The Application of Pearson and Mutual Information Correlation Network Structure Analysis in Mining Pathogenic Mechanism

The structural parameters of Pearson correlation (PC) and mutual information correlation (MUC) network of gens are used to study the pathogenic mechanism and the difference between the two correlations in the study of biological function. As an example, the PC and MUC networks of bipolar disorder (BD) are constructed, and the top 30 genes (namely, SKGs) with large difference in the average degree of the networks are analyzed. It is found that BD is significantly correlated with nervous system, and is related to immune system, genetic regulation, cell growth/apoptosis and angiogenesis. In addition, PC has universality in revealing biological functions, but the effect of MUC is obviously greater than that of PC. This suggests that the influence of non-linear components on biological function attributes is greater than that of linear components. Therefore, research methods based on linear correlation PC are not enough to reveal the comprehensive information of biological mechanism, and research methods only using MUC also omit linear components.