Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates
Mathematics and Computers in Simulation - IMACS sponsored Special issue on the second IMACS seminar on Monte Carlo methods
Pani: a novel algorithm for fast discovery of putative target nodes in signaling networks
Proceedings of the 2nd ACM Conference on Bioinformatics, Computational Biology and Biomedicine
| Hi-index | 0.00 |
Recent functional genomics research has yielded a large in silico gene regulatory network model (622 nodes) for endomesoderm development of sea urchin, a model organism for embryonic development. The size of this network makes it challenging to determine which genes are most responsible for a given biological effect. In this paper, we explore feasibility and accuracy of existing in silico techniques for identifying key genes that regulate Endo16, a widely-accepted gastrulation marker. We apply target prioritization tools (sensitivity analysis and PANI) to the endomesoderm network to identify key regulators of Endo16 and validate the results by comparing against a set of benchmark Endo16 regulators collated from literature survey. Our study reveals that global sensitivity analysis methods are prohibitively expensive and inappropriate for large networks. We show that PANI efficiently produces superior prioritization results compared to both random prioritization and local sensitivity analysis (LSA) techniques. Specifically, the area under the ROC curve was 0.625, ~0.5, and 0.549 for PANI, random prioritization, and LSA, respectively. Our study reveals that certain unique characteristics of the endomesoderm network affect the performance of target prioritization techniques. In addition to identifying many known regulators of Endo16, PANI also discovered additional regulators (e.g., Snail) that did not appear initially in the benchmark regulators set.