Differential analysis of biological networks

摘要: In cancer research, the comparison of gene expression or DNA methylation networks inferred from healthy controls and patients can lead to discovery biological pathways associated disease. As a progresses, its signalling control are subject some degree localised re-wiring. Being able detect disrupted interaction patterns induced by presence progression disease novel molecular diagnostic prognostic signatures. Currently there is lack scalable statistical procedures for two-network comparisons aimed at detecting topological differences. We propose dGHD algorithm, methodology differential in comparisons. The algorithm relies on statistic, Generalised Hamming Distance (GHD), assessing difference between evaluating significance. builds non-parametric permutation testing framework but achieves computationally efficiency through an asymptotic normal approximation. show that GHD more subtle differences compared standard distance networks. This results achieving high performance simulation studies as measured sensitivity specificity. An application problem co-methylation subnetworks ovarian demonstrates potential benefits proposed discovering network-derived biomarkers with trait interest.