A signaling pathway analysis method based on information divergence

Abnormal regulation of signaling pathways is the key factor causing disease. For better understanding disease mechanisms, many methods have been proposed to identify the significantly differential pathways between diseases and normal individuals via microarray gene expression datasets. Unlike previous common analysis processes, which is focused on merging gene difference into difference of pathway indirectly. In this paper, the idea of information divergence is introduced and a novel signaling pathway analysis method from a holistic view is presented to improve the detection results. We identify significantly differential pathways directly via computing the KL divergence between real and simulated probability distributions of gene-gene regulatory ability. We test our method on four human microarray expression datasets. The results illustrate that the capability of our approach in detecting significantly differential pathways between two sample groups is superior to other three classical pathway analysis methods.