Inferring the Structure of Genetic Regulatory Networks Using Information Theoretic Tools

By combining the mutual information and conditional mutual information, a practical metric is proposed to capture the inference confidence of direct connectivity between two genes. This metric helps to avoid the disadvantage of general schemes, i.e., the dichotomy of either being connected or disconnected. Based on data sets generated by synthetic networks, the performance of proposed algorithm is compared favorably with respect to other schemes in the literature. The proposed algorithm is also applied on realistic cutaneous melanoma data set to recover a genetic network containing 470 genes