Kernel-based visualisation of genes with the gene ontology

With the development of microarray-based high-throughput technologies for examining genetic and biological information en masse, biologists are now faced with making sense of large lists of genes identified from their biological experiments. There is a vital need for "system biology" approaches which can allow biologists to see new or unanticipated potential relationships which will lead to new hypotheses and eventual new knowledge. Finding and understanding relationships in this data is a problem well suited to visualisation. We augment genes with their associated terms from the Gene Ontology and visualise them using kernel Principal Component Analysis with both specialised linear and Gaussian kernels. Our results show that this method can correctly visualise genes by their functional relationships and we describe the difference between using the linear and Gaussian kernels on the problem.