A phase diagram for gene selection and disease classification

Identifying a small subset of discriminate genes is important for predicting clinical outcomes and facilitating disease diagnosis. Based on the model population analysis framework, we present a method, called PHADIA, which is able to output a phase diagram displaying the predictive ability of each variable, which provides an intuitive way for selecting informative variables. Using two publicly available microarray datasets, its demonstrated that our method can selects a few informative genes and achieves significantly better or comparable classification accuracy compared to the reported results in the literature. The source codes are freely available at: www.libpls.net.