A signaling pathway analysis model based on Kullback-Leibler divergence

Abnormal regulation of signaling pathways is the key factor to cause disease. Many works focus on identifying the significantly differential pathways between diseases and normal samples via microarray gene expression datasets. However, it is general for exiting methods to concentrate on the difference of pathway components, either the expression or correlation among genes in a given pathway. Thus this will ignore the overall change of pathway. Here we present a powerful analysis model based on the concept of Kullback-Leibler divergence, which mainly measure the difference between two probability distributions of regulation capacity well. We compared our approach with other three classical algorithms on four different human expression datasets, and the results indicate that the capability of our method in detecting disturbed pathways is superior to previous approaches. In conclusion, via introducing the idea of Kullback-Leibler divergence, measure the whole difference of pathway from an overall perspective will provide a complementary analysis framework of pathway analysis.