In light of the fast growth in DNA technology there is a compelling demand for tools able to perform efficient, exhaustive and integrative analyses of multiple microarray datasets. Specifically, what is particularly evident is the need to link the results obtained from these new tools with the wealth of clinical information. The final goal is to bridge the gap existing between biomedical researchers and pathologists or oncologists providing them with a common framework of interaction.
To overcome such difficulty the SING and BISITE groups have developed geneCBR, a freely available Bioinformatics software tool that allows the use of combined techniques that can be applied to gene selection, clustering, knowledge extraction and prediction. In diagnostic mode, geneCBR employs a case-based reasoning model that incorporates a set of fuzzy prototypes for the retrieval of relevant genes, a growing cell structure network for the clustering of similar patients and a proportional weighted voting algorithm to provide an accurate diagnosis.
Specifically geneCBR is a model that can perform cancer classification based on microarray data. In order to store the information belonging to each sample, the system uses a fuzzy codification to represent the gene expression levels of each sample. This operation permits the generalization over the whole case base in order to tackle intra-experimental and inter-experimental variations in the data. Based on the fuzzy discretization of real gene expression data into a small number of fuzzy membership functions, the system is capable of constructing a set of prototypes that are able to represent the main characteristics of previously ascertained classes.