Fig. 1.

Cancer classification using high-throughput screening technologies. a An example of gene expression profiling data. The experimental dataset contains n samples and each sample has m interrogated genes (m≥n). b The extracted samples of an experiment are labeled according to their phenotypes or different types of cell lines. Both a gene expression matrix and a class vector form the input of gene selection for cancer classification. c A subset G of significant genes is obtained as an output for a certain cancer classification, which is a so-called gene signature