An Integrative Scoring Approach to Identify Transcriptional Regulations Controlling Lung Surfactant Homeostasis

Transcriptional regulatory network identification is both a fundamental challenge in systems biology and an important practical application of data mining and machine learning. In this study, we propose a semi-supervised learning-based integrative scoring approach to tackle this challenge and predict transcriptional regulations. Our approach out-performs a state-of-the-art label propagation method and reaches AUC scores above 0.96 for three datasets from microarray experiments in the validation. A map of the transcriptional regulatory network controlling lung surfactant homeostasis was constructed. The predicted and prioritized transcriptional regulations were further validated through experimental verifications. Many other predicted novel regulations may serve as candidates for future experimental investigations.