Quantifying Perturbed Biological Processes By Analyzing High-Throughput Data Using Causal Networks Models

      Sewer, A.; Martin, F.; Thomson, T. M.; Vasilyev, D.; Park, J.; Frushour, B.; Westra, J. W.; Hoeng, J.; Peitsch, M. C.
      Conference date
      Aug 30, 2013
      Conference name
      The International Conference on Systems Biology (ICSB) 2013

      Background: The scientific community faces an ongoing challenge in the analysis of highthroughput omics data to accurately characterize the biological processes and molecular mechanisms that are perturbed by diseases, drug treatments and environmental agents. A large number of biological processes taking place in healthy lung and vascular tissues have been recently captured into literature-based causal network models, which include cell proliferation, cellular stress responses, and inflammation. Results: In order to leverage both the measured data and the prior biological knowledge contained in the network models, we developed a novel systems-level scoring approach called network perturbation amplitude (NPA). The NPA algorithm computes the amplitudes of treatment- or diseaseinduced perturbations in a network model using transcriptomic data as an input, enabling the identification of activated molecular mechanisms. Targeted perturbations on in vitro systems, including normal human bronchial epithelial (NHBE) cells treated with a cyclin-dependent kinase inhibitor, were scored by the NPA approach, which was able to identify relevant mechanisms and to provide a comparative quantitation of the impacted biological processess. When we evaluated perturbations in more complex experimental systems, such as rats exposed to formaldehyde in vivo, the NPA results captured the degree of activation of known mechanistic effects mediated by the exposures. Conclusions: The ability of the NPA approach to infer the degree of activation of molecular mechanisms and cellular processes provides a powerful means to broadly assess biological activity using a variety of transcriptomic datasets. We have developed and demonstrated the value of a novel integrative, systems biology approach to evaluate disease progression. We expect the range of applications of the NPA approach to be extended to drug safety and discovery as well as to guide biomarker development.