Abstract
Caenorhabditis elegans vulval development is a paradigmatic example of animal organogenesis with extensive experimental data. During vulval induction, each of the six multipotent vulval precursor cells (VPCs) commits to one of three fates (primary, secondary, tertiary). The precise primary-secondary-tertiary formation of VPC fates is controlled by a network of intercellular signaling, intracellular signal transduction and transcriptional regulation. The construction of mathematical models for this network will enable hypothesis generation, biological mechanism discovery and system behavior analysis.
We have developed a mathematical model based on dynamic Bayesian networks to model the biological network that governs the VPC primary-secondary-tertiary pattern formation process. Our model has six interconnected subnetworks corresponding to six VPCs. Each VPC subnetwork contains 20 components. The causal relationships among network components are quantitatively encoded in the structure and parameters of the model. Statistical machine learning techniques were developed to automatically learn both the structure and parameters of the model from data collected from literatures. The learned model is capable of simulating vulval induction under 36 different genetic conditions. Our model also contains a few hypothetical causal relationships between network components, and hence can serve as guidance for designing future experiments. The statistical learning nature of our methodology makes it easy to not only handle noise in data but also automatically incorporate new experimental data to refine the model.
Supplementary data are available at Bioinformatics online.