provided as an optional function. Here, we analyzed here the outcome prediction of 220 DLBCL patients with high heterogeneity using GeneFIS .